cAMP Signaling: Methods and Protocols [1 ed.] 1493925369, 9781493925360

Citation preview

Methods in Molecular Biology 1294

Manuela Zaccolo Editor

cAMP Signaling Methods and Protocols

METHODS

IN

MOLECULAR BIOLOGY

Series Editor John M. Walker School of Life and Medical Sciences University of Hertfordshire Hatfield, Hertfordshire, AL10 9AB, UK

For further volumes: http://www.springer.com/series/7651

cAMP Signaling Methods and Protocols

Edited by

Manuela Zaccolo Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, UK

Editor Manuela Zaccolo Department of Physiology, Anatomy, and Genetics University of Oxford Oxford, UK

ISSN 1064-3745 ISSN 1940-6029 (electronic) Methods in Molecular Biology ISBN 978-1-4939-2536-0 ISBN 978-1-4939-2537-7 (eBook) DOI 10.1007/978-1-4939-2537-7 Library of Congress Control Number: 2015934164 Springer New York Heidelberg Dordrecht London © Springer Science+Business Media New York 2015 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. Cover Illustration: Epifluorescence image of a neonatal cardiac myocyte expressing a FRET-based reporter for detection of cAMP levels in real-time. The panels on the left and centre show the pseudocolor image of the cell in the cyan (480nm) and yellow (540nm) channels, respectively, acquired on excitation of the sample at 430nm. The panel on the right shows the image of the 480nm/535 nm FRET signal from the same cell, showing in red subcellular compartments with higher cAMP levels. The myocytes was treated with 10nM isoproterenol to trigger generation of cAMP. Image: M Zaccolo (University of Oxford, UK). Printed on acid-free paper Humana Press is a brand of Springer Springer Science+Business Media LLC New York is part of Springer Science+Business Media (www.springer.com)

Preface Adenosine 3′,5′-monophosphate (cAMP), the prototypical intracellular second messenger, regulates a large variety of cellular functions and biological processes, including gene transcription, cell metabolism, proliferation, development, as well as more specialized functions depending on the cell type. In its simpler formulation, the cAMP signaling pathway involves a hormone (the “first” messenger) that binds and activates a specific G protein-coupled receptor that in turn activates adenylyl cyclases to synthesize cAMP. The intracellular (or “second”) messenger cAMP then binds to a limited number of intracellular effectors, most notably to protein kinase A (PKA), which phosphorylates downstream targets leading to a specific functional outcome. Signal termination is mediated by phosphodiesterases (that hydrolyze cAMP) and phosphatases (that dephosphorylate PKA targets), enzymes that are modulated by complex regulatory mechanisms. In the last 15 years, the field of cAMP signaling has witnessed an exciting development with accumulating evidence demonstrating that cAMP is compartmentalized and that spatial regulation of cAMP signals is critical for faithful signal propagation and for specificity of response. This realization has changed our understanding of cAMP signaling from a model where a linear pathway connects the receptor located at the plasma membrane with an effector and its function to a model where signal propagation occurs within a complex network of cAMP-dependent signaling pathways simultaneously operating within the same cell. The pathway or pathways the cAMP signal travels along are dictated by the overall state of the cell at the time the cAMP signal is generated, depending on the activity of on/off signals that operate on individual routes at that particular time. Based on this new model, the functional outcome of a signal mediated by cAMP depends strictly on local and temporal regulation. The hormonal specificity of cAMP action results from the generation of distinct pools of the second messenger which in turn mediate different functional outcomes via activation of different subsets of the cAMP effector PKA. PKA is largely localized to different subcellular compartments via binding to a family of scaffolding proteins known as A Kinase Anchoring Proteins (AKAPs). Apart from their common ability to anchor PKA, AKAPs show a high degree of structural variability which allows for different subcellular localization and binding to a variety of other signaling components. As a result, AKAPs serve as signaling centers, where elements of the cAMP signaling pathway and other regulatory molecules are organized for a particular task. The realization of this extremely complex spatial organization and local regulation is providing novel mechanistic insight into cell physiology and is producing a novel framework for the identification of disease mechanisms. This new model also offers the potential to establish original avenues for the treatment of disease. New approaches have been developed that allow researchers to gain information that goes beyond a measure of cAMP activity at the whole cell or cell population level. In preparing this volume, I have tried to encompass new technological developments that specifically address questions related to cAMP compartmentalization, that probe relevant protein–protein interactions, that increase the spatial and temporal resolution of cAMP signals detection, and that can facilitate integration of the mounting complexity of the information that is becoming available on this signaling system.

v

vi

Preface

I am extremely grateful to all authors for living with my deadlines and providing excellent and comprehensive methods and extensive notes with essential “tricks of the trade” that are so precious when troubleshooting a new technique. Finally, I thank the Senior Editor, John Walker, for giving me the opportunity to compile this volume in the excellent series, Methods in Molecular Biology. I hope the selection of methods will prove appealing and will be a real resource to researchers in the field. Oxford, UK

Manuela Zaccolo

Contents Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Simultaneous Assessment of cAMP Signaling Events in Different Cellular Compartments Using FRET-Based Reporters . . . . . . . . . . . . . . . . . . . . . . . . . Alex Burdyga and Konstantinos Lefkimmiatis 2 Recording Intracellular cAMP Levels with EPAC-Based FRET Sensors by Fluorescence Lifetime Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Marcel Raspe, Jeffrey Klarenbeek, and Kees Jalink 3 A Novel Approach Combining Real-Time Imaging and the Patch-Clamp Technique to Calibrate FRET-Based Reporters for cAMP in Their Cellular Microenvironment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Andreas Koschinski and Manuela Zaccolo 4 Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET Reporters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Matías Machado and Sergio Pantano 5 Automated Image Analysis of FRET Signals for Subcellular cAMP Quantification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Silas J. Leavesley, Arie Nakhmani, Yi Gao, and Thomas C. Rich 6 Channel-Based Reporters for cAMP Detection . . . . . . . . . . . . . . . . . . . . . . . . Thomas C. Rich, Wenkuan Xin, Silas J. Leavesley, and Mark S. Taylor 7 Imaging Sub-plasma Membrane cAMP Dynamics with Fluorescent Translocation Reporters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anders Tengholm and Olof Idevall-Hagren 8 Adenoviral Transduction of FRET-Based Biosensors for cAMP in Primary Adult Mouse Cardiomyocytes . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oliver Lomas, Marcella Brescia, Ricardo Carnicer, Stefania Monterisi, Nicoletta C. Surdo, and Manuela Zaccolo 9 Generation of Transgenic Mice Expressing FRET Biosensors. . . . . . . . . . . . . . Daniela Hübscher and Viacheslav O. Nikolaev 10 Photoactivatable Adenylyl Cyclases (PACs) as a Tool to Study cAMP Signaling In Vivo: An Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Marina Efetova and Martin Schwärzel 11 Selective Disruption of the AKAP Signaling Complexes. . . . . . . . . . . . . . . . . . Eileen J. Kennedy and John D. Scott 12 Screening for Small Molecule Disruptors of AKAP–PKA Interactions . . . . . . . Carolin Schächterle, Frank Christian, João Miguel Parente Fernandes, and Enno Klussmann

vii

v ix 1

13

25

41

59 71

85

103

117

131 137 151

viii

Contents

13 Structure-Based Bacteriophage Screening for AKAP-Selective PKA Regulatory Subunit Variants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ryan Walker-Gray and Matthew G. Gold 14 A Yeast-Based High-Throughput Screen for Modulators of Phosphodiesterase Activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ana Santos de Medeiros and Charles S. Hoffman 15 Separation of PKA and PKG Signaling Nodes by Chemical Proteomics . . . . . . Eleonora Corradini, Albert J.R. Heck, and Arjen Scholten 16 Development of Computational Models of cAMP Signaling . . . . . . . . . . . . . . Susana R. Neves-Zaph and Roy S. Song Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

167

181 191 203 219

Contributors MARCELLA BRESCIA • Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, UK ALEX BURDYGA • Department of Physiology, Anatomy and Genetics, Burdon Sanderson Cardiac Science Centre, BHF Centre of Research Excellence, University of Oxford, Sherrington Building, Parks Road, Oxford, UK RICARDO CARNICER • Department of Cardiovascular Medicine, University of Oxford, Oxford, UK FRANK CHRISTIAN • Institute of Cardiovascular and Medical Sciences, University of Glasgow, Glasgow, UK ELEONORA CORRADINI • Biomolecular Mass Spectrometry and Proteomics, Bijvoet Center for Biomolecular Research, Netherlands Proteomics Centre and Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Domplein, Utrecht, The Netherlands MARINA EFETOVA • Institute for Biology Neurobiology, Freie Universität Berlin, Berlin, Germany YI GAO • Department of Biomedical Informatics, Stony Brook University, Stony Brook, NY, USA MATTHEW G. GOLD, PH.D. • Department of Neuroscience, Physiology, & Pharmacology, University College London, London, UK OLOF IDEVALL-HAGREN • Department of Medical Cell Biology, Biomedical Centre, Uppsala University, Uppsala, Sweden ALBERT J.R. HECK • Biomolecular Mass Spectrometry and Proteomics, Bijvoet Center for Biomolecular Research, Netherlands Proteomics Centre and Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Utrecht, The Netherlands CHARLES S. HOFFMAN • Biology Department, Boston College, Chestnut Hill, MA, USA DANIELA HÜBSCHER • Heart Research Center Göttingen, Georg August University Medical Center, University of Göttingen, Göttingen, Germany KEES JALINK • Department of Cell Biology, The Netherlands Cancer Institute, Amsterdam, The Netherlands EILEEN J. KENNEDY • Department of Pharmaceutical and Biomedical Sciences, University of Georgia College of Pharmacy, Athens, GA, USA JEFFREY KLARENBEEK • Department of Cell Biology, The Netherlands Cancer Institute, Amsterdam, The Netherlands ENNO KLUSSMANN • Max Delbruck Center for Molecular Medicine Berlin (MDC), Berlin, Germany ANDREAS KOSCHINSKI • Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, UK SILAS J. LEAVESLEY • Department of Chemical and Biomolecular Engineering, Center for Lung Biology, University of South Alabama, Mobile, AL, USA; Department of Pharmacology, Center for Lung Biology, University of South Alabama, Mobile, AL, USA KONSTANTINOS LEFKIMMIATIS • Department of Physiology, Anatomy and Genetics, Burdon Sanderson Cardiac Science Centre, BHF Centre of Research Excellence, University of Oxford, Sherrington Building, Parks Road, Oxford, UK

ix

x

Contributors

OLIVER LOMAS • Department of Cardiovascular Medicine and Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, UK MATÍAS MACHADO • Group of Biomolecular Simulations, Institut Pasteur de Montevideo, Montevideo, Uruguay ANA SANTOS DE MEDEIROS • Biology Department, Boston College, Chestnut Hill, MA, USA STEFANIA MONTERISI • Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, UK ARIE NAKHMANI • Department of Electrical and Computer Engineering, University of Alabama at Birmingham, Birmingham, AL, USA SUSANA R. NEVES-ZAPH, PH.D. • Departments of Pharmacology & Systems Therapeutics, System Biology Center, Friedman Brain Institute, NY Icahn School of Medicine at Mount Sinai, New York, NY, USA VIACHESLAV O. NIKOLAEV • Institute of Experimental Cardiovascular Research, University Medical Center Hamburg-Eppendorf, Hamburg, Germany SERGIO PANTANO • Group of Biomolecular Simulations, Institut Pasteur de Montevideo, Montevideo, Uruguay JOÃO MIGUEL PARENTE FERNANDES • Max Delbruck Center for Molecular Medicine Berlin (MDC), Berlin, Germany MARCEL RASPE • Department of Cell Biology and Department of Biochemistry, The Netherlands Cancer Institute, Amsterdam, The Netherlands THOMAS C. RICH • Department of Pharmacology and Center for Lung Biology, University of South Alabama, Mobile, AL, USA CAROLIN SCHÄCHTERLE • Max Delbruck Center for Molecular Medicine Berlin (MDC), Berlin, Germany ARJEN SCHOLTEN • Biomolecular Mass Spectrometry and Proteomics, Bijvoet Center for Biomolecular Research, Netherlands Proteomics Centre and Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Utrecht, The Netherlands MARTIN SCHWÄRZEL • Institute for Biology Neurobiology, Freie Universität Berlin, Berlin, Germany JOHN D. SCOTT • Department of Pharmacology, Howard Hughes Medical Institute, University of Washington School of Medicine, Seattle, WA, USA ROY S. SONG • Departments of Pharmacology & Systems Therapeutics, Icahn School of Medicine at Mount Sinai, New York, NY, USA NICOLETTA C. SURDO • Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, UK MARK S. TAYLOR • Department of Physiology, University of South Alabama, Mobile, AL, USA ANDERS TENGHOLM • Department of Medical Cell Biology, Biomedical Centre, Uppsala University, Uppsala, Sweden RYAN WALKER-GRAY • Department of Neuroscience, Physiology, & Pharmacology, University College London, London, UK WENKUAN XIN • Department of Drug Discovery and Biomedical Sciences, South Carolina College of Pharmacy, University of South Alabama, Mobile, AL, USA MANUELA ZACCOLO • Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, UK

Chapter 1 Simultaneous Assessment of cAMP Signaling Events in Different Cellular Compartments Using FRET-Based Reporters Alex Burdyga and Konstantinos Lefkimmiatis Abstract Several aspects of the cAMP signaling cascade, including the levels of the messenger itself and the activity of its main effector protein kinase A (PKA), can be measured in living cells, thanks to genetically encoded probes based on fluorescence resonance energy transfer (FRET). While these biosensors enable the assessment of cAMP or PKA activity with great spatial and temporal resolution, concomitant events triggered by the same stimuli at the same or other cellular compartments are not easily assessed. In this chapter we present a simple approach that allows the simultaneous measurement of cAMP and its actions in subcellular compartments of neighboring cells. As proof of principle, we compare cAMP signals and PKA activity in the cytosol of neighboring HEK cells. We propose that this flexible and powerful method can significantly improve the direct comparison of cAMP signals and their action in specific cellular domains. Key words cAMP, PKA, FRET, Imaging, Fluorescence, Microdomains, Compartmentalization

1

Introduction The second messenger cyclic AMP (cAMP) is a multifunctional molecule involved in the control of many, sometimes contradictory, cellular functions [1]. For instance, cAMP can induce or inhibit cell migration [2–4], promote or block cell death [5], and regulate functions as different as cardiac excitation-contraction coupling and gene transcription [1, 6]. Some of these actions can be explained by the activation of different cAMP effectors; however, only three such proteins have been described (protein kinase A (PKA), exchange proteins activated by cAMP (EPACs), and cyclic nucleotide-gated ion channels), a number insufficient to explain the functional complexity of this messenger. An alternative explanation for the multiple actions of cAMP, proposed early after its discovery, is the spatial and temporal confinement of its cascade [7–9]. Direct evidence for the existence of subcellular cAMP signaling domains and the confirmation of

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_1, © Springer Science+Business Media New York 2015

1

2

Alex Burdyga and Konstantinos Lefkimmiatis

compartmentalization as a regulatory mechanism of cAMP action was provided more than a decade ago, thanks to the development of molecular tools for the detection of cAMP changes [10–13] and PKA activity [14–16]. These probes are based on the principle of fluorescence resonance energy transfer (FRET), where the distance or orientation between a donor and a suitable acceptor fluorophore varies in response to cAMP (or PKA activity)-induced conformational changes (Fig. 1) [17]. This technology revolutionized the studies on cAMP signaling because it combined sensitivity of detection with very high spatial and temporal resolution. One important characteristic of the FRET-based sensors is that they can be genetically engineered allowing optimization both of the FRET efficiency [16, 18–20] and spatial resolution [21–23]. Taking advantage of this quality and thanks to the addition of specific targeting peptides, a large array of targeted FRET sensors has been developed. These constructs allow, in principle, the direct measure of cAMP or its actions at different subcellular compartments. Arguably, expression of two or more FRET sensors [24] in the same cell is desirable [25], as it would allow the contemporaneous measure of multiple signaling parameters (e.g., free cAMP and PKA activity) and decrease the cell-to-cell variability. However, this very attractive approach is technically challenging. For instance, the majority of FRET sensors have been developed using the

EpacH90 Adenylyl cyclases Phosphodiesterases

AKAR4 T

PKA

Tp Phosphatases

Fig. 1 Principle of action of EpacH90 and AKAR4. Both sensors exploit a protein domain that can undergo a large conformational change upon cAMP binding for EpacH90 or PKA-dependent phosphorylation for AKAR4. In steady state EpacH90 displays maximum FRET that decreases in the presence of cAMP. On the contrary AKAR4 reaches maximum FRET when an intramolecular portion of the sensor is phosphorylated by PKA

Monitoring of Synchronous cAMP and PKA Events

3

efficient FRET pair of cyan fluorescent protein (and its derivatives Cerulean and mTurquoise) [18, 19] and yellow fluorescent protein (and its pH-insensitive version of circularly permutated Venus) [18]. Two different reporters can be used in the same cell only if they are physically segregated, but even when this requirement is met, some overlap of targeted sensors is virtually unavoidable in the highly dynamic milieu of a living cell, where structures can rapidly change position and shape. A solution to these issues may be the development of novel FRET sensors based on fluorophores that do not show spectral overlap with CFP/YFP [24, 26]. However only a limited number of efficient red-shifted FRET pairs have been developed [24], and their use in cAMP or PKA activity sensors has been limited. An alternative possibility that would allow the simultaneous monitoring of cAMP and/or PKA in multiple cellular domains would be to perform measures in neighboring cells from the same sample [21, 27]. Albeit a compromise, this approach has several positive aspects: (a) it is easy to set up, (b) it does not present the problem of overlapping signals from different reporters and can be used with probes based on the same FRET pair, (c) it is very flexible and can be used with virtually any FRET-based sensor, and (d) because multiple measures are performed simultaneously and the comparison is direct, each experiment provides, at least, twice the amount of information (see Note 1). Here, as proof of principle, we monitor cAMP and PKA activity in mixed populations of HEK293 cells (see Fig. 2), expressing a cytosolic cAMP sensor called EpacH90 [19] or the PKA activity probe AKAR4 [16]. This pair of sensors has particular demonstrative value, both at the technical and biological levels. Both probes are FRET based, however, while AKAR4 works on the principle of “gain of FRET” and reaches maximum FRET ratio when PKA activity is high, EpacH90 works based on “loss of FRET,” and its FRET ratio decreases in response to cAMP elevation (Fig. 1). We present four separate ways to distinguish between the two sensors and a comprehensive protocol for obtaining mixed populations expressing different probes. We propose this simple but powerful method as an alternative approach for monitoring simultaneously multiple signaling aspects of the cAMP cascade.

2

Materials

2.1 Cell Culture and Transfection

1. Human embryonic kidney 293 (HEK293) cells (see Note 2). 2. Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10 % fetal bovine serum (FBS), 100 U/mL penicillin, 100 μg/mL streptomycin, and 2 mM glutamine.

Alex Burdyga and Konstantinos Lefkimmiatis

4

a

i

YFP

mCherry

Frsk 20mM + IBMX 200mM H89 30mM

Frsk 5mM

b

ii

YFP / CFP ratio

1.2

1.0

AKAR4

0.8

0.6

H90 0

300

600

900

1200

Time (s) Frsk 5mM

c

Frsk 20mM + IBMX 200mM

1.4

H89 30mM

H90 Intensity

1.2

AKAR4 1.0

0.8

CFP Ch. YFP Ch.

H90

0.6 0

300

600

900

1200

Time (s) Fig. 2 Simultaneous measurement of cAMP levels and PKA activity using specific FRET sensors. (a) HEK293 cells were transfected with either AKAR4 alone or with H90 together with mCherry. Twenty-four hours after transfection, cells were mixed and plated onto glass coverslips. The day after, cells were mounted onto a perfusion chamber, and the expression of the two sensors was assessed at the microscope stage using YFP fluorescence (i). The cellular distribution of the two sensors could be used as parameter for distinguishing between the two constructs. Acquiring mCherry fluorescence (ii) confirmed that the cell displaying no nuclear

Monitoring of Synchronous cAMP and PKA Events

5

3. Dulbecco’s phosphate-buffered solution (DPBS): NaCl 138 mM, KCl 2.67 mM, KH2PO4 1.47 mM, Na2HPO4-7H2O 8.06 mM, pH 7.1. 4. Trypsin-EDTA solution (0.05 %): NaCl 138 mM, KCl 5.33 mM, KH2PO4 0.44 mM, Na2HPO4-7H2O 0.34 mM, NaHCO3 4.17 mM, dextrose 5.56 mM, EDTA4Na2H2O 0.48 mM, phenol red 0.0251 mM, trypsin 0.0210 mM. 5. 25 or 75 cm2 cell culture flasks. 6. 6 cm culture dishes. 7. 12-well culture plates. 8. Sterile 15 mm glass coverslips. 9. Chloroform/EtOH (1:1) solution (see Note 3). 10. Lipofectamine 2000, TransFectin, FuGENE, or any other reagent that can achieve 15 % or greater transfection efficiency while maintaining minimal cellular toxicity (see Note 4). 11. Mammalian expression vectors encoding the sensors of interest. We used pcDNA3-AKAR4 encoding the PKA activity sensor AKAR4 [16] and the cAMP-sensitive sensor EpacH90 encoded by pcDNA3-EpacH90 [19]. 12. Mammalian expression vector encoding a marker fluorophore (see Note 5). For this purpose, we utilized pcDNA3-mCherry [21, 28]. 2.2 FRET Measures and Analysis

1. Imaging buffer: NaCl 125 mM, KCL 5 mM, Na3PO4 1 mM, MgSO4 1 mM, HEPES 20 mM, glucose 5.5 mM, CaCl2 1 mM, pH adjusted to 7.4 using 1 M NaOH (see Note 6). 2. Microscope equipped for FRET. We use a FRET system built around an Olympus IX71 inverted microscope equipped with beam splitter (OPTOSPLIT II) and a CCD camera (CoolSNAP HQ2) (see Note 7).

Fig. 2 (continued) fluorescence was transfected with the H90 sensor (see Note 25). (b) Representative traces corresponding to one cell expressing H90 (red trace) together with the marker mCherry and two cells expressing AKAR4 (green and blue traces). Treatment with low doses of the cAMP-generating agonist forskolin induced a rise in cAMP which as expected induced a decrease in the FRET ratio of H90 (red trace) and a contemporaneous increase in the FRET ratio of AKAR4, indicating increase in PKA activity (green and blue traces). Washout of the agonists resulted in a drop of cAMP and PKA back to basal levels. Application of forskolin and IBMX was used to saturate both sensors and reveal the maximal FRET shift possible. PKA inhibitor H89 (see Note 26) was applied at the end of the experiment and resulted in complete inhibition of PKA activity without any significant changes to the cAMP levels, as measured by the cAMP sensor H90. (c) Traces show individual background subtracted CFP (solid line) and YFP (dotted line) intensity values for one cell transfected with AKAR4 (green trace) and one cell transfected with H90 + mCherry (red trace), which were used to compute ratio traces shown in b. As expected, the AKAR4 sensor displays an increase in FRET with increasing PKA activity, while the H90 sensor displays loss of FRET upon elevation of cAMP

6

Alex Burdyga and Konstantinos Lefkimmiatis

3. Imaging chamber. We use a ready-made open bath perfusion chamber RC-25F assembled to a PH-4 platform (with optional temperature control) and a stage adapter SA-OLY/2 (all Warner Instruments). 4. Perfusion system (optional) (see Note 8). 5. Data analysis software (see Note 9).

3

Methods

3.1 Cell Culture and Transfection

1. Maintain HEK293 cells in DMEM and grow them in 25–75 cm2 size culture flasks. Passage approximately every 2 days upon reaching 70–80 % confluence. 2. The day before transfection, rinse the cells briefly with DPBS and use trypsin-EDTA solution to detach them from the flask’s surface. Thereafter, plate cells onto two 6 cm petri dishes at a confluence of 45–60 %. 3. The day after seeding, transfect the two plates using Lipofectamine 2000 (or equivalent reagent). For the experiments illustrated in this chapter, we used 1.4 μg of pcDNA3AKAR4 for one plate, while for the other we used 1 μg of pcDNA3-EpacH90 together with 400 ng of pcDNA3mCherry (see Notes 10 and 11). 4. Twenty-four hours after transfection, rinse the cells with DPBS thoroughly and then detach them using trypsin-EDTA solution. Mix the two populations in a proportion of 1:1 and seed at a confluence of 50–60 % into a 12-well culture plate containing sterilized 15 mm glass coverslips (see Note 12). 5. Allow the cells to attach to the coverslips for 16–20 h before experimentation.

3.2 Mounting the Coverslip onto the Imaging Chamber

1. Apply a thin ring of vacuum grease to the bottom of the imaging chamber (see Note 13). 2. Remove the glass coverslip using a pair of sharp-tipped forceps and place it onto the tissue paper, with the cells facing up. 3. Hold down the coverslip with the tip of the forceps and use a small piece of filter paper to quickly blot the outer 1–2 mm of the coverslip, in order to create a dry ring on the periphery of the glass (see Note 14). 4. Press down the imaging chamber against the coverslip in such a way that the vacuum grease ring on the imaging chamber makes contact with the dry ring on the coverslip. 5. Add 500–800 μL of imaging buffer to the coverslip to prevent it from drying (see Note 15).

Monitoring of Synchronous cAMP and PKA Events

7

6. The following step is specific to the type of chamber and perfusion system that you will be using and may not apply to you: attach the imaging chamber to the PH-4 platform, and then fix this assembly to the Olympus stage adapter. 7. Place the imaging chamber and stage adapter onto the microscope stage. 8. Attach the 6-way manifold as well as the suction line to the imaging chamber in order to enable the use of the perfusion system. 3.3 Detecting cAMP Levels in Multiple Microdomains

1. Switch on the microscope and find a field of view that contains cells transfected with different sensors (see Notes 11 and 16). 2. Confirm the identity of each sensor using the co-expressed marker protein. In our case this can be achieved by monitoring the 610 nm emission of mCherry upon excitation with 560 nm light (see Fig. 1). 3. Position the cells as required and use the imaging software to draw regions of interest (ROI) around individual cells (see Fig. 1). 4. Configure the system to generate excitation light corresponding to the donor fluorophore and to record the emission light of the donor and acceptor. 5. It is helpful to view live FRET ratio trace during the experiment, so enable this option if it is available in your acquisition software (see Note 17). 6. Configure the acquisition time and the time-lapse interval as required (see Note 18). 7. Begin the experiment by establishing a stable baseline during the first 3–10 min. 8. Apply a stimulus to the cells. Different combination of stimuli can be applied and rinsed away using the perfusion system, which allows a large degree of flexibility in the experimental protocol (see Note 19). 9. Apply a positive control at the end of the experiment to maximally activate the FRET sensor. This can be achieved by raising cAMP levels using the phosphodiesterase (PDE) inhibitor 3-isobutyl-1-methylxanthine (IBMX, 100 μM–1 mM) together with the adenylyl cyclase (AC) activator forskolin (10 μM–25 μM) (see Fig. 1 and Note 20). 10. Subsequently, and if possible, try to maximally inhibit the sensor. This will allow you to see the maximal FRET range and also understand whether the starting baseline was elevated. To achieve this, we used H89 to inhibit the PKA sensor (see Fig. 1). 11. Once the experiment is finished, stop the time-lapse acquisition and make sure to save your data (see Note 21).

8

Alex Burdyga and Konstantinos Lefkimmiatis

12. Clean up the perfusion system, including all the tubing, first using deionized water, then 70 % ethanol, and finally deionized water again. 3.4 Analyzing Recorded FRET Data

1. Open the saved time-lapse files using the imaging software (see Note 22). 2. Draw ROIs around individual cells as well as in the background (area devoid of any cells) (see Notes 23 and 24). 3. Export the intensity values of each ROI into a format that is recognized by Excel (Microsoft). 4. Using Excel, subtract the background values from the corresponding donor and acceptor emission values for each individual ROI. Now calculate the FRET ratio (see Note 17). 5. Plot graphs from the calculated FRET ratios using Excel or other suitable software such as OriginPro (OriginLab), Prism, or KaleidaGraph.

4

Notes 1. This is particularly relevant for studies using primary cells. In our experience with isolated neonatal rat ventricular myocytes, we find that this method decreased the number of animals needed for each experimental protocol by 50 %. 2. Many different cell lines can be used for this approach. A limitation, however, is the transfection efficiency. We find that in order to ensure successful application, transfection efficiency has to be 15 % or greater. One solution to low transfection efficiency is the use of viral-based expression vectors that ensure high infection levels in immortalized as well as primary cells. 3. Care must be taken to prepare the chloroform/ethanol mix (1:1) in a chemical extraction hood according to the laboratory safety procedures. 4. Regardless of the particular transfection reagent used, follow the manufacturer’s instructions in order to optimize transfection efficiency and minimize cellular toxicity for each cell type. 5. The use of a marker fluorophore becomes necessary if the FRET sensors target different domains of the same structure that cannot be distinguished at the optical microscope (e.g., mitochondrial matrix vs. outer mitochondrial membrane). It is important to choose a fluorophore that does not interfere with the sensors. We find that for cyan-yellow-based sensors, the far-red protein mCherry is optimal. 6. We make 0.5 L of a 10× concentrated imaging buffer solution that is not pH adjusted and does not contain CaCl2 or glucose,

Monitoring of Synchronous cAMP and PKA Events

9

which we sterilize using a 0.22 μm filter and store at 4 °C for up to a month. On the day the cells are imaged, we prepare a 1× concentrated buffer which is complemented with glucose and calcium, and its pH is adjusted to 7.4 using 1 M NaOH. 7. Many different types of imaging systems can be used to successfully record FRET signals, provided that the donor can be selectively excited, while separate emission can be collected for both the donor and acceptor fluorophores. For example, for the widely used FRET pair (CFP-YFP), the system would require the ability to excite CFP (430 nm) and the capability to record the emission of both donor and acceptor, 480 nm and 545 nm, respectively. 8. We use a homebuilt perfusion system with the ability to inject a maximum of six separate solutions, thanks to a 6-way manifold MPP-6 (Warner Instruments). Perfusion is achieved via gravity and we calculate its velocity to be 1 mL/min. Excess solution is removed with the aid of a vacuum suction, collected in a Buchner flask, and properly disposed of at the end of each experimental session. 9. We use the MetaFluor software (Molecular Devices) to collect the images of the cyan and yellow fluorescent proteins. A more sophisticated version of this program allowing for automation and image analysis called MetaMorph is also available (http:// www.moleculardevices.com/systems/metamorph-researchimaging/metamorph-microscopy-automation-and-imageanalysis-software). 10. Transient transfection often leads to heterogeneous level of expression. Therefore it is important to validate each sensor used. Basic validation should test the minimal, maximal, and basal FRET as a function of expression. In the case of cAMP, this can be achieved by treating the cells with an AC inhibitor (SQ22536) (250 μM) [28] or AC activator (forskolin) (1–20 μM) in the presence of nonspecific phosphodiesterase inhibitor (IBMX) (100 μM–1 mM) depending on the cell type. The aim is to identify the optimal expression “window” where cells are expressing relatively low amounts of sensor, produce clear, near maximal FRET changes above the background noise, and display no cellular toxicity. 11. We recommend to image cells that have a similar level of expression. 12. Different modes of sterilization of the coverslips can be used. We find that brief immersion of the coverslips in a solution of chloroform/ethanol (1:1) followed by air-drying under a laminar flow hood is a fast and efficient way. However coverslips can also be autoclaved in bulk or immerged in absolute ethanol and then flamed. For all these procedures, relevant protective measures have to be followed.

10

Alex Burdyga and Konstantinos Lefkimmiatis

13. Depending on the type of imaging chamber you decide to use, application of vacuum crease may not be necessary. 14. We found that this step minimizes the possibility of the perfusion system developing a leak during the experiment. 15. Try and perform this procedure relatively quickly within 60 s, as leaving the coverslip out of the 12-well plate will cause it to dry within 5–10 min. 16. In our experience, the cAMP H90 and the PKA AKAR4 sensors used in this chapter reach optimal expression level 24–36 h after transfection and are well tolerated by the cells up to 72 h post transfection. It is advised to set a titration experiment using different amounts of plasmid DNA in order to identify the optimal rate of expression of each sensor in respect to specific cell types. 17. Depending on whether the particular FRET probe works on the principle of loss or gain of FRET, it may be necessary to divide the donor over acceptor to see an increase in FRET ratio, which will correspond to a rise in cAMP/PKA activity. 18. Excessive excitation time of the fluorophores might produce a drop of their fluorescence intensity (photobleaching) and affect the starting FRET ratio. Similarly, for some types of fluorescent proteins, prolonged excitation might result in photoisomerization, causing a change in the emission intensity. Photoisomerization is usually reversible, and, if it occurs, it will appear as a drifting FRET baseline until the fluorophores recover back to their native state. To accelerate recovery, stop imaging for 5–15 min, then confirm that the baseline has stabilized, and begin imaging the experiment. We find that in our setup, exposure times of 100 ms, with LED light source power set at approximately 0.7 W and acquisition intervals of 5–15 s, ensure minimum bleaching and photoisomerization of the probes. 19. Agonists might be dissolved in solvents such as DMSO and ethanol, which can have adverse effects on the cells. Therefore, corresponding amount of solvents should be applied to all the solutions (i.e., if the final concentration of DMSO in the imaging solution on application of an agonist is going to be 3 μL/ mL, then adjust all solutions, including the control solution, to contain DMSO at a final concentration of 3 μL/mL). 20. We noted significant cell-type-dependent variability of the responses to IBMX and to a lesser extent forskolin. In order to identify the optimal working concentration of each of these compounds, we recommend the setting of a dose-response assay in the cell type of interest. 21. MetaFluor software requires for raw data to be saved into a specified folder directory before staring the experiment; if this

Monitoring of Synchronous cAMP and PKA Events

11

is not initially performed, then all of the data acquired during the experiment will be lost. 22. Third-party software can also be used to open saved imaging file formats, such as the free image analysis program, ImageJ (http://imagej.nih.gov/ij/). 23. Care must be taken when monitoring signals within ROIs drawn around individual cells, as they can change shape and move over time. For this reason, you should quickly scroll through the entire length of the experiment and double-check that the ROI stays true to the original cell and modify its outline accordingly if the cell changes shape/position. If the cell moves too much to enable the use of a fixed-shape ROI, then dynamic ROIs can be used with some software packages, which can automatically track individual cells over time. In our experience empty space within the ROI (introduced due to uneven cellular shapes) will not generate any variability in the final measurements after background subtraction. 24. Check that no fluorescent debris passes through any of the ROIs during the entire length of the experiment, as this will likely introduce FRET artifacts. 25. Due to the exclusion of H90 from the nucleus, unlike AKAR4 which enters the nucleus, the two sensors can be clearly distinguished without the cotransfection of mCherry with H90. 26. Although H89 is a potent inhibitor of PKA, and therefore it serves a good purpose as a control in this type of experiments, it is important to note that this drug inhibits many other kinases (ROCKII, S6K1, etc.) at concentrations comparable to those required to inhibit PKA [29]. The use of other more selective inhibitors of PKA such as protein kinase inhibitor (PKI) is recommended. References 1. Lefkimmiatis K, Zaccolo M (2014) cAMP signaling in subcellular compartments. Pharmacol Ther 143:295–304. doi:10.1016/j.pharmthera. 2014.03.008 2. Lim CJ, Kain KH, Tkachenko E et al (2008) Integrin-mediated protein kinase A activation at the leading edge of migrating cells. Mol Biol Cell 19:4930–4941 3. Burdyga A, Conant A, Haynes L et al (2013) cAMP inhibits migration, ruffling and paxillin accumulation in focal adhesions of pancreatic ductal adenocarcinoma cells: effects of PKA and EPAC. Biochim Biophys Acta 1833:2664–2672 4. Zimmerman NP, Roy I, Hauser AD et al (2013) Cyclic AMP regulates the migration

and invasion potential of human pancreatic cancer cells. Mol Carcinog 54:203–215. doi: 10.1002/mc.22091 5. Insel PA, Zhang L, Murray F et al (2012) Cyclic AMP is both a pro-apoptotic and antiapoptotic second messenger. Acta Physiol (Oxf) 204:277–287 6. Beavo JA, Brunton LL (2002) Cyclic nucleotide research—still expanding after half a century. Nat Rev Mol Cell Biol 3:710–718 7. Hayes JS, Brunton LL, Mayer SE (1980) Selective activation of particulate cAMPdependent protein kinase by isoproterenol and prostaglandin E1. J Biol Chem 255: 5113–5119

12

Alex Burdyga and Konstantinos Lefkimmiatis

8. Hayes JS, Brunton LL, Brown JH et al (1979) Hormonally specific expression of cardiac protein kinase activity. Proc Natl Acad Sci U S A 76:1570–1574 9. Buxton IL, Brunton LL (1983) Compartments of cyclic AMP and protein kinase in mammalian cardiomyocytes. J Biol Chem 258: 10233–10239 10. Zaccolo M, De Giorgi F, Cho CY et al (2000) A genetically encoded, fluorescent indicator for cyclic AMP in living cells. Nat Cell Biol 2:25–29 11. Ponsioen B, Zhao J, Riedl J et al (2004) Detecting cAMP-induced Epac activation by fluorescence resonance energy transfer: Epac as a novel cAMP indicator. EMBO Rep 5: 1176–1180 12. DiPilato LM, Cheng X, Zhang J (2004) Fluorescent indicators of cAMP and Epac activation reveal differential dynamics of cAMP signaling within discrete subcellular compartments. Proc Natl Acad Sci U S A 101: 16513–16518 13. Nikolaev VO, Bünemann M, Hein L et al (2004) Novel single chain cAMP sensors for receptor-induced signal propagation. J Biol Chem 279:37215–37218 14. Zhang J, Ma Y, Taylor SS, Tsien RY (2001) Genetically encoded reporters of protein kinase A activity reveal impact of substrate tethering. Proc Natl Acad Sci U S A 98:14997–15002 15. Zhang J, Hupfeld CJ, Taylor SS et al (2005) Insulin disrupts beta-adrenergic signalling to protein kinase A in adipocytes. Nature 437: 569–573 16. Depry C, Allen MD, Zhang J (2011) Visualization of PKA activity in plasma membrane microdomains. Mol Biosyst 7:52–58 17. Klarenbeek J, Jalink K (2014) Detecting cAMP with an EPAC-based FRET sensor in single living cells. Methods Mol Biol 1071:49–58 18. Klarenbeek JB, Goedhart J, Hink MA et al (2011) A mTurquoise-based cAMP sensor for both FLIM and ratiometric read-out has improved dynamic range. PLoS One 6: e19170

19. Van der Krogt GNM, Ogink J, Ponsioen B, Jalink K (2008) A comparison of donoracceptor pairs for genetically encoded FRET sensors: application to the Epac cAMP sensor as an example. PLoS One 3:e1916 20. Allen MD, Zhang J (2006) Subcellular dynamics of protein kinase A activity visualized by FRET-based reporters. Biochem Biophys Res Commun 348:716–721 21. Lefkimmiatis K, Leronni D, Hofer AM (2013) The inner and outer compartments of mitochondria are sites of distinct cAMP/PKA signaling dynamics. J Cell Biol 202:453–462 22. Terrin A, Di Benedetto G, Pertegato V et al (2006) PGE(1) stimulation of HEK293 cells generates multiple contiguous domains with different [cAMP]: role of compartmentalized phosphodiesterases. J Cell Biol 175:441–451 23. Stangherlin A, Koschinski A, Terrin A et al (2014) Analysis of compartmentalized cAMP: a method to compare signals from differently targeted FRET reporters. Methods Mol Biol 1071:59–71 24. Miranda JG, Weaver AL, Qin Y et al (2012) New alternately colored FRET sensors for simultaneous monitoring of Zn2+ in multiple cellular locations. PLoS One 7:e49371 25. Sample V, DiPilato LM, Yang JH et al (2012) Regulation of nuclear PKA revealed by spatiotemporal manipulation of cyclic AMP. Nat Chem Biol 8:375–382 26. Shaner NC, Campbell RE, Steinbach PA et al (2004) Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nat Biotechnol 22:1567–1572 27. Hofer AM, Curci S, Doble MA et al (2000) Intercellular communication mediated by the extracellular calcium-sensing receptor. Nat Cell Biol 2:392–398 28. Lefkimmiatis K, Srikanthan M, Maiellaro I et al (2009) Store-operated cyclic AMP signalling mediated by STIM1. Nat Cell Biol 11:433–442 29. Lochner A, Moolman JA (2006) The many faces of H89. Cardiovasc Drug Rev 24(3–4): 261–274

Chapter 2 Recording Intracellular cAMP Levels with EPAC-Based FRET Sensors by Fluorescence Lifetime Imaging Marcel Raspe, Jeffrey Klarenbeek, and Kees Jalink Abstract Eukaryotic cells use second messengers such as Ca2+, IP3, cGMP, and cAMP to transduce extracellular signals like hormones, via membrane receptors to downstream cellular effectors. FRET-based sensors are ideal to visualize and measure these rapid changes of second messenger concentrations in time and place. Here, we describe the use of EPAC-based FRET sensors to measure cAMP with spatiotemporal resolution in cells by fluorescence lifetime imaging (FLIM). Key words FLIM, FRET, cAMP, EPAC, Live-cell FLIM

1

Introduction Second messengers are critical intermediates that relay signals from membrane-bound receptors to intracellular effectors. The second messenger cyclic adenosine monophosphate (cAMP) is used by many organisms in a wide variety of signaling pathways. In rare cases, such as the slime mold Dictyostelium [1], cAMP can also convey extracellular signals. Intracellular concentrations of cAMP change rapidly upon synthesis from ATP by a family of adenyl cyclases (ACs) or degradation by a subgroup of phosphodiesterases (PDEs). Mammalian ACs are either cytosolic or membrane bound and regulated via (I) phosphorylation by PKA, PKC, and calmodulin-dependent protein kinases (CaMK) and (II) second messenger molecules such as Ca2+ and (III) via protein-protein interactions. A prominent mode of activation for mammalian membrane-bound ACs is the interaction with heteromeric G proteins. Extracellular signals, such as hormones and neurotransmitters, bind to G protein-coupled membrane receptors. The Gα protein exchanges GDP for GTP, and as a consequence, the Gαβγ complex dissociates into Gα and Gβγ. The G proteins then bind ACs, and depending on the specific G protein subunits and the type of AC, they are either activated or inhibited [2, 3].

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_2, © Springer Science+Business Media New York 2015

13

14

Marcel Raspe et al.

The cAMP concentration is reduced by several PDEs. Some PDEs are selective for cGMP or cAMP; the majority however is capable of degrading both cGMP and cAMP. Just as the ACs, PDEs are regulated by a variety of mechanisms such as phosphorylation, localization, small signaling molecules, and protein-protein interactions. cAMP itself binds to the GAF-domain* of PDEs and activates or inactivates the cAMP hydrolysis depending on the specific PDE. PDEs have been associated with several diseases, including heart failure, depression, asthma, and inflammation, and many drugs are developed that selectively target PDEs [4, 5]. It is, therefore, important to elucidate the regulation of cAMP concentrations in these diseases, as well as in healthy cells. Förster resonance energy transfer (FRET) is the non-radiative transfer of energy from a donor fluorophore to an acceptor fluorophore [6]. FRET is extremely sensitive to distance: typically, a fluorescent protein needs to be in the range of 1–10 nm for FRET to occur. This characteristic distance range makes FRET ideal to measure protein-protein interactions as well as conformational changes within proteins in living cells. The most common techniques to read out FRET are sensitized emission (SE) and fluorescence lifetime imaging (FLIM). SE can be measured with relatively simple and widely available equipment by recording both donor and acceptor emissions. However, SE is not fully quantitative unless endpoint calibrations are performed or quite elaborate corrections are carried out [7, 8]. FLIM recording, on the other hand, is a much more robust and quantitative technique. FLIM reports on FRET because FRET shortens the donor lifetime. FLIM has the important advantages that lifetimes generally are independent of concentration, bleaching, or excitation fluctuations. However, fluorescence lifetimes of excited fluorophores typically are a few nanoseconds, and FLIM therefore requires complex and dedicated machinery. Time-correlated single photon counting (TCSPC) [9] and frequency-domain (FD)-FLIM [10–12] are the most widely used techniques to measure fluorescent lifetimes. Here, we describe the use of FD-FLIM to measure cAMP levels in multiple cells with a high sample rate. Genetically encoded intramolecular FRET sensors are extremely useful tools to study second messenger concentrations such as Ca2+ with calmodulin sensor [15] and troponin C-sensor [16], cGMP with the regulatory domain of PDEs and the cGMP-binding domain B from cGMP-dependent protein kinase (GKI) [17, 18], and cAMP concentrations with protein kinase A (PKA) [19], the protein “exchange proteins activated directly by cyclic AMP” (EPAC)1 [20, 21], and EPAC2 [22]. Previously, we described cAMP sensing with SE using our EPAC1-based sensor [23].

*

GAF is an acronym for mammalian cGMP-binding PDEs, Anabaena adenylyl cyclases, and Escherichia coli F hlA [13, 14].

Recording Intracellular cAMP Levels with EPAC-Based FRET Sensors by FLIM

15

EPAC1 is a guanine nucleotide exchange factor for Rap1 that is activated by direct binding of cAMP [21]. EPAC1 has an N-terminal DEP (Dishevelled, Egl, Pleckstrin) domain that is essential for membrane localization, a cAMP-binding domain, a REM domain (Ras exchanger motif), and a C-terminal GEF catalytic domain (guanine exchange factor) that regulates the GDP/ GTP binding affinity of the Ras-like protein Rap. The EPAC2 protein is identical in domain structure except for a second N-terminal cyclic nucleotide monophosphate (cNMP) binding domain [24]. Several groups reported on cAMP FRET sensors. Nikolaev et al. [25] made a compact FRET sensor by fusing the cNMP domains of EPAC1 and EPAC2 in-between the donor and acceptor fluorophores. The full-length EPAC1 protein was also used [20]. In our design, we deleted the membrane-binding DEP domain (ΔDEP) of EPAC1, and we also introduced point mutations to render it catalytically inactive (CD, T781A, and F782A) so as to prevent unwanted downstream signaling to Rap1 and Rap2 [21, 26]. This proved a good strategy as this configuration yields very robust FRET changes upon cAMP binding. In our original sensor, EPAC(ΔDEP, CD) was sandwiched between CFP and YFP donor and acceptor fluorophores [21]. Since then, we have reported several rounds of optimization [27, 28]. In this report, we illustrate a reporter where a truncated version of mTurquoise2 [29] (cyan fluorescent protein analogue) is used as donor and a tandem of circular permutated Venus and Venus (yellow fluorescent protein analogue) as an acceptor [27, 28]. The donor has an exceptionally high quantum yield and brightness, allowing for dim excitation and thus minimizing bleaching and phototoxicity. Moreover, it has an outstanding signal-to-noise ratio and good fluorophore maturation, and it is biochemically inert (i.e., it does not interact with other proteins, and its illumination minimally perturbs cell function). Unlike most other fluorescent proteins, the decay of excited mTurquoise2 is well fitted with a single exponent, which makes it especially suited for FLIM. This construct will be referred to as mT2EPACcpVV. We describe the step-by-step protocol to perform a cAMP recording using a Lambert Instruments FD-FLIM setup. The protocol described here should, however, be equally applicable to recordings with other intramolecular FRET sensors.

2

Materials All solutions are made with deionized water.

2.1

Stock Solutions

1 M NaCl2; 1 M NaOH; 2.5 M CaCl2; 1 M CaCl2; 1 M MgCl2; 0.1 M KCl; 1 M Glucose*;1 M HEPES* (*see Note 1).

16

2.2

Marcel Raspe et al.

Materials

2.3 Working Solutions

0.22 μm filters; Attofluor cell chamber; 24 mm Ø, 0.15 mm glass coverslips; forceps; non-pyrogenic polystyrene tubes; 6-well plate. 1. 2× HEPES buffered saline (HBS-buffer): 280 mM NaCl, 10 mM KCl, 20 mM HEPES, pH = 7.2 at 37 °C. Magnesium (1 mM MgCl2), calcium (1 mM CaCl2), and glucose (10 mM) are added when the dilution to 1× HBS-buffer is made. 2. 2× HBS-buffer (for transfection): 280 mM NaCl, 50 mM HEPES, 1.5 mM Na2HPO4, pH 7.2. The optimal pH depends on the cell line that is used (see Note 2). 3. 1.5:1 W/V polyethylenimine (PEI) (MW ~ 25.000) ethanol solution. Store in glass at −20 °C (see Note 3). 4. 1× phosphate buffered saline (PBS): 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, and 1.8 mM KH2PO4.

2.4

Cell Media

1. DMEM supplemented with 10 % FCS and penicillin/streptomycin (pen/strep). Use the appropriate medium with additives if other cell lines are used. 2. 0.05 % trypsin-EDTA solution. 3. DMEM/F-12 and/or phenol red-free Leibovitz’s L15 (see Note 4).

2.5

3

Microscope

Leica DMIRE2 with Lambert Instruments software and hardware. A 1 W, 442 nm LED is used as excitation light source, and a Leica CFP filter cube (BP 436/20 nm, dichroic 455 nm, BP 480/40 nm) is used in the light path (see Note 5). The entire microscope is kept at 37 °C.

Methods

3.1

Cell Culture

24 mm Ø, 0.15 mm glass coverslips are sterilized with 70 % alcohol or UV-C and placed in a 6 wells plate with 2 ml of DMEM with FCS and pen/strep. Cells are trypsinized and resuspended in medium, and approximately 150.000 cells are added to each well (see Note 6). Depending on the cell line, cells are transfected 8–24 h after plating and cultured for another 24–72 h prior to FLIM.

3.2

Transfection

1. PEI transfection: 2 μl of the PEI solution is mixed with 1 μg of plasmid DNA in 100 μl serum-free medium in a polystyrene tube. Mix gently, and incubate for 15 min at room temperature. Drop the PEI/medium mixture to the cells, and place the 6 wells plate, after gently swirling, back in the incubator (see Note 7). 2. Calcium phosphate transfection: put 86 μl of 2× HBS in a polystyrene tube, and add 2–5 μg of plasmid DNA diluted in

Recording Intracellular cAMP Levels with EPAC-Based FRET Sensors by FLIM

17

H2O (final volume is 194.9 μl). Mix gently and add 5.1 μl of 2.5 M CaCl2 and mix again. Incubate for 20 min at room temperature, and add dropwise to the cells (see Note 8). Place the cells back in a CO2 incubator. Optionally replace the medium after 24 h. 3. Any commercial transfection reagent can be used according to the manufacturer’s protocol. 3.3

Imaging

3.3.1 General Preparations

1. Cell chambers, for example, Attofluor cell chambers, and imaging medium are preheated at 37 °C. The coverslip is taken from the 6 wells plate with sharp forceps and carefully mounted in a cell chamber (see Note 9). 2. Wash the cells once with PBS, and immediately add 1–2 ml HBS+/+ (=1× HBS with glucose, CaCl2, and MgCl2) to the cell chamber (see Note 10). 3. Place the loaded cell chamber on the microscope. Use the appropriate immersion liquid on the objective, and focus on the cells. After focusing on the cells, wait up to 5 min to equilibrate the temperature.

3.3.2 FrequencyDomain FLIM

1. Turn all the hard- and software (FLIM control box, microscope, LED cooling, camera, and computer) on, 30–60 min prior to imaging (see Note 11). 2. Check if the software and hardware settings are correct for EPAC-sensor imaging. 440 nm excitation light source, modulation frequency at ~40 MHz (see Note 12), MCP between 600 and 800 V, and phase angle set at maximum intensity. 3. Measure an appropriate reference dye that is at microscope temperature. Frequently used reference dyes are erythrosin B (τ = 69 ps at 37 °C) and Rhodamine 6G (τ = 3.83 ns at 37 °C) (see Notes 13 and 14). 4. Mount the cells on the microscope. Focus by looking through the eyepiece onto the cells (see Subheading 3.3.1). 5. Optimize the exposure time and sampling rate by measuring a time series in unstimulated cells. The exposure time should be set so that the camera chip is below 50 % of full-well capacity (midrange in gray values) (see Note 15). 6. Select a region with one or more healthy cells (see Note 16). Start the actual experiment by recording a baseline for a few frames before the stimulation of the cells with a concentrated stock reagent (see Note 17). Follow the changes in lifetime. In case of transient effects, optionally, a second stimulus can be added. When the plateau is reached after the final (calibration) stimulus, wait for another few frames. In Fig. 1, the initial baseline is recorded for approximately 4 min. Then isoproterenol, a β-adrenergic receptor agonist, is

18

Marcel Raspe et al.

a Intensity

b Int. (a.u.) Lifetime

max

τ (ns) 4.0

c 3.4

Tau (ns)

3.2

3.0

Isoproterenol

IBMX/Forskolin

2.8 0

2.0

2.6

2.4 0

10

20

30

40

Time (min)

Fig. 1 Time-lapse FLIM imaging of U2OS cells transfected with mT2EPACcpVV. (a) mTurquoise2 fluorescence intensity at t = 0, 7.5, and 40 min. (b) Corresponding fluorescence lifetimes. (c) Fluorescence lifetime curve. Cells were stimulated with isoproterenol (0.25 nM) and with a mixture of IBMX (100 μM) with forskolin (25 μM), as indicated with the arrowheads

added resulting in an increase of intracellular cAMP and an increase of fluorescence lifetime due to the opening of the FRET sensor. The second stimulation with IBMX/forskolin activates the ACs and inhibits the PDEs and irreversibly increases the intracellular cAMP concentration to the maximum level. 7. Stop the experiment, and make sure the data is properly saved. The data can be analyzed with manufacturer-supplied software, or the raw data can be extracted and saved for processing with your own macros in, e.g., ImageJ (NIH, USA). 8. Post-image processing: all commercial FD-FLIM setups come with their own software package giving “apparent” weighted average lifetimes calculated from the phase (φ) and with modulation (m) properties. For mono-exponentially decaying fluorophores, τφ and τm are equal. With FRET sensors, the τφ and τm are weighted averages and are not equal. The values of τφ and τm can be plotted in a polar plot (Fig. 2). The position within the polar plot gives information about several aspects of the sensor such as the extent of opening/closing of the sensor, bleaching, and autofluorescence (see Note 18). If you decide to calculate τφ and τm for yourself, extract the raw data, and analyze the data in ImageJ, Visual Basic,

Recording Intracellular cAMP Levels with EPAC-Based FRET Sensors by FLIM

Donor-only

4 ns

5 ns

0.5

3 ns

19

mT1

EPAC cpVV

mT1

EPAC tdTomato

mT2

EPAC cpVV

Open

2 ns

0.4 M*SIN(Ф)

mono-exponential decay

Closed

0.3 Acceptor-only

0.2

1 ns

Background

0.1

0 0

0.2

0.4

0.6

0.8

1

M*COS(Ф)

Fig. 2 Polar Plot. Three EPAC sensors with different donors and acceptors are plotted in the polar plot. The EPAC sensors expressed in U2OS cells are induced to open by a mixture of IBMX/forskolin. mT2EPACcpVV is mTurquoise2-EPACcpVenusVenus, mT1EPACcpVV is mTurquoise1-EPAC-cpVV, and mT1EPACtdTomato is mTurquoise1-EPAC-tdTomato

MATLAB, or any other software program you prefer to use (see Note 19). 9. Remove the cell chamber from the objective. Clean the objective with a tissue. Carefully clean the cell chamber with repeated cycles of water and 70 % alcohol to prevent any residual compound from sticking to the metal and affect future experiments. Store the cell chamber in 1 M NaOH (see Note 20). 10. After all experiments are (successfully) performed, turn off all equipment in reverse order (as compared to the start-up).

4

Notes 1. Glucose and HEPES cannot be autoclaved and have to be filtered with a 0.22 μM filter. 2. The optimal pH of the 2× HBS-buffer depends on the cell type used for the experiments. Therefore, a series of different pH buffers is made, ranging from 6.8 to 7.2 in steps of 0.05 and tested on exponentially growing cells (~50 % confluent). 3. After 3 months, transfection efficiency drops dramatically. Storage at 4 °C is possible, but 20 °C is recommended to prevent evaporation. It is strongly recommended to use glass

20

Marcel Raspe et al.

containers, since ethanol taken from an Eppendorf tube by itself sometimes affects cells. 4. When imaging for extended times, cell culture medium is preferred over 1× HBS-buffer. Keeping the autofluorescence minimal is important during FLIM recordings. DMEM/F-12 and/or phenol red-free Leibovitz’s L15 is a good option. DMEM/F-12 is buffered with CO2. If no CO2 is present at the microscope, Leibovitz’s L15 is a good choice. If other media are used, riboflavins and phenol red are a major source of autofluorescence. 5. Any inverted microscope with a modulated light source and image-intensified camera is suitable for FD-FLIM. There are various commercial providers of FD-FLIM equipment. 6. For a homogenous layer of cells, move the plate twice in south/north direction followed by twice in west/east direction. This will prevent cells from piling up in the middle of the well. 7. After 10 h, the first cells will be fluorescent, and 72 h after transfection, expression is highest. Note that PEI can be toxic to cells, and replacing the medium with fresh medium after 24 h is for some cell types advisable. 8. Do not exceed 20 min of incubation since large precipitates are formed which will decrease transfection efficiency. 9. If the forceps is not sufficiently sharp, a bent needle can be useful for lifting up the coverslip. Prevent leakage by placing the coverslip exactly in the middle, and screw the ring tight. However, screwing too tightly can result in breakage of the coverslip. 10. Clean the bottom of the coverslip with a paper tissue. Gently press the paper for a second time to the bottom of the coverslip to check if no leakage occurs. If the cell chamber is leaky, the chamber is not screwed sufficiently tight, the coverslip is not in the middle of the ring, or the coverslip is broken. 11. The Lambert Instruments control box needs time before it becomes stable. Therefore, it is especially important to switch the Lambert Instruments control box on, minimally 30 min before imaging. 12. The optimal modulation frequency for phase and modulation fluorescence lifetime is determined with the following equations [30]: f phase　 optimum =

1 2 × p ×t

f modulation　 optimum =

2 2 × p ×t

Recording Intracellular cAMP Levels with EPAC-Based FRET Sensors by FLIM

21

13. Any reference material can be used when the lifetime is accurately measured. We use the TCSPC FLIM for measuring the “real” fluorescence lifetime of a reference dye. 14. Because of the spectral properties of these dyes, use a long pass 514 nm filter cube instead of the CFP filter cube. Besides the filter cube and exposure time, use exactly the same settings as for the EPAC sensor. After collecting the reference, measure a sample with the same dye, and check the lifetime of a few regions of interest (ROIs). 15. High sample rates improve the accuracy of the cAMP response curve but require more excitation. Too much excitation light may affect stability of the recording, presumably due to the light-dependent production of reactive oxygen species (ROS). Start with an initial exposure time far below full-well capacity of the CCD-chip, because stimulation of the EPAC sensor results in an increase of emitted light due to the opening of the FRET sensor. 16. Make sure the fluorescent cell looks like the none-transfected cells as an indication that they are healthy. Very bright cells often have different morphology, e.g., are more rounded, and these are to be avoided. An intermediate bright fluorescent cell is often a good choice for measuring. 17. Usually, a highly concentrated (~1/1,000) reagent is used to prevent lifetime changes due to the change in volume, temperature, and/or osmolarity. We mix the reagent by first pipetting a small volume with a yellow tip on a P20. Then we transfer the yellow tip carefully to a P200. Now a small drop of reagent is present in the middle of the yellow tip. To stimulate the cells, we carefully put the tip of the pressed pipette in the medium of the cell chamber and release the plunger. Then we gently mix while avoiding air bubbles. Do not mix directly above the cells as you may wash them away. 18. During FD-FLIM, two apparent lifetimes τphase and τmodulation are simultaneously acquired. In some cases, it can be useful to make a polar plot with the acquired phases (Φ) and modulation depths (M). A polar plot can show to what extent the FRET sensor is opened, if fluorophore bleaching occurs, and the contribution of autofluorescence [31–33]. In Fig. 2, three different EPAC sensors stimulated with IBMX/forskolin are inserted in the polar plot. Perfect mono-exponentially decaying fluorophores are located on the semicircle. Lower lifetimes are moving to the right part of the semicircle. FRET sensors are always a mixture of multiple decays and form a line within the semicircle. The mT2 EPACcpVV sensor starts in a closed conformation and opens after the addition of IBMX/forskolin. If the opening of the sensor completely prevents energy transfer, the points end on the

22

Marcel Raspe et al.

semicircle (when the donor decays mono-exponentially). If FRET-energy transfer is 100 % efficient, the point will be at the acceptor-only point (which is impossible to measure with FLIM, as the donor does not emit light). An EPAC sensor with a different donor mTurquoise1 (mT1EPACcpVV) and with different acceptor (mT1EPACtdTomato) is plotted as well. Because of the different donors, these sensors cross the semicircle at a different place. The dynamic range of the sensor is affected by the acceptor (e.g., tdTomato has a smaller range than the cpVV). 19. The sequence for calculating τφ and τm should be, for example, (a) Acquire data from reference and experiment, both with a background image. (b) Subtract background from all images. This is important for acquiring the proper τmodulation. (c) Calculate from a reference with known fluorescence lifetime, system phase, and modulation with the following Equations [11, 12, 34]: K

Fsin = å sin(2 × p × n × k / K ) × I k

(1)

1 K

Fcos = å cos(2 × p × n × k / K ) × I k

(2)

1 K

FDC = åI k

(3)

1

K is the number of phases recorded. Typically, we use 12 phases. k is the kth image. n the number of harmonics that are included in the calculations. n = 1 is usually sufficient. Ik is the intensity of the kth image. The system phase (i.e., phase of the light without sample) is calculated with Fsin and Fcos and Eqs. 4 and 5. æF ö ftotal ( ref ) = fsystem + freference = TAN -1 ç cos ÷ è Fsin ø

(4)

fsystem = ftotal ( ref ) - TAN -1 (w × t ref )

(5)

ω, the angular frequency is 2·π·f. Here, f is the modulation frequency of the excitation source. And the system modulation is calculated with Fsin, Fcos, and FDC and a theoretical reference modulation with Eqs. 6–8.

Recording Intracellular cAMP Levels with EPAC-Based FRET Sensors by FLIM

1

mref =

(w × t ) mtotal ( ref ) =

2

(6) +1

2 Fsin 2 + Fcos 2

msystem =

23

(7)

FDC mtotal ( ref )

(8)

mref

(d) Next, calculate Fsin Eq. 1, Fcos Eq. 2, and FDC Eq. 3 for a sample with unknown lifetime, and determine φtotal(sample) and mtotal(sample) with Eqs. 4 and 7. τphase and τmodulation are now calculated with Eqs. 9 and 10.

t phase =

1 × TAN fsample - fsystem w

(

)

(9)

2

t mod

1 = w

æ msystem ö ÷ -1 ç ç mtotal ( sample ) ÷ ø è

(10)

(e) Once τphase and τmodulation are calculated, they can be put in a graph. 20. Some organic compounds bind to the steel ring and affect the follow-up experiment. Repetitive washing with water and alcohol is sufficient to get rid of most compounds. Some compound are however not fully removed by washing, and therefore, we place the steel cell chamber in concentrated NaOH for prolonged times. References 1. Wang Y, Chen CL, Iijima M (2011) Signaling mechanisms for chemotaxis. Dev Growth Differ 53:495–502 2. Halls ML, Cooper DMF (2011) Regulation by Ca2+-signaling pathways of adenylyl cyclases. Cold Spring Harb Perspect Biol 3:1–22 3. Gancedo JM (2013) Biological roles of cAMP: variations on a theme in the different kingdoms of life. Biol Rev Camb Philos Soc 88:645–668 4. Francis SH, Blount MA, Corbin JD (2011) Mammalian cyclic nucleotide phosphodiesterases : molecular mechanisms and physiological functions. Physiol Rev 91:651–690 5. Soderling SH, Beavo JA (2000) Regulation of cAMP and cGMP signaling: new phosphodiesterases and new functions. Curr Opin Cell Biol 12:174–179

6. Forster T (1948) Zwischenmolekulare energiewanderung und fluoreszenz. Ann Phys 437:55–75 7. Van Rheenen J, Langeslag M, Jalink K (2004) Correcting confocal acquisition to optimize imaging of fluorescence resonance energy transfer by sensitized emission. Biophys J 86: 2517–2529 8. Jalink K, Van Rheenen J (2009) FRET and FLIM techniques. In: Laboratory techniques in biochemistry and molecular biology. 1st ed., 33, Gadella TWJ (ed) Elesvier B.V. Amsterdam, Netherlands pp. 289–349 9. Becker W, Bergmann A, Hink M, König K, Benndorf K, Biskup C (2004) Fluorescence lifetime imaging by time-correlated singlephoton counting. Microsc Res Tech 63:58–66

24

Marcel Raspe et al.

10. Spencer RD, Weber G (1969) Measurements of subnanosecond fluorescence with a crosscorrelation phase fluorometer. Ann N Y Acad Sci 158:361–376 11. Gadella TWJ, Jovin TM, Clegg RM (1993) Fluorescence lifetime imaging microscopy (FLIM): spatial resolution of microstructures on the nanosecond time scale. Biophys Chem 48:221–239 12. Schneider PC, Clegg RM (1997) Rapid acquisition, analysis, and display of fluorescence lifetime-resolved images for real-time applications. Rev Sci Instrum 68:4107–4119 13. Aravind L, Ponting C (1997) The GAF domain: an evolutionary link between diverse phototransducing proteins. Trends Biochem Sci 22:458–459 14. Zoraghi R, Corbin JD, Francis SH (2004) Properties and functions of GAF domains in cyclic nucleotide phosphodiesterases and other proteins. Mol Pharmacol 65:267–278 15. Miyawaki A, Llopis J, Heim R, Mccaffery JM, Adams JA, Ikura M, Tsien RY (1997) Fluorescent indicators for Ca2+ based on green fluorescent proteins and calmodulin. Nature 388:882–887 16. Mank M, Reiff DF, Heim N, Friedrich MW, Borst A, Griesbeck O, Zellula AG (2006) A FRET-based calcium biosensor with fast signal kinetics and high fluorescence change. Biophys J 90:1790–1796 17. Honda A, Sawyer CL, Cawley SM, Dostmann WRG (2005) Cygnets In: Phosphodiesterase Methods and Protocols, 307, Lugnier C (ed.). Humana Press Inc., Totowa, NJ, pp. 27–43 18. Nikolaev VO, Gambaryan S, Lohse MJ (2006) Fluorescent sensors for rapid monitoring of intracellular cGMP. Nat Methods 3:23–25 19. Adams RS, Harootunian TA, Buechler YJ, Taylor SS, Tsien RY (1991) Fluorescence ratio imaging of cyclic AMP in single cells. Nature 349:694–697 20. Dipilato LM, Cheng X, Zhang J (2004) Fluorescent indicators of cAMP and Epac activation reveal differential dynamics of cAMP signaling within discrete subcellular compartments. Proc Natl Acad Sci U S A 101: 16513–16518 21. Ponsioen B, Zhao J, Riedl J, Zwartkruis F, van der Krogt G, Zaccolo M, Moolenaar WH, Bos JL, Jalink K (2004) Detecting cAMP-induced Epac activation by fluorescence resonance energy transfer: Epac as a novel cAMP indicator. EMBO Rep 5:1176–1180 22. Zhang CL, Katoh M, Shibasaki T, Minami K, Sunaga Y, Takahashi H, Yokoi N, Iwasaki M, Miki T, Seino S (2009) The cAMP sensor

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

Epac2 is a direct target of antidiabetic sulfonylurea drugs. Science 325:607–610 Klarenbeek J, Jalink K (2014) Detecting cAMP with an epac-based FRET sensor in single living cells. In: Zhang J, Ni Q, Newman RH Methods in molecular biology, (eds) 1071. Humana Press, Totowa, NJ, pp 49–58 Rehmann H, Prakash B, Wolf E, Rueppel A, de Rooij J, Bos JL, Wittinghofer A, Rooij JD (2003) Structure and regulation of the cAMPbinding domains of Epac2. Nat Struct Biol 10:26–32 Nikolaev VO, Bünemann M, Hein L, Hannawacker A, Lohse MJ (2004) Novel single chain cAMP sensors for receptor-induced signal propagation. J Biol Chem 279:37215–37218 De Rooij J, Rehmann H, van Triest M, Cool RH, Wittinghofer A, Bos JL (2000) Mechanism of regulation of the Epac family of cAMPdependent RapGEFs. J Biol Chem 275: 20829–36 Klarenbeek JB, Goedhart J, Hink MA, Gadella TWJ, Jalink K (2011) A mTurquoise-based cAMP sensor for both FLIM and ratiometric read-out has improved dynamic range. PLoS One 6:e19170 Van der Krogt GNM, Ogink J, Ponsioen B, Jalink K (2008) A comparison of donoracceptor pairs for genetically encoded FRET sensors: application to the Epac cAMP sensor as an example. PLoS One 3:e1916 Goedhart J, von Stetten D, Noirclerc-Savoye M, Lelimousin M, Joosen L, Hink MA, van Weeren L, Gadella TWJ, Royant A (2012) Structure-guided evolution of cyan fluorescent proteins towards a quantum yield of 93%. Nat Commun 3:751, ncomms1738 Elder AD, Matthews SM, Swartling J, Yunus K, Frank JH, Brennan CM, Fisher AC, Kaminski CF (2006) Application of frequency-domain fluorescence lifetime imaging microscopy as a quantitative analytical tool for microfluidic devices. Opt Express 14:5456–5467 Digman MA, Caiolfa VR, Zamai M, Gratton E (2008) The phasor approach to fluorescence lifetime imaging analysis. Biophys J 94:L14–L16 Redford GI, Clegg RM (2005) Polar plot representation for frequency-domain analysis of fluorescence lifetimes. J Fluoresc 15:805–815 Eichorst, JP, Teng KW, Clegg, RM (2014) Polar plot representation of time-resolved fluorescence. Methods in molecular biology. 1076, Engelborghs V, Visser A (eds), Humana Press Inc., Totowa, NJ, pp. 97–112 Lakowicz JR, Szmacinski H, Nowaczyk K, Berndt KW, Johnson M (1992) Fluorescence lifetime imaging. Anal Biochem 202:316–330

Chapter 3 A Novel Approach Combining Real-Time Imaging and the Patch-Clamp Technique to Calibrate FRET-Based Reporters for cAMP in Their Cellular Microenvironment Andreas Koschinski and Manuela Zaccolo Abstract Fluorescence resonance energy transfer (FRET)-based reporters are invaluable tools to study spatiotemporal aspects of cyclic adenosine monophosphate (cAMP) signaling and compartmentalization in living cells. These sensors allow estimation of relative changes of cAMP levels in real-time and intact cells. However, one of their major shortcomings is that they do not easily allow direct measurement of cAMP concentrations. This is mainly due to the fact that the methods to calibrate these sensors in their physiological microenvironment are not generally available. All published approaches to calibrate FRET-based reporters rely at least in part on data derived under nonphysiological conditions. Here, we present a protocol to calibrate FRET reporters completely “in cell.” We introduce a combination of FRET imaging of cAMP and the whole-cell patch-clamp techniques to microinfuse or dilute intracellular cAMP to known concentrations. This method represents a general tool to accurately estimate intracellular cAMP concentrations by allocating concentration values to FRET ratio changes. Key words Imaging, Fluorescence resonance energy transfer (FRET), cAMP, FRET reporter calibration, Patch clamp, Microinfusion

1

Introduction Since their invention about 25 years ago, FRET-based cAMP biosensors [1] have undergone a constant development. Initially, only delivered into cells by microinjection [2], today, a multitude of sophisticated genetically encoded FRET reporters are available [3]. By targeting these sensors to specific subcellular compartments such as the plasma membrane [4], the nucleus [4], or mitochondria [5], functionally defined compartments, as, for example, the RI or RII compartments [6], microdomains centered on A-kinase anchoring proteins (AKAPs) [6], or other protein complexes [7, 8], these sensors now have reached an unprecedented spatial resolution. They are indispensable tools for the investigation of cAMP–PKA signaling pathways and subcellular cAMP dynamics.

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_3, © Springer Science+Business Media New York 2015

25

26

Andreas Koschinski and Manuela Zaccolo

All the available sensor variants share the same basic working mechanism. Two fluorophores are used to sandwich a cAMPbinding domain that undergoes a conformational change upon binding of the second messenger. If the emission spectrum of the donor fluorophore overlaps with the excitation spectrum of the acceptor fluorophores, on donor excitation, energy can be transferred non-radiatively from the donor fluorophore to the nearby acceptor. This phenomenon involves a direct dipole–dipole interaction and is called Förster resonance energy transfer (FRET) [9]. As the FRET efficiency is inversely proportional to the sixth power of the donor–acceptor distance, even small conformational changes due to binding or release of cAMP can result in pronounced alterations of energy transfer efficiency. Thus, the ratio of the emission intensities of the donor and the acceptor fluorophore (FRET ratio) will change accordingly and provide an excellent readout for the change of global or—in the case of targeted sensors—local intracellular cAMP concentrations. A major drawback is that it is very difficult to allocate real intracellular cAMP concentrations to the FRET change measurements. Not being able to link a change in FRET ratio to concentration changes poses a major limitation when comparing responses detected by different FRET reporters. In the majority of studies in which a comparison between different reporters is required, normalization to the maximal response of the respective sensor with a saturating stimulus (e.g., a combination of forskolin, 10–25 μmol/L, and IBMX, 100 μmol/L) is normally applied. However, unless the sensors that are being compared show superimposable dose–response curves [4, 6], this practice should be avoided as it may lead to gross misinterpretation of the data. If the compared sensors have different saturation levels, different affinities for the signal molecule, or different slopes of their cAMP dose– response curves, the results obtained with this approach may be misleading. Unfortunately, for most sensors, the respective dose– response curves are not available. The same applies to sensors that are modified by the fusion to targeting sequences, as this may affect the FRET signal via a number of mechanisms. Altered interaction between the fluorophores, altered cAMP-binding properties due to conformational change of the sensor secondary to the fusion, or exposure of the fluorophores to different microenvironmental conditions (e.g., local variations in pH, Cl− concentration, etc.) may affect fluorescence emission independently of cAMP signals. Due to all these factors, FRET efficiency might be altered unpredictably or even be abolished [8]. In these cases, the dose–response curves might also be altered significantly. Thus, absolute values of FRET ratio changes for different sensors, or even for the same parental sensor that has been modified with different targeting sequences, are not necessarily directly comparable.

“In-cell” Calibration of FRET-Based cAMPsensors

27

Besides that, the expression of FRET ratio changes simply as a percentage (e.g., of maximal FRET change at sensor saturation) suggests a linearity in the FRET response which for most sensors does not exist. In the attempt to overcome some of these problems, sensors have been characterized either “in cell” by generating dose– response curves with, for example, forskolin [10], or in vitro with known concentrations of cAMP [11, 12]. It should be noted that in vitro sensor characterizations often are performed in nonphysiologic salt solutions [11–13], with ionic, pH, and osmolal conditions that are different from physiological and that may significantly influence the sensor performance [14, 15] (and own observations). In addition, for some tested sensors, we noticed aggregation phenomena under such artificial conditions. Recently, an approach based on an in-cell two-point calibration has been published [16]. This method relies on blocking intracellular adenylate cyclase with the inhibitor MDL-12,330A (100 μM) to deplete intracellular cAMP and reach a minimal (“zero”) value for the FRET ratio, followed by a saturating stimulus using the membrane-permeable cAMP analogue 8-Br-2′-OMe-cAMP-AM (20 μM), which then should elicit the maximum response of the sensor. In case EC50 and Hill coefficient of the sensor are available, these minimal and maximal FRET ratio values allow the extrapolation of the corresponding in-cell calibration curve. However, as EC50 and Hill coefficient typically are determined under artificial conditions (in vitro), the abovementioned concerns still apply. To overcome such limitations, we developed a microinfusion technique which is based on a combination of FRET imaging [3] and the whole-cell patch-clamp technique [17]. A crucial prerequisite to this approach is to match exactly the infused solution to the cytosolic pH of the cells, to avoid artifacts. Therefore, in the following method, we also determine the cytosolic pH by exploiting the ph-dependent differential quenching of fluorophores and the resulting FRET ratio change of the FRET sensor itself. In brief, cells expressing the FRET-based reporter to be calibrated are attached to a glass pipette such that a tight connection is formed between cell membrane and the pipette, the so-called Giga-seal. When this connection has been established, the membrane under the pipette tip is ruptured in a controlled way. This establishes a direct flow between cytosol and the solution in the patch pipette. The solution in the patch pipette contains a known concentration of cAMP which will diffuse, depending on the concentration gradient, either into the cytosol or out of the cell into the pipette, resulting in a shift in FRET ratio that is dependent on the change in cAMP concentration. As the volume in the pipette typically is several orders of magnitude larger than the cell volume

28

Andreas Koschinski and Manuela Zaccolo

Fig. 1 Main principle of the calibration by the microinfusion technique. (a) A tight connection between cell membrane and pipette tip is established. The strength of this connection (“Giga-seal”) is monitored by the patch-clamp setup and expressed in Giga-Ohm. Simultaneously, the FRET ratio of the sensor is monitored by the FRET setup, indicated by the beam coming from the objective. (b) After the recording of the FRET-baseline ratio, the membrane under the tip is ruptured. Due to the concentration gradient, cAMP will immediately start to diffuse into the cytosol. (c) After a few seconds, the cAMP concentration equilibrates between cytosol and patch pipette; the FRET ratio reaches a new value (plateau). FRET ratio changes, strength of the connection (seal resistance), and quality of the direct connection between cytosol and patch pipette (access resistance) are constantly monitored. Note that the volume of the intracellular buffer in the patch pipette is about two million times more than the volume of the cytosol

(10 μl, compared to approximately 5 pl of the cytosol of a CHO cell), the cytosolic cAMP concentration will equilibrate rapidly to the adjusted cAMP concentration in the pipette solution. During the entire course of the experiment, the FRET ratio is monitored and recorded by the FRET recording setup. Simultaneously, the quality of the seal between membrane and pipette is monitored by the patch-clamp setup to ensure that there is no leakage from the extracellular medium into the cell or vice versa. In addition, the access resistance is also monitored continuously to ensure a constantly open connection between the patch pipette and cytosol (see Fig. 1). Each cell is microinfused with a given concentration of cAMP. For each concentration, a number of cells are microinfused. Then the average FRET ratio change for each concentration is calculated and plotted against the concentration to generate the calibration curve for the sensor. This curve then can be used to analyze FRET ratios or ratio changes in the respective cells in terms of real cAMP concentrations. This calibration protocol can be applied to any cell type that can be subjected to the whole-cell patch-clamp technique and is not limited to any specific type of sensor or signaling molecule. The accuracy of this approach is only limited by the variability of the basal ratios of the cells and can be increased by measuring a representative pool of cells to generate a more representative mean.

“In-cell” Calibration of FRET-Based cAMPsensors

2

29

Materials

2.1 Cell Culture and Transfection

1. Chinese hamster ovary cells (CHO) (see Note 1). 2. CHO growth medium: HAM’S-F12, 10 % fetal bovine serum, 100 U/mL penicillin, 100 μg/mL streptomycin, and 2 mM glutamine. 3. 0.05 % trypsin. 4. Dulbecco’s phosphate buffer solution (DPBS): 137.9 mM NaCl, 2.7 mM KCl, 8.1 mM Na2HPO4, 1.5 mM KH2PO4. 5. Transfection reagent (MIR 2300, Mirus). 6. Geneticin G418 or other appropriate selection antibiotic for the used vector. 7. Cell counting chamber or equivalent. 8. A microscope with fluorescence capability and long working distance objective to allow inspection of fluorescent cells in cell culture dishes.

2.2 Determination of Intracellular pH

1. CHO cells, transfected with the sensor of interest. 2. FRET setup, acquisition and analysis software [3, 16]. Alternatively, a setup capable of measuring intracellular pH with, e.g., BCECF [18] or SNARF-1 [19]. 3. Extracellular buffer: 140 mM KCl, 4 mM NaCl, 1 mM MgCl2, 2 mM CaCl2, 1 mM Na-pyruvate, 15 mM glucose, 10 mM HEPES. Buffer adjusted with KOH to various pH (6.8 to 7.8) (see Note 2). 4. Nigericin, final concentration 3.3 μM, and valinomycin, final concentration 5 μM.

2.3 Microinfusion Experiments and Analysis

1. Extracellular buffer (modified Tyrode solution): 140 mM NaCl, 3 mM KCl, 2 mM MgCl2, 2 mM CaCl2, 15 mM glucose, 10 mM HEPES. Buffer adjusted with NaOH to pH 7.2. 2. Intracellular buffer: 20 mM NaCl, 140 mM KGlutamate, 2 mM MgCl2, 0.00404 mM CaCl2, 0.1 mM BAPTA (yielding a calculated free Ca2+ concentration of 10 nM, low buffered), 10 mM HEPES. Buffer adjusted with KOH according to the intracellular pH of the respective cells (here pH 7.64). Supplemented with various cAMP-Na concentrations. 3. cAMP-natrium salt. 4. Patch-clamp pipette puller and glass capillaries. 5. Combined patch-clamp [17] and FRET setup [3, 16], with acquisition and analysis software. 6. Graph Pad Prism™ software.

30

3

Andreas Koschinski and Manuela Zaccolo

Methods

3.1 Generation of Stably Transfected Monoclonal Cell Lines (See Note 3)

1. Seed cells into 35 mm cell culture Petri dishes, and transfect with the vector construct carrying the reporter-encoding sequence. In our example, we used CHO cells and transfected them with a newly developed, targeted cAMP sensor (Inf2AKAP79) (see Note 4). 2. Apply the selection antibiotic 24–48 h after transfection. For the CHO cells, we used 800 μg/ml G418. 3. When the cells under antibiotic selection start growing normally again and have reached an adequate density, plate an aliquot of them into 96-well cell culture plates at a dilution of one cell per well (see Note 5). 4. After approximately 2 or 3 days, fluorescent colonies should become visible. If a fluorescent clone has grown to a number of at least a few hundred cells, transfer the cells to a 25 cm2 cell culture flask, let them proliferate, and start normal passaging (see Note 6).

3.2 Determination of the Intracellular pH (See Note 7)

Perform a series of pH measurements using the FRET reporter as a pH sensor [18, 19]. The necessary equations and examples are shown below (Subheading 3.3). 1. Seed stably transfected cells on high-quality glass coverslips, and let them grow to your preferred density (see Note 8). Transfer the coverslip with the cells into the KCl-based buffer of desired pH. 2. Excite at 435 nm, and record the emission intensities at 480 nm (CFP emission) and 535 nm (YFP emission) over ztime. 3. After recording the basal ratio, apply nigericin and valinomycin (see Note 9). 4. Record the subsequent pH-induced apparent FRET change until the changing ratio reaches a new plateau. 5. Calculate the time course of FRET change by computing the ratio (R) value for each acquisition time point, using Eq. 1. Then calculate the FRET changes using Eq. 2, and plot the values against time. Repeat steps 1–5 with different extracellular pH values. Example traces are shown in Fig. 2a. 6. Plot the maximal FRET change values against the applied pH. Fit a curve to the values, and determine the zero-crossing (Fig. 2b). The value at which the fitted curve crosses the x-axis represents the mean intracellular pH of the cells. 7. Match the pipette solutions for the calibration experiments (Subheading 3.3) to the determined pH value.

“In-cell” Calibration of FRET-Based cAMPsensors

31

Fig. 2 Determination of the intracellular pH. (a) Plot of the pH-induced FRET change over time from five experiments at the indicated pH of the bath solution. Curves represent the mean of 21–27 cells for each pH value. Nigericin (3.3 μM) and valinomycin (5 μM) were bath applied at the time point indicated by the arrow. (b) Curve fit to the FRET change values obtained when applying bath solutions at different pH as shown in (a). Zerocrossing of the curve at pH 7.64 indicates the mean cytosolic pH of the cells (Note 10)

3.3 Generation of cAMP Dose– Response Curves for the cAMP Sensor

1. Seed stably transfected cells on high-quality glass coverslips, and let them grow to your preferred density (see Note 8). 2. Perform a series of microinfusion measurements, infusing different concentrations of cAMP into different cells, e.g., 1,000 μM, 100 μM, etc., down to 0.01 μM and 0 μM. If necessary, perform additional experiments at intermediate concentrations (e.g., 30 μM, 3 μM, 0.3 μM, etc.) (see Note 11). For each cell under examination: Fill the patch pipette with intracellular solution containing the appropriate cAMP concentration (see Note 12). Establish a tight connection to the cell (the so-called Giga-seal). Start FRET monitoring by exciting the cell at 435 nm, and record the emission intensities at 480 nm (CFP emission) and 535 nm (YFP emission) over time (Fig. 3a) (see Note 13). Monitor and record the FRET ratio for some minutes before establishing the whole-cell configuration. Establish the direct access to the cytosol (the “whole-cell” patch-clamp configuration) by rupturing the membrane under the pipette by electrical or suction pulse. Keep on monitoring the FRET ratio until the changing ratio reaches a new plateau. Simultaneously monitor access resistance, seal/membrane integrity, and membrane potential during the entire experiment (see Note 14). For a more detailed description of the patch-clamp technique, see ref. [17].

32

Andreas Koschinski and Manuela Zaccolo

Fig. 3 Generation of FRET change curves and calculation of ΔR[%]. Example curves from an experiment performed in CHO cells stably transfected with the cAMP sensor and infused with 100 μM cAMP. (a) Plot of background-subtracted CFP and YFP emission intensity values against time. (b) Plot of the R values calculated from (a) over time. The values used to average the data and to calculate Rt0 and Rtx are indicated by rectangular shaded areas. (c) Calculation of the ratio, (d) calculation of the ratio change. R, Em480 nm/Em535 nm; Rt0, R value calculated before the infusion; Rtx, R value calculated at the plateau of the response; ΔR, change in R value. Arrows indicate the establishment of the whole-cell configuration, providing direct access to the cytosol

Calculate the time course of FRET change by computing the ratio (R) value for each acquisition time point, using Eq. 1. R=

Emission 480 nm Emission 535 nm

(1)

3. Plot the R values against time to obtain the ratio change curve (Fig. 3b). 4. Calculate the induced FRET change, ΔR[%], generated for each concentration of infused cAMP using Eq. 2 (see Note 15):

DR [ % ] =

( Rtx - Rt 0 ) Rt 0

´ 100

(2)

Here, ΔR[%] is the ratio change expressed as percent, Rt0 is the average of 5–10 R values before establishment of the whole-cell configuration, and Rtx is the average of 5–10 R values after the ratio value has stabilized to a new value. Averages should be calculated when the ratio reaches a plateau, and the signal is stable. 5. Plot the calculated ΔR[%] values against the respective infused cAMP concentration, and fit a curve to the points by applying the fit function for a sigmoidal dose–response curve with variable slope (Fig. 4). This can be done using Graph Pad Prism™

“In-cell” Calibration of FRET-Based cAMPsensors

33

Fig. 4 Dose–response curve for the cAMP sensor. Dose–response curve for the cAMP-dependent FRET change of the cAMP sensor. Note that the x-crossing of the curve at 1.021 μM indicates the basal intracellular cAMP concentration. The EC50 value is indicated by the dashed line. The horizontal bar shows the useful sensitivity range (5–95 % of the maximal FRET change), covering a concentration range of 0.66–78.7 μM. The vertical bar indicates the absolute dynamic range of the sensor (29.6 %). All values represent mean ± SEM, n = 5–14 Table 1 Parameters of the dose–response curves for the cAMP sensor R0 [%]

−2.464

Rmax [%]

27.14

n (Hill coefficient)

1.230

EC50 [μM]

7.179

Sensitivity range(5–95 %)

0.66 to 78.7 μM

Dynamic range [%]

29.6

x-crossing [μM]

1.021

Parameters for the curve shown in Fig. 4 as calculated by Graph Pad Prism™. R0 is the minimal FRET change, Rmax the maximal FRET change. EC50 is the concentration of cAMP that generates a half-maximal FRET change, n is the Hill coefficient. Also shown are the useful sensitivity range (5–95 % of Rmax), the absolute dynamic range, and the calculated x-crossing of the curve (see Note 18)

or similar software (see Note 16), which will then automatically calculate the specific parameters EC50, Hill coefficient (n), R0 (the value of FRET change in the absence of cAMP), and Rmax (the value of FRET change at saturating concentration of cAMP) (see Note 17). The values obtained for our example experiments are given in Table 1.

34

Andreas Koschinski and Manuela Zaccolo

3.4 Calculation of cAMP Concentrations Using the Dose–Response Curve

Below, we illustrate how to calculate the concentration of cAMP generated in stably transfected CHO cells when treated with a cAMP-raising stimulus (10 μM forskolin). In our example experiment, the FRET change on application of this stimulus was 21.8 ± 1.2 % (SEM, n = 3).

Fig. 5 Conversion of FRET changes into cAMP concentration changes. (a) Calculation of cAMP concentrations using the graphical approach. Dashed red lines indicate SEM limits. (b) Calculation of cAMP concentrations using the mathematical approach following Eq. 3. Note that because of the logarithmic relationship between dose and response, the errors have to be added and subtracted to the respective mean concentration and calculated separately. This can be avoided by first converting each single response into a concentration and then calculating the mean

“In-cell” Calibration of FRET-Based cAMPsensors

35

The cAMP concentration that would lead to the respective FET change can be determined directly from the curve using the graphical approach illustrated in Fig. 5a, or it can be calculated by applying Eq. 3:

x [mM ] = 10

ö æ æ R - R0 ö -1 ÷÷ ÷ ç log çç max DR - R0 ÷ ç ø è - log EC50 ÷ -ç n ÷ ç ÷ ç ø è

(3)

Here, n is the Hill coefficient, ΔR is the FRET change value for which the cAMP concentration is calculated, R0 the minimal and Rmax the maximal FRET change for the given sensor, and x is the unknown concentration in μmol/L. Because of the nonlinear relationship between cAMP concentration and FRET response, the upper and lower limits of the respective errors (SD or SEM) will differ. In our example, the cAMP concentration was calculated to be 24.6 + 6.9/−4.6 μM (Fig. 5b), indicating that 10 μM forskolin leads to an increase of the intracellular cAMP concentration from the calculated basal value of around 1 μM to 24.6 μM, representing a cAMP increase of nearly 25 times.

4

Notes 1. For our calibrations, we choose the CHO cell line. This cell line is widely used and easily available, and in vitro growth and transfection procedures are simple. In addition, this cell line shows only a very low basal phosphodiesterase (PDE) activity, so that PDEs will not influence the measurements by degrading cAMP. Therefore, reproducibility and comparability of sensor data, even if obtained from different labs, should be granted. Calibration of different sensors in this standard cell line might lead to a pool of comparable calibration data for a multitude of different sensors, which could then be used as a general database for “in cell” sensor characteristics. However, we would like to remark that a dose–response curve of a given sensor must not necessarily be the same in all cell types. Characteristics (shape of the curve, EC50, max. FRET change) might vary in different cell types. Therefore, it is important to verify if the generated reference curve is still valid if a different cell system is used. A rough four-point calibration in the target system (0, max, and two points in the linear part of the curve) should provide sufficient data to verify or, if necessary, fit the respective curve. 2. This buffer is not suitable for cardiomyocytes, as it will depolarize the cells, and the resulting Ca2+ influx will lead to cell death. For cardiomyocytes, the buffer should contain no Ca2+ and an adequate amount of a Ca2+-chelating agent to keep

36

Andreas Koschinski and Manuela Zaccolo

extracellular Ca2+ near zero. In our experience, 3 mM EGTA works well. Do not transfer the cells directly into this buffer, but preincubate for a few minutes in Ca2+-free Tyrode solution to prevent damage to the cells that may result from an initial influx of Ca2+ during the process of exchanging the cell culture medium against the KCl-based buffer. 3. It is not absolutely necessary to create a poly- or monoclonal cell line stably expressing the sensor. If working with transiently transfected cells, skip steps 2–4. The advantage of using stable polyclonal or especially monoclonal cell lines is that there tends to be less cell-to-cell variability. 4. This is one member of a whole family of targeted sensors, based on a newly developed, parental cAMP sensor (manuscript in preparation). With this, sensor FRET increases with increasing cAMP concentrations. However, the protocol is not limited to this sensor, but works essentially in the same way with other sensors, e.g., the EPAC1-camps [11] or a PKAbased sensor [20]. 5. The suitable concentration of antibiotics, as well as the survival time of non-transfected cells under treatment, is cell type and antibiotic specific. It has to be determined for each cell type and plasmid separately. As a rule of thumb, after about 1 week of treatment, no non-transfected cell should have survived. When the number of cells expressing the sensor is at least in the range of 5–10 % (or better above), trypsinize the cells, suspend, and separate in medium, count, and dilute to a density of 1 cell per 200 μl (“limiting dilution” method). Then transfer 200 μl aliquots of this diluted solution into a 96-well cell culture plate. 6. In our experience, transfected cells tend to selectively suppress or minimize the expression of the fluorescent construct. Before you select a clone, make sure that all the cells in the colony are fluorescent. Do not choose extremely bright clones, as the level of expression of the sensor might be too high and may lead to significant cAMP-buffering effects. Also, avoid colonies that show even a small percentage of nonfluorescent cells, as these cells will have a survival advantage and might overgrow the fluorescent cells quickly. Monoclonal cell lines are preferable over polyclonal lines for the same reason. However, if at some time you realize that the number of non- or weakly fluorescent cells increases, you may repeat the clone-selection step. Therefore, it may be helpful to freeze some aliquots of transfected cells directly after the selection step to have a backup. 7. The most convenient and the most direct approach for these measurements is to use the FRET sensor itself as a sensor for intracellular pH. As especially the widely used EYFP fluorophore is quenched significantly by acidification, whereas, e.g., the ECFP fluorophore in the physiological range of pH

“In-cell” Calibration of FRET-Based cAMPsensors

37

shows significantly less quenching [21], a change in pH alters the FRET ratio of the sensor and can be exploited to generate in-cell pH-dependency curves. The principal method to determine the intracellular pH is adapted from Thomas et al. [22] and modified for the use with FRET reporters. However, conventional intracellular pH monitoring, e.g., with BCECF or SNARF-1, can be applied as well and is indeed the method of choice if the FRET sensor only shows a limited pH dependency. The pH determination with these dyes can be performed essentially as described here for FRET reporters, except for an additional loading step where the pH-sensitive dye is introduced into the cells by diffusion. Appropriate excitation and emission wavelengths which will be required to detect the specific dye must be available on the imaging setup. Be aware that you may need to correct for the cross-excitation/bleedthrough of the sensor fluorescence into the dye emissions. For reference regarding BCECF or SNARF-1 measurements, see refs. [18, 19]. 8. It is advisable to standardize growth conditions. Cells grown to different density or different age after seeding may show significant differences in basal cAMP levels or pH. The same applies to added supplements, especially serum. For our experiments, cells were grown at 37 °C, 5 % CO2, and 95 % humidity and measured only on the second day after seeding. Cell density at the time of measurement was 60–80 % confluence. Care should be taken not to change the FBS batch during a series of experiments, as the composition of different lots of serum can vary quite substantially and affect cell behavior. 9. As nigericin is a H+/K+-antiporter, a high pH gradient might built up a K+ gradient which then would prevent a further equilibration of the pH. Therefore, valinomycin, a K+ ionophore, is added to prevent any potential K+ gradient. The final concentrations have to be determined for the specific cell line and range from 0.5 to 10 μM for nigericin and 0.3–10 μM for valinomycin. In our experience, for CHO cells, final concentrations of 3.3 μM (2.5 μg/mL) nigericin and 5 μM valinomycin worked reliably without visibly stressing the cells. 10. This unexpectedly alkaline intracellular pH was not due to the presence of the sensor, the technique used, or the transfection. Control experiments with non-transfected CHO cells as well as CHO cells transfected with different constructs were tested using SNARF-1 as pH-sensitive dye and essentially showed the same results. These findings confirm the importance of measuring intracellular pH, as actual values may differ significantly from the expected “textbook” range of 6.8–7.4. However, even within this range, it is essential to precisely match the intracellular solution to the cytosolic pH, as a difference of only

38

Andreas Koschinski and Manuela Zaccolo

0.1 pH values is sufficient to give an apparent FRET change of about 5 % (compare values at pH 7.2 and 7.3 in Fig. 2b). 11. Be aware that different ionic conditions may also affect the sterical conformation or even directly act on fluorophores of sensors and therefore introduce a bias, e.g., Cl− is well known for its quenching effect on YFP. Other parameters like different ionic strengths and osmolalities of solutions also may affect FRET responses [23, 24] (and our own observations). Try to match ionic composition and osmolality of the intracellular buffer to in vivo intracellular conditions as closely as possible. In case the sensor needs additional factors (ligands, substrate, e.g., ATP, GTP) which might be dialyzed out of the cell, make sure to also include these factors at adequate concentrations in the intracellular buffer. 12. Glass pipettes are pulled from borosilicate glass capillaries (Hilgenberg GmbH, Malsfeld, Germany,) with a Sutter P-2000 puller (Sutter Instrument, Novato, USA). They have a tip diameter of about 2 μm and an initial resistance of 7.4 MOhm when filled with the intracellular solution. 13. It is essential to correct all signals for background fluorescence and drifting baselines. Useful information to correction procedures can be found in [16]. We recommend performing all experiments without any a priori correction, as some acquisition programs, e.g., MetaFluor, will otherwise store only processed data. The original data will be irrevocably lost in this case. The stored original data should then be exported into any adequate spreadsheet program (e.g., MS Excel™ or Open Office Calc™) and analyzed off line. Correction for bleed-through and cross-excitation is not essential [25]. As long as always the same equipment is used, the signals will be altered by an equipment-specific constant factor which will not influence internal quantitative comparability or the calibration itself. However, if you want to generate generally comparable data, you will have to also correct for spectral bleed-through and cross-excitation. Useful information for these correction procedures can also be found in [15]. 14. Electrophysiological parameters vary with experimental conditions and cell types. For our CHO-derived clones and the described buffer combination, seal resistance typically is in the range of 5 GOhm and whole-cell membrane resistance is in the range of 1 GOhm. As increasing access resistance or a leaky seal may introduce artifacts, it is necessary to monitor access and membrane resistance as well as membrane potential during the entire experiment. For example, a membrane potential decreasing toward zero may be an indication of a potential loss of membrane integrity which may also introduce artifacts. In our experiments, the membrane potential of the CHO cells typically was −39.3 ± 2.3 mV (SEM, n = 27).

“In-cell” Calibration of FRET-Based cAMPsensors

39

15. Equation 2 directly expresses the FRET change as percentage. The equation ΔR = (Rtx − Rt0)/Rt0 which expresses the ratio change as a numeric value can be applied as well. 16. Graph Pad Prism™ uses an iterative approach according to the least square fit method to calculate and fit curves. The basis of this approach is the following equation which has been partly adapted to the terms used in this protocol: DR = R0 +

Rmax - R0 1 + 10((

log EC50 - y )´n )

Here, y is the logarithm of the concentration of the ligand (cAMP) and ΔR is the FRET change. ΔR starts at a minimum (R0 or “bottom”) and goes with a sigmoid shape to the maximum (Rmax or “top”). 17. EC50 and Hill coefficient, together with R0 and Rmax, are the essential parameters to characterize dose-dependency curves. EC50 characterizes the concentration of a stimulus needed to generate a half-maximal FRET change and is a measure of the sensitivity of a sensor. The lower this concentration, the higher is the sensitivity of the sensor to its ligand. The Hill coefficient (or Hill slope) (n) correlates with the steepness of the curve. A higher numeric value of the Hill coefficient determines a steeper curve, also providing an estimate of cooperativity effects in the reaction. 18. As most biological sensors possess a sigmoidal concentration– response curve, the resolution of these sensors at very low and very high concentrations typically is poor. Although minimal (ΔR0) and maximal (ΔRmax) FRET changes define the absolute sensitivity range (or “dynamic range”) of the sensor, the useful sensitivity range strongly depends on the shape of the curve and is smaller than this absolute range. Typically, the useful sensitivity range is between 5 % and 95 % of the maximal response of the sensor. However, in case of a very high dynamic range, the signal-to-noise ratio might be improved, and the useful dynamic range might be extended. In our example, the useful sensitivity range (5–95 %) of this cAMP sensor was determined to be 0.66 to 78.7 μmol/L cAMP. This sensitivity range can be calculated either graphically, using the curve shown in Fig 5a, or mathematically, using Eq. 3.

Acknowledgments The work described in this paper was supported by the British Heart Foundation (PG/10/75/28537 and RG/12/3/29423) and the NSF-NIH CRCNS program (NIH R01 AA18060).

40

Andreas Koschinski and Manuela Zaccolo

References 1. Miyawaki A, Llopis J, Heim R et al (1997) Fluorescent indicators for Ca2+ based on green fluorescent proteins and Calmodulin. Nature 388:882–887 2. Adams SR, Harootunian AT, Buechler YJ et al (1991) Fluorescence ratio imaging of cyclic AMP in single cells. Nature 349:694–697 3. Gesellchen F, Stangherlin A, Surdo N et al (2011) Measuring Spatiotemporal Dynamics of Cyclic AMP Signaling in Real-Time Using FRET-Based Biosensors. Methods Mol Biol 746:297–316 4. Terrin A, Di Benedetto G, Pertegato V et al (2006) PGE(1) stimulation of HEK293 cells generates multiple contiguous domains with different [cAMP]: role of compartmentalized phosphodiesterases. J Cell Biol 175:441–451 5. Lefkimmiatis K, Leronni D, Hofer A (2013) The inner and outer compartments of mitochondria are sites of distinct cAMP/PKA signaling dynamics. J Cell Biol 202:453–462 6. Di Benedetto G, Zoccarato A, Lissandron V et al (2008) Protein kinase A type I and type II define distinct intracellular signaling compartments. Circ Res 103:836–844 7. Sin YY, Edwards HV, Li X et al (2011) Disruption of the cyclic AMP phosphodiesterase-4 (PDE4)-HSP20 complex attenuates the beta-agonist induced hypertrophic response in cardiac myocytes. J Mol Cell Cardiol 50: 872–883 8. Herget S, Lohse MJ, Nikolaev VO (2008) Realtime monitoring of phosphodiesterase inhibition in intact cells. Cell Signal 20:1423–1431 9. Förster T (1948) Zwischenmolekulare Energiewanderung und Fluoreszenz. Ann Phys 437: 55–7 10. Stangherlin A, Koschinski A, Terrin A et al (2014) Analysis of compartmentalized cAMP: a method to compare signals from differently targeted FRET reporters. Methods Mol Biol 1071:59–71 11. Nikolaev VO, Bünemann M, Hein L et al (2004) Novel single chain cAMP sensors for receptor-induced signal propagation. J Biol Chem 279:37215–37218 12. Violin J, DiPilato LM, Yildirim N et al (2008) β2-adrenergic receptor signaling and desensitization elucidated by quantitative modeling of real time cAMP dynamics. J Biol Chem 283: 2949–2961 13. Lam A, St-Pierre F, Gong Y et al (2012) Improving FRET dynamic range with bright

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

green and red fluorescent proteins. Nat Methods 9(10):1005–1012 Beavo JA, Bechtel PJ, Krebs EG (1974) Activation of Protein Kinase by Physiological Concentrations of Cyclic AMP. Proc Natl Acad Sci U S A 71(9):3580–3583 Reimann EM, Walsh D, Krebs EG (1971) Purification and Properties of Rabbit Skeletal Muscle cAMP dependent Protein Kinases. J Biol Chem 246:1986–1995 Boerner S, Schwede F, Schlipp A et al (2011) FRET measurements of intracellular cAMP concentrations and cAMP analog permeability in intact cells. Nat Protoc 6:427–438 Hamill OP, Marty A, Neher E et al (1981) Improved Patch-Clamp Techniques for HighResolution Current Recording from Cells and Cell-Free Membrane Patches. Pflugers Arch 391:85–100 Rink TJ, Tsien RY, Pozzan T (1982) Cytoplasmic pH and free Mg2+ in lymphocytes. J Cell Biol 95:189–196 Buckler KJ, Vaughan-Jones RD (1990) Application of a new pH-sensitive fluoroprobe (carboxy-SNARF-1) for intracellular pH measurement in small, isolated cells. Pflugers Arch 417:234–239 Zaccolo M, De Giorgi F, Cho CY et al (2000) A genetically encoded fluorescent indicator for cyclic AMP in living cells. Nat Cell Biol 2:25–29 Llopis, J., McCaffery, J. M., Miyawaki, A., et al (1998) Measurement of cytosolic, mitochondrial, and Golgi pH in single living cells with green fluorescent proteins. Proc Natl Acad Sci 95, 6803–6808d Thomas JA, Buchsbaum RN, Zimniak A et al (1979) Intracellular pH-measurements in Ehrlich Ascites tumor cells utilizing spectroscopic probes generated in situ. Biochemistry 18(11):2210–2218 Wachter RM, Remington SJ (1999) Sensitivity of the yellow variant of green fluorescent protein to halides and nitrate. Curr Biol 9: 628–629 Felber LM, Cloutier SM, Kündig C et al (2004) Evaluation of the CFP-substrate-YFP system for protease studies: advantages and limitations. Biotechniques 36:878–885 Evellin S, Mongillo M, Terrin A et al (2004) Measuring dynamic changes in cAMP using fluorescence resonance energy transfer. Methods Mol Biol 284:259–270

Chapter 4 Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET Reporters Matías Machado and Sergio Pantano Abstract A significant contribution to the research in cAMP signaling has been made by the development of genetically encoded FRET sensors that allow detection of local concentrations of second messengers in living cells. Nowadays, the availability of a number of 3D structures of cyclic nucleotide-binding domains (CNBD) undergoing conformational transitions upon cAMP binding, along with computational tools, can be exploited for the design of novel or improved sensors. In this chapter we will overview some coarse-grained geometrical considerations on fluorescent proteins, CNBD, and linker peptides to draw simple qualitative rules that may aid the design of novel sensors. Finally, we will illustrate how the application of these simple rules can be used to describe the mechanistic basis of cAMP sensors reported in the literature. Key words Fluorescent protein, Allosteric mechanism, CNBD, Rational design, Protein engineering, Coarse grain, SIRAH

1

Introduction Monitoring cascades of events in living cells requires nondestructive methods to follow biological processes. Fluorescent imaging in vivo has become one of the most powerful tools to follow the temporal and spatial localization of a variety of intracellular molecular events. Most of the astonishing developments in this field are based on the use and manipulation of the green fluorescent protein (GFP) from Aequorea victoria [1, 2]. One of the main reasons for the widespread use of GFP is that it only requires molecular oxygen to produce an autocatalytic cyclization reaction generating an intrinsic chromophore from a sequence of three natural amino acids, Ser65-Tyr66-Gly67. This feature, added to strong folding properties, makes this protein perfectly suitable to be used as genetically encoded bioluminescent marker [3]. The fluorescent properties of the chromophore can also be substantially influenced by their local protein microenvironment [4], and tailor-made mutations can shift the excitation and emission spectra of the

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_4, © Springer Science+Business Media New York 2015

41

42

Matías Machado and Sergio Pantano

protein, covering a range of nearly 300 nm, spanning most of the visible spectrum [5]. This feature allows for a vast multiplicity of applications, for instance, to follow simultaneously the localization of proteins fused to different GFP variants. Furthermore, the combined use of appropriate couples of spectral variants of GFP may give rise to the fluorescence resonance energy transfer (FRET) effect, which can be used as a detector for short-range intermolecular interactions. The FRET effect is a non-radiative, distancedependent, energy transfer from a donor to an acceptor chromophore [6]. Being a dipole-dipole interaction, the energy transfer efficiency (E) varies with the spatial separation (R) and relative orientation of the dipole moments of donor and acceptor, often called orientational or kappa-squared factor (κ2), as defined in Eqs. 1 and 2: E=

1

(1)

6

1 + ( R / R0 )

1/ 6

R0 = 9.78 ´ 103 ´ ( Q d k 2n -4 J )

(2)

where the Förster distance (R0) depends on the quantum yield (Qd), the refractive index (n), and the emission spectrum overlap integral (J), which are specific parameters for a given FRET couple. Thus, if a given donor is excited by incident light and an acceptor is in close proximity, part of the donor’s excited state energy is transferred to the acceptor. This results in a measurable reduction of both the donor’s fluorescence emission intensity and its exited state lifetime with an increase in the acceptor’s emission intensity (reviewed in [6]). Because of the strong dependency on the donor-acceptor distance, the FRET efficiency steeply decays as the two chromophores separate, becoming negligible at 10 nm. To exploit such behavior, engineered FRET sensors must include at least two parts. The first, referred to as the detector, must have one or more components that come in close proximity or undergo a conformational transition in response to the molecular event of interest. The second part acts as a transducer, conveying the signal originated by the detector to be measured by a given instrument. In genetically encoded, in vivo applicable, fluorescent sensors, the transducer is most frequently composed by appropriate pairs of spectral variants of GFP fused to the detector, which function as donor and acceptor chromophores for FRET [7]. In this chapter we will provide a series of geometrical considerations, which combined with molecular visualization tools and structural information on proteins of interest may help engineering a protein scaffold to construct a FRET sensor.

Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET…

2

43

Geometric Considerations Given the stringent geometrical conditions required for FRET to happen, it may be useful to briefly review some fundamentals of structural biology from the cAMP sensors perspective. This will provide some characteristic distances involved in the different components of a protein detector. Since we are mainly interested in cAMP sensors, we will limit these considerations to the cAMPbinding domains (referred more generally as cyclic nucleotidebinding domains, CNBD), fluorescent proteins (FP; we will use this term only for GFP spectral variants and not for other proteins possessing fluorescence), and short peptides frequently used as linkers. Whenever a macromolecular structure is solved by any experimental method, its coordinates are deposited in the Protein Data Bank (PDB, http://www.pdb.org). Each structure is univocally identified by a 4-character alphanumeric code, called the PDB ID. Structural information regarding the chemical nature and position of each atom in x, y, z coordinates, authors, related publication, experimental conditions, etc. is written in plain text files usually denoted with the extension .pdb or .ent. These files are easily available from the PDB and self-explanatory. Moreover, any molecular visualization program recognizes this file format to render molecular images. It is of outmost importance to look at the information contained in PDB files before starting work with a structure1. Structural information will guide us to decide suitable scaffolds for detectors, sequence positions for FP fusion, and reasonable length and sequence for peptide insertion linking the functional modules of the sensor. For this task, the reader needs to become familiar with some tools for molecular visualization. Many excellent software packages for most operative systems are freely available on the web including comprehensive tutorials. A very short, and obviously incomplete, list of software and web sites includes: VMD http://www.ks.uiuc.edu/Research/vmd/ PyMol http://www.pymol.org/ Yasara http://www.yasara.org/ Jmol http://jmol.sourceforge.net/ DeepView http://spdbv.vital-it.ch/ Chimera http://www.cgl.ucsf.edu/chimera/ Rasmol http://rasmol.org/

1

Throughout this chapter, text in italics contains practical information that the reader may find particularly useful if they intend to apply this approach.

44

Matías Machado and Sergio Pantano

In this chapter we use VMD. For readers that intend to use this software, the work by Hsin et al. [8] is highly recommended. However, any molecular visualization software able to measure interatomic distances would be adequate. 2.1 Detector Modules: CNBDs

There are at least four molecular architectures which can be taken as representative of CNBDs (Fig. 1). Historically, the first reported structure was the catabolite gene activator protein CAP, which contains a dimeric CNBD at the N-terminus followed by DNAbinding domains at the C-terminus. Then, tandem modules of CNBD were structurally resolved for different isoforms of the regulatory domain of the protein kinase A (PKA). Subsequently, the tetrameric structure of the hyperpolarized cyclic nucleotide (HCN)-gated channel was solved. In this last case, the N-terminal portion of the protein forms a transmembrane channel regulated by a CNBD placed at the C-terminal of the polypeptide chain. Finally, the regulatory domain of the guanine nucleotide exchange factors specific for the small GTP-binding proteins Rap1 and Rap2 (EPAC) was crystallized in the absence and presence of cAMP (see ref. [9] for a recent and complete review). The structure of the

Fig. 1 Representative protein architectures containing CNBDs. The region of the CNBD is highlighted in orange cartoons. If additional CNBDs are present in the structure, they are colored in green. The remaining portion of the protein is shown in light gray, while other existing chains are in black. If present, cAMP is represented in blue spheres. All structures are drawn on the same scale (right down corner). (a) Dimer of CAP subunits bound to a cognate DNA target (PDB ID: 2CGP). (b) Regulatory subunit of the PKA type-IIβ (PKA-RIIβ, PDB ID: 1CX4). (c) HCN (PDB ID: 1Q43). The side view (top) shows the relative orientation of HCN to the cell membrane. The membrane patch was manually placed as a reference and is not part of the X-ray structure. The down view (bottom) displays the fourfold symmetry of the subunits. (d) Structure of EPAC2 in complex with cAMP and RAP1B (PDB ID: 3CF6). (e) Tetrameric holoenzyme complex of PKA (PDB ID: 3TNP), which is formed by two PKA-RIIβ and two catalytic subunits (PKA-C)

Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET…

45

holotetramer of the isoform RIIβ of PKA containing two copies of catalytic and regulatory subunits was recently determined [10], giving rise to a rather complete set of 3D information that can be used to rationalize the structure and allosteric transitions of cAMP sensors. The general organization of a CNBD is an 8-stranded β-barrel, which forms the binding site, flanked by two helices, which are frequently referred as A and B (Fig. 2a). After helix B, all known proteins but EPAC2 present a variable helical segment frequently called helix C, which closes the binding pocket making contacts with the nucleobase of cAMP [11]. The cAMP-bound form of EPAC2, however, presents a β-hairpin, which poses onto the cyclic nucleotide [12]. The global shape of a CNBD has an approximate radius of gyration of 1.5 nm. The radius of gyration can be taken as a spherical approximation of the size of a given macromolecule. The square radius of gyration is the average squared distance of any atom in the polymer from its

Fig. 2 Molecular building blocks of cAMP sensors. The protein backbone is represented in cartoons. Secondary structural elements are classified as α-helix (black), β-strand (gray), and extended or coiled conformations (white). The N- and/or C-termini (NT and CT, respectively) are indicated. All the figures except for the insets are presented at the same scale. (a) CNBD structure taken from PKA-RIIβ (residues 277–412, PDB ID: 1CX4). The semitransparent helical segment (orange) indicates a flexible segment in the absence of ligand. The conformational space accessible to this flexible segment is indicated by a blurred zone. (b) High-resolution structure of GFP (PDB ID: 2WUR) showing the approximate height and width of the molecule and the position of the chromophore (green balls). The inset shows the chromophore (balls and sticks) and the orientation of its dipole moment, which is relevant for the FRET effect. (c) Structure of a 20-alanine peptide in an extended conformation. The inset shows the distance between consecutive Cα atoms. (d) Same sequence as c but in α-helix conformation. The inset denotes the characteristic pitch of one α-helix turn

46

Matías Machado and Sergio Pantano

center of mass. For proteins it is frequently calculated using the position of the Cα carbons. Using VMD, the gyration radius can be calculated using the tk console, which can be opened from the “Extensions” pulldown in the main menu. The command line: measure rgyr [atomselect 0 “name CA”] returns the gyration radius calculated on the Cα carbons measured in Å. In this expression “measure” is a command to compute structural properties, “rgyr” is the argument specifying the property to evaluate, and the specification between brackets is the selection to which the “measure” command will be applied. A selection is defined by using the command “atomselect”, the index of the loaded molecule and a keyword, which in this case stands for all Cα atoms. For more details on command line syntaxes, see [8]. The allosteric mechanism in CNBDs works in such a way that, in the absence of cAMP, the β-barrel increases its plasticity but keeping the global folding, while the C-terminal element opens up acquiring a disordered or movable conformation [13]. Experimental information and molecular dynamics simulations suggest that the allosteric change can vary according to the protein context [13–15]. However, we can think that isolated CNBDs remain in a stiff and compact conformation in the presence of cAMP. The absence of cAMP drives the structure to a more flexible state in which the C element detaches from the body of the protein (Fig. 2a). The global mobility of the C-terminal segment is difficult to characterize since it is simply not visible in experimental structures of cAMP-free CNBDs. To overcome this limitation, we will apply some coarsegrained structural criteria to anticipate this conformational transition, which is exploited by unimolecular cAMP sensors [16, 17]. 2.2 Models Without Experimental Structures

As above outlined, experimental structures of several CNBDs have been resolved, and some of them will be used as practical examples in the forthcoming section. However, the 3D structures on proteins of interest may not be available for other systems. This limitation can be partially surmounted by the use of a large variety of modeling techniques. A full description of these methods goes beyond the scope of this chapter. However, a number of userfriendly online tools are available for nonspecialists. A list of structural bioinformatics tools can be found at the web Portal of the Swiss Institute of Bioinformatics (www.expasy.org) or at http://en.wikipedia.org/wiki/List_of_protein_structure_ prediction_software. We will just mention HHpred (http://toolkit.tuebingen.mpg. de/hhpred). This is an interactive and well-documented server, which integrates identification of remote protein homologues, sequence alignment, secondary structure prediction, and homology modeling. This server allows obtaining 3D models in PDB format in a few steps, requiring only the primary sequence of the protein. 3D models produced by these online tools can be used for rough distance estimations in absence of direct experimental data.

Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET…

47

2.3 Fluorescent Proteins

The wide range of colors contained in the GFP palette is obtained by introducing a limited number of mutations. Therefore, all FP variants present a very well-conserved and highly stable 11-stranded β-barrel structure, which confers them a roughly cylindrical shape with a height of ~5 nm and a diameter of ~3 nm (Fig. 2b). In terms of excluded volume, it can be approximated with a radius of gyration of ~1.5 nm. It is worth to note that, since the detection of FRET effect is limited to a maximum inter-chromophore distance of nearly 10 nm, an intermolecular separation of nearly 3 FP molecular diameters is enough to abolish FRET, underlining the relative accurateness required in the design. This GFP folding contributes to both the formation and stabilization of the different conjugated ring systems that are responsible for the spectral properties of the FP variants [5]. Quantum chemistry calculations indicate that the orientation of the dipole moment is almost the same in all cases [18]. This vector can be simply estimated as the imaginary line that connects the two farthest atoms on the π-conjugated element of the chromophore moiety (Fig. 2b, inset). The wild-type GFP has the N- and C-termini in close proximity both pointing to one of the bases of the cylinder. There are also circular permutations of the FP variants in which the location of the terminals varies around the folding [19]. This potentially allows for designing a multiplicity of conformational arrangements, some of which may constrain the FPs’ orientation and impact the FRET efficiency of the sensor. However, FPs are commonly placed at one or both terminals of the detector. Such disposition confers a large conformational freedom. Since rotations around single peptide bonds involve energy barriers below the thermal energy, room temperature randomizes the reciprocal orientations of donor and acceptors. As a consequence, a spatiotemporal average of κ2 = 2/3 is a frequently used approximation. Despite subtle differences, we can assume that the main determinant for the FRET efficiency is the inter-chromophore distance.

2.4

Frequently, linker peptides varying in length and sequence are introduced between the sensor and the chromophores to improve FRET efficiency. This may be a good strategy to either increase flexibility or help alleviate steric hindrance between the detector and fluorescent proteins. Unfortunately, there are no general rules that can aid in the choice of the optimal peptide length or sequence. However, a necessary starting point to make a rational prediction about the linker’s nature is to consider the dimensions of the elementary building block, i.e., the single amino acid, in comparison with typical sizes of the protein modules used to build up the sensor. The separation between two consecutive Cα carbons in an extended peptide is of 0.38 nm, while the pitch of a single turn of α-helix constituted, on average, by 3.6 amino

Linker Peptides

48

Matías Machado and Sergio Pantano

acids is of 0.54 nm (Fig. 2c, d). In the absence of experimental data, secondary structure predictions can offer rough estimations of the propensity to form a given structure or to remain in a disordered state. There are many servers for secondary structure prediction from primary sequence (see, e.g., those listed at http://expasy.org/proteomics/protein_structure). The results of these predictions have to be considered as propensities and not rigid structural data. For unstructured peptides (random coil-like), a persistence length of nearly 0.4–0.6 nm has been reported [20, 21]. This is very close to the size of a single amino acid in extended configuration, meaning that the linearity of unstructured peptides is lost already after two consecutive amino acids. For arbitrary unstructured peptides, polymer theory predicts the average end-to-end distance to be approximated as the square root of the number of amino acids times the length of the repetitive unit [22]. In the simplest case we can approximate the end-to-end distance of a peptide using the freely joined chain model as: L*√N, where L is the residue length (0.38 nm) and N is the number of residues. Although valid for long chains, this is already a good approximation for polymers longer than 10 residues. Notice, however, that this is only an indicative size for the average end-to-end distance, and completely extended or more compact conformations cannot be ruled out. Clearly, helical segments are expected to behave as rather rigid rods, while unstructured peptides are subjected to large variations, posing more challenges to the prediction of their conformational behavior. For instance, peptides with high helical content containing repeats of the motif – KAAAE – have been employed as rigid spacers between anchoring domains and cAMP sensors [23]. Finally, it is very important to keep in mind that in many reported structures there are flexible regions that are not “visible” from the experimental data. Disordered residues present in the protein but unresolved in the 3D structure are reported in PDB files after the remark “MISSING RESIDUES” and correspond typically to terminals and/or loopy regions. Several amino acids can be “MISSING”. Hence, including these missing residues in a construct will introduce “per se” a flexible linker region between two protein modules. The existence of these flexible regions has to be taken into account in the design of the linker lengths.

3

Practical Rules for Designing a FRET Sensor An accurate in silico design of recombinant protein may require bioinformatics, modeling, and simulation techniques. However, from the above geometrical considerations, we can draw simple rules to aid molecular and cellular biologists to identify FP linking points in proteins of interest to be used as molecular detectors.

Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET…

49

For practical reasons we will split the 10 nm FRET cutoff on each FP and represent it with two spheres of 5 nm radius centered on each chromophore. Under this scheme, the maximum distance from the linking point at which a fluorescent donor will be able to establish FRET with an acceptor can be estimated as an effective average radial distance (reff). This radius can be defined by the sum of the linker length l (calculated as 0.38 nm *√N; see above, Subheading 2.4) plus 1.5 nm (i.e., the gyration radius of a FP; see above, Subheading 2.3) plus 5 nm (half of the FRET cutoff, Fig. 3a). Under the strong hypothesis that the FPs can sample the space with uniform probability, reff will describe a sphere centered on the linking point (Fig. 3b). The relative sizes and shapes of the linker and detector protein will determine excluded volumes. Unfortunately, it is impossible to draw general rules to consider

Fig. 3 Schematic representation of the geometric criteria used to estimate the FRET effect. (a) Description of the distances used to estimate the maximum effective radius (reff) up to which a FP will be able to establish FRET with a second partner. l = linker length, and rc = chromophore radius. (b) Owing to the intrinsic flexibility, the FP is assumed to widely sample the conformational space describing a sphere (dashed line) centered on the linking point (black dot) within reff and limited by steric interactions with the detector protein. (c) For a construct to behave as a good sensor, the FRET changes upon allosteric transition must involve the passage from a low to a high overlap between the spheres defined by each FP. Images are drawn approximately on scale

50

Matías Machado and Sergio Pantano

the excluded volumes for arbitrary molecular systems, and these have to be evaluated in each case (see below). Keeping these rules in mind, the choice of the relative position for a second linking point has to be made considering the allosteric variation of the distance d to produce a measurable FRET change upon a conformational change in the detector (Fig. 3c). For the sake of simplicity, distances between amino acids are usually considered between the Cα carbons of each amino acid, as they are insensible to small conformational changes in the side chains. Intuitively, the probability for FRET to happen is proportional to the intersection of the two spheres defined by the reff of each FP divided by the sum of the volumes of each of the spheres separated by d. This distance d can be measured from the 3D structure of a single polypeptide chain (unimolecular sensor) or a protein-protein complex (bimolecular sensor). Clearly, for a protein-protein complex that dissociates/associates upon cAMP binding, fulfilling this condition is enough for the complex to behave as a FRET detector. However, for a single polypeptide chain, the change in d upon cAMP binding needs to be sizeable enough to be detectable. In summary, a practical protocol for analyzing/designing a FRET sensor for cAMP can be outlined as follows: (a) Determination of suitable CNBD structures. For this task an exhaustive search has to be performed in the PDB. If the protein of interest is not present in this database, theoretical models can be constructed as explained in Subheading 2.2. (b) Identification of suitable FP linking points, evaluating conformational changes upon ligand binding. (c) Estimation of the average linker length and inter-chromophore distance, following the procedure indicated in this section. Finally, we notice that simple analytical geometry can be used to calculate the relative volumes of intersection of both spheres, leading to a quantitative estimation of the FRET probability. However, owing to the coarse nature of our approach, we would advise extreme caution in a quantitative interpretation. Therefore, in the next section we will restrict our analysis to a qualitative level based on visual inspection of the detector’s structure and the regions available to the FPs.

4

Practical Examples In this section we will analyze the mechanistic bases of a couple of popular genetically encoded FRET sensors. At the time in which both architectures were proposed, the availability of 3D experimental data on the protein modules involved in cAMP recognition was rather limited. For this reason the design of cAMP sensors

Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET…

51

relied largely on intuition. Currently, 3D structures of several CNBDs are deposited in the PDB in their apo and bound conformations, helping us to rationalize (a posteriori) the viability of different constructs to work as cAMP sensors. 4.1 Bimolecular cAMP Sensor

The first example is the sensor based on PKA including cyan and yellow fluorescent proteins (CFP and YFP, respectively) tagged at the C-termini of the regulatory and catalytic subunits of the PKA, respectively [24]. The C-terminal of the catalytic subunit of PKA (PKA-C) is linked to a FP by an 11-amino acid peptide of sequence IDYPYDVPDYA. Similarly, the C-terminal of the regulatory subunit (isoform IIβ, PKA-R) is linked to a donor FP by the peptide KLYPYDVPDYA (Fig. 4a). Such linkers were originally selected for detection/purification purposes as they include the immunodominant sequence of the influenza virus (HA) to be used as an epitope tag. In basal conditions both subunits form a complex in which energy can be transferred from the CFP to the YFP, producing the FRET effect. Upon a rise in cAMP concentration, PKA-C and PKA-R detach from each other, hampering the FRET signal. We will apply the protocol outlined in the Subheading 3 to achieve a mechanistic description of this sensor: (a) Determination of suitable CNBD structures The first step is to check for experimental 3D structures of the proteins of interest at the PDB (see above, Subheading 2). Using the keywords “cAMP dependent Protein Kinase II beta” on the search engine, several hits are obtained. In particular, the structure 3TNP, which corresponds to the PKA holotetramer of Mus musculus reported in 2012 [10], can be identified. Since the number of reported structures in the PDB grows continuously, the number of hits will increase with time. Moreover, the PDB is a human-curated database, so, protein codes may change. However, structural information is not removed from the database. Although the entire polypeptide chain of the regulatory subunit was used in the crystallization assay, the conformational flexibility resulting from the absence of cAMP limits the structural determination to the amino acid 393 (see above, Subheading 2.4). Inspection of the information contained in the PDB file indicates that missing residues are comprised between positions 1–13, 122–129, 325–336, and 394–416. However, the only segment relevant for the sensor includes the two tandem CNBDs, located at amino acids 158–277 and 278–416. Notice that, assuming a complete separation of the two subunits upon cAMP binding, there is no need for further characterization of the allosteric transition. Therefore, only this structure will suffice to analyze the sensor.

52

Matías Machado and Sergio Pantano

Fig. 4 Bimolecular cAMP sensor. (a) Schematic representation of the two parts composing the bimolecular sensor. (b) Structural determinants in the FRET sensor in the cAMP-free conformation based on the complex of PKA-RIIβ (gray) and PKA-C (black) taken from the PDB structure 3TNP. The Cα of residues linking to YFP and CFP on which each reff is centered are represented in the structure as yellow and blue dots, respectively. Transparent spheres are drawn as indicated in Subheading 4.1, using the corresponding values of reff for each FP. A GFP molecule is illustrated in the bottom left corner of the figure as a reference for the scale size. (c) Zoom-in image illustrating the local environment surrounding the residue Phe350 in PKA-C (indicated with a white rectangle in b). Salt bridges between Phe350 and Lys92 or Lys111 are indicated as dashed lines

(b) Identification of suitable FP insertion points Visualization of the structure of the PKA-C/PKA-RIIβ complex suggests that the C-terminal of both proteins may be indeed an appropriate point to fuse FP, as the distance between them is about 4 nm (Fig. 4b). It is also worth noticing that Phe350, the last residue of PKA-C, is buried in a hydrophobic pocket, and the COO− moiety is coordinated by a double salt bridge with Lys92 and Lys111 (Fig. 4c). Therefore, the presence of a flexible linker may be needed to lighten possible steric hindrance with the bulky FP. (c) Estimation of the average linker length and inter-chromophore distance The structure of the linkers has not been determined experimentally. However, secondary structure predictions on both linker sequences using, for instance, psipred (http://bioinf.cs.ucl.ac.uk/psipred/) suggest a prevalence of coil conformations. So, we can estimate the linker length for a segment of 11 amino acids at the end of the catalytic subunit to be ~1.3 nm (see above, Subheading 2.4). On the other hand, assuming that

Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET…

53

residues between position 394 and 416 in the regulatory subunit plus 11 residues in the linker are unstructured (34 amino acids in total), we obtain an average linker length of 2.2 nm. In the presence of cAMP, PKA-C and PKA-R subunits are supposed to dissociate [24], abolishing the FRET signal. Hence, it is interesting to predict if in the absence of cAMP this construction can work as a FRET sensor. For this to happen, the cAMP-free state must produce a sizeable FRET effect. A rough geometric consideration of the putative complex (Fig. 4b) suggests that both FPs will be, on spatiotemporal average, in a good reciprocal position to produce FRET in the absence of cAMP. This is evident from the high overlap between both spheres representing the region accessible to both FPs (Fig. 4b). This suggests that the simple procedures explained in Subheading 3 would lead to the prediction of this construct as a viable sensor. To draw a transparent sphere of reff = 7.8 nm, which center is the Cα of Phe350 in PKA-C using VMD, do the following: – Load the PDB ID 3TNP. – Open the tcl/tk console. – Execute the following command lines: set YFP [atomselect 0 "chain C and resid 350 and name CA"] draw material Transparent draw color yellow draw sphere [lindex [$YFP get {x y z}] 0] radius 78 resolution 40 The first line uses the command “set” to assign the value of the selection to the variable arbitrarily called “YFP”. The second and third lines use the command “draw” to set the display settings of the object to be created. The fourth line generates the sphere. Within this line, the second argument specifies the center of the sphere; in this case we use the selection “YFP” and the command “get” to obtain the coordinates of the Cα atom. Then, the tcl command “lindex” is used to format the data in a proper way. Notice that in VMD distances are measured in Å. The argument “resolution” impacts in the definition of the rendering. 4.2 Unimolecular cAMP Sensor

As a second example, let us consider the set of unimolecular cAMP sensors reported by Nikolaev et al. in 2004 [17]. In this case the CFP-YFP couple was fused at the N- and C-termini of a single CNBD of PKA or EPAC. These sensors exhibited a FRET signal that decreased upon increasing concentrations of cAMP. For the sake of simplicity and consistency with the previous paragraph, we will consider constructs based on the CNBD of PKA.

54

Matías Machado and Sergio Pantano

Fig. 5 Unimolecular cAMP sensors. (a) Schematic composition of individual fusion proteins based on PKA-RIIβ. (b) Structural determinants in the FRET sensor PKA-camps at the cAMP-bound conformation. Cartoon representation of residues 264–403 from PKA-RIIβ in complex with cAMP (green spheres) (PDB ID: 1CX4). The Cα of residues linking to YFP and CFP are represented in the structure as yellow and blue dots, respectively. Transparent spheres are drawn according to the estimated reff for each FP and centered at the Cα atom of the corresponding linking residue. A GFP molecule is shown in the bottom left corner of the figure as a reference of the size scale. (c) Same as b but considering residues 264–393 from PKA-RIIβ at the cAMP-free conformation (PDB ID: 3TNP)

Originally three constructs were presented spanning different segments on PKA-RIIβ (Fig. 5a). Let’s first analyze the best performing construct called PKA-camps to rationalize the allosteric mechanism of this sensor. This construct uses amino acids 264–403 of PKA-RIIβ as a detector module linked to the FP couple. (a) Determination of suitable CNBD structures Being a unimolecular sensor, we need to look for structures representative of the free and bound conformations in order to foresee the changes in the reciprocal position of the FPs upon cAMP binding to the detector. In the previous example, we identified the structure 3TNP as representative of the free state. When conducting the same search in the PDB, we also find the structure 1CX4 [25]. This corresponds to a truncated version of PKA-RIIβ from Rattus norvegicus spanning amino acids 130–412 with two cAMP molecules present at binding sites A and B. Within the coarseness of our approach and considering the high sequence identity between mammalian PKA proteins,

Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET…

55

we can overlook the fact that structures 3TNP and 1CX4 correspond to proteins of different species. As a practical rule, homologous proteins with an identity degree over ~50 % can be considered disregarding a possible conformational influence. Missing residues are reported in this structure between amino acids 112–129 (N-terminal), 326–333 (4–5 loop), and 413–416 (C-terminal). However, only the C-terminal region is relevant for our purposes. (b) Identification of suitable FP insertion points Being a unimolecular sensor, we have to compare the differences in distances between FP insertion points in the free and bound state. Linking points are placed at the beginning and end of the second CNBD of PKA-RIIβ (representing the N- and C-termini of the detector). In the structure 1CX4 representing the bound state, the Cα carbons of amino acids 264 and 403 are separated by ~4 nm (Fig. 5b). In contrast, in the structure 3TNP, the absence of cAMP results in a flexible configuration of the C-terminal helix, and the 3D information is lost after Asp393. So we will consider this last structurally determined residue as linking point, and the missing residues will be considered as an unstructured linker peptide. Using this criterion, the distance d between both linking points now shortens to 2.8 nm in the cAMP-bound state. Although both distances are clearly within the FRET range, it is still important to evaluate possible changes in reff originated by the presence of flexible segments in the absence of cAMP. (c) Estimation of the average linker length and inter-chromophore distance In the bound state (structure 1CX4), this construct presents no flexible linkers. We can, hence, estimate that each chromophore will be located at ~1.5 nm from their respective linking points on the detector, and the distance d measured between the two linking points is ~4 nm. The absence of linkers suggests that both chromophores will remain at a rather stiff position with a separation of ~7 nm in the cAMP-bound state. Visual inspection suggests that although there is a clear overlap between FRET productive conformations, most of the overlapping space is not available because of the excluded volume imposed by the detector itself (Fig. 5b). To estimate the average inter-chromophore distance in the absence of cAMP (structure 3TNP), we can consider that the CFP is fused at residue 393 linked by a flexible 10-residue-long peptide with an average length of ~1.3 nm, and the distance between both fusion points now shortens to ~2.8 nm. Notice that assuming a helical but delocalized conformation for the C-helix would only slightly change the final conclusion, as the distance measured on the 1CX4 structure between residues 391 and 403 is ~2 nm.

56

Matías Machado and Sergio Pantano

There is a small change in the relative distances between the FP fusion points (~1 nm), which increases the FRET signal, as expected. More important, however, is the fact that the increased flexibility in the absence of cAMP results in a larger reff for the CFP linked at the C-terminal of the detector, which may overcome the steric hindrance present in the bound state (Fig. 5c). Thus, the combined effect of a shorter linking point distance (d), with a longer linker, allows for a higher overlap between the conformational spaces available to both FPs and explains the increase of FRET measured in the absence of cAMP. The second construct in Fig. 5a using as a detector amino acids 264–416 in the CNBD plus the HA linker peptide shows a reduced change in FRET upon cAMP binding [17]. Following the same reasoning, in the cAMP-bound structure 1CX4, the last amino acid determined is Val412, which is at 3.6 nm from Met264 and is linked to the FP by a 15-residue-long peptide (average length ~1.5 nm). In the cAMP-free form, the conformation of the detector is similar to the previous construct, but a longer unstructured linker is present. It is easy to verify (not shown) that the shortdistance differences in the structured regions in combination with longer flexible linkers result in less relevant changes in the FRET productive overlap region, which translate in less marked response of the sensors to cAMP binding. Moreover, the reader can confirm that same conclusions can be drawn for the construct spanning amino acids 255–416 + HA, as the region comprised by amino acids 255–264 is compact and well folded in the X-ray structures, not altering significantly the relevant distances that determine the FRET change upon cAMP binding. Finally, we turn our attention to the construct using as a detector amino acids 103–416 plus the HA linker peptide, which shows no change in FRET upon cAMP binding. First, we notice that amino acid positions below 130 in PKA-RIIβ are visible in the cAMP-free structure (3TNP) because it is partially bound to the catalytic subunit but are expected to be flexible in the free protein [26]. With this in mind we could repeat the reasoning as for the previous examples. However, it is worth to notice that the distance d between the FP fusion points is nearly 4.5 nm in both the cAMPbound and cAMP-free conformations (compare structures 1CX4 and 3TNP). Hence we can anticipate that in this construction the inter-chromophore distances are not sensible to the allosteric change, resulting in a nonfunctional sensor.

5

Conclusions Along this chapter we showed how combining very simple rules from structural biology with molecular visualization and online tools can achieve a rough molecular-level description of commonly

Structure-Based, In Silico Approaches for the Development of Novel cAMP FRET…

57

used cAMP reporters. More refined predictions require advanced computational techniques in molecular visualization, comparison, editing, and simulations. An example of sensor optimization by means of molecular dynamics simulation of the linker peptides can be found in Lissandron et al. [27]. The approximate nature of the approach presented here cannot, obviously, guarantee an optimal design for other molecular system. However, it is expected that application of these criteria could be useful in early stages of the design of new FRET sensors for cAMP, helping to rule out constructs that are not likely to be functional because of rather obvious structural reasons, thus saving time and effort. It is expected that the increasing number of experimental structures along with faster simulation techniques and accessibility of user-friendly online tools will boost the design of more sensible and versatile cAMP sensors in the near future.

Acknowledgments This work was partially funded by FOCEM (MERCOSUR Structural Convergence Fund), COF 03/11, and Intramural Transversal Program 2013, Institut Pasteur de Montevideo. M.M and S.P are members of the SNI, ANII, Uruguay. References 1. Tsien RY (1998) The green fluorescent protein. Annu Rev Biochem 67:509–544 2. Zimmer M (2002) Green fluorescent protein (GFP): applications, structure, and related photophysical behavior. Chem Rev 102:759–781 3. Newman RH, Fosbrink MD, Zhang J (2011) Genetically encodable fluorescent biosensors for tracking signaling dynamics in living cells. Chem Rev 111:3614–3666 4. Remington SJ (2006) Fluorescent proteins: maturation, photochemistry and photophysics. Curr Opin Struct Biol 16:714–721 5. Day RN, Davidson MW (2009) The fluorescent protein palette: tools for cellular imaging. Chem Soc Rev 38:2887–2921 6. Lakowicz JR (1999) Energy transfer. Fluorescence spectroscopy. Kluwer Academic/ Plenum, New York, pp 368–391 7. Sipieter F, Vandame P, Spriet C et al (2013) From FRET imaging to practical methodology for kinase activity sensing in living cells. Prog Mol Biol Transl Sci 113:145–216 8. Hsin J, Arkhipov A, Yin Y et al. (2008) Using VMD: an introductory tutorial. Curr Protoc Bioinformat Chapter 5, Unit 5.7

9. Taylor SS, Ilouz R, Zhang P et al (2012) Assembly of allosteric macromolecular switches: lessons from PKA. Nat Rev Mol Cell Biol 13:646–658 10. Zhang P, Smith-Nguyen EV, Keshwani MM et al (2012) Structure and allostery of the PKA RIIbeta tetrameric holoenzyme. Science 335:712–716 11. Berman HM, Ten Eyck LF, Goodsell DS et al (2005) The cAMP binding domain: an ancient signaling module. Proc Natl Acad Sci U S A 102:45–50 12. Rehmann H, Arias-Palomo E, Hadders MA et al (2008) Structure of Epac2 in complex with a cyclic AMP analogue and RAP1B. Nature 455:124–127 13. Berrera M, Pantano S, Carloni P (2007) Catabolite activator protein in aqueous solution: a molecular simulation study. J Phys Chem B 111:1496–1501 14. Berrera M, Pantano S, Carloni P (2006) cAMP Modulation of the cytoplasmic domain in the HCN2 channel investigated by molecular simulations. Biophys J 90:3428–3433 15. Pantano S, Zaccolo M, Carloni P (2005) Molecular basis of the allosteric mechanism of

58

16.

17.

18.

19.

20.

21.

Matías Machado and Sergio Pantano cAMP in the regulatory PKA subunit. FEBS Lett 579:2679–2685 Pantano S (2008) In silico description of fluorescent probes in vivo. J Mol Graph Model 27:563–567 Nikolaev VO, Bunemann M, Hein L et al (2004) Novel single chain cAMP sensors for receptor-induced signal propagation. J Biol Chem 279:37215–37218 Ansbacher T, Srivastava HK, Stein T et al (2012) Calculation of transition dipole moment in fluorescent proteins–towards efficient energy transfer. Phys Chem Chem Phys 14: 4109–4117 Topell S, Glockshuber R (2002) Circular permutation of the green fluorescent protein. Methods Mol Biol 183:31–48 Rief M, Oesterhelt F, Heymann B et al (1997) Single Molecule Force Spectroscopy on Polysaccharides by Atomic Force Microscopy. Science 275:1295–1297 Carrion-Vazquez M, Oberhauser AF, Fowler SB et al (1999) Mechanical and chemical unfolding of a single protein: a comparison. Proc Natl Acad Sci U S A 96:3694–3699

22. Flory PJ (1975) Spatial configuration of macromolecular chains. Science 188: 1268–1276 23. Di BG, Zoccarato A, Lissandron V et al (2008) Protein kinase A type I and type II define distinct intracellular signaling compartments. Circ Res 103:836–844 24. Zaccolo M, De GF, Cho CY et al (2000) A genetically encoded, fluorescent indicator for cyclic AMP in living cells. Nat Cell Biol 2:25–29 25. Diller TC, Madhusudan, Xuong NH et al. (2001) Molecular basis for regulatory subunit diversity in cAMP-dependent protein kinase: crystal structure of the type II beta regulatory subunit. Structure 9, 73–82 26. Smith FD, Reichow SL, Esseltine JL et al (2013) Intrinsic disorder within an AKAPprotein kinase A complex guides local substrate phosphorylation. Elife 2:e01319 27. Lissandron V, Terrin A, Collini M et al (2005) Improvement of a FRET-based indicator for cAMP by linker design and stabilization of donor-acceptor interaction. J Mol Biol 354: 546–555

Chapter 5 Automated Image Analysis of FRET Signals for Subcellular cAMP Quantification Silas J. Leavesley, Arie Nakhmani, Yi Gao, and Thomas C. Rich Abstract A variety of FRET probes have been developed to examine cAMP localization and dynamics in single cells. These probes offer a readily accessible approach to measure localized cAMP signals. However, given the low signal-to-noise ratio of most FRET probes and the dynamic nature of the intracellular environment, there have been marked limitations in the ability to use FRET probes to study localized signaling events within the same cell. Here, we outline a methodology to dissect kinetics of cAMP-mediated FRET signals in single cells using automated image analysis approaches. We additionally extend these approaches to the analysis of subcellular regions. These approaches offer an unique opportunity to assess localized cAMP kinetics in an unbiased, quantitative fashion. Key words Förster resonance energy transfer, Image cytometry, Microscopy, Cyclic nucleotide

1

Introduction Förster resonance energy transfer (FRET) is a process in which energy is nonradiatively transferred from one fluorescence molecule (donor fluorophore) to a second fluorescence molecule (acceptor fluorophore) [1]. Because the efficiency at which energy is transferred is highly dependent on the intermolecular spacing [2], FRET has played a key role in allowing the study of intracellular biological events, such as protein–protein interactions, protein folding, and enzyme–substrate kinetics [3]. By utilizing appropriate fusion protein reporters (such as Epac-based cAMP probes), FRET assays can be employed to measure bulk and intracellular second messenger concentrations—the most widespread of which have been the measurement of cAMP [4–6] and cGMP [7, 8]. Coincident with the development of new FRET reporters, there have been a large number of fluorescence microscopy approaches developed to measure FRET efficiency, including the use of 1, 2, or 3 fluorescence filter cubes, fluorescence lifetime, spectral imaging, and acceptor photobleaching. Each of these approaches has

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_5, © Springer Science+Business Media New York 2015

59

60

Silas J. Leavesley et al.

associated advantages and limitations. However, when performed appropriately, the final outcome of any of these fluorescence microscopy-based FRET assays should be the ability to sample the FRET efficiency for any pixel or region in a fluorescence microscopy image. In theory, FRET efficiency image data should yield a wealth of information describing localized cAMP concentrations. In practice, partly due to signal-to-noise limitations, many FRETbased cAMP assays have simply focused on measuring a bulk cAMP response for the entire population (field of view or large region of interest) or for a single (representative) cell of interest. However, manual selection of regions or cells of interests can introduce two key limitations: use of a small number of cells to represent the population response and introduction of operator bias in the selection of the region. The purpose of this chapter is to present a methodology for automated image analysis to allow selection of many or all cells in a field of view, as well as subcellular regions, to allow more accurate estimations of FRET efficiencies and cAMP concentrations at a cellular and subcellular level, without operator bias. In addition, the automated image analysis approaches, once implemented in a script or pipeline format, significantly reduce analysis time of large fluorescence microscopy image datasets, as operator interaction is greatly reduced.

2

Materials

2.1 FRET-Based cAMP Measurement Components

1. Fluorescence widefield or confocal microscope configured for FRET measurements: We have found that an inverted widefield microscope (such as TE2000, Nikon Instruments, or similar) or an inverted spectral confocal microscope (such as A1R, Nikon Instruments, or similar) equipped with spectral imaging capabilities provides the lowest coefficient of variation (CV) when conducting FRET measurements [9] and is preferable to approaches utilizing one or several fluorescence filter cubes (see Note 1). (a) Light path. The light path consists of a broadband excitation source (Xe arc lamp), an excitation filter (405 nm with 20 nm bandwidth is appropriate for CFP-YFP reporters), dichroic beamsplitter (450 nm long pass), objective, emission filter (450 nm long pass), tunable filter for hyperspectral imaging, and CCD camera. (b) A high numerical aperture, achromatically corrected, objective. Although air and oil immersion can be used, a water-immersion objective (40×, 60×, or similar) will provide the best match of refractive index of the immersion fluid and cellular buffer, for live-cell studies.

Automated Image Analysis of FRET Signals for Subcellular cAMP Quantification

61

(c) A tunable filter for hyperspectral imaging, such as HSi-300, Chromodynamics, or VersaChrome filter-based system, Semrock, with appropriate wavelength range for the FRET reporter being used (450–700 nm for CFP-YFP reporters). 2. Extracellular buffer solution: 145 nM NaCl, 4 mM KCl, 20 mM HEPES, 10 mM D-Glucose, 1 mM MgCl2, 1 mM CaCl2, pH 7.3. 3. Cells expressing the donor alone, acceptor alone, or the donor– acceptor FRET-based cAMP sensor prepared on glass coverslips (see Note 2). Cell culture and infection protocols have been previously described [9, 10]. Note that the multiplicity of infection (MOI) is likely dependent on the cell line used. 4. Cellular and membrane labels: A nuclear label, such as Hoechst (Hoechst 33342, Life Technologies), should be used to enable automated image analysis algorithms to automatically locate nuclei. Further labels may be added for identifying other cellular structures (membrane, mitochondria, etc.) or for simultaneous study of other signaling events (see Note 3). 5. Calibration tools: A broadband NIST-traceable light source is used to calibrate the system to flat spectral response (LS-1-CAL, Ocean Optics). A multi-ion discharge lamp (MIDL) can be used to verify spectral resolution (MIDL, Lightform, Inc.). 2.2 Image Analysis Components

1. Computer workstation, appropriately equipped for image analysis (see Note 4 for general hardware recommendations). 2. Software for spectral image analysis: Traditionally, analysis of spectral/hyperspectral image data has been performed using a least-squares linear unmixing algorithm with nonnegative constraints [9, 11]. It should be noted, however, that alternatives for spectral image analysis are available (see Note 5). While several software programs contain linear unmixing algorithms, such as ENVI (Exelis Visual Information Solutions) and image analysis suites from the major microscope manufacturers (Nikon, Olympus, Zeiss), we typically employ custom linear unmixing scripts using the Matlab (MathWorks) programming environment. An example linear unmixing algorithm, implemented in MATLAB, has been previously described for analysis of spectral image data of FRET-based cGMP probes that is also applicable to analysis of FRET-based cAMP probes [10]. 3. Software for automated cell segmentation and quantification: Several software programs are available for automated image analysis and quantification, also called image cytometry (see Note 6). We have used Cell Profiler (The Broad Institute— freely available) and custom scripts written in the MATLAB programming environment.

62

3

Silas J. Leavesley et al.

Methods The automated image analysis process consists of the following general steps: image acquisition, image segmentation, and feature extraction (Subheadings 3.1–3.3). There are many approaches available for performing automated image segmentation and feature extraction, and many software programs are available—from specialized programs such as Cell Profiler (The Broad Institute [12]), to general image analysis packages such as MATLAB (The MathWorks). Hence, in addition to the description of steps provided below, we have provided an example spectral image dataset and image analysis pipeline (http://www.southalabama.edu/ centers/bioimaging/Resources.html) that can be run using Cell Profiler, an open-source software package for automated image analysis. The example image analysis pipeline performs the basic task of quantifying single-cell FRET response for each cell within an image—the topic of this book chapter. However, the pipeline can easily be edited to suit a range of additional needs (e.g., allowing automated measurement of cellular geometry, intensities of additional fluorescent labels, etc.). Where appropriate, in Subheading 3, we have referenced the corresponding steps in the example Cell Profiler pipeline with “*CP*.” It should be noted that these same steps could be coded in alternative software packages (such as MATLAB or ImageJ), but are shown using Cell Profiler for ease of use for a broad audience. Additional steps can also be added to the image analysis process to allow identification of subcellular regions and quantification of subcellular FRET efficiencies (Subheadings 3.4 and 3.5). A diagram of the image processing steps is shown in Fig. 1.

3.1 Image Acquisition: Spectral Imaging Fluorescence Microscopy of FRETBased cAMP Reporters

Spectral/hyperspectral imaging of FRET-based cGMP reporters has been previously described in Methods in Molecular Biology [10]. The equipment configuration and image acquisition methods are identical for FRET-based cAMP reporters if the same donor and acceptor are used in the reporter construct (donor— CFP, acceptor—YFP). A brief summary of the image acquisition methodology is given below. 1. The fluorescence excitation should be configured so as to preferentially excite the donor (405 nm for CFP). 2. The spectral imaging fluorescence detection should be configured to acquire fluorescence over a spectral range that includes both the donor and acceptor emission (450–650 nm for CFP and YFP). 3. Prepare a background/blank sample using a blank coverslip mounted in a coverslip holder and extracellular buffer. 4. Correct the fluorescence emission to a flat spectral response using a NIST-traceable light source [9, 13].

Automated Image Analysis of FRET Signals for Subcellular cAMP Quantification

63

Hyperspectral Image

Hoechst

CFP

YFP

CFP+YFP

All Nuclei

FRET Estimation

Expressing Cell Regions

Expressing Nuclei

Expressing Cells

Nuclear Boundary

Cell Boundary

Perinuclear ROIs

Whole Cell ROIs

NearMembrane ROIs

Perinuclear FRET

Whole Cell FRET

NearMembrane FRET

Filtering (Optional)

Fig. 1 Summary of steps in the methodology for automated single-cell cAMP measurements using a FRET reporter: image acquisition, image segmentation (region selection), and feature extraction, as well as secondary subcellular region selection and feature extraction. Each main set of steps is indicated by blue dashed lines and a label that indicates the corresponding methods subsection

(a) Acquire a spectral image stack of the NIST-traceable lamp, as projected through the coverslip. (b) Acquire a spectral image stack with the NIST-traceable lamp turned off. This will be used as the background spectrum. (c) Calculate the correction coefficient as: CC =

ILamp IMeasured - IBackground

(1)

where ILamp is the known (NIST-traceable) lamp spectrum, IMeasured is the measured spectrum, and IBackground is the background spectrum. 5. The spectral (wavelength) resolution can be verified using a multi-ion discharge lamp [13, 14]. 6. Acquire spectral image data for single-labeled samples to use as spectral controls. A separate spectral image scan is needed for each label used. The scan for each label should be performed using identical excitation wavelength as is used for FRET studies (e.g., 405 nm). In some cases, the integration time may have to be increased to allow sampling of the pure label spectrum with minimal noise artifact (e.g., when exciting YFP

64

Silas J. Leavesley et al.

directly with 405 nm excitation, a longer integration time will be needed than is required when exciting CFP with 405 nm excitation and measuring YFP emission that results from energy transfer). As an example, for a relatively simple preparation utilizing the cAMP FRET reporter and Hoechst 33342, 3 scans are needed (one of the donor only, one of the acceptor only, and one for Hoechst). 7. Perform measurements on intact cells with the FRET-based cAMP reporter. (a) Place a coverslip containing adherent cells expressing the cAMP FRET reporter (and additional fluorescent labels) on the microscope. (b) Select a field of view with appropriately expressing cells. We recommend selecting a field of view with moderately expressing cells (see Note 7). (c) Begin the time-lapse study. Acquire images at several baseline time points before adding reagents. (d) Add reagents to trigger changes in cAMP levels (see Note 8). (e) Acquire further time-lapse images for a period of time longer than the expected cAMP response time (e.g., until the system has reached steady state; see Note 9). 8. Measure FRET response at minimal and maximal cAMP levels. (a) Minimal cAMP levels can often be approximated by basal conditions. We recommend measuring 30 s–2 min of basal activity (usually 5–20 time points) to allow calculation of the standard error of the FRET measurements. Use of an adenylyl cyclase inhibitor, such as 100 μM MDL-12,330A hydrochloride, has also been described for ensuring minimal cAMP levels [15]. (b) Maximal cAMP levels may be approximated by addition of 500 μM IBMX (PDE inhibitor) + 50 μM Forskolin (adenylyl cyclase activator). 9. Correct images to a flat spectral response. This is achieved by subtracting the background spectrum from the measured image data and then multiplying by the correction coefficient (Eq. 1). This relatively simple correction can be performed using a simple MATLAB script or ImageJ routine. 10. Linearly unmix the spectral image data (see Note 5 for alternative algorithms for spectral analysis). Unmixing the spectral images from each time point will result in an abundance image for each fluorescent label at each time point. Abundance refers to the amount of a fluorophore that is detected after spectral unmixing. The abundance images at each time point are used as the inputs to the automated image analysis procedure (described below).

Automated Image Analysis of FRET Signals for Subcellular cAMP Quantification

3.2 Image Segmentation: Automated Whole-Cell cAMP Measurements

65

1. Save the signal from each fluorescent label (the linearly unmixed signal or abundance) as a separate image in tif format (tagged image file format). Examples of linearly unmixed images are given in the example data file (http://www.southalabama.edu/centers/bioimaging/Resources.html). 2. Locate nuclei by thresholding the nuclear image. *CP* In Cell Profiler, use the “IdentifyPrimaryObjects” module to perform nuclear thresholding. (a) Select the thresholding level using a mixture of Gaussians (MoG) algorithm with two intensity populations (foreground and background). (b) Apply the threshold to generate a binary image. This creates a mask which allows geometric analysis of the nuclei and exclusion of inappropriate (incorrect) objects. These binary images are used further for nuclei labeling and matching to the cell boundary. (c) Exclude nuclei that are too large, too small, or touching the border of the image. (d) Assign a label to each nucleus. 3. Locate expressing cell regions by thresholding the donor and acceptor image. *CP* Use the “IdentifyPrimaryObjects” module. (a) Create a new image that is the sum of the donor and the acceptor images (the linearly unmixed donor and acceptor signals). Save this image in tif format. 4. Filter whole-cell regions so that each cell contains only one nuclei and so that each nuclei has a corresponding cell. *CP* Use the “MaskObjects” module so that only nuclei within expressing cell regions are considered for analysis. Then use the “IdentifySecondaryObjects” module to identify a single cell corresponding to each masked nucleus. (a) Additional filtering can be performed to remove abnormally shaped regions. For example, regions with a diameter of less than a typical cell may be excluded as fragments. In addition, regions with very high intensity of both CFP and YFP correspond to cells with very high FRET reporter expression, which are usually nonviable and may be excluded from analysis. In addition, regions that have a very high circularity are often nonviable or detached cells. These regions can be automatically excluded from analysis by thresholding based on cell radius or circularity. *CP* Use the “MeasureObjectSizeShape” module to extract geometric measurements from each expressing cell. Then use the “FilterObjects” module to identify only cells within the desired cell radius range (e.g., identify cells with a mean radius between 3 and 15 pixels). 5. Label whole-cell regions with unique identification tags.

66

Silas J. Leavesley et al.

3.3 Feature Extraction: Quantitative Measurement of Whole-Cell Data

1. Calculate the FRET efficiency for each region. (a) Measure the pixel-averaged donor intensity (abundance from linear unmixing, ADonor) and acceptor intensity (AAcceptor) for each region. *CP* Use the “MeasureObjectIntensity” module to measure the donor and acceptor intensity within each cell. These measurements can then be exported to Excel (Microsoft Corporation), using the “ExportToSpreadsheet” module. The following equation is then entered into excel. (b) Calculate the FRET efficiency [9] as: ESpectral =

A Acceptor A Donor + A Acceptor

(2)

2. Relate FRET efficiency to cAMP concentrations using information gathered in step 9 of Subheading 3.1. Spectrofluorimetry data may also be used to relate FRET efficiency and cAMP concentration (see Note 10). 3. Extract other quantitative whole-cell or cytoplasmic measurements as necessary for the assay (see Note 11). 3.4 Automated Subcellular cAMP Measurements

1. Perform the steps in Subheadings 3.2 and 3.3 for whole-cell cAMP measurements before quantifying subcellular features. 2. Identify the cytosolic space. *CP* use the “IdentifyTertiaryObjects” module to define the region between the nucleus and cell border as the cytoplasm on a per cell basis. 3. Locate subcellular regions by thresholding each organelle image. (a) Select the thresholding level or threshold calculation algorithm as appropriate for each organelle/region to be identified. (b) Apply the threshold to generate a binary image. This image is used further for extraction of statistical properties of the intensity for the selected subcellular regions. (c) Exclude regions that are too large, too small, or touching the border of the image. This process is done automatically by the mathematical morphology tools of closing and opening applied to binary masks of the subcellular regions. 4. Assign each subcellular region as belonging to only one wholecell region. (a) Assign subcellular regions completely contained within one whole-cell region to the respective cell ID. (b) Subcellular regions overlapping more than one whole-cell region can be either split or removed from consideration.

Automated Image Analysis of FRET Signals for Subcellular cAMP Quantification

67

Alternatively, the boundaries of whole-cell regions can be adjusted to better contain appropriate subcellular regions. (c) Discard subcellular regions touching the edge of the image. 3.5 Feature Extraction: Quantitative Subcellular Measurements

1. Calculate the FRET efficiency for each subcellular region. (a) Measure the pixel-averaged donor intensity (abundance from linear unmixing, ADonor) and acceptor intensity (AAcceptor) for each region. (b) Calculate the FRET efficiency for each subcellular region as described in step 1b of Subheading 3.3. 2. Relate FRET efficiency to cAMP concentrations using information gathered in step 9 of Subheading 3.1. Spectrofluorimetry data may also be used to relate FRET efficiency and cAMP concentration (see Note 10). 3. Extract other quantitative measurements of subcellular regions as necessary for the assay (see Note 11).

4

Notes 1. Several methods have been developed for quantifying FRET in fluorescence microscopy experiments. These include the use of multiple fluorescence filter cubes [16], fluorescence lifetime [17], acceptor photobleaching [18], and spectral/hyperspectral imaging [9, 10, 19]. The effectiveness of the various FRET measurement approaches has been compared in several publications [9, 16, 20]. We have found that spectral imaging allows FRET measurements with reduced coefficients of variation, when compared to multiple fluorescence filter cube approaches [9]. In addition, spectral imaging FRET measurement strategies can facilitate measurement of other fluorescent labels, such as nuclear, organellar, or signaling labels, to give spatial or functional context to FRET-based cAMP measurements. 2. We have previously used a CFP-Epac-YFP FRET reporter that was originally described by Nikolaev and colleagues [6]. We have found this construct to provide a FRET efficiency of approximately 45 % at basal conditions and 32 % at saturating cAMP concentrations (50 μM), using HEK 293 cells [9]. More recently, FRET-based cAMP probes have been developed using alternative constructs that have shown potential for higher larger changes in FRET efficiency upon cAMP binding [21]. Because these probes have different fluorescent protein mutants for the donor and acceptor, care should be taken to appropriately configure the detector settings of the microscope to ensure adequate measurement of the fluorescence emission.

68

Silas J. Leavesley et al.

3. A variety of labels are available for identification of different subcellular structures. Upon identifying these structures using automated image analysis algorithms, it is possible to quantify the changes in FRET within or adjacent to these structures. 4. Computer workstation hardware specifications are continually being improved. The following recommendations represent minimum requirements for capabilities for quantitative image processing: dual core processor, 2 GHz processor speed, 4 GB RAM memory, and 20 GB available hard drive space. 5. Although nonnegatively constrained linear unmixing is the standard approach for analysis of spectral microscopy data, alternative approaches are available. Of special interest may be approaches that account for the noise characteristics of image data [22]. 6. Alternative, freely available software programs for automated analysis of cellular microscopy image data include: ImageJ (National Institutes of Health), FIJI [23] (a modified distribution of ImageJ), and Ilastik [24]. 7. It is important to select cells with a FRET reporter expression level that is suited for the measurement being made. For example, the FRET reporter level must be sufficient to allow accurate detection of changes in FRET efficiency. Likewise, cells expressing excessive amounts of FRET reporter may substantively buffer free cAMP levels [25]. 8. Agonist concentration dependence should be examined. Typical agonists concentration ranges include 1–100 μM Forskolin, 10–1,000 nM Isoproterenol, and 10–1,000 nM Prostaglandin E2. 9. cAMP signaling kinetics may occur over a range of time scales, from tens of seconds to tens of minutes or hours [25]. 10. Spectrofluorimetry measurements of cell lysate may be used to construct a cAMP dose–response curve [9]. In brief, cells expressing the FRET probe are suspended in buffer, lysed with a dounce, and the cell lysate placed in a quartz cuvette. Cell lysate is pretreated with PDE inhibitors (50 μM Rolipram and 500 μM IBMX). A straightforward calculation suggests that baseline cAMP in the cuvette will be two to three orders of magnitude lower that the cAMP concentration required to elicit half maximal FRET response and ~100-fold lower than threshold cAMP levels detectable with current FRET probes. In brief, 10^6 cells have a volume of ~3 × 10−6 L [26]; this volume is diluted into 1 mL of buffer (~333 dilution). Basal cAMP levels (free and bound) are typically in the order of 1 μM in cells and tissues [27]; thus, baseline cAMP levels in a cuvette are ~3 × 10−3 μM, substantially lower than the 0.5–2 μM K1/2 of EPAC-based cAMP sensors.

Automated Image Analysis of FRET Signals for Subcellular cAMP Quantification

69

A cuvette containing cell lysates is placed in a spectrofluorimeter and cell lysates are treated with PDE inhibitors (50 μM Rolipram and 500 μM IBMX). A baseline scan of the fluorescence emission spectrum is then made while preferentially exciting the donor. Increasing concentrations of cAMP are added to the cuvette, with an emission scan acquired after each sequential addition. Upon completion of the experiment, the emission spectrum is linearly unmixed, and the resultant donor and acceptor abundances are used to calculate the FRET efficiency (Eqn. 2) for each cAMP concentration. This data is then fit using the Hill equation, establishing a quantitative relationship between FRET efficiency and cAMP concentration. 11. Other geometric, intensity, or texture features may provide additional information about the cellular physiology [12, 28]. These parameters may be measured for each region (cell, organelle, etc.) using automated image analysis approaches and data extracted for each time point in the study.

Acknowledgement This work was supported by NIH grants P01 HL066299 and S10 RR027535, and the Abraham Mitchell Cancer Research Fund. References 1. Förster T (1948) Zwischenmolekulare energiewanderung und fluoreszenz. Ann Phys 437: 55–75 2. Clegg RM (1995) Fluorescence resonance energy transfer. Curr Opin Biotechnol 6: 103–110 3. Zaccolo M (2004) Use of chimeric fluorescent proteins and fluorescence resonance energy transfer to monitor cellular responses. Circ Res 94:866–873 4. Mongillo M, McSorley T, Evellin S et al (2004) Fluorescence resonance energy transfer–based analysis of cAMP dynamics in live neonatal rat cardiac myocytes reveals distinct functions of compartmentalized phosphodiesterases. Circ Res 95:67–75 5. Ponsioen B, Zhao J, Riedl J et al (2004) Detecting cAMP-induced Epac activation by fluorescence resonance energy transfer: Epac as a novel cAMP indicator. EMBO Rep 5:1176–1180 6. Nikolaev VO, Bünemann M, Hein L et al (2004) Novel single chain cAMP sensors for receptor-induced signal propagation. J Biol Chem 279:37215–37218

7. Nikolaev VO, Gambaryan S, Lohse MJ (2006) Fluorescent sensors for rapid monitoring of intracellular cGMP. Nat Methods 3:23–25 8. Honda A, Adams SR, Sawyer CL et al (2001) Spatiotemporal dynamics of guanosine 3′, 5′-cyclic monophosphate revealed by a genetically encoded, fluorescent indicator. Proc Natl Acad Sci 98:2437–2442 9. Leavesley SJ, Britain A, Cichon LK et al (2013) Assessing FRET using spectral techniques. Cytometry A 83:898–912 10. Rich TC, Britain AL, Stedman T et al (2013) Hyperspectral imaging of FRET-based cGMP probes. In: Krieg T, Lukowski R (eds) Guanylate cyclase and cyclic GMP: methods and protocols, vol 1020, 1st edn, Methods in molecular biology. Springer Science+Business Media, LLC, New York. ISBN 1627034587 11. Zimmermann T, Rietdorf J, Girod A et al (2002) Spectral imaging and linear un-mixing enables improved FRET efficiency with a novel GFP2– YFP FRET pair. FEBS Lett 531:245–249 12. Carpenter AE, Jones TR, Lamprecht MR et al (2006) CellProfiler: image analysis software for

70

13.

14.

15.

16.

17.

18.

19.

20.

Silas J. Leavesley et al. identifying and quantifying cell phenotypes. Genome Biol 7:R100 Leavesley SJ, Annamdevula N, Boni J et al (2012) Hyperspectral imaging microscopy for identification and quantitative analysis of fluorescently-labeled cells in highly autofluorescent tissue. J Biophotonics 5:67–84 Favreau PF, Hernandez C, Lindsey AS et al (2014) Tunable thin-film optical filters for hyperspectral microscopy. J Biomed Opt 19: 011017-1–011017-11 Börner S, Schwede F, Schlipp A et al (2011) FRET measurements of intracellular cAMP concentrations and cAMP analog permeability in intact cells. Nat Protoc 6:427–438 Gordon GW, Berry G, Liang XH et al (1998) Quantitative fluorescence resonance energy transfer measurements using fluorescence microscopy. Biophys J 74:2702–2713 Sun Y, Hays NM, Periasamy A et al (2012) Monitoring protein interactions in living cells with fluorescence lifetime imaging microscopy. Methods Enzymol 504:371 Gu Y, Di W, Kelsell D et al (2004) Quantitative fluorescence resonance energy transfer (FRET) measurement with acceptor photobleaching and spectral unmixing. J Microsc 215:162–173 Leavesley SJ, Gao Y, Nakhmani A (submitted) Spectral image cytometry for automated subcellular cyclic nucleotide measurements. Front Physiol Vasc Physiol Berney C, Danuser G (2003) FRET or no FRET: a quantitative comparison. Biophys J 84:3992–4010

21. Klarenbeek JB, Goedhart J, Hink MA et al (2011) A mTurquoise-based cAMP sensor for both FLIM and ratiometric read-out has improved dynamic range. PLoS One 6:e19170 22. Novo D, Grégori G, Rajwa B (2013) Generalized unmixing model for multispectral flow cytometry utilizing nonsquare compensation matrices. Cytometry A 83A:508–520 23. Schindelin J, Arganda-Carreras I, Frise E et al (2012) Fiji: an open-source platform for biological-image analysis. Nat Methods 9: 676–682 24. Sommer C, Straehle C, Kothe U et al (2011) Ilastik: interactive learning and segmentation toolkit. In: IEEE. ISBN: 1424441277, pp 230–233 25. Rich TC, Webb KJ, Leavesley SJ (2014) Can we decipher the information content contained within cyclic nucleotide signals? J Gen Physiol 143:17–27 26. Feinstein WP, Zhu B, Leavesley SJ et al (2012) Assessment of cellular mechanisms contributing to cAMP compartmentalization in pulmonary microvascular endothelial cells. Am J Physiol Cell Physiol 302:C839–C852 27. Beavo J, Bechtel P, Krebs E (1974) Activation of protein kinase by physiological concentrations of cyclic AMP. Proc Natl Acad Sci 71: 3580–3583 28. Ljosa V, Carpenter AE (2009) Introduction to the quantitative analysis of two-dimensional fluorescence microscopy images for cell-based screening. PLoS Comput Biol 5:e1000603

Chapter 6 Channel-Based Reporters for cAMP Detection Thomas C. Rich, Wenkuan Xin, Silas J. Leavesley, and Mark S. Taylor Abstract In the last 15 years, tremendous progress has been made in the development of single-cell cAMP sensors. Sensors are based upon cAMP-binding proteins that have been modified to transduce cAMP concentrations into electrical or fluorescent readouts that can be readily detected using patch clamp amplifiers, photomultiplier tubes, or cameras. Here we describe two complementary approaches for the detection and measurement of cAMP signals near the plasma membrane of cells. These probes take advantage of the ability of cyclic nucleotide-gated (CNG) channels to transduce small changes in cAMP concentrations into ionic flux through channel pores that can be readily detected by measuring Ca2+ and/or Mn2+ influx or by measuring ionic currents. Key words Cyclic nucleotide-gated channel, cAMP, GPCR, Adenylyl cyclase, Phosphodiesterase

1

Introduction The development of real-time, single-cell cAMP sensors has allowed an understanding of the cellular and molecular mechanisms underlying signaling specificity in the cAMP pathway. Two main classes of cAMP probes have been developed, probes based on cyclic nucleotide-gated (CNG) channels and probes based upon cAMP-binding proteins sandwiched between fluorescent proteins that form Förster resonance energy transfer (FRET) pairs. These probes have complementary strengths and weaknesses that have been described elsewhere [1, 2]. Here we outline complementary protocols for the use of CNG channels as cAMP sensors. The first approach takes advantage of Ca2+ (Mn2+) permeability of CNG channels. This approach is typically used to test protocols prior to more technically difficult electrophysiological experiments. This approach is also well suited for high-throughput/high-content screening for agents that trigger increases or decreases in intracellular cAMP levels (see Note 1). The second approach utilizes more direct (and accurate) electrophysiological measurements. Electrophysiological measurements are better suited to take advantage of the strengths of CNG

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_6, © Springer Science+Business Media New York 2015

71

72

Thomas C. Rich et al.

channel-based sensors—their fast activation kinetics and ability to amplify small changes in cyclic nucleotide concentration [3–5]. Importantly, CNG channels have been genetically modified to have high sensitivity and specificity for cAMP [6–9]. It should be noted that electrophysiological measurements of CNG channel activity are technically more difficult than imaging experiments and cannot be used to study Ca2+-mediated regulation of cAMP signals due to the Ca2+ permeability of CNG channels [2, 10]. Even with these limitations, electrophysiological measurements of CNG channel activity are utilized by several groups for the measurement of cAMP signals. They have the highest kinetic resolution for the measurement of cAMP signals near the plasma membrane and, as such, offer unique insights into the mechanisms of specificity within the cAMP-signaling pathway.

2

Materials

2.1 Fluorometric cAMP Measurements

1. Spectrofluorimeter. A stirred cuvette spectrofluorimeter such as the PTI Quanta Master 40. 2. Disposable cuvettes (see Note 2). 3. Extracellular buffer solution 1: 145 mM NaCl, 4 mM KCl, 20 mM HEPES, 10 mM D-Glucose, 1 mM MgCl2, and 1 mM CaCl2, pH 7.4. 4. Cell type of interest (see Note 3). 5. Adenovirus or plasmids encoding CNG channel constructs. The following constructs are based on CNG2 (the wild-type olfactory α subunit) and are commonly used for cAMP measurements: WT CNGA2, K1/2 ~ 36 μM cAMP; E583M, K1/2 ~ 10 μM cAMP; C460W/E583M, K1/2 ~ 1 μM cAMP; Δ61–90/C460W/E583M, K1/2 ~ 12 μM cAMP. 6. Reagents: Stock solutions are stored in single-use aliquots at −20 °C. Vehicle: dimethyl sulfoxide (DMSO) or extracellular buffer; 1 mM fura-2/AM in DMSO; 50 mM forskolin in DMSO; 1–10 mM isoproterenol in buffer containing the antioxidants 0.1 mM ascorbic acid and 1 mM thiourea; 1–10 mM prostaglandin E1 in DMSO; 500 mM 3-isobutyl-1methylxanthine (IBMX) in DMSO; 10 mM rolipram in DMSO; 10–100 mM cAMP in extracellular buffer or DMSO (see Note 4).

2.2 Components for Electrophysiological cAMP Measurements

1. Electrophysiology setup for whole-cell and perforated patch experiments (see Note 5). 2. Rapid perfusion switch system (see Note 6). 3. Pipette puller and polisher (see Note 7).

CNG Channels as Real Time cAMP Sensors

73

4. Extracellular buffer solution 2: 145 mM NaCl, 4 mM KCl, 20 mM HEPES, 10 mM D-Glucose, 0.1 mM MgCl2, pH 7.4. 5. Extracellular buffer solution 3: 145 mM NaCl, 4 mM KCl, 20 mM HEPES, 10 mM D-Glucose, 10 mM MgCl2, pH 7.4. 6. Extracellular buffer solution 4: 145 mM KCl, 4 mM NaCl, 20 mM HEPES, 10 mM D-Glucose, 10 mM MgCl2, pH 7.4. 7. Capillary tubing for patch pipettes (see Note 8). 8. Pipette solution 1: 140 mM KCl, 0.5 mM MgCl2, 10 mM HEPES, 5 mM Na2ATP, 0.5 mM Na2GTP, 1 mM cAMP, pH 7.4. 9. Pipette solution 2: 70 mM KCl, 70 mM potassium gluconate, 4 mM NaCl, 0.5 mM MgCl2, 10 mM HEPES, pH 7.4, and 50–200 μg/mL of nystatin. 10. Cell types of interest. 11. Adenovirus or plasmids encoding CNG channel constructs as outlined in Subheading 2.1, item 5. 2.3 Data Analysis Software

1. Software such as Excel (Microsoft), SigmaPlot (Systat Software), or MATLAB (MathWorks) for the analysis of spectrofluorimeter measurements (see Note 9). 2. PulseFit, Clampfit, or MATLAB software for analysis of electrophysiological experiments (see Note 9).

3

Methods

3.1 cAMP Measurements Using a Spectrofluorimeter

1. Adjust spectrofluorimeter settings for fura-2 measurements. Several settings in the spectrofluorimeter acquisition software must be set prior to data acquisition: (a) The spectrofluorimeter should be set for at least two of the following excitation wavelengths: 340, 360, or 380 nm. The combination of 340 and 380 nm is used for Ca2+ measurements, and the combination of 360 and 380 nm is used for Mn2+ quench measurements (see Note 10). We have typically used a dwell time of 0.05–0.10 s at each wavelength. (b) The emission wavelength should be set to 510 nm. (c) The excitation and emission slit widths should be adjusted to minimize cross talk (see Note 11). (d) The speed of the stir bar should be adjusted to maintain cells in suspension without causing cell lysis. A stirred solution is required to maintain the density of cells in the light path and for adequate mixing of reagents. 2. Measure the background/blank signal using a cuvette containing 3 mL extracellular buffer solution 1 and stir bar only.

74

Thomas C. Rich et al.

3. Measure Ca2+ responses. Intact cell measurements are made using cells expressing one or more of the CNG channel constructs. The steps required to make spectrofluorimeter-based measurements in cell populations are as follows. (a) Add 3 mL of extracellular buffer solution 1 containing intact cells loaded with fura-2 (1 × 106 cells/mL) and a stir bar to a cuvette, place the cuvette in the spectrofluorimeter, and allow temperature to equilibrate. Note 11 describes conditions for loading fura-2. (b) Start measurement of intracellular Ca2+ (or Mn2+, see Note 10). Record baseline levels for at least 1 min prior to addition of agonists/antagonists. (c) Add agonists/antagonists to trigger changes in intracellular cAMP levels. GPCR agonists include 10–1,000 nM isoproterenol or PGE1. Antagonists of PDEs include 100– 500 μM IBMX, a broadband PDE inhibitor, and 10 μM rolipram, a PDE4-specific inhibitor. Experiments should be conducted in a stirred cuvette to ensure adequate distribution of agonists/antagonists. (d) Measure the time course of the cAMP signal. Measurements are continued until the response has reached steady state. In response to GPCR agonists, this typically occurs within 5–10 min [5, 11–17]. 4. Correct for background fluorescence by subtracting the background/blank signal (step 2) from all experimental trials. 5. Calculate fluorescence changes by calculating the ratio of intensities measured at two excitation wavelengths (either F340/F380 or F360/F380, see Note 10). We recommend this approach rather than estimating free Ca2+ because ideally cells will be loaded with high concentrations of fura-2 (see Note 12). Responses are often presented as R/R0 where R0 represents the average baseline fluorescence intensity ratio. 3.2 cAMP Measurements Using Electrophysiological Approaches

1. Pull patch pipettes. Place capillary tubing into the pipette puller. Choose settings that yield a pipette resistance of 1–1.5 MΩ (this may take a few iterations to achieve). Polish pipettes to achieve a smooth surface and a final resistance of 1.5–2 MΩ. Make certain that the filament wire has been coated with glass prior to polishing. 2. Place coverslip with cells expressing CNG channel constructs into the perfusion chamber (see Notes 13 and 14). Add buffer to cover cells (typically 0.5–1 mL). Place chamber on microscope stage. Hook up the bulk perfusion system (inflow and outflow tubes). Turn on the system and check for fluid leaks. The flow rate should be sufficient to exchange the bulk bath solution in 30–45 s (this can be readily checked using colored solutions).

CNG Channels as Real Time cAMP Sensors

75

3. Fill 0.5–1 mm of the patch pipette tip with the nystatincontaining Pipette Solution 2 (see Note 15). Suction may be required. Backfill the remainder of the pipette with Pipette Solution 1 (which does not contain nystatin). Place the patch pipette into the pipette holder. 4. Lower the pipette into the bath until the pipette tip is touching a preselected cell. Apply gentle suction until a highresistance seal (>5 GΩ) is formed. This should occur quickly (2–60 s) for most cell types. Subsequently, nystatin will form small holes in the plasma membrane allowing only monovalent ions to cross; this will provide electrical access to the cell. It may take 10–15 min for electrical access to reach a stable level. 5. Capacitive transients are elicited by applying 20 mV steps from the holding potential and recorded at 40 kHz (filtered at 10 kHz) for calculation of series resistance. Capacitance measurements should be made periodically throughout the experiment to track potential changes in series resistance. 6. Wait until a stable electrical access to the cell is achieved. The final series resistance (pipette resistance plus access resistance to the cell) should ideally be between 10 and 20 MΩ. Higher access resistances lead to excessive voltage error. Lower access resistances become unstable leading to a rupture of the small membrane patch beneath the patch pipette and thus transition to the whole-cell configuration. The concentration of nystatin used in Pipette Solution 2 may need to be altered to reach appropriate series resistances. 7. While waiting for the series resistance to stabilize, position the rapid solution switcher near the cell of interest. It is recommended that the actual switch time be estimated by switching from extracellular buffer solution 3 to extracellular buffer solution 4 during a 2 s step from a holding potential of −80–0 mV. The voltage protocol will activate endogenous voltage-gated K+ channels. Thus, a switch in extracellular K+ concentration will cause a rapid change in the reversal potential; the resultant change in current through K+ channels will accurately track the solution exchange time. We routinely observe switch times ≤60 ms using the Warner SF-77B fast-step system. 8. Electrical recordings should be sampled at ~5 times the fastest signal being measured. We typically sample at 1–5 kHz. 9. Typically the holding potential is maintained at 0 mV. Currents are elicited with 50 ms pulses to +50 and −50 mV and are sampled at 1 kHz. This is done to minimize (inactivate) currents through voltage-gated channels. Residual endogenous currents may need to be inhibited pharmacologically. In cells with higher series resistance, lower magnitude voltage steps may be used to minimize voltage errors due to series resistance.

76

Thomas C. Rich et al.

10. Collect baseline current levels for at least 2 min using the above protocol. At least once per minute briefly (2 s) expose the cell to extracellular solution 3 (using a rapid solution switcher). The high Mg2+ concentration will block currents through CNG channels and allow tracking of baseline currents (through other ion channels) throughout the experiment. Maintain bulk perfusion throughout the experiment to prevent buildup of Mg2+ or experimental reagents in the bath solution. 11. After recording baseline currents, expose cells to experimental reagents (e.g., PDE inhibitors, GPCR agonists or antagonists) for specified amounts of time using the rapid perfusion system. Recordings are typically maintained until responses have reached steady state. In the case of oscillatory responses, several periods should be recorded if possible. 12. After experimental protocols are finished, brief suction is applied to the patch pipette to rupture the patch membrane beneath the pipette (whole-cell configuration). This will be readily observed as a broadening of capacitive transients. When the current reaches steady state, block currents with extracellular buffer solution 3 (see Note 16). The Mg2+ blockable current represents maximal current through CNG channels (Imax). 13. Calibrate responses. Responses may be calibrated using the relation: I/Imax = [cAMP]N/([cAMP]N + K1/2N), where I/Imax is the fraction of maximal current, K1/2 is the cAMP concentration that gives a half-maximal current, and N is the Hill coefficient. Imax is estimated as described in step 12. The K1/2 for different CNG channels has been estimated previously [4, 5]. For example, if I/Imax were found to be 0.7 in an experiment using C460W/E583M channels (K1/2 ~ 1 μM), the estimated cAMP concentration would be ~1.5 μM.

4

Notes 1. High-throughput/high-content screening. Measurement of Ca2+ influx through CNG channels is readily adaptable to high-throughput/high-content screening platforms such as the FLIPR2 fluorometric imaging plate reader (Molecular Devices). This system is designed for measurement of intracellular Ca2+ in 96- and 384-well plate formats. The implementation of CNG channel-based assays on high-throughput systems allows rapid screening for compounds that regulate the cAMP pathway, including agonists and antagonists of G proteincoupled receptors, adenylyl cyclase, and phosphodiesterase. 2. In our experience SARSTEDT disposable cuvettes (No. D-51588) minimally attenuate UV illumination and as such are well suited for the measurement of Ca2+ signals using

CNG Channels as Real Time cAMP Sensors

77

fura-2 as the Ca2+ indicator. However, standard 10 × 10 × 48 mm plastic cuvettes would be appropriate for Ca2+ indicators in the visible range such as fluo-4. 3. Cells can be either immortalized cell lines such as HEK-293 cells or primary cultures such as mouse embryonic fibroblasts. Cells should be plated on coverslips or 100 mm dishes prior to transfection/infection. Transfection protocols for HEK-293 cells are outlined below. Fugene 6 reagent-based transfection: HEK-293 cells are plated at ~60 % confluence in either 35 mm dishes with coverslips (electrophysiology) or in 100 mm dishes (spectroscopy). Cells are transfected with constructs encoding either WT, E583M, C460W/E583M, or Δ61–90/C460W/E583M channel constructs using the Fugene 6 reagent (Promega, Madison, WI) with 1 μg cDNA and 3 μL Fugene 6 reagent per 35 mm dish or 6 μg cDNA and 18 μL Fugene 6 reagent per 100 mm dish. Cells are assayed 48–56 h post-transfection. Typically >90 % HEK-293 cells express CNG channel constructs. Adenovirus-mediated gene transduction: HEK-293 cells are plated at ~60 % in 35 mm dishes with coverslips or in 100 mm dishes. Cells are infected with adenovirus encoding CNG channel constructs with a MOI of 10 pfu/cell. Two hours postinfection, 1 mM hydroxyurea is added to the media to inhibit viral replication. Cells are assayed 24 h postinfection. Typically, >90 % of HEK-293 cells express functional CNG channels. A similar procedure is used for adenovirus-mediated gene transduction in other cell types, with the exceptions that the MOI required for expression is typically higher, 100 pfu/ cell, and hydroxyurea is not required (Table 1). 4. Either extracellular buffer or DMSO may be used as a solvent for cAMP; however, at concentrations of cAMP greater than ~20 mM, cAMP tends to precipitate out of salt solutions. At concentrations ~100 mM or higher, cAMP tends to precipitate out of DMSO. 5. We typically use a system based upon the HEKA EPC-10 patch clamp amplifier with Sutter MP-225 micromanipulators. The system should allow perfusion of the bulk bath solution within ~30 s to avoid accumulation of reagents entering the bath solution via the rapid solution switcher. 6. There are a number of fast solution switchers on the market. We use the Warner SF-77B fast-step system with the VC6 valve controller. We have also successfully implemented fast solution switch systems using model railroad track switchers obtained from model train/hobby shops. 7. Several commercially available pipette pullers and polishers are available such as the Sutter Instruments P-2000 puller and the Narishige MF-830 polisher.

78

Thomas C. Rich et al.

Table 1 Conditions for expression of CNG channels using adenovirus constructs in various cell types Cultured cell lines

MOI (pfu/cell)

Incubation time (h)

References

HEK-293

10a

24

[4]

C6-2B

100–200

48–72

[4, 38]

GH4C1

50

24–48

[8]

PC12

100

48

b

A7R5

50–100

24–48

b

Adult ventricular myocytes (rat, rabbit)

500–3,000, 100–200

24–48

[11, 39]

Neonatal rat cardiac myocytes (rat)

20–50

24–48

[40]

Vascular smooth muscle (rat)

100–200

48–72

[18]

Airway smooth muscle (human)

20

48

[16]

Pulmonary endothelial cells (rat, mouse)

100–200, 50–100

48

b

Embryonic fibroblasts (mouse)

500

48

[14]

Primary cultures

MOI = multiplicity of infection. pfu = plaque-forming units Expression of CNG channels in HEK-293 cells requires addition of 1–2 mM hydroxyurea 2 h postinfection to inhibit viral replication (see Note 2) b Rich laboratory, unpublished data a

8. The outer diameter of the tubing must be compatible with the pipette holder. We typically use thin-wall borosilicate capillary tubing with filament such as World Precision Instruments TW150-4, 1.5/1.12 mm outer/inner diameter. 9. We typically use custom scripts coded in MATLAB for analysis of emission intensities measured using a spectrofluorimeter and electrophysiological experiments. Alternatively, software for analysis of electrophysiological experiments (e.g., PulseFit or Clampfit) is often provided with purchase of the patch clamp amplifier. 10. Measurement of CNG channel activity by monitoring Mn2+ influx. It is possible that in some systems Ca2+ influx through CNG channels will trigger Ca2+-induced Ca2+ release (CICR), which would contribute to the observed Ca2+ responses. A different experimental protocol may be utilized to ensure that altered Ca2+ handling properties of the cells do not contribute to the observed responses: monitoring Mn2+ influx through CNG channels and subsequent quenching of fura-2 [18]. In this protocol, Mn2+ quench of fura-2 is measured at an excitation wavelength of 360 nm (the isosbestic point for fura-2 at different Ca2+ concentrations). Start the protocol as outlined

CNG Channels as Real Time cAMP Sensors

79

in Subheading 3.1, steps 1–3. Monitor fura-2 fluorescence at excitation wavelengths of 360 and 380 nm. After measuring the baseline fluorescence (F0) for 1 min, add 5 μM MnCl2. A slow decrease in fluorescence due do to basal channel activity may be observed. Then add agents that alter cAMP levels. Activation of CNG channels leads to Mn2+ influx that is readily detected as a loss of fura-2 fluorescence monitored at an excitation wavelength of 360 nm. Importantly, it is unlikely that changes in free intracellular Mn2+ will significantly alter adenylyl cyclase activity given the low extracellular concentrations of Mn2+ (5 μM) and the high Mn2+ affinity of fura-2, ~2.8 nM [19]. Normalize data to the pre-stimulus fluorescence (F0) to correct for variations in dye concentration and to allow for comparison of results on different batches of cells. To quantify data, perform linear fits to the slope of the agonist- and/or antagonist-induced change in fluorescence over time. This protocol describes an approach to measure CNG channel activity (changes in the rate of Mn2+ influx) that is largely independent of changes in intracellular Ca2+ levels. However, in some cell types (e.g., cardiac myocytes) basal Mn2+ influx through endogenous channels is too high to accurately assess agonist- and/or antagonist-mediated changes in cAMP levels. 11. Steps to adjust spectrofluorimeter slit width. Adjust excitation and emission slit widths to achieve ~5 nm full width at half height. Load a cuvette containing cells expressing fluorescent donor only. Perform emission scan. Adjust (reduce) excitation and emission slit widths equally and rescan. Continue until the crosstalk is ≤20 % of peak donor emission intensity. 12. Fura-2/AM loading. A cell-loading solution is prepared by combining 4 μL fura-2/AM stock (1 mM in DMSO) with 2 μL pluronic F-127 (20 % solution in DMSO) in an Eppendorf tube and adding 1 mL of extracellular buffer solution 1 (or similar buffer solution) to yield a solution containing 4 μM fura-2/AM, 0.04 % pluronic F-127. Cells are loaded at room temperature for 15–30 min, rinsed with extracellular solution buffer 1, and allowed to sit in the dark at room temperature for an additional 15 min. This provides adequate opportunity for de-esterification of the dye, thereby trapping it in the cells. Loading at temperatures above 35 °C should be avoided as it tends to compartmentalize indicator within intracellular organelles of many cell types. In this assay, changes in the rate of Ca2+ influx are indexes of changes in cAMP concentration (cAMP is proportional to the rate of Ca2+ influx through CNG channels). Intracellular fura-2 concentrations should be adequate to displace endogenous Ca2+ buffers and ensure excess of unbound indicator over the experimental time course (i.e., following maximal Ca2+ influx through CNG channels).

80

Thomas C. Rich et al.

Thus, higher fura-2/AM loading concentrations (i.e., 16 μM) help preserve fidelity of Ca2+ influx measurements and avoid nonlinear signals at higher Ca2+ concentrations [8]. Lower intracellular fura-2 concentrations are useful for other assays when maintenance and measurement of physiologically meaningful free Ca2+ levels—preserving downstream Ca2+-mediated signaling events—is a priority [20]. 13. Electrophysiological measurements are made in single cells and, as such, can be made immortalized cell lines such as HEK-293 cells or primary cultures such as mouse embryonic fibroblasts and a variety of primary cultures such as airway smooth muscle cells and cardiac myocytes. Cells should be plated on glass coverslips prior to transfection/infection. Note 3 outlines transfection protocols for HEK-293 cells. 14. Cyclic AMP measurements using endogenous cyclic nucleotide regulated channels. Three electrophysiological approaches have been used to detect cyclic nucleotide signals using endogenous ion channels: monitoring protein kinase A (PKA)mediated regulation of L-type Ca2+ channels, monitoring endogenous CNG channel activity, and monitoring endogenous hyperpolarization-activated cyclic nucleotide-gated (HCN) channel activity. (a) Measurement of PKA-mediated regulation of L-type Ca2+ channel activity. The use of PKA-mediated regulation of L-type Ca2+ channel activity to deduce underlying changes in cAMP concentration has been most successfully implemented in cardiac myocytes [11, 21–23]. This approach uses whole-cell electrophysiological approaches (see Note 15) to monitor PKA-mediated potentiation of L-type Ca2+ channel activity via well-established voltage protocols to measure L-type Ca2+ channel activity. Two difficulties arise from monitoring PKA-mediated regulation of voltagegated channels as a surrogate for cAMP measurements. First, most cell types lack sufficient levels of PKA-regulated ion channels for accurate and consistent measurements of channel regulation and thus accurate inference of underlying cyclic nucleotide levels. Second, more importantly, it is often unclear whether observed agonist-/antagonistinduced changes in current occur due to changes in PKA activity, phosphatase activity, or both. Thus, these measurements more accurately reflect agonist-/antagonistinduced changes in the balance between PKA and phosphatase activities than underlying cAMP signals. (b) Measurement of endogenous CNG channel activity. Endogenous CNG channels have been used to monitor cyclic nucleotides in rod and cone outer segments [24– 28] and olfactory cilia [29–32]. The concentration of

CNG Channels as Real Time cAMP Sensors

81

CNG channels in these cellular domains is high, allowing for accurate assessment of currents through CNG channels and the underlying cyclic nucleotide signals. Endogenous CNG channels are low-abundance proteins in most other cellular systems, making it difficult to accurately assess whether observed agonist-induced changes in current are due to cyclic nucleotide-mediated regulation of the endogenous CNG channels. Thus, endogenous CNG channels are not practical for cAMP measurements in most cellular systems. (c) Measurement of HCN channel activity. Some cell types express high enough levels of HCN channels to monitor cAMP-mediated changes in HCN activity, or HCN channels can be overexpressed for cAMP measurement [33, 34]. In either case, HCN channels are activated by membrane hyperpolarization and regulated by cAMP binding. Thus, accurate measurement of cAMP-mediated changes in voltage-dependent kinetic properties (primarily the voltage dependence of channel activation) of HCN channels is required for quantitative cAMP measurements. In cell types that do not express sufficient endogenous HCN channel levels, the CNG channel constructs described in this chapter are in general more desirable because their gating properties are not highly dependent on membrane potential. 15. Alternate electrophysiological approaches for measurement of CNG channel activity. Two alternate electrophysiological approaches for the use of CNG channel as cyclic nucleotide sensors have been utilized: patch cram [35, 36] and whole-cell measurements [5, 11–13]. (a) Patch cram measurements: CNG channels are expressed at high levels in Xenopus oocytes. Excised membrane patches expressing high levels of CNG channels are removed from the oocytes and crammed into recipient cells. The advantage of this approach is that the sensitivity of the CNG channels can be measured in the same patches that subsequent measurements of intracellular cyclic nucleotide signaling are made. However, two limitations have precluded this approach from widespread use. First, the patch that is excised from Xenopus oocytes contains a variety of signaling proteins in addition to CNG channels. This makes it difficult to ensure that measured responses are not influenced by signaling proteins within the Xenopus oocyte patch. Second, few cell types are both large enough and robust enough to survive being impaled by a patch pipette. (b) Whole-cell measurements. Whole-cell measurements of cAMP signals are possible largely because in many cell types, near-membrane cAMP levels appear to equilibrate slowly with the bulk cytosol; thus dialysis of cAMP from

82

Thomas C. Rich et al.

the cytosol into the patch pipette is slow [4, 5, 11, 16]. In whole-cell experiments, the membrane directly underneath the pipette is ruptured by applying light suction or a brief electrical pulse (the “zap” feature available on many patch clamp amplifiers). This provides electrical access to cells and allows dialysis of solutions from patch pipettes into the cytosol. Thus, the whole-cell configuration is particularly useful for allowing known intracellular concentrations of compounds that affect signal transduction (e.g., inhibitors and small proteins) to be introduced into cells [13, 16, 17]. Additional information about electrophysiological approaches can be found in [37]. 16. Calibration of CNG channel measurements. In some cell types—e.g., cardiac myocytes—the introduction of small concentrations of nystatin triggers a rapid induction of leak currents and loss of plasma membrane integrity (presumably due to interactions with the membranes of intracellular organelles such as the mitochondria). For experiments in these cell types, cAMP should be omitted from pipette solution 2, and Imax should be estimated by exchanging the bath solution for extracellular buffer containing high concentrations of membrane permeant CNG channel agonists, e.g., 100 μM pCPT-cAMP.

Acknowledgments This work was supported by NIH P01HL066299, the Center for Lung Biology, and the Colleges of Medicine and Engineering, University of South Alabama. References 1. van der Krogt GN, Ogink J, Ponsioen B et al (2008) A comparison of donor-acceptor pairs for genetically encoded FRET sensors: application to the Epac cAMP sensor as an example. PLoS One 3:e1916. doi:10.1371/journal.pone.0001916 2. Rich TC, Webb KJ, Leavesley SJ (2014) Perspectives on: cyclic nucleotide microdomains and signaling specificity: can we decipher the information content contained within cyclic nucleotide signals? J Gen Physiol 143:17–27 3. Finn JT, Grunwald ME, Yau K-W (1996) Cyclic nucleotide-gated ion channels: an extended family with diverse functions. Annu Rev Physiol 58:395–426 4. Rich TC, Fagan KA, Nakata H et al (2000) Cyclic nucleotide-gated channels colocalize with adenylyl cyclase in regions of restricted cAMP diffusion. J Gen Physiol 116:147–161

5. Rich TC, Fagan KA, Tse TE et al (2001) A uniform extracellular stimulus triggers distinct cAMP signals in different compartments of a simple cell. Proc Natl Acad Sci U S A 98: 13049–13054 6. Liu M, Chen TY, Ahamed B et al (1994) Calcium-calmodulin modulation of the olfactory cyclic nucleotide-gated cation channel. Science 266:1348–1354 7. Varnum MD, Black KD, Zagotta WN (1995) Molecular mechanism for ligand discrimination of cyclic nucleotide-gated channels. Neuron 15:619–625 8. Rich TC, Tse TE, Rohan JG et al (2001) In vivo assessment of local phosphodiesterase activity using tailored cyclic nucleotide-gated channels as cAMP sensors. J Gen Physiol 118:63–77

CNG Channels as Real Time cAMP Sensors 9. Fagan KA, Schaack J, Zweifach A et al (2001) Adenovirus encoded cyclic nucleotide-gated channels: a new methodology for monitoring cAMP in living cells. FEBS Lett 500:85–90 10. Rich TC, Karpen JW (2002) Cyclic AMP sensors in living cells: what signals can they actually measure? Ann Biomed Eng 30:1088–1099 11. Rochais F, Vandecasteele G, Lefebvre F et al (2004) Negative feedback exerted by cAMPdependent protein kinase and cAMP phosphodiesterase on subsarcolemmal cAMP signals in intact cardiac myocytes: an in vivo study using adenovirus-mediated expression of CNG channels. J Biol Chem 279:52095–52105 12. Rochais F, Abi-Gerges A, Horner K et al (2006) A specific pattern of phosphodiesterases controls the cAMP signals generated by different Gs-coupled receptors in adult rat ventricular myocytes. Circ Res 98:1081–1088 13. Xin W, Tran TM, Richter W et al (2008) Functional roles of GRK and PDE activities in the regulation of β2 adrenergic signaling. J Gen Physiol 134:349–364, PMCID: PMC2279169 14. Blackman BE, Heimann J, Horner K et al (2011) PDE4D and PDE4B function in distinct subcellular compartments in mouse embryonic fibroblasts. J Biol Chem 286:12590–12601 15. Willoughby D, Wong W, Schaack J et al (2006) An anchored PKA and PDE4 complex regulates subplasmalemmal cAMP dynamics. EMBO J 25:2051–2061 16. Horvat SJ, Deshpande DA, Yan H et al (2012) A-kinase anchoring proteins regulate compartmentalized cAMP signaling in airway smooth muscle. FASEB J 26:3670–3679 17. Rich TC, Xin W, Conti M et al (2007) Cellular mechanisms underlying prostaglandin-induced transient cAMP signals near the plasma membrane of HEK-293 cells. Am J Physiol Cell Physiol 292:C319–C331 18. Piggott LA, Hassell KA, Berkova Z et al (2006) Natriuretic peptides and nitric oxide stimulate cGMP synthesis in different cellular compartments. J Gen Physiol 128:3–14 19. Kwan CY, Putney JW (1990) Uptake and intracellular sequestration of divalent cations in resting and methacholine-stimulated mouse lacrimal acinar cells. Dissociation by Sr2+ and Ba2+ of agonist-stimulated divalent cation entry from the refilling of the agonist-sensitive intracellular pool. J Biol Chem 265:678–684 20. Neher E, Augustine GJ (1992) Calcium gradients and buffers in bovine chromaffin cells. J Physiol 450:273–301 21. Jurevicius J, Fischmeister R (1996) cAMP compartmentation is responsible for a local activation of cardiac Ca2+ channels by β-adrenergic agonists. Proc Natl Acad Sci U S A 93:295–299

83

22. Frace AM, Mery P-F, Fischmeister R et al (1993) Rate-limiting steps in β-adrenergic stimulation of cardiac calcium current. J Gen Physiol 101:337–353 23. Hartzell HC, Mery PF, Fischmeister R et al (1991) Sympathetic regulation of cardiac calcium current is due exclusively to cAMP-dependent phosphorylation. Nature 351:573–576 24. Yau K-W, Lamb TD, Matthews G et al (1979) Current fluctuations across single rod outer segments. Vision Res 19:387–390 25. Baylor DA, Yau K-W, Lamb TD et al (1978) Properties of the membrane current of rod outer segments. Sens Processes 2:300–305 26. Yau K-W, Lamb TD, Baylor DA (1977) Lightinduced fluctuations in membrane current of single toad rod outer segments. Nature 269: 78–80 27. Pugh EN Jr, Lamb TD (1993) Amplification and kinetics of the activation steps in phototransduction. Biochim Biophys Acta 1141:111–149 28. Molday RS (1998) Photoreceptor membrane proteins, phototransduction, and retinal degenerative diseases: the Friedenwald Lecture. Invest Ophthalmol Vis Sci 39:2493–2513 29. Gold GH (1999) Controversial issues in vertebrate olfactory transduction. Annu Rev Physiol 61:857–871 30. Lowe G, Gold GH (1993) Contribution of the ciliary cyclic nucleotide-gated conductance to olfactory transduction in the salamander. J Physiol 462:175–196 31. Nakamura T, Gold GH (1987) A cyclic nucleotide-gated conductance in olfactory receptor cilia. Nature 325:442–444 32. Chen CH, Nakamura T, Koutalos Y (1999) Cyclic AMP diffusion coefficient in frog olfactory cilia. Biophys J 76:2861–2867 33. Heine M, Ponimaskin E, Bickmeyer U et al (2002) 5-HT-receptor-induced changes of the intracellular cAMP level monitored by a hyperpolarization-activated cation channel. Pflugers Arch 443:418–426 34. Ponimaskin EG, Heine M, Zeug A et al (2007) Monitoring receptor-mediated changes of intracellular camp level by using ion channels and fluorescent proteins as biosensors. In: Chattopadhyay A (ed) Serotonin receptors in neurobiology. CRC Press, Boca Raton, FL, Chapter 2 35. Trivedi B, Kramer RH (1998) Real-time patchcram detection of intracellular cGMP reveals long-term suppression of responses to NO and muscarinic agonists. Neuron 21:895–906 36. Trivedi B, Kramer RH (2002) Patch cramming reveals the mechanism of long-term suppression of cyclic nucleotides in intact neurons. J Neurosci 22:8819–8826

84

Thomas C. Rich et al.

37. Hille B (2001) Ionic channels of excitable membranes. Sinauer Associates, Sunderland, MA 38. Fagan KA, Rich TC, Tolman S et al (1999) Adenovirus-mediated expression of an olfactory cyclic nucleotide-gated channel regulates the endogenous Ca2+-inhibitable adenylyl cyclase in C6-2B glioma cells. J Biol Chem 274:12445–12453

39. Xin W, Yang X, Rich TC et al (2012) All preconditioning-related G protein-coupled receptors can be demonstrated in the rabbit cardiomyocyte. J Cardiovasc Pharmacol Ther 17:190–198 40. Walsh KB, Rich TC, Coffman Z (2009) Development of a high throughput assay for monitoring cAMP levels in cardiac ventricular myocytes. J Cardiovas Pharm 53:223–230

Chapter 7 Imaging Sub-plasma Membrane cAMP Dynamics with Fluorescent Translocation Reporters Anders Tengholm and Olof Idevall-Hagren Abstract Imaging cAMP dynamics in single cells and tissues can provide important insights into the regulation of a variety of cellular processes. In recent years, a large number of tools for cAMP measurements have been developed. While most cAMP reporters are designed to undergo changes in fluorescence resonance energy transfer (FRET), there are alternative techniques with advantages for certain applications. Here, we describe protocols for cAMP measurements in the sub-plasma membrane space based on the detection of the cAMP-induced translocation of engineered fluorescent protein-tagged subunits of protein kinase A between the cytoplasm and the plasma membrane. Total internal reflection fluorescence (TIRF) imaging of the changes in reporter localization yields robust signal changes and has contributed to the discovery of cAMP oscillations in the sub-plasma membrane space of insulin-secreting β-cells stimulated with glucose and gluco-incretin hormones. We also demonstrate how the technique can be combined with measurements of the cytosolic Ca2+ concentration or with recordings of the subcellular localization of the cAMP effector protein Epac2. The translocation reporter approach provides a valuable complement to other methods for imaging sub-membrane cAMP dynamics in various types of cells. Key words cAMP oscillations, Plasma membrane, Protein kinase A, Translocation, Total internal reflection fluorescence

1

Introduction Second messenger signaling is highly dynamic and often involves distinct temporal patterns and spatial compartmentalization, which help to achieve efficiency and specificity in the control of downstream cellular functions. Such spatiotemporal dynamics have been particularly well recognized for Ca2+, thanks to the early advent of fluorescent indicators to measure the cytoplasmic Ca2+ concentration in individual cells [1]. Measurements of intracellular cAMP dynamics have been more challenging and were for a long time restricted by a lack of suitable tools. There are now a number of available methods for real-time measurements of cAMP (reviewed in [2, 3] and described elsewhere in this volume). Most cAMP indicators are based on fluorescence resonance energy transfer

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_7, © Springer Science+Business Media New York 2015

85

86

Anders Tengholm and Olof Idevall-Hagren

(FRET). The first version required cumbersome microinjections of fluorescently labeled regulatory and catalytic subunits of protein kinase A (PKA) [4] and suffered from distorted signal as the probe moieties gradually became paired with endogenous subunits during repeated cycles of activation. These problems were overcome by genetically encoded versions of the sensor [5, 6]. FRET probes with the additional advantage of being expressed as a single polypeptide chain were soon generated from the other cAMP-target proteins Epac1 and Epac2 as well as from isolated cAMP-binding domains from PKA, Epac, and the hyperpolarization-activated cyclic nucleotide-gated potassium channel 2 [7–10]. Detection of FRET is not always trivial, however, and signal changes are often rather small and the readout prone to various artifacts [2, 11, 12]. Alternative classes of cAMP sensor include cyclic nucleotide-gated (CNG) channels [13], the activities of which are monitored with patch-clamp current recordings or as changes of the intracellular Ca2+ concentration. Such channels have the advantage of responding quickly and of reporting cAMP in the sub-plasma membrane space where the adenylate cyclases are located and where many important signaling and physiological events take place. Drawbacks with the CNG channel sensors are that their expression may affect the electrophysiological activity of the cell and that the Ca2+ entry may modulate adenylate cyclase and phosphodiesterase activities, thereby affecting the cAMP signal that is being measured. Another method, particularly useful for recording sub-plasma membrane cAMP dynamics, is based on protein translocation as a readout [14]. Like the early versions of the FRET biosensors [4, 5], it takes advantage of the fact that the regulatory and catalytic subunits of PKA dissociate following binding of cAMP to the regulatory subunit. By targeting the regulatory subunit to the plasma membrane, the holoenzyme becomes located at the membrane under basal conditions [14]. As the sub-membrane cAMP concentration increases, the holoenzyme dissociates, and the change in localization of the catalytic subunit can be monitored with fluorescence microscopy simply as a change in localization. This translocation causes a signal change that is much larger than the simultaneously occurring loss of FRET. We have generated such a translocation sensor with a modified PKA RIIβ regulatory subunit, similar to that in the original genetically encoded FRET-based sensor. The regulatory subunits of PKA dimerize, and in addition, subunits of the RII type are often targeted to A-kinase anchoring proteins (AKAPs) by interactions with their N-terminal region [15]. To avoid mistargeting of the sensor, ninety amino acid residues, including both the AKAP-binding and the dimerization regions, were deleted. Cyan fluorescent protein (CFP) was added to the C-terminus (see Note 1), and the fluorescent protein was in turn extended by a C-terminal polybasic

Imaging Sub-Membrane cAMP Dynamics

87

stretch and a CAAX motif from K-Ras (see Note 2). The CAAX motif becomes posttranslationally farnesylated, and this lipid modification together with the basic amino acid residues targets the protein to the plasma membrane (Fig. 1). The PKA catalytic subunit Cα was used in its full-length wild-type form and tagged with a yellow fluorescent protein (YFP) at its C-terminus (see Note 3). The localization of the reporter can be visualized with confocal microscopy, with which the translocation of the fluorescent Cα subunit can be detected either as changes in plasma membrane or cytoplasmic fluorescence (Fig. 2). However, a better approach is to use total internal reflection fluorescence (TIRF) microscopy, which selectively illuminates a volume within 1.40) objective and beam focusing and positioning optics. Such TIRF illumination devices are available from most of the major microscope manufacturers. An alternative TIRF configuration uses a prism

96

Anders Tengholm and Olof Idevall-Hagren

for illumination, which has the advantage of allowing imaging with a low magnification (10–20×) objective to obtain information from many cells in parallel (Fig. 3b). The authors use a custom-built system, but commercial illuminators are available from, e.g., TIRF Labs Inc (Cary, NC). 6. Commercial confocal microscopes are usually equipped with the necessary peripheral instrumentation, but TIRF systems may require custom configuration. 7. A variety of diode lasers, diode-pumped solid-state lasers, and gas lasers suitable for exciting translocation reporters based on CFP/YFP or GFP/mCherry are available from several companies. CFP is ideally excited with 445 nm diodes or with the 442 nm line of a helium-cadmium laser or even the 458 nm line of an argon laser. The 514.5 nm line of an argon laser or similar lines of solid-state lasers are the best for exciting YFP. GFP is excited with the 488 nm line of the argon laser or with solid-state lasers at, e.g., 491 or 473 nm. The red fluorescent protein mCherry is efficiently excited at 561 nm provided by many solid-state lasers. 8. Filters can be obtained from Chroma Technology (Bellows Falls, VT), Omega Optical (Brattleboro, VT), or Semrock (Rochester, NY). Dual-channel recordings require a device for excitation wavelength switching, such as a filter wheel (e.g., from Sutter Instruments, Novato, CA) or an acousto-optic tunable filter (e.g., from AA Opto-electronic, Orsay, France). Dual-wavelength emission recording requires either a filter wheel as above or an image splitter, such as the Photometrics Dual View DV-2 system (Tucson, AZ). 9. Data acquisition and analysis software are usually included with commercial microscope systems but may have to be purchased separately for custom-built setups. The authors use MetaFluor from Molecular Devices (Sunnyvale, CA) and the free ImageJ software [26]. 10. It may be critical to maintain a temperature of 37 °C during an experiment as well as to add and remove test substances. We use a peristaltic pump in combination with a custom-built superfusion chamber, chamber heater, and microscope objective heater. Similar equipment is available from, e.g., Warner Instruments (Hamden, CT). 11. As an alternative to poly-L-lysine, cover slips can be coated with collagen, fibronectin, or laminin. 12. This protocol typically yields 30 % transfection efficiency 1 day after transfection. The protocol is optimized for insulinsecreting MIN6 β-cells and may have to be modified for other types of cells. For example, for some primary cells, including

Imaging Sub-Membrane cAMP Dynamics

97

pancreatic β-cells, transduction with adenoviral vectors may be required. 13. All cDNA should be transfected at the same time to obtain maximal cotransfection. If more than two plasmids are used, it is not recommended to use more than a total of 3 μg plasmid DNA per cover slip as higher concentrations may have adverse effects and contribute to reduction of the transfection efficiency and interfere with cell function. Adjust the volume of liposomes when changing the amount of DNA to maintain the ratio of DNA-Lipofectamine 2000 at 1:2.5 (w/v). The plasmid DNA used for transfection must be of high quality. Our experience is that the quality of plasmid DNA correlates positively with the scale of production. We therefore recommend producing plasmid DNA from at least 200 mL bacterial cultures, corresponding to a standard “maxi-prep”. 14. Do not allow cells to express the biosensor protein for more than 24 h, since excessive levels of the fusion proteins may affect cell function. This is because the cAMP-binding part of the reporter may buffer cAMP changes and the catalytic subunit remains catalytically active. We have tried to generate a kinase-dead mutant, but unfortunately, the mutation somehow affected the ability of the protein to interact with the membrane-localized regulatory subunit. 15. Although cells can be stimulated by adding medium into the bath with a pipette, it is preferable to add medium using a superfusion system. Such a system not only permits convenient washout of the stimulus but also eliminates problems with evaporation that otherwise would occur, since the ambient temperature in many cases should be maintained at 37 °C to observe normal physiological responses. 16. There will be cells expressing the fluorescent biosensors over a wide range of expression levels. High levels of cAMP-binding proteins and catalytically active PKA may potentially interfere with processes in the cells as mentioned in Note 14. Therefore, choose cells with relatively low expression levels, approximately in the lower third of the brightness range of all cells. One should also consider that the two translocation reporter components are expressed at roughly equal levels. If there is a deficiency of the membrane-anchoring component, the catalytic subunit may bind exclusively to endogenous regulatory subunits at other subcellular locations. When these complexes dissociate following an elevation of the cAMP concentration, the signal at the plasma membrane may increase rather than decrease giving the impression of an inverted response. 17. Select camera exposure time and gain settings so that no pixels in the image will be saturated. It is important to consider that

98

Anders Tengholm and Olof Idevall-Hagren

the fluorescence intensity of the translocating PKA-Cα component might change severalfold during an experiment. It is good to keep the exposure time as short as possible without compromising signal-to-noise ratio, since excessive exposure to excitation light may result in photobleaching and phototoxic effects. If the fluorescence collected from the cells is very low, it is possible with most CCD cameras to combine charges in adjacent pixels to form one pixel in a process named binning. This will enhance the signal at the expense of optical resolution. On the contrary, if the signal is too bright, it indicates an excessive excitation light intensity. Reduce the laser power or attenuate the light with neutral density filters in the excitation beam path. The laser beam should be completely blocked with a shutter between image captures to avoid adverse effects of the light on the specimen. 18. (a) For confocal microscopy images, plasma membrane translocation and dissociation of the biosensor can be quantified by placing a region of interest over the area corresponding to the plasma membrane, either manually or using a segmentation algorithm, such as a threshold or edge detection function. Measurements of membrane fluorescence may be difficult since the cell shape sometimes changes over the time course of an experiment, in particular after cell stimulation. An alternative is to quantify the changes in cytoplasmic fluorescence by defining a region of interest inside the cell that excludes the nucleus, membrane, or other conspicuous organelles. This approach only works for analyzing changes in Cα-YFP fluorescence but not for the ΔRIIβ-CFP-CAAX reference channel or for ratio images. (b) In TIRF microscopy images, a region of interest over the cell will always show fluorescence in the plasma membrane, and translocation or dissociation is simply recorded as changes of intensity or ratio in this region. (c) An increase of the sub-membrane cAMP concentration results in the loss of Cα-YFP intensity but does not affect the ΔRIIβ-CFP-CAAX signal. The CFP/YFP ratio thus increases. If a ratio change is associated with changes in the ΔRIIβ-CFP-CAAX signal, it indicates that there are also changes in cell adhesion or other unspecific effects of the stimulus. (d) The cAMP translocation biosensor is based on the dissociation of the catalytic and regulatory subunits of PKA, a process with a reported dissociation constant in the nanomolar to low micromolar range. This means that at very high cAMP concentrations, most of the subunits will have dissociated from each other, and the sensitivity for detecting changes in cAMP will be reduced. This will not be a problem

Imaging Sub-Membrane cAMP Dynamics

99

under most physiological conditions, but if saturation is suspected, we recommend complementing the experiments with an alternative biosensor based on the low affinity cAMP-binding Epac proteins [7, 9, 10]. (e) To compensate for the differences in the expression of the two reporter components, ratio changes can be expressed in relation to the baseline by dividing the ratio value at each time point with the prestimulatory level. In the case, only the Cα reporter component is fluorescent; the intensity changes can be expressed as fluorescence intensity changes in relation to baseline (F/F0 or F0/F). 19. Dissolve Fura Red at 10 mM in DMSO and store at −20 °C protected from light. Mix before use and incubate together with cells at 2–20 μM during 20–45 min. The optimal indicator concentrations and incubation time vary depending on the cell type and cell density. Excessive loading may reduce the Ca2+ response by buffering the concentration changes, whereas too little loading gives poor signal-to-noise ratio. 20. Fura Red is readily excited by the same light source as GFP or YFP, such as the 488 or 514.5 nm lines of the argon ion laser. Fluorescence emission is detected with a 630 nm long-pass filter. Upon Ca2+ binding, the excitation spectrum of Fura Red becomes blue shifted resulting in loss of fluorescence excited at 488 and 514.5 nm. Alternation between the Fura Red and cAMP translocation reporter emission filters using a filter wheel or similar device allows simultaneous measurements of cytoplasmic Ca2+ concentration and cAMP dynamics in the same cell. However, as YFP shows some emission even above 600 nm, there may be a slight crossover of YFP fluorescence into the Fura Red channel in cells with very bright YFP fluorescence and poor Fura Red loading. The degree of overlap is easily estimated by imaging cells expressing YFP, but lacking the Ca2+ indicator, using the Fura Red filter set. Alternatively, cAMP can be measured non-ratiometrically with an unlabeled regulatory subunit and a CFP-labeled Cα subunit as illustrated in Fig. 4b. 21. In recent years, there has been substantial expansion of the genetically encoded Ca2+-indicator family, including the addition of two red-shifted versions that could be used together with the cAMP translocation biosensor. These two Ca2+ indicators, R-GECO1 [27] and RCaMP [28], are readily excited by 561 nm excitation light, while emission is measured >620 nm, providing complete separation from both CFP and YFP. The dynamic range, brightness, and response amplitude of these indicators are significantly better than those of the Ca2+ dye Fura Red, but the genetically encoded indicators require the transfection of an additional plasmid.

100

Anders Tengholm and Olof Idevall-Hagren

22. The mCherry fluorophore is excited by 561 nm laser light, and emission is collected through a 584 nm long-pass filter (available from, e.g., Semrock). Simultaneous recordings of mCherry, CFP, and YFP fluorescence is difficult on most confocal or TIRF microscopy setups due to lack of suitable dichroic mirrors. Therefore, changes in cAMP will be non-ratiometrically recorded simply as changes in plasma membrane YFP fluorescence, while the distribution of mCherry-Epac2 are recorded in parallel. Fluorescence from the CFP-tagged regulatory subunit anchored to the plasma membrane will not be recorded, precluding ratiometric measurements. Alternation between the mCherry and cAMP translocation reporter emission filters using a filter wheel or similar device allows simultaneous measurements of mCherry fluorescence and cAMP dynamics in the same cell.

Acknowledgments We thank Drs Hongyan Shuai and Geng Tian for their help with the preparation of Figs. 3 and Fig. 4. The authors’ work is supported by grants from the European Foundation for the Study of Diabetes, the family Ernfors Foundation, the Novo Nordisk Foundation, the Swedish Diabetes Association, and the Swedish Research Council (67X-14643, 67P-21262, 325-2012-6778, 524-2013-298). References 1. Berridge MJ, Bootman MD, Roderick HL (2003) Calcium signalling: dynamics, homeostasis and remodelling. Nat Rev Mol Cell Biol 4:517–529 2. Willoughby D, Cooper DM (2008) Live-cell imaging of cAMP dynamics. Nat Methods 5:29–36 3. Berrera M, Dodoni G, Monterisi S et al (2008) A toolkit for real-time detection of cAMP: insights into compartmentalized signaling. Handb Exp Pharmacol 186:285–298 4. Adams SR, Harootunian AT, Buechler YJ et al (1991) Fluorescence ratio imaging of cyclic AMP in single cells. Nature 349:694–697 5. Zaccolo M, De Giorgi F, Cho CY et al (2000) A genetically encoded, fluorescent indicator for cyclic AMP in living cells. Nat Cell Biol 2:25–29 6. Zaccolo M, Pozzan T (2002) Discrete microdomains with high concentration of cAMP in stimulated rat neonatal cardiac myocytes. Science 295:1711–1715 7. Nikolaev VO, Bunemann M, Hein L et al (2004) Novel single chain cAMP sensors for

8.

9.

10.

11.

12.

receptor-induced signal propagation. J Biol Chem 279:37215–37218 Nikolaev VO, Bunemann M, Schmitteckert E et al (2006) Cyclic AMP imaging in adult cardiac myocytes reveals far-reaching beta1adrenergic but locally confined beta2-adrenergic receptor-mediated signaling. Circ Res 99: 1084–1091 Ponsioen B, Zhao J, Riedl J et al (2004) Detecting cAMP-induced Epac activation by fluorescence resonance energy transfer: Epac as a novel cAMP indicator. EMBO Rep 5: 1176–1180 DiPilato LM, Cheng X, Zhang J (2004) Fluorescent indicators of cAMP and Epac activation reveal differential dynamics of cAMP signaling within discrete subcellular compartments. Proc Natl Acad Sci U S A 101:16513–16518 Piston DW, Kremers GJ (2007) Fluorescent protein FRET: the good, the bad and the ugly. Trends Biochem Sci 32:407–414 Vogel SS, Thaler C, Koushik SV (2006) Fanciful FRET. Sci STKE 2006:re2. doi:10.1126/stke.3312006re3312002

Imaging Sub-Membrane cAMP Dynamics 13. Rich TC, Fagan KA, Nakata H et al (2000) Cyclic nucleotide-gated channels colocalize with adenylyl cyclase in regions of restricted cAMP diffusion. J Gen Physiol 116:147–161 14. Dyachok O, Isakov Y, Sågetorp J et al (2006) Oscillations of cyclic AMP in hormone-stimulated insulin-secreting β-cells. Nature 439:349–352 15. Scott JD, Dessauer CW, Tasken K (2013) Creating order from chaos: cellular regulation by kinase anchoring. Annu Rev Pharmacol Toxicol 53:187–210 16. Steyer JA, Almers W (2001) A real-time view of life within 100 nm of the plasma membrane. Nat Rev Mol Cell Biol 2:268–275 17. Dyachok O, Idevall-Hagren O, Sågetorp J et al (2008) Glucose-induced cyclic AMP oscillations regulate pulsatile insulin secretion. Cell Metab 8:26–37 18. Tian G, Sandler S, Gylfe E et al (2011) Glucose- and hormone-induced cAMP oscillations in α- and β-cells within intact pancreatic islets. Diabetes 60:1535–1543 19. Tian G, Sågetorp J, Xu Y et al (2012) Role of phosphodiesterases in the shaping of sub-plasma membrane cAMP oscillations and pulsatile insulin secretion. J Cell Sci 125:5084–5095 20. Li J, Shuai HY, Gylfe E et al (2013) Oscillations of sub-membrane ATP in glucose-stimulated beta cells depend on negative feedback from Ca2+. Diabetologia 56:1577–1586 21. Hansen C, Howlin J, Tengholm A et al (2009) Wnt-5a-induced phosphorylation of

22.

23.

24.

25.

26.

27.

28.

101

DARPP-32 inhibits breast cancer cell migration in a CREB-dependent manner. J Biol Chem 284:27533–27543 Malmersjö S, Liste I, Dyachok O et al (2010) Ca2+ and cAMP signaling in human embryonic stem cell-derived dopamine neurons. Stem Cells Dev 19:1355–1364 Idevall-Hagren O, Barg S, Gylfe E et al (2010) cAMP mediators of pulsatile insulin secretion from glucose-stimulated single β-cells. J Biol Chem 285:23007–23018 Idevall Hagren O, Jakobsson I, Xu Y et al (2013) Spatial control of Epac2 activity by cAMP and Ca2+-mediated activation of Ras. Sci Signal 6:ra29. doi:10.1126/ scisignal.2003932 Miyazaki J, Araki K, Yamato E et al (1990) Establishment of a pancreatic β cell line that retains glucose-inducible insulin secretion: special reference to expression of glucose transporter isoforms. Endocrinology 127:126–132 Schneider CA, Rasband WS, Eliceiri KW (2012) NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9:671–675 Zhao Y, Araki S, Wu J et al (2011) An expanded palette of genetically encoded Ca2+ indicators. Science 333:1888–1891 Akerboom J, Carreras Calderon N, Tian L et al (2013) Genetically encoded calcium indicators for multi-color neural activity imaging and combination with optogenetics. Front Mol Neurosci 6:2. doi:10.3389/fnmol.2013.00002

Chapter 8 Adenoviral Transduction of FRET-Based Biosensors for cAMP in Primary Adult Mouse Cardiomyocytes Oliver Lomas, Marcella Brescia, Ricardo Carnicer, Stefania Monterisi, Nicoletta C. Surdo, and Manuela Zaccolo Abstract Genetically encoded biosensors that make use of fluorescence resonance energy transfer (FRET) are important tools for the study of compartmentalized cyclic nucleotide signaling in living cells. With the advent of germ line and tissue-specific transgenic technologies, the adult mouse represents a useful tool for the study of cardiovascular pathophysiology. The use of FRET-based genetically encoded biosensors coupled with this animal model represents a powerful combination for the study of cAMP signaling in live primary cardiomyocytes. In this chapter, we describe the steps required during the isolation, viral transduction, and culture of cardiomyocytes from an adult mouse to obtain reliable expression of genetically encoded FRET biosensors for the study of cAMP signaling in living cells. Key words Primary cardiomyocyte, Mouse, Adenovirus, cAMP, FRET, Live-cell imaging

1

Introduction Cyclic adenosine monophosphate (cAMP) is a second messenger that mediates many physiological functions, such as the force (inotropy) and frequency (chronotropy) of the contraction of cardiomyocytes [1], phototransduction in the retina [2], and glycogenolysis in the liver [3]. The visualization of real-time changes in cyclic nucleotide signaling within distinct cellular compartments was not possible until the development of live-cell imaging techniques that have improved spatial and temporal resolution over biochemical assays [4]. The first fluorescent biosensors for cAMP were developed in the early 1990s [5] but were superseded by genetically encoded biosensors, principally those that make use of the phenomenon of fluorescence resonance energy transfer (FRET) to measure dynamic, ratiometric changes in cyclic nucleotide signaling [6]. FRET describes the non-radiative transmission of

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_8, © Springer Science+Business Media New York 2015

103

104

Oliver Lomas et al.

energy from a donor to an acceptor fluorophore. This process of energy transfer is highly dependent on the distance between the fluorophores and therefore can be modulated by the conformation of the protein structure to which the fluorophores are bound. In the case of cAMP, sensors have been developed whereby binding of the cyclic nucleotide leads to a change in protein conformation, which modulates the degree of FRET, from which the cAMP concentration can be inferred. The advent of transgenic mouse models that allow the germ line or tissue-specific control of gene expression has made the need to use FRET biosensors in primary mouse cells of great importance. Indeed, animals with transgenic FRET biosensors have been generated [7]. However, this approach is limited by the time-consuming development of a transgenic animal, which is a significant barrier to the use of novel sensor constructs. Adenoviral transduction of genetically encoded FRET biosensors represents a relatively straightforward and flexible approach to using FRET biosensors in primary adult mouse cardiomyocytes. Adenoviral gene transfer is readily achievable in certain cell types, such as the neonatal rat cardiomyocyte—as described in other chapters. However, cardiomyocytes from the adult mouse have proved challenging to infect and maintain in culture for the expression of the biosensor to occur. Adult murine cardiomyocytes are not only difficult to subject to gene transduction, whether plasmid or adenoviral; they also present a significant challenge to maintain in culture long enough to allow expression of the desired fluorescent protein. Pavlović et al. [8] have shown that the rate of loss of contractile function in rodent cardiomyocytes is species dependent, with up to 50 % of mouse cardiomyocytes losing the ability to contract after 24 h in culture, whereas >75 % of rat cardiomyocytes can contract after 48 h of culture. Here, we present a method for the effective transduction of genetically encoded FRET biosensors by adenoviral vectors that allow the investigation of cAMP signaling in adult murine cardiomyocytes in less than 24 h in culture. The chapter provides details of how to isolate adult murine cardiomyocytes, how to prepare the isolation for culture with an adenovirus bearing a FRET biosensor, and when to perform the imaging experiments. This method allows the cost-effective, reproducible, and rapid investigation of cAMP signaling in the adult murine cardiomyocyte.

2

Materials 1. Cover slips—15 mm diameter borosilicate glass. 2. Culture well plates to accommodate cover slips. 3. Laminin (100 μg/ml).

cAMP Imaging in Adult Mouse Cardiac Myocytes

105

Fig. 1 Example Langendorff perfusion apparatus with benchtop microscope and dissection equipment

4. Heparin sodium salt for injection (see Note 1). 5. Langendorff apparatus—including heated water circulation and 100 % oxygen supply (Fig. 1). 6. 2× scissors, one straight Vannas, one standard 11 cm straight blade. 7. 2× forceps, one small to allow tie to be placed around aorta and one large to aid gross dissection of skin and surround tissues. 8. Cannula—21 gauge. 9. 5/0 gauge silk wax suture material—approximately 4 cm in length. 10. 4× 23 gauge needles for limb transfixion—taping of limbs also possible. 11. Adult mouse usually >8 weeks of age for sufficiently large caliber of aorta to allow cannulation. 12. Dissection pad. 13. Benchtop microscope or other magnifying equipment. 14. Supply of calcium-free deionized water.

106

Oliver Lomas et al.

15. Tyrode buffer solution: 130.0 mM NaCl, 5.6 mM KCl, 3.5 mM MgCl2, 5.0 mM HEPES buffer, 10.0 mM glucose, 20.0 mM taurine, 0.4 mM Na2HPO4. pH 7.4, filter sterilized. 16. 1 % bovine serum albumin (BSA)—three 10 ml preparations using the Tyrode buffer described above with calcium-free, 600, and 1,200 μM Ca2+ as three separate solutions. 17. Isolation solution: collagenase and protease enzyme in isolation solution with calcium as per manufacturer’s instructions (see Note 2). 18. Modified Eagle’s medium (MEM) with Hank’s salts, 9 mM NaHCO3, 1 % L-glutamine, 1 % penicillin/streptomycin, 2.5 % solution by volume fetal bovine serum (FBS). 19. Water bath preheated to 37.5 °C. 20. Fine gauze and clip to attach to funnel. 21. Benchtop centrifuge. 22. Plastic collection tubes—10–15 and 50 ml. 23. Funnel to fit in 10–15 ml collecting tube. 24. 3 ml teat pipettes. 25. Adenoviral vector construct of biosensors required. 26. Fluorescence microscopy setup [9].

3

Methods 1. At least 2 h prior to cardiomyocyte isolation, coat sterile borosilicate glass cover slips with laminin (e.g., 100 μl of 100 μg/ ml on a 15 mm cover slip). 2. 20–30 min prior to culling of the mouse, perform an intraperitoneal injection with 5,000 heparin per kg by mass (see Note 1). 3. When materials and Langendorff are prepared (Fig. 1), perform cervical dislocation of an adult mouse according to local rules and protocols (see Note 3). Immediately place the mouse supine on the dissection pad with limbs held out laterally to leave the thorax clearly visible. 4. Perform a midline skin incision using scissors and larger forceps to reveal the sternum. 5. Two lateral thoracic/costal incisions—midaxillary line. 6. Reflect xiphisternum to reveal underlying precordium. 7. Excise sternum and ribs by manubrial incision. 8. Switch to microscope or other magnification equipment magnifying spectacles (loops) to aid visualization of the aorta.

cAMP Imaging in Adult Mouse Cardiac Myocytes

107

Fig. 2 Photograph of the hub of a cannula to show the presence of an air bubble which needs to be removed prior to attachment to the Langendorff perfusion apparatus

9. Excise thymus to reveal great vessels (see Note 4). 10. Place suture deep to the aorta using small forceps. 11. Make a single loose tie around the aorta. 12. Perform a single incision into the aorta, just proximal to the first bifurcation of the aorta using straight Vannas scissors (see Note 5). 13. Insert a 21 gauge cannula into the now open aorta, enough to allow the cannula to rest in place, but not too deep to pass into the left ventricle (see Note 6). 14. Tighten knot around the cannula and apply second tie. 15. Lift knot by pulling the ends of the suture, and use larger scissors to free the heart and cannula from the surrounding tissue. 16. With the heart now attached to the cannula and free from the surrounding tissue, check that any air bubbles in the hub of the cannula are removed (see Note 7 and Fig. 2). 17. Open the tap at the base of Langendorff apparatus and mount cannula and heart (Fig. 3). 18. Allow the Tyrode buffer to perfuse the heart to start with to ensure that perfusion takes place. The first few drops of perfusate should be heavily bloodstained, which will clear within approximately 1–2 min.

108

Oliver Lomas et al.

Fig. 3 Cannulated primary adult mouse heart undergoing Langendorff perfusion with digestive enzyme solution. Note the homogenous pale pink color to the myocardium. In practice, the warming jacket is positioned around the preparation but is lowered in this instance to allow the preparation to be seen clearly

19. When satisfactory perfusion is taking place, switch to a perfusion with digestive enzyme solution (see Note 8 and Fig. 3). 20. Digestion is complete when the drop rate from the perfused heart increases in speed rapidly to >1 drop per second from an initial drop rate of approximately 1 drop every 2–3 s. Collect digestive enzyme solution, as 2–4 ml are required in the next step. 21. Place the heart in the Petri dish with a few milliliters of the collected enzyme solution, and excise atria and right ventricle with fine scissors and forceps. 22. Finely chop the remaining left ventricular tissue with scissors, and place the suspension into a plastic centrifugation tube. Triturate the myocardial tissue using a 3 ml teat pipette for 30 s. 23. Filter through gauze into another centrifugation tube and add Tyrode + 1 % BSA to terminate action of digestive enzymes. 24. Centrifuge lysate ≈40 × g for 3 min with slow acceleration/ deceleration of the rotor (see Note 9). 25. A pellet of cells should be clearly visible after the first centrifugation (see Fig. 4). 26. Remove and discard supernatant and resuspend with Tyrode + 1 % BSA supplemented with 600 μM calcium (see Note 10).

cAMP Imaging in Adult Mouse Cardiac Myocytes

109

Fig. 4 Example of pellet of cells achieved after the first centrifugation of isolation suspension from a heart extracted from a 10-week-old male C57/BL6 mouse

27. Leave to settle at room temperature and pressure for 7 min. During this time, a clearly visible pellet should form, similar but smaller than after step 24. 28. Carefully remove supernatant, and replace with 1 % BSA supplemented with 1.2 µM calcium (see Note 11). 29. Leave to suspension to settle at room temperature for 30 min. 30. After this second wash, remove supernatant and resuspend in MEM culture medium supplemented with 2.5 % FBS and plate on cover slips that have been coated with laminin (see Note 12). 31. After 2 h of recovery and adherence to cover slips, gently remove medium and replace with fresh MEM medium (without FBS) and infect cells with adenovirus of choice using a multiplicity of infection (MOI) 1:100 (see Note 13). 32. After 3 h of the incubation of cells with virus, wash the cells carefully three times with fresh MEM medium, this time supplemented with 0.5 μM cytochalasin D (see Note 14). 33. Incubate at 37 °C in 5 % CO2 with expression within the next 24 h (see Note 15). 34. Conduct experiments in a Tyrode buffer solution with 1.4 mM [Ca2+] (Fig. 5).

110

Oliver Lomas et al.

Fig. 5 Graph to show an experiment involving cytosolic EPAC-based FRET biosensor expressed in a primary adult mouse cardiomyocyte. The yellow and blue lines represent the changes in background-subtracted fluorescence intensity (right axis) of yellow fluorescent protein (YFP) and cyan fluorescent protein (CFP), respectively. The red line represents the CFP/YFP ratio, which has been normalized to the baseline during the first 240 s of the experiment (left axis). The dynamic range of the biosensor is revealed by the addition of agents that maximally elevate the intracellular concentration of cAMP, forskolin to activate adenylyl cyclase to increase cAMP production, and IBMX to inhibit phosphodiesterases that hydrolyze cAMP. Upon addition of forskolin and IBMX, the rising cAMP binds the EPAC-based sensor and changes its conformation such that the CFP and YFP fluorophores separate. The separation of fluorophores results in a loss of FRET between the fluorophores. Therefore, excitation of CFP results in greater CFP emission at the expense of FRET to the YFP, with a concomitant fall in YFP intensity. The reversibility of the process demonstrated the rapid reversal of the fluorescence intensities of YFP and CFP to baseline, upon removal (wash) of the forskolin and IBMX

4

Notes We would like to emphasize the need to be fastidious with the isolation of adult cardiomyocytes as the better the initial yield of cells, the greater the number and quality of cells obtained, ultimately improving the reliability of the subsequent fluoroscopic experiments. In particular, any glassware should be washed thoroughly in deionized, calcium-free water to prevent contamination with calcium which can damage the cardiomyocytes during the isolation process.

cAMP Imaging in Adult Mouse Cardiac Myocytes

111

1. The anticoagulant heparin is used to prevent the formation of blood clots that may cause myocardial infarction [10]. The procedure can be performed in the absence of heparin, but the time from circulatory status to perfusion by the Langendorff apparatus is shortened to less than 5 min rather than up to 9 min when heparin is used. 2. Numerous tissue dissociation enzyme products are available. In our experience, we have found Liberase® TH Research Grade (Roche) to be highly effective and reliable, and it is associated with good expression of fluorophore. We use a concentration of 1 mg Liberase in 20 ml of Tyrode (50 μg/ml) with calcium at 100 μM. 3. Use a manual technique of cervical dislocation to avoid trauma to thoracic blood vessels. By avoiding significant hemorrhage prior to the division of the aorta, there is a clearer surgical field and less physiological stress to the preparation, which ultimately increases the chances of a good yield of fluorescent cells after viral infection. 4. In situ aortic cannulation is described in this chapter. Ex situ cannulation of the aorta is possible and has been used to generate good quality isolations of cardiomyocytes. Be familiar with the anatomy of the great vessels of the mouse. The aorta is located centrally, between the superior vena cava and the pulmonary vein. 5. The use of Vannas scissors with straight blades is very important as the incision in the aorta needs to be perpendicular to the vessel wall. A clean cut of the aorta facilitates the insertion of the cannula into the aorta. 6. Leave 1–2 mm of aorta visible above the aortic root and make sure that the tip of the cannula remains visible. If the cannula has entered the left ventricle, the principal of Langendorff retrograde perfusion via the coronary arteries is lost and the isolation fails. Equally, division of the aorta too close to the first bifurcation can allow the vessel to collapse, which delays cannulation or makes it impossible. 7. Irrigate the cannula with Tyrode buffer to eliminate air bubbles that would otherwise interrupt the flow of perfusate and hence severely hinder the isolation of cardiomyocytes (Fig. 2). 8. During perfusion, the myocardium should remain pale pink in color; if it becomes white, infarction has occurred and the isolation procedure has failed. 9. The centrifugation must be gentle. If the rotor speed or its rate of acceleration is too high, the cardiomyocytes become severely damaged. Gravity sedimentation may be used throughout the isolation procedure, but it is slower compared to centrifugation.

112

Oliver Lomas et al.

Fig. 6 Freshly isolated adult mouse cardiomyocytes after serial washing Tyrode solutions containing 1 % bovine serum albumin (BSA) and a rising concentration of calcium from one wash to the next

10. The aim of the serial washes is twofold. First is to allow gravitational separation of the heavier viable cells from the fibroblasts and lysed, dead cardiomyocytes which remain in suspension. Secondly, the rising concentration of calcium aims to prevent toxic calcium overload of the cells which occurs on placing the cells in the MEM culture medium which has a calcium concentration of 1.8 mM. 11. At this point of resuspension, it is a good opportunity to count the cells using a hemocytometer. From our practice, we obtain approximately 1.2 million viable cells from an excellent isolation, which represents about 80 % of the total cellular population (Fig. 6). The remaining 20 % of rounded, distorted cells are considered to be nonviable cardiomyocytes. 12. The rationale for the use of MEM-based culture medium is based upon observations by Li et al. [11] in which the expression of the Coxsackie adenovirus receptor (CAR), responsible for allowing entry of the adenovirus into the cardiomyocyte, is enhanced compared to other media such as M-199. 13. In order to drive the expression of the large proteins that constitute the FRET biosensors, a high ratio of viral particle to cardiomyocyte (multiplicity of infection or MOI) of 100 is required. For example, 1 μl (1 × 10−6 l) of a viral preparation with a titer of 1010 viral particles per ml is applied to culture of 100,000 (105) cells, i.e., a ratio of 107 viral particles to 105 cells. 14. Cytochalasin D in sub-micromolar concentrations has been found to preserve structural integrity and contractile function in adult rat cardiomyocytes [12]. The principal mechanism of action is thought to be via actions on the cytoskeleton that

cAMP Imaging in Adult Mouse Cardiac Myocytes

113

Fig. 7 Graphs to show changes in (a) fractional shortening (FS). n = 3 animals 9–18 cells per group + 95 % confidence interval. Two-tailed unpaired T-test performed. Fresh FS = 4.7 % ± 0.54 (SEM), n = 12. Cells cultured with virus alone FS = 3.1 % ± 0.31 (SEM), n = 9. Cells cultured with virus and cytochalasin D FS = 5.1 % ± 0.41 (SEM), n = 18. Fresh vs culture with virus alone—p = 0.04 (*). Fresh vs culture with virus and cytochalasin D—p = 0.42 (b) return velocity (RV). n = 3 animals 9–18 cells per group + 95 % confidence interval. Two-tailed unpaired T-test performed. Fresh RV = 1.4 ± 0.2 (SEM), n = 12. Cells cultured with virus alone RV = 1.2 ± 0.3 (SEM), n = 9. Cells cultured with virus and cytochalasin D RV = 1.7 ± 0.3 (SEM), n = 18. Fresh vs culture with virus alone—p = 0.64

prevents the loss of T-tubules that has been observed in cultured rodent ventricular cardiomyocytes [8]. We sought to identify whether this applies to adult mouse cardiomyocytes cultured with adenoviral preparations encoding a FRET biosensor for cAMP. Isolated cells, adherent to laminin-coated cover slips were subjected to 1 Hz electrical field stimulation at 30 °C, and the characteristics of fractional shortening and return velocity were recorded in three groups: freshly isolated cells, cultured cells with virus alone, and cultured cells with virus and 0.5 μM cytochalasin D (Fig. 7). Viral infection with the cAMP FRET biosensor was verified by the observation of an appropriate FRET response to the addition of forskolin, an activator of adenylyl cyclase, and IBMX, a specific, nonselective phosphodiesterase inhibitor. When adult mouse cardiomyocytes are cultured with adenovirus alone, a decrease in fractional shortening is observed compared to freshly isolated cells. However, this effect is abolished in the presence of 0.5 μM cytochalasin D (Fig. 7a). The rate of relaxation of cardiomyocytes, as measured by the “return velocity,” is similar across all groups with no significant difference between the groups (Fig. 7b). From these observations, 0.5 μM cytochalasin D is associated with improved cardiomyocyte contractile properties without affecting viral transduction efficiency. 15. The timing of imaging experiments depends upon the intensity of the fluorophore. Constructs with greater numbers of

114

Oliver Lomas et al.

Fig. 8 Both images acquired using the same microscopy apparatus with the same excitation light intensity and an exposure time of 100 ms. (a) Adult murine cardiomyocytes at 22 h post isolation exhibiting autofluorescence. (b) Membranetargeted EPAC-based sensor at 22 h post isolation showing plasma and nuclear membrane localization

fluorescent subunits or brighter fluorophores will be observed earlier than those that are less bright. When assessing for fluorescence, be aware of the localization of the sensor as a guide to whether the sensor is being expressed or the cell is exhibiting autofluorescence (see Fig. 8). NADPH has long been recognized as being autofluorescent [13], which can pose a problem with FRET biosensors as the emission spectrum of NADH is 477 nM, which is similar to YFP-Venus (a common fluorophore used in FRET biosensors) at 530 nM [14]. The pattern of NADH-based autofluorescence shows longitudinal striations to the cell, which differ from the classical transverse striations of cardiomyocytes.

Acknowledgments This study was supported by the British Heart Foundation (PG/10/75/28537 and RG/12/3/29423) and the NSF-NIH CRCNS program (NIH R01 AA18060) to MZ and Wellcome Trust PhD Programme for Clinicians at the University of Oxford to OCL. References 1. Tsien RW (1977) Cyclic AMP and contractile activity in heart. Adv Cyclic Nucleotide Res 8:363–420 2. Miki N, Baraban JM, Keirns JJ, Boyce JJ, Bitensky MW (1975) Purification and properties of the light-activated cyclic nucleotide

phosphodiesterase of rod outer segments. J Biol Chem 250(16):6320–6327 3. Northrop G, Parks RE Jr (1964) 3′, 5′-Ampinduced hyperglycemia in intact rats and in the isolated perfused rat liver. Biochem Pharmacol 13:120–123

cAMP Imaging in Adult Mouse Cardiac Myocytes 4. Zaccolo M, Pozzan T (2002) Discrete microdomains with high concentration of cAMP in stimulated rat neonatal cardiac myocytes. Science 295(5560):1711–1715. doi:10.1126/ science.1069982 5. Adams SR, Harootunian AT, Buechler YJ, Taylor SS, Tsien RY (1991) Fluorescence ratio imaging of cyclic AMP in single cells. Nature 349(6311):694–697 6. Zaccolo M, De Giorgi F, Cho CY, Feng L, Knapp T, Negulescu PA, Taylor SS, Tsien RY, Pozzan T (2000) A genetically encoded, fluorescent indicator for cyclic AMP in living cells. Nat Cell Biol 2(1):25–29. doi:10.1038/71345 7. Nikolaev VO, Bunemann M, Schmitteckert E, Lohse MJ, Engelhardt S (2006) Cyclic AMP imaging in adult cardiac myocytes reveals farreaching beta1-adrenergic but locally confined beta2-adrenergic receptor-mediated signaling. Circ Res 99(10):1084–1091. doi:10.1161/01. RES.0000250046.69918.d5 8. Pavlovic D, McLatchie LM, Shattock MJ (2010) The rate of loss of T-tubules in cultured adult ventricular myocytes is species dependent. Exp Physiol 95(4):518–527. doi:10.1113/expphysiol.2009.052126 9. Gesellchen F, Stangherlin A, Surdo N, Terrin A, Zoccarato A, Zaccolo M (2011) Measuring spatiotemporal dynamics of cyclic

10.

11.

12.

13.

14.

115

AMP signaling in real-time using FRET-based biosensors. Methods Mol Biol 746:297–316. doi:10.1007/978-1-61779-126-0_16 Louch WE, Sheehan KA, Wolska BM (2011) Methods in cardiomyocyte isolation, culture, and gene transfer. J Mol Cell Cardiol 51(3): 288–298. doi:10.1016/j.yjmcc.2011.06.012 Li Z, Sharma RV, Duan D, Davisson RL (2003) Adenovirus-mediated gene transfer to adult mouse cardiomyocytes is selectively influenced by culture medium. J Gene Med 5(9):765–772. doi:10.1002/jgm.405 Tian Q, Pahlavan S, Oleinikow K, Jung J, Ruppenthal S, Scholz A, Schumann C, Kraegeloh A, Oberhofer M, Lipp P, Kaestner L (2012) Functional and morphological preservation of adult ventricular myocytes in culture by sub-micromolar cytochalasin D supplement. J Mol Cell Cardiol 52(1):113–124 Eng J, Lynch RM, Balaban RS (1989) Nicotinamide adenine dinucleotide fluorescence spectroscopy and imaging of isolated cardiac myocytes. Biophys J 55(4):621–630. doi:10.1016/s0006-3495(89)82859-0 Nagai T, Ibata K, Park ES, Kubota M, Mikoshiba K, Miyawaki A (2002) A variant of yellow fluorescent protein with fast and efficient maturation for cell-biological applications. Nat Biotechnol 20(1):87–90. doi:10.1038/nbt0102-87

Chapter 9 Generation of Transgenic Mice Expressing FRET Biosensors Daniela Hübscher and Viacheslav O. Nikolaev Abstract Transgenic mice play a significant role in modern biomedical research. They allow not only mechanistic insights into the functions of specific genes and proteins. Recent strategies have also established the use of transgenic mice as an exciting tool for the expression and in vivo or in situ analysis of fluorescent biosensors, which are capable of directly reporting second messenger levels and biochemical processes in real time and in living cells. In this chapter, we present a detailed protocol for the generation of plasmid vectors and transgenic mice expressing a Förster resonance energy transfer (FRET)-based biosensor for the second messenger 3′,5′-cyclic adenosine monophosphate. These tools and techniques should provide great potential for the analysis of second messenger dynamics in a more physiologically relevant context. Key words Transgenic mice, FRET, Cardiomyocyte, cAMP, Biosensor

1

Introduction Transgenic mice play an important role in contemporary biomedical research. Many biological and medical questions can be addressed in transgenic animals. For example, they allow detailed analysis of individual gene functions and can be used as genetic animal models of human disease [1–3]. Generally, a specific exogenous DNA fragment can be integrated into a host animal genome to overexpress, to knock in, or to knock out a specific gene in a constitutive or regulated fashion [4]. There are many approaches to establish transgenic animals, one of which is the microinjection of DNA [5–7]. Murine zygotes before the cell fusion can be harvested and manipulated with specific external DNA by injecting it in the male pronucleus. In this way, the external DNA can be permanently integrated into the host genome. However, the position where the external DNA is integrated is random. The injected zygotes are transferred into a surrogate mother, and some of the pups born carry the transgene integrated into their genome.

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_9, © Springer Science+Business Media New York 2015

117

118

Daniela Hübscher and Viacheslav O. Nikolaev

Apart from using transgenic mice to study gene function, they can be an exciting tool for the expression of fluorescent biosensors capable of directly reporting second messenger levels and biochemical processes in living cells. For example, they have enabled direct visualization of the second messenger 3′,5′-cyclic adenosine monophosphate (cAMP) and its real-time monitoring in adult cardiomyocytes [8, 9]. Such biosensors are based on the principle of Förster resonance energy transfer (FRET) which occurs between the donor and acceptor fluorophores (e.g., cyan (CFP)- and yellow (YFP)-fluorescent proteins) when they come into close spatial proximity [10]. The simplest cAMP biosensor Epac1-camps is comprised of a cAMP-binding domain (derived, e.g., from the Epac1 protein) sandwiched between CFP and YFP [11, 12]. cAMP binding leads to a conformational change in the sensor molecule which can be visualized by a decrease of the FRET signal. In this chapter, we describe a detailed protocol for the generation of plasmid vectors and transgenic mice expressing the cAMP biosensor Epac1-camps in a tissue-specific and ubiquitous manner.

2 2.1

Materials Equipment

1. Thermocycler (e.g., SensoQuest, Göttingen). 2. Regular DNA electrophoresis chamber and power supply. 3. Gel documentation system. 4. Laboratory scale. 5. Microcentrifuge (e.g., Pico 17, Thermo Scientific). 6. Thermomixer (e.g., Eppendorf). 7. NanoDrop 2000 device (Thermo Scientific) or regular laboratory photometer. 8. FRET imaging system comprised of an inverted microscopy, light source, Dual-View beam splitter, CCD camera, and imaging software. An example can be found in a previously described comprehensive protocol [11].

2.2

Mice

2.3

Vectors

FVB/NRj mice (Janvier Labs, Saint-Berthevin, France). 1. α-MHC vector, empty [9]. 2. CAG vector, empty [13, 14]. 3. Epac1-camps plasmid [12].

2.4

Cells

1. Competent E. coli cells—Top 10 (C4040-10, Invitrogen). 2. 293A cells (R70507, Invitrogen).

Transgenic Mice Expressing FRET Reporters

2.5 Substances and Solutions

119

1. Agarose peqGOLD Universal (35-1020, Peqlab). 2. Ampicillin: 100 mg/ml stock solution in aqua bidest. 3. DirectPCR Tail (31-102T, PeqLab). 4. dNTPs: 100 mM (U1240, Promega) in nuclease-free water. 5. Ethidium bromide 1 % solution in water. 6. GoTaq DNA polymerase and GoTaq buffer (M3175, Promega). 7. KCM 5×: 500 mM potassium chloride, 150 mM calcium chloride, 250 mM magnesium chloride dissolved in nuclease-free water. 8. LB agar: powder by Miller (A0927, Applichem), dissolve in deionized water and autoclave. After cooling the solution down to ~50 °C, add the desired antibiotic (e.g., 0.1 mg/ml ampicillin) and mix well. Pour this solution into 10 cm petri dishes, cool down until the agar becomes solidified, and store at 4 °C. 9. LB medium: powder by Miller (A0954, Applichem), dissolve in deionized water, autoclave, and store at 4 °C. 10. Loading dye buffer DNA IV (A3481.0025, Applichem). 11. Midori Green advance (MG05, Nippon Genetics Europe GmbH). 12. TAE buffer for DNA electrophoresis (A1416, Applichem). 13. Proteinase K (A3830.0500, Applichem): 100 mg/ml solution in deionized water and store the aliquots at −20 °C. 14. Pfu polymerase and 10× Pfu buffer (M7745, Promega). 15. Quick-Load DNA Ladder (N0467S, New England Biolabs). 16. T4 DNA ligase and buffer (M0202S, New England Biolabs). 17. TE buffer: 5 mM Tris–HCl, pH 7.4, 0.1 mM EDTA. 18. Isopropanol. 19. Isoproterenol (I6504, Sigma). Prepare a 10 mM stock solution in water, aliquot, and store at −20 °C until use. 20. Nuclease-free water. 21. Transfection reagent Lipofectamine 2000 (11668-027, Life Technologies).

2.6

Kits

1. QIAquick Gel Extraction Kit (28706, Qiagen). 2. QIAfilter Plasmid Midi Kit (12243, Qiagen).

2.7

Primers

All primers were ordered from Eurofins MWG Operon, dissolved in deionized water upon arrival at final concentration of 100 pmol/ μl, and subsequently stored at −20 °C. For cloning of the CAG-Epac1-camps vector (italic = linker sequence): Forward: 5′ AAA GAT ATC TGC AGC GCC ACC ATG GTG AGC AAG GGC G 3′

120

Daniela Hübscher and Viacheslav O. Nikolaev

Reverse: 5′AAA GAA TTC CTT GTA CAG CTC GTC CAT GC 3′ For cloning of the α-MHC-Epac1-camps vector (italic = linker sequence): Forward: 5′ AAA GAT ATC ATG GTG AGC AAG GGC G 3′ Reverse: 5′AAA GAA TTC CTT GTA CAG CTC GTC CAT GC 3′ For genotyping: Forward: 5′ TGA CAG ACA GAT CCC TCC TAT 3′ Reverse: 5′ CAT GGC GGA CTT GAA GAA GT 3′ 2.8 Restriction Enzymes

All restriction enzymes and buffers were from New England Biolabs and used in the following buffers, according to the manufacturer’s protocol. BamHI (R3136S): buffer 3 + bovine serum albumin (BSA). EcoRI-HF (R3101S): buffer 4. EcoRV-HF (R3195S): buffer 4. KpnI-HF (R3142S): buffer 4. SpeI (R0133S): buffer 4 + BSA. XhoI (R0146S): buffer 4 + BSA.

2.9 Other Materials for DNA Purification

3 3.1

1. Slide-A-Lyzer Dialysis Cassette (66383, Thermo Scientific). 2. 0.2 μm filter FP 013/AS (462945, Schleicher & Schuell).

Methods Cloning Strategy

To generate the ubiquitous transgenic Epac1-camps mice, we used the empty CAG vector (Fig. 1) as a backbone for the transgenic construct [13, 14]. CAG acts as a constitutively active promoter (any other well-established constitutively active promoter can be used instead), so that the genes downstream the promoter are transcribed in virtually every cell type. In this vector, a multiple cloning site is provided with several single-cutter restriction sites. For selection in bacteria, a gene encoding for ampicillin antibiotic resistance is also located in the vector backbone. As template DNA, we used the previously described Epac1camps sensor plasmid (Fig. 1), which includes the cAMP-binding site flanked by two fluorophores for FRET measurements [12]. Subcloning can be performed in one step but with two separate inserts which need to be ligated into one vector. First, we cut out the first insert DNA fragment (Epac-Cfp) with the restriction enzymes EcoRI and XhoI. These enzymes cut the DNA just before the Epac and after the Cfp fluorophore sequence. Second, the Yfp

Transgenic Mice Expressing FRET Reporters

121

Fig. 1 Cloning strategy for the CAG-Epac1-camps construct. The empty CAG vector serves as a backbone. It is cut with the restriction enzymes EcoRV and XhoI. As an insert, two fragments are ligated into this backbone. They include the Epac-Cfp fragment cut out from the Epac1-camps plasmid with EcoRI and XhoI and the Yfp sequence amplified by PCR and digested using EcoRV and EcoRI

sequence was amplified by the polymerase chain reaction (PCR). We designed the forward primer which includes the EcoRV restriction site and a linker sequence in front of the Kozak sequence and the start codon (ACC ATG). The reverse primer contained an EcoRI restriction site which can be ligated together with the EpacCfp fragment. The CAG backbone vector was digested with the restriction enzymes EcoRV and XhoI. The vector backbone was then used for a triple ligation with the digested Yfp and Epac-Cfp fragments (Fig. 1). For transgenic mouse generation, the resultant construct was digested with the restriction enzymes SpeI and BamHI and ran in a 1 % agarose gel. The upper band (2,467 b.p.) was cut out and further used (see Subheading 3.7), while the lower two bands (2,010 and 337 b.p., which contain the antibiotic resistance and other unneeded sequences) were discarded. To prove the functionality of the new construct, we transfected it into 293A cells using Lipofectamine 2000, according to the manufacturer’s protocol. 24 h after transfection, these cells showed fluorescence and measurable FRET responses to the β-adrenergic receptor agonist isoproterenol, indicative of an increase in intracellular cAMP (Fig. 2). For generating transgenic animals with cardiac-specific expression of Epac1-camps, a tissue-specific promoter is required. Alpha myosin heavy chain (α-MHC) is a contractile protein expressed in

122

Daniela Hübscher and Viacheslav O. Nikolaev

Fig. 2 Proof of functionality of the CAG-Epac1-camps construct. (a) shows the fluorescence of transfected 293A cells with the CAG-Epac1-camps vector. A FRET measurement is shown in (b). The CFP/YFP ratio is increasing after the addition of isoproterenol (Iso) which is indicative of a cAMP increase in cells

Fig. 3 Cloning scheme of the α-MHC-Epac1-camps construct. The empty α-MHC vector serves as a backbone. It is cut with the restriction enzymes KpnI and XhoI. As an insert, two fragments are ligated into this backbone. They include the Epac-Cfp fragment cut out from the Epac1-camps plasmid with EcoRI and XhoI and the Yfp sequence amplified by PCR and digested using KpnI and EcoRI

adult cardiomyocytes, and its promoter is routinely used for tissue-specific expression in adult myocardium. To clone such a construct, we choose the α-MHC vector as a backbone (Fig. 3) [9]. We digested this vector with KpnI and XhoI. The restriction sites are located behind the α-MHC promoter and before the

Transgenic Mice Expressing FRET Reporters

123

Fig. 4 Example of a genotyping PCR analysis. Shown are representative results from a genotyping PCR analysis of transgenic founder offspring. Mice with the numbers 1, 3, 4, and 7 contain the Epac1-camps construct in their genome. NC = negative control, PC = positive control, bp = base pairs

poly-A tail. The Epac1-camps vector was digested as described above with EcoRI and XhoI to excise the Epac-Cfp fragment. The Yfp sequence was again amplified by PCR. In this case, the KpnI restriction site was introduced in front of the start codon and an EcoRI restriction site after the Yfp sequence to triple ligate it with Epac-Cfp fragment and the α-MHC vector backbone. The resulting vector α-MHC-Epac1-camps (Fig. 3) was control digested with BamHI. This enzyme has three restriction sites in this vector, so three bands are detectable on a gel. For microinjections, we digested the α-MHC-Epac1-camps construct with the restriction enzyme SpeI and ran in a 1 % agarose gel. The upper band (7,885 b.p.) was cut out and further used (see Subheading 3.7), while the lower band (2,916 b.p., which contains the antibiotic resistance and other unneeded sequences) was discarded. The linearized and purified DNA sequences were injected into the male pronuclei of the FVB/N mouse zygotes. Microinjections are normally performed by transgenic mouse facilities which we supply with the specially prepared transgenic construct DNA (see Subheading 3.7). Please, refer to a previously published comprehensive protocol if you need to establish this procedure in your own laboratory [15]. The genotypes of the newborn mice were analyzed by PCR (Fig. 4). The primers for the genotyping selected to bind both parts of the ligated constructs. We choose a forward primer which binds in the α-MHC promoter sequence and the reverse primer which binds in the middle of the Yfp gene. Mice where the construct was integrated in the host genome were classified as founder and used for further breedings. The integration of the external DNA is random. Therefore, the F1 generation has to be analyzed by PCR and FRET measurements to test the functionality of these construct in the descendants. We usually chose one or two founders within the F1 generation showing robust sensor fluorescence and FRET responses. This founder is then selected to establish a new transgenic mouse line. Below, the cloning steps are described in more detail.

124

Daniela Hübscher and Viacheslav O. Nikolaev

3.2 PCR and DNA Extraction

The specific DNA fragment to be integrated into the host DNA must first be amplified by a polymerase chain reaction (PCR). Therefore, the required DNA fragment is amplified using specific primer pairs. 1. For the PCR mix, pipette 10 μl Pfu buffer, 2 μl dNTPs (10 mM), 2.5 μl each primer forward and reverse (pre-diluted to 10 pmol/μl), 0.5 μl template DNA (0.1–0.3 μg), 83.5 μl nuclease-free water, and 1 μl Pfu polymerase in a PCR reaction tube. Prepare this mix on ice. 2. Mix carefully and spin down the solution. 3. Start the following program in a thermocycler: 94 °C—3 min. 94 °C—15 s.

55 °C—15 s

×30 cycles.

72 °C—1.2 min (see Note 1). 72 °C—7 min. 4 °C—∞. 4. Purify the PCR product by running the PCR product in a 1 % agarose gel supplemented with Midori Green (4 μl/50 ml of the gel) or ethidium bromide stock solution (3 μl/50 ml of the gel). 5. Mix 100 μl PCR product with 10 μl loading dye and transfer all on the agarose gel. Start the electrophoresis for 30–60 min at 100 V. 6. Cut out the specific DNA fragment under UV light (see Note 2), determine the weight of the gel piece, and transfer it into a 1.5 ml reaction tube. 7. Extract the DNA out of the gel piece with the QIAquick Gel Extraction Kit. Add a threefold amount of the QG buffer to the gel piece and melt it in a thermomixer at 50 °C. Add onefold amount of isopropanol to the reaction sample and transfer it to a spin column. Follow the instruction provided by the manufacturer. Elute in 40 μl nuclease-free water. For insertion of a specific DNA fragment in a target vector, both components have to be digested with the same restriction enzymes. Since restriction enzymes generate “sticky ends” (in the majority of cases), the specific DNA fragment (insert) can be easily ligated into the target vector.

Transgenic Mice Expressing FRET Reporters

3.3 Restriction Digest and Ligation

125

1. Prepare two reaction samples: (a) 6 μg of the target vector (fill up to 40 μl with nuclease-free water) and (b) 40 μl of the insert (PCR product; see above) in a reaction tube. 2. Add 5 μl of a 10× restriction buffer specific to the selected restriction enzyme to each preparation. Add 0.5 μl 100× BSA, if required by the restriction enzyme. 3. Add 2.5 μl of each of the two reaction enzymes required. 4. Mix the samples and incubate it for at least 2 h at 37 °C in a thermomixer. 5. Visualize the DNA digestion on a 1 % agarose gel. Cut out the desired fragments under UV light. 6. Purify the DNA with the QIAquick Gel Extraction Kit. Elute the vector in 50 μl and the insert in 25 μl EB buffer. 7. For ligation, mix 11.5 μl of the digested insert and 1 μl of the linearized vector (corresponds to a ~5:1 insert/vector ratio) in a separate reaction tube. 8. Add 1.5 μl of 10× ligase buffer and 1 μl of the T4 DNA ligase. 9. Mix the sample and incubate for at least 2 h (optimally overnight) at 14 °C. For amplification of the ligated reaction sample, transform them into competent E. coli cells. Only bacteria, which absorbed a correctly ligated vector, are able to grow under antibiotic specific pressure.

3.4

Transformation

1. To transform the ligation reaction product into competent E. coli cells, add 65 μl nuclease-free water and 20 μl of the 5× KCM buffer to the ligation sample. 2. Store this preparation for 5 min on ice. 3. Then add 100 μl of competent E. coli cells, mix gently, and incubate 20 min on ice and afterwards 10 min at room temperature. 4. Add 1 ml LB medium (without antibiotic) and incubate for 50 min in a thermomixer at 37 °C by 700 rpm. 5. Centrifuge the suspension for 2 min at 2,400 × g. 6. Collect 100 μl of the supernatant and remove the rest of it. Resuspend the pellet with 100 μl of the supernatant and then transfer and spread the suspension on LB agar plates (with ampicillin). Incubate the plates by 37 °C overnight. 7. Pick individual bacteria colonies with a sterile pipette tip and transfer it into a 15 ml reaction tube with 3 ml LB medium. 8. Add the desired antibiotic in a 1:1,000 dilution. 9. Incubate this suspension at 37 °C in a shaker overnight.

126

Daniela Hübscher and Viacheslav O. Nikolaev

To harvest the vector DNA, purify the plasmid DNA from bacteria by using QIAfilter Plasmid Midi Kit (Qiagen) according to the manufacturer’s protocol. Elute the DNA with 100 μl of nuclease-free water. To make sure that you have picked a positive colony with the correctly integrated construct, perform a control digest. 3.5

Control Digest

1. Take 8 μl out of a plasmid DNA sample and add 1 μl of 10× specific restriction enzyme buffer. 2. Add two restriction enzymes (0.5 μl each) to the reaction tube, which leads to a specific banding pattern after digestion. 3. Incubate the tubes for at least 30–60 min at 37 °C. 4. Add 1.5 μl loading dye to each sample and visualize the specific banding pattern on a 1 % agarose gel via electrophoresis. Positive samples can be amplified to obtain higher amount of plasmid.

3.6 Proliferation of the Constructed Vector

1. Take 100 μl out of the bacterial suspension after transformation (Subheading 3.4, step 4) and inoculate 50–60 ml LB media containing the specific antibiotic. Incubate this bacteria suspension at 37 °C in a shaker overnight (see Note 3). 2. Centrifuge the whole suspension at 11,385 × g and 4 °C for 5 min. 3. Remove the supernatant and purify the DNA with the QIAfilter Plasmid Midi Kit following the manufacturer’s protocol. This generated purified plasmid DNA could be now used for different applications. For generating transgenic animals via microinjection, the DNA has to be linearized and purified in a special way.

3.7 Preparation of the DNA for the Microinjection

1. Linearize 40 μg of the DNA with a specific restriction enzyme (see Note 4). 2. Separate the linearized DNA on a 1 % agarose gel, 30–60 min at 80–100 V. 3. Cut out the band of the right size under UV light (see Note 5). 4. Purify the gel pieces by the QIAfilter Plasmid Midi Kit. Use 4 spin columns and elute each in 100 μl sterile TE buffer. 5. Sterile filter the DNA with a 1 ml syringe and a 0.2 μM filter carefully (see Note 6). 6. For effective removal of salts, ethidium bromide, and further contaminations, the DNA has to be dialyzed. Use the Slide-ALyzer cassettes (Thermo Scientific) according to the manufacturer’s protocol (see Note 7). 7. Measure the concentration of the dialyzed DNA with a NanoDrop device or a photometer.

Transgenic Mice Expressing FRET Reporters

127

8. Adjust the concentration of the sample with sterile TE buffer. For pronucleus injection, a DNA concentration of ~30 ng/μl in TE buffer and a minimum volume of 200 μl have to be submitted to the transgenic mouse unit. This is usually diluted down to 3 ng/μl with the same buffer for injections. After injections and embryo transfer, the mouse pups are genotyped at the age of 3–4 weeks to identify the founder animals for the further breeding. 3.8

Genotyping

1. Take tail biopsies (by cutting a 3–5 mm piece of tail) from mice aged 3–4 weeks and store them at −20 °C. 2. Extract the DNA out of the tissue with 190 μl direct PCR tail lysis buffer and 10 μl Proteinase K in the thermomixer at 55 °C overnight under vigorous mixing. 3. Heat the samples for 45 min to 85 °C without shaking. 4. Cool down the samples for 30 min at 4 °C. 5. Pipette 0.5 μl of the sample in a PCR reaction tube and add 14.7 μl nuclease-free water, 4 μl 5× GoTaq buffer, 0.5 μl dNTPs, 0.05 μl of the forward primer, 0.05 μl of the reverse primer (both at 100 pmol/μl concentration), and 0.2 μl of the GoTaq polymerase (see Notes 8 and 9). Prepare these steps on ice. 6. Mix carefully and spin down the reaction mix. 7. Start the PCR in the thermocycler: 94 °C—4 min. 94 °C—30 s. 56 °C—30 s

×35 cycles.

72 °C—1.2 min. 72 °C—7 min. 4 °C—∞. 8. Analyze the results on a 1 % agarose gel. Usually, about 20–30 % of the newly generated mice are PCR positive. Positively identified founders can then be used for further breedings with wild-type mice to establish several heterozygous transgenic mouse lines. The functionality of the construct in the F1 generation can be analyzed by FRET measurements in single isolated cells. Founders which give rise to newborns with a sufficient amount of fluorescence and good FRET responses can be chosen to establish new transgenic mouse lines. There might be a certain variability in the percentage of fluorescent cells and in the intensity of fluorescence. We normally prefer working with the lines in which virtually all desired cells are brightly fluorescent to obtain good signal-to-noise ratios in FRET experiments.

128

4

Daniela Hübscher and Viacheslav O. Nikolaev

Notes 1. The PCR elongation time is calculated as 2 s per kilobase of DNA. 2. Make sure that you switch on the UV light only for a very short time. Otherwise, strand disruptions may occur. 3. Check the cloudiness of the suspension. For optimal results, the suspension should become cloudy and have an optical density above 1. 4. Due to high amount of DNA, mix this reaction in the total volume of 200 μl. Add water at 160 μl to the DNA, 20 μl of the restriction enzyme buffer, and 20 μl of the restriction enzyme or enzyme mixture. 5. Make sure that you only use the UV light for a very short time. Otherwise, strand disruptions could occur. 6. Rinse the reaction tube, the syringe, and the filter with 50 μl TE buffer to make sure that you do not lose DNA. Pipette and filtrate very carefully; otherwise, the DNA may shear. 7. The membrane of the dialysis cassette has to be hydrated in the TE buffer for 30 s before the sample is loaded into the cassette. Make sure that you use an 18–19 gauge needle for adding the sample; otherwise, the DNA could shear. Dialyze for 2 h at room temperature in TE buffer, change the dialysis buffer, and dialyze for another 2 h. Change again the dialysis buffer and dialyze overnight at 4 °C. 8. It is advisable to prepare a master mix containing all PCR components except for the tissue DNA. Upscale the quantities of the components according to the number of samples and pipette 19.5 μl of master mix to each DNA sample right before starting the reaction. 9. The primer pair should include parts of the original vector and of the insert. Each primer should have a base pair length between 15 and 30, resulting in a melting temperature of around 55 °C. Avoid presence of repeats containing more than four identical nucleotides, especially guanosine phosphates. Use also a positive (plasmid or transgenic mouse DNA) and a negative (water or a wild-type mouse DNA) control for the PCR reaction.

Acknowledgment The work in authors’ laboratories is supported by the Deutsche Forschungsgemeinschaft (grants NI 1301/1, FOR 2060, SFB 1002 TP A01, IRTG 1816), the German Centre for Cardiovascular Research (DZHK), and the University of Göttingen Medical Center (“pro futura” grant).

Transgenic Mice Expressing FRET Reporters

129

References 1. Klohs J, Rudin M, Shimshek DR et al (2014) Imaging of cerebrovascular pathology in animal models of Alzheimer’s disease. Front Aging Neurosci 6:32 2. Zheng Y, Kong Y, Li F (2014) Kruppel-like transcription factor 11 (KLF11) overexpression inhibits cardiac hypertrophy and fibrosis in mice. Biochem Biophys Res Commun 443: 683–688 3. Levy E, Spahis S, Garofalo C et al (2014) Sar1b transgenic male mice are more susceptible to high-fat diet-induced obesity, insulin insensitivity and intestinal chylomicron overproduction. J Nutr Biochem 25:540 4. Scharfenberger L, Hennerici T, Kiraly G et al (2014) Transgenic mouse technology in skin biology: generation of complete or tissuespecific knockout mice. J Invest Dermatol 134:e16 5. Liu C, Xie W, Gui C et al (2013) Pronuclear microinjection and oviduct transfer procedures for transgenic mouse production. Methods Mol Biol 1027:217–232 6. Gordon JW, Scangos GA, Plotkin DJ et al (1980) Genetic transformation of mouse embryos by microinjection of purified DNA. Proc Natl Acad Sci U S A 77:7380–7384 7. Tian XL, Wang QK (2006) Generation of transgenic mice for cardiovascular research. Methods Mol Med 129:69–81 8. Lohse MJ, Bünemann M, Hoffmann C et al (2007) Monitoring receptor signaling by

9.

10.

11.

12.

13.

14.

15.

intramolecular FRET. Curr Opin Pharmacol 7:547–553 Nikolaev VO, Bünemann M, Schmitteckert E et al (2006) Cyclic AMP imaging in adult cardiac myocytes reveals far-reaching beta1adrenergic but locally confined beta2-adrenergic receptor-mediated signaling. Circ Res 99: 1084–1091 Zaccolo M (2004) Use of chimeric fluorescent proteins and fluorescence resonance energy transfer to monitor cellular responses. Circ Res 94:866–873 Börner S, Schwede F, Schlipp A et al (2011) FRET measurements of intracellular cAMP concentrations and cAMP analog permeability in intact cells. Nat Protoc 6:427–438 Nikolaev VO, Bunemann M, Hein L et al (2004) Novel single chain cAMP sensors for receptor-induced signal propagation. J Biol Chem 279:37215–37218 Calebiro D, Nikolaev VO, Gagliani MC et al (2009) Persistent cAMP-signals triggered by internalized G-protein-coupled receptors. PLoS Biol 7:e1000172 Niwa H, Yamamura K, Miyazaki J (1991) Efficient selection for high-expression transfectants with a novel eukaryotic vector. Gene 108:193–199 Cho A, Haruyama N, Kulkarni AB (2009) Generation of transgenic mice. Curr Protoc Cell Biol. Chapter: Unit–19.11. doi: 10.1002/ 0471143030.cb1911s42

Chapter 10 Photoactivatable Adenylyl Cyclases (PACs) as a Tool to Study cAMP Signaling In Vivo: An Overview Marina Efetova and Martin Schwärzel Abstract Photoactivatable adenylyl cyclases (PACs) are proteins that combine the capacity of a photoreceptor with that of an adenylyl cyclase. When ectopically expressed under the control of specific promoters, these naturally occurring proteins become potent transgenic tools that facilitate the increase of cellular cAMP levels by the use of light. Currently, three different PAC transgenes—the euglenoid euPACα and euPACβ, as well as the beggiatoan bPac—are available. These transgenic tools provide cyclase activity capable of increasing cellular cAMP levels up to a hundredfold with either phasic- or tonic-like kinetic characteristics. Here, we consider the functional features of different cyclases and provide operating guidelines to optimize the use of PACs in vivo. Key words Photoactive adenylyl cyclases, PAC–euPACα–euPACβ–bPac–BlaC, cAMP-signaling pathway, Optogenetics, Artificial signaling

1

Introduction Photoactivatable adenylyl cyclases (PACs) were first identified in the unicellular flagellate Euglena gracilis [1]. This naturally occurring euglenoid cyclase comprised a hetero-tetramer composed of two PACα and two PACβ subunits. Each subunit exhibited adenylyl cyclase activity strongly enhanced by blue light of 455 nm and contained two BLUF-type photoreceptor domains involved in the binding of flavin adenine nucleotides, as well as two catalytic domains homologous to class III adenylyl cyclases [1]. Ectopic expression of either euPACα or euPACβ subunits alone reconstituted functional PACs that allowed for fast and reversible manipulation of cellular cAMP levels by light in a variety of model systems [2]. An additional PAC was isolated from the sulfide-oxidizing bacterium Beggiatoa [3, 4]. Unlike euPACs, the beggiatoan Pac (bPAC) consisted of a singular blue light-sensing BLUF domain that was C-terminally linked to a type III adenylyl cyclase domain yielding a photoactive cyclase of only 350 amino acids in length.

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_10, © Springer Science+Business Media New York 2015

131

132

Marina Efetova and Martin Schwärzel

Fig. 1 Cellular cAMP levels of Drosophila brain tissue as revealed by ELISA. Darkand light-induced activities of PAC transgenes can be separated by the use of the PDE inhibitor IBMX in combination with illumination. PACα exhibits strong cyclase activity that is only partially controlled by light. Note that adding a C-terminal GFP tag to the α transgene annihilates light dependency leaving only dark activity. PACβ is devoid of dark activity while showing light-dependent cyclase activity. Data represent means ± SEM (N = 4); *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ns not significant

PAC transgenes exhibit different amounts of light-induced cyclase activity, and euPACβ is the least effective cyclase [2]. In contrast, euPACα and bPAC are strong cyclases that increase cellular cAMP levels hundredfold upon saturating illumination [2, 3]. At first sight, minor inducibility of euPACβ might look like a disadvantage. However, it may become useful when fine-tuned increase in cellular cAMP levels is desirable. Moreover, due to the reduced cyclase activity, euPACβ seems to be devoid of dark activity—and hence increased basal cAMP levels (see Fig. 1). While light-dependent onset of cyclase activity is a fast process in either of those transgenes, its offset dynamics is highly variable. Note that bPAC returns to the inactive state rather slowly with a decay constant of τ ≈ 14 s [3, 4] while onset dynamics is in the range of milliseconds. Dependent on illumination strength, the slow offset dynamics of bPAC will result in a cAMP signal that far exceeds the illumination period. Like any transgene, PACs are not completely silent in the absence of light. This dark activity is proportional to, firstly, the amount of transgene expressed and, secondly, to the efficiency of the cyclases. Consequently, dark activity can be

Generation of cAMP by PACs

133

reduced by carefully controlling the expression strength of the transgene [4]. Dark activity is not necessarily obvious when cellular cAMP levels are quantified as it can be counteracted by cyclic nucleotide-specific phosphodiesterases (PDEs). We recommend use of the PDE blocker IBMX in appropriate quantification experiments as well as in physiological preparations to determine its functional impact (see Fig. 1). Currently available PACs are activated by blue light (λmax, approx. 450 nm), a spectral band also used by FRET-based cAMP sensors [5], thus complicating the simultaneous use of both transgenes. Moreover, blue light is well absorbed by flavins and porphyrins, i.e., compounds that are highly abundant in animal tissue and therefore can produce undesired side effects. However, when carefully controlled, the use of PAC is well suited to modulate cellular cAMP levels in freely moving animals like Drosophila and C. elegans [2, 3, 6, 7] as well as in cell culture and dissected tissue [3, 8].

2

Operating Guidelines for PAC Transgenes Uncovering regulatory functions of the cAMP-signaling cascade with cell-type-specific resolution is not trivial in a multicellular tissue or organ. There are two reasons for this: firstly, the pharmacological agents used for interference with cAMP signaling do not provide cell-type specificity, and, secondly, genetic approaches based on cell-type-specific promoters lack the fine temporal control over transgene activity—typically seconds—needed for physiological studies. Transgenic PAC overcomes these obstacles since it can be expressed under the control of cell-type-specific promoters and its illumination results in upregulation of cellular cAMP levels on a millisecond timescale. Moreover, the use of PAC can be combined with further genetic and/or pharmacological tools that target other components of the cAMP-signaling pathway [8]. In order to do so, there are several rather practical hurdles that need to be surmounted: firstly, an efficient and reproducible illumination of the specimen needs to be installed; secondly, a reliable quantification of cellular cAMP levels within the specimen is required; and thirdly, experimental conditions need to be adjusted to ensure efficient modulation of cellular cAMP levels by blue light.

2.1 Controlling Light Conditions in the Lab

Wavelengths beyond 550 nm (yellow) do not activate PAC transgenes and are thus safe to be used for ambient illumination of the lab. In contrast, blue light should be locked out of the laboratory as much as possible. It is more or less necessary to transform the facility into a darkroom, i.e., it should be impossible to read a newspaper if lights are switched off. Windows should be covered with curtains or roller shutters. Aluminum foil taped directly to the window is an easy and efficient way to lock out sunlight.

134

Marina Efetova and Martin Schwärzel

We also suggest the use of red LED light bulbs to be fitted in regular sockets and adhesive LED light strips for illumination of incubators or housing cages. In case cold light sources are needed, e.g., for dissection purposes, make sure to avoid accidental activation of PAC transgenes by using red filters. We have good experience with red filters that are available for the Schott KL1500 LCD and the KL2500 LCD. 2.2 Customizing Illumination of the Specimen

It is straightforward to use a fluorescent microscope as a means to activate PAC within the specimen and to quantify the biological function of interest. It is noteworthy that the specifications of a wild-type GFP filter cube usually fit the need to activate PAC (λex = 445 ± 45 nm). If inappropriate, a variety of LEDs or laser diodes can be used, either in combination with collimation lenses to illuminate open field arenas or couple LEDs or lasers to fiber optics in order to use them in cannulas for implantation. In any case, light intensity should be calibrated by the use of a power meter. Various companies, e.g., Thorlabs, have meanwhile specialized in distributing equipment for optogenetics.

2.3 Determine Cellular cAMP Concentrations in the Specimen

We have used commercially available colorimetric ELISA kits to determine cellular cAMP concentrations. We experienced that adding 1 N HCl as recommended by many of the manufacturers procedures was a good means to stop enzymatic reactions and preserve cAMP from degradation by PDEs. Once the protocol measures have been established to set the lower detection threshold, we had good experience with determining the upper threshold by adjusting dilution of the samples. Note that the linear range of an ELISA is rather limited; thus, fine-grained dilutions, e.g., twofold, might be appropriate.

2.4 How to Optimize the Use of the PAC Transgene

The PAC transgene will not be completely silent. As a consequence, the target cells are likely to experience extra cyclase activity that will increase tonic cAMP levels even in the absence of light. Dark activity must not inevitably result in an increase in net cAMP levels detected by the ELISA but rather might provide a constant challenge to the PDE system; hence, we recommend using a PDE inhibitor like IBMX to critically test for dark activity. Note that cyclase activity is temperature dependent [4] so lowering temperature of the specimen might be an additional option to optimize your preparation in addition to adjusting transcript levels. Kinetic aspects of PAC transgenes should also be considered. While onset of cyclase activity is fast in bPAC and euPACs, their offset kinetics differs tremendously [3]. Note that due to the slow decay constant illumination of bPAC, this enzyme is very likely to produce a cAMP signal persisting far beyond the illumination period. It might be reasonable to monitor the return to basal cAMP levels after illumination of the specimen.

Generation of cAMP by PACs

2.5

Case Study

135

We exemplify functional features of PAC transgenes using the Drosophila study case (Fig. 1). We expressed different PAC transgenes under the control of the neuron-specific elav promoter and monitored cAMP levels in dissected Drosophila brains by the use of ELISA. Comparison of non-illuminated control specimen (black bars) to illuminated specimen (white bars) revealed the unimpeded potency of the particular transgenes. The use of the PDE inhibitor IBMX in non-illuminated samples (gray bars) revealed their dark activity. It is noteworthy that fusion of GFP to the C-terminus of euPACα annihilated light induction without affecting dark activity. Regulation of cellular cAMP levels appeared to be optimally controlled by the euPACβ transgene, as these samples do not indicate background activity (comparing white to gray bars) yet significant light induction of the cyclase. Which of these cyclases perform best in your physiological assay needs to be determined individually.

References 1. Iseki M, Matsunaga S, Murakami A et al (2002) A blue-light-activated adenylyl cyclase mediates photoavoidance in Euglena gracilis. Nature 415:1047–1051 2. Schroder-Lang S, Schwarzel M, Seifert R et al (2007) Fast manipulation of cellular cAMP levels by light in vivo. Nat Methods 2007(4): 39–42 3. Stierl M, Stumpf P, Udwari D et al (2011) Light modulation of cellular cAMP by a small bacterial photoactivated adenylyl cyclase, bPAC, of the soil bacterium Beggiatoa. J Biol Chem 286:1181–1188 4. Ryu MH, Moskvin OV, Siltberg-Liberles J et al (2010) Natural and engineered photoactivated nucleotidyl cyclases for optogenetic applications. J Biol Chem 285:41501–41508

5. Berrera M, Dodoni G, Monterisi S et al (2008) Handbook Exp Pharmacol 186:285–298 6. Bellmann D, Richardt A, Freyberger R et al (2010) Optogenetically induced olfactory stimulation in Drosophila larvae reveals the neuronal basis of odor-aversion behavior. Front Behav Neurosci 2010(4):27 7. Weissenberger S, Schultheis C, Liewald JF et al (2010) PACalpha – an optogenetic tool for in vivo manipulation of cellular cAMP levels, neurotransmitter release, and behavior in Caenorhabditis elegans. J Neurochem 116(4):616–625 8. Efetova M, Petereit L, Rosiewicz K et al (2013) Separate roles of PKA and EPAC in renal function unraveled by the optogenetic control of cAMP levels in vivo. J Cell Sci 2013(126): 778–788

Chapter 11 Selective Disruption of the AKAP Signaling Complexes Eileen J. Kennedy and John D. Scott Abstract Synthesis of the second messenger cAMP activates a variety of signaling pathways critical for all facets of intracellular regulation. Protein kinase A (PKA) is the major cAMP-responsive effector. Where and when this enzyme is activated has profound implications on the cellular role of PKA. A-Kinase Anchoring Proteins (AKAPs) play a critical role in this process by orchestrating spatial and temporal aspects of PKA action. A popular means of evaluating the impact of these anchored signaling events is to biochemically interfere with the PKA–AKAP interface. Hence, peptide disruptors of PKA anchoring are valuable tools in the investigation of local PKA action. This article outlines the development of PKA isoform-selective disruptor peptides, documents the optimization of cell-soluble peptide derivatives, and introduces alternative cell-based approaches that interrogate other aspects of the PKA–AKAP interface. Key words AKAP, cAMP signaling, Protein kinase A (PKA), Compartmentalization, Anchoring proteins, RIAD, Ht31, AKAP-IS, STAD, Rselect

1

Introduction A-Kinase Anchoring Proteins (AKAPs) play a fundamental role in the spatial and temporal regulation of protein kinase A (PKA), yet how these protein–protein interactions influence normal and pathological signaling in the cell is just beginning to be understood. Although AKAPs differ greatly in sequence, subcellular localization, and repertoire of enzyme binding partners, they all share the defining commonality of a direct interaction with the regulatory subunits (RI or RII) of the PKA holoenzyme [1]. PKA anchoring proceeds through an amphipathic helix that inserts into a customized groove formed by the docking and dimerization (D/D) of R-subunit protomers [2–4]. When tethered to AKAPs, the PKA holoenzyme is spatially restricted with access to a limited number of cellular substrates (Fig. 1). This offers a mechanism to selectively promote cellular events that proceed through the ubiquitous second messenger molecule cAMP [5, 6]. However, this PKAbinding module denotes only one facet of AKAP action as other regions of the anchoring protein interact with additional enzymes

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_11, © Springer Science+Business Media New York 2015

137

138

Eileen J. Kennedy and John D. Scott

Fig. 1 Signaling through AKAP complexes. When intracellular concentrations of cAMP are low, the PKA holoenzyme complex is largely bound to AKAPs within the cell. AKAPs are localized to various intracellular sites including the plasma membrane and organelles, thereby concentrating PKA to particular locations within the cell. Upon stimulation, intracellular cAMP levels rise. Each R-subunit of PKA binds two cAMP molecules and undergoes an allosteric conformational change to release the activated catalytic subunits

to integrate other second messenger signals within distinct multivalent assemblies [7–9]. Accordingly, these signaling complexes can include other kinases, protein phosphatases, adenylyl cyclases, phosphodiesterases, and target substrates [10–14]. The complexity of this cellular system is further compounded by the utilization of four distinct regulatory subunit isoforms of PKA: RI (RIα and RIβ) and RII (RIIα and RIIβ) which differ in tissue distribution, cAMP sensitivity, and AKAP-mediated localization. These additional layers finely tune when and where PKA activity is applied [15]. The vast majority of AKAPs selectively bind the RII isoform; however, a limited number of dual-specific AKAPs can also interact with RI [4, 16, 17]. Due to the spatial and temporal nature of interactions with AKAPs, uncovering the intricacies of AKAP-mediated signaling events has proven to be a substantial challenge. To complicate matters further, the human genome encodes about fifty AKAPs and most cell types express at least 10–15 different anchoring proteins [18]. Added to this, most anchoring proteins are expressed as families of alternatively spliced transcripts [19, 20]. This degree of complexity makes it difficult to elucidate each of their individual roles. However, one strategy to study the role of anchoring in these signaling events is to selectively displace PKA subtypes from the AKAP platform. Accordingly, numerous isoform-specific disruptors have been developed (Fig. 2; Table 1) [21, 22]. Although these compounds are valuable tools to study AKAP–PKA signaling, the major drawback is that these inhibitors will nonspecifically inhibit all AKAP interactions with either the RI or RII isoforms by binding to and occluding the anchoring site on the regulatory subunits.

Disruption of AKAP Complexes

139

Fig. 2 Engineered peptide disruptors of AKAP complexes. Isoform-selective disruptors were developed to have specificity of targeting toward either the RI or RII isoform of PKA. Despite considerable sequence divergence between the different disruptor peptides, they all share the common feature of forming an amphipathic helix with a largely hydrophobic binding interface (shown in gray) that complements the binding surface of the D/D domain of the R-subunits. Asterisks represent incorporation of the nonnatural amino acid (S)-2-(4′-pentenyl)alanine to form an all-hydrocarbon bridge within the sequence Table 1 PKA inhibitor compounds for inhibition of AKAP-mediated signaling

1.1 RII-Selective Disruptors of AKAP Complexes

PKA inhibitors

Mechanism of action

PKI peptide

Blocks the catalytic site of PKA

H89

ATP-competitive inhibitor of PKA

Rp-cAMPS

Prevents cAMP binding to R-subunits

The first AKAP disruptor peptide, Ht31, was derived from the PKA-anchoring domain of AKAP-Lbc [23]. In this study, a 23-amino-acid amphipathic helix was identified from a screen seeking to find peptide antagonists of PKA anchoring. The discovery of this peptide set the precedence for defining canonical docking interactions between AKAPs and RII. Although Ht31 has limited cell permeability, chemical modification of the peptide was performed to increase its overall hydrophobicity [24]. The addition of stearic acid to the N-terminus of the peptide was found to greatly enhance cellular permeability. However, there may be concern that the conjugation of such a lipid moiety contributes to retention of Ht31 in cell membranes. Stearated forms of Ht31 and the negative proline analog control (St-Ht31 and St-Ht31P) are widely available as commercial reagents.

140

Eileen J. Kennedy and John D. Scott

Since there is considerable amino acid divergence between the PKA-anchoring helices of various AKAPs, a bioinformatics approach was used to identify an RII-specific consensus sequence [25]. This sequence was then optimized by peptide array screening to identify a more potent RII inhibitor peptide, AKAP-in silico (AKAP-IS) [25]. This peptide was shown to have improved affinity for RII as compared to Ht31 peptide. The Kd value of AKAP-IS is less than 1 nM for RII, while it has a Kd value for RI in the mid-high nM range. The initial AKAP-IS peptide was not cell permeable and also had limited solubility in aqueous solution. However, a subsequent modification introduced a TAT sequence at the N-terminus of AKAP-IS to greatly improve cell permeability for cell-based experiments [26]. Despite the hydrophilicity of the TAT sequence, the conjugated peptide, TAT–AKAP-IS, is still highly hydrophobic and requires solubilization in an aqueous 10 % DMSO solution. Using a structure-based approach, AKAP-IS was further optimized to improve the affinity and selectivity to yield SuperAKAP-IS [4]. In order to achieve this, the crystal structure of the AKAP docking site on RIIα was solved either alone or in complex with the inhibitor peptide AKAP-IS [4]. The identification of key residues involved in binding to the RII isoform and the use of further peptide screening arrays allowed for the design of a peptide disruptor with significantly enhanced RII selectivity that had fourfold higher affinity for RII and approximately 12-fold less affinity for RI as compared to AKAP-IS. Based on the biological observation that AKAP18 has a high affinity for RIIα and that an N-terminally truncated form, AKAP18δ, has an even higher affinity, a new class of disruptor peptides was derived [27]. This class of peptides demonstrated high affinity for RIIα with dissociation constants as low as 0.4 nM. Analysis of sequence divergence between these peptides helped to further define important residues for engagement with the RII docking site. Analogous to Ht31, the AKAP18δ peptides were also modified with the addition of a stearate moiety in order to promote cellular uptake. Within the last 5 years, small molecules were developed to disrupt AKAP–RII interactions [28, 29]. Very large, relatively flat surfaces, such as the protein–protein interaction interface between the amphipathic helix of an AKAP and the RII D/D docking site, are notoriously difficult to target using small molecule approaches. These small molecule scaffolds are an exciting new area for further investigation. Although these different compounds have limited potency (IC50 = 20–40 μM), this is a promising starting point for compound optimization using a small molecule targeting approach. Moreover, development of more selective small molecule scaffolds could yield anchoring disruptors with improved efficacy as they may evade some of the shortcomings inherent in peptides including limited cell permeability, low stability, and loss of secondary structural folds in solution.

Disruption of AKAP Complexes

141

Perhaps the most promising development in anchoring disruptor peptides is the recent introduction of Stapled AKAP Disruptor (STAD) peptides. Chemically modified RII-specific AKAP disruptors were developed where nonnatural amino acids were incorporated into A-kinase binding (AKB) sequences to bestow small-molecule-like properties onto the peptide sequences [21]. Synthetic libraries were designed based on previously identified AKB or AKB-like sequences, where nonnatural olefinic amino acids were incorporated and cyclized so as to conformationally constrain an alpha-helical fold. This chemical modification was previously shown to promote cellular permeability and proteolytic stability to peptides [30]. The STAD peptides developed in this study are highly cell permeable and effectively block interactions between AKAPs and RII inside cells. The incorporation of the peptide “staple” introduced significant hydrophobicity to an already hydrophobic sequence, so the addition of a small PEG-3 linker was added to the N-terminus to notably improve water solubility for cell-based experiments. The rapid cellular uptake, resistance to degradation, and relatively long half-lives in cells of the STAD peptides provide a more flexible platform for studying dynamic AKAP signaling events under a variety of conditions. All of the PKA-anchoring disruptor reagents discussed thus far have been patterned after an AKAP motif. Recently, a phage selection procedure was employed that exploits high-resolution structural information to engineer RII D/D domain mutants that are selective for a particular AKAP [31]. Competitive selection screening revealed RII sequences (RSelect) that were preferential for interaction with an individual AKAP. Biochemical and cell-based experiments validated the efficacy of RSelect mutants for AKAP2 and AKAP18. This new class of engineered proteins based on the reciprocal surface of the AKAP–PKA interaction has the potential to be used to dissect the contributions of different AKAP-targeted pools of PKA and aid in the design of compounds targeting these subset populations. 1.2 RI-Selective Disruptors of AKAP Complexes

Although numerous RII-specific AKAP disruptors have been identified, designing peptides for RI-selective interactions has proven to be more elusive. The first RI-selective peptide inhibitors were identified through peptide array screening nearly a decade after the design of Ht31 [32]. The prototype used for the peptide array was derived from the A-kinase binding (AKB) domain of AKAP10 [32]. Although the crystal structure of the AKB domain of RI was not solved at the time, the minimal sequence required and surface residue interactions involved in docking to RI were described through systematic analysis. Based on this study, the AKB binding site on RI was shown to involve multiple interactions with charged residues, while the analogous binding site on RII was shown to largely provide a hydrophobic patch for AKB binding. A major limitation of the peptides identified in this study, as with many

142

Eileen J. Kennedy and John D. Scott

unmodified peptides, is that they lack cell permeability and therefore require transfection or genetic encoding in order to characterize their activity in cells. Subsequent studies employed a bioinformatics approach coupled with peptide array screening to yield the RI-selective peptide, RIAD [33]. The binding sequences from several dual-specific AKAPs were used as a starting point to steer toward RI specificity. RIAD was found to have a notably improved binding affinity for RI as well as greater specificity for RI over RII. While the RIAD peptide alone was not cell permeable, the C-terminal addition of 11 arginine residues afforded this property. While transfection can result in artifacts and compensatory expression changes within the cell, the cell-permeable version of RIAD was utilized to illustrate disruption of RI-specific AKAP interactions in intact, non-modified cells. RIAD analogs were later developed that incorporated nonnatural and natural amino acids into the sequence to improve proteolytic stability [34]. However, cell permeability of the RIAD analogs remains an issue. The crystal structure of the docking/dimerization (D/D) domain of RIα was solved in recent years [35]. Numerous structural differences were identified between RI and RII that dictate engagement with various AKB sequences including the depth of the binding groove, the presence of a disulfide bridge within the binding site, and a shift of registry for binding by the AKB sequence. These structural insights will undoubtedly lead to the development of optimized peptide-based or synthetic scaffolds that can discriminate against RII interactions while maintaining high-affinity binding with RI. Additional selectivity in RI anchoring may involve a separate RI-binding interface that is upstream of the amphipathic helix. A distinct region upstream of the docking helix was identified on RI-specific AKAPs [36]. This RI-specific region (RISR) was also shown to disrupt RI binding and may serve as an additional targeting site for RI-specific disruption. 1.3 cAMPStimulating Conditions

As a means to interrogate AKAP signaling events in cell-based studies, multiple strategies can be applied to stimulate increased levels of intracellular cAMP (Table 2). While some reagents stimulate cAMP to physiological levels, many cause inappropriately high concentrations of cAMP. Forskolin is perhaps the most widely used stimulator of cAMP production by activating adenylyl cyclase (AC) activity. To date, nearly 10,000 citations list the use of forskolin as a PKA activator. Forskolin is a diterpene natural product isolated from Coleus forskohlii [37] and was found to stimulate cAMP concentrations in diverse tissue types in a reversible manner [38]. Eight of the nine membrane-bound isoforms of AC are stimulated by forskolin [39], with AC9 being the exception [40]. Further, the potency of stimulation varies among the different isoforms [41]. Since expression and regulation of the AC isoforms vary among

Disruption of AKAP Complexes

143

Table 2 cAMP-stimulating agents for activation of AKAP complexes cAMP-stimulating agents

Mechanism of action

Forskolin

Activates adenylyl cyclases

IBMX

Inhibits PDEs

Isoproterenol

Indirectly activates adenylyl cyclases

PGE2

Indirectly activates adenylyl cyclases

DB-cAMP

Activates PKA

cell and tissue types, the extent of forskolin-induced stimulation of cAMP can vary considerably and often to levels that are not physiologically relevant [39]. However, since forskolin acts as an agonist for the majority of the AC isoforms, it is considered to be a general, potent stimulator of intracellular cAMP across diverse cell types. Another approach for increasing intracellular cAMP levels is through inhibition of phosphodiesterase (PDE) activity. A nonspecific PDE inhibitor, 3-isobutyl-1-methylxanthine (IBMX), was first identified from a panel screen of various xanthine derivatives to have inhibitory effects on PDEs [42]. IBMX is a moderately potent inhibitor against the majority of PDE isoforms but appears to have no effect on PDE8 or PDE9 [43]. Due to its broad inhibitory activity on PDEs, IBMX is routinely used in conjunction with an AC-stimulating agent such as forskolin to further increase overall intracellular cAMP concentrations. Additional caution must be taken when interpreting results from experiments that use a forskolin/IBMX cocktail to stimulate PKA as this combination treatment stimulates cAMP production to supraphysiological levels and prolongs the second messenger response well beyond its normal time course. A much more physiologically relevant means to stimulate cAMP production is through activation of β1- and β2-adrenergic receptors by isoproterenol (isoprenaline) [44]. Isoproterenol is a synthetic catecholamine that acts as an agonist for this subclass of G proteincoupled receptors (GPCRs). Upon stimulation of β-adrenergic receptors, Gs proteins are activated inside cells, thereby leading to stimulation of AC activity. After isoproterenol stimulation, cAMP levels rise significantly, but then fall back to near background levels and are resistant to further stimulation even in the presence of persistent isoproterenol treatment [45]. Although β-adrenergic receptors are widely expressed in a variety of cells and isoproterenol can elicit a notable effect on cAMP levels, isoproterenol-stimulated cAMP production is useful for short time-course studies but is not effective as a cAMP-stimulating agent for sustainable periods.

144

Eileen J. Kennedy and John D. Scott

Another physiologically relevant method for cAMP stimulation involves prostaglandin E2 (PGE2). PGE2 is a hormonelike biological compound that binds to a subclass of G protein-coupled receptors called the E prostanoids (EP) [46]. Both EP2 and EP4 can stimulate cAMP production, but EP4 is broadly expressed while EP2 is only expressed in limited tissue/cell types [47]. Although isoproterenol and PGE2 activate different classes of GPCRs, both compounds ultimately lead to activation of Gs proteins and ACs. A chemical strategy to induce cAMP-sensitive signaling was developed using the cell-permeable cAMP analog, dibutyryl cyclic adenosine monophosphate (DB-cAMP). Although the compound enters the cell in an inactive form, hydrolysis of one of the butyrate groups permits the compound to activate PKA [48]. The butyrate cleavage product was also found to cause significant, unintended secondary effects in cells including differentiation, activation of cell signaling pathways, and growth inhibition [48]. Subsequent monobutyrated analogs of DB-cAMP are now available that have reduced hydrolysis and therefore have limited off-target effects caused by the butyrate side product [49]. Although there are clear advantages of these DB-cAMP analogs, it remains unclear whether they are resistant to all of the cAMP phosphodiesterase subtypes that exist in a typical cell, in particular PDE8, PDE10, and PDE11 [50]. 1.4

PKA Inhibitors

Over 40 years ago, a protein inhibitor of PKA was identified [51]. Protein kinase inhibitor (PKI) is expressed as three isoforms [52] that differ in expression in different tissues and cells and has an affinity for PKA in the sub-micromolar range [53]. A short, 20-aminoacid sequence was identified as the inhibitory component of PKI, and a synthetic peptide spanning this sequence was shown to act as a highly selective, potent inhibitor of PKA [54–56]. Multiple analogs derived from this 20-mer sequence were synthesized and tested so as to define the residues that are critical for its inhibitory activity [57–59]. This sequence was also found to be highly specific for PKA with no inhibitory effect on PKG [58]. PKI acts as a substrate mimic to block the catalytic site on PKA, thereby preventing substrate phosphorylation [60]. This mechanism provides greater target specificity of PKI for PKA; however, at high concentrations of PKI treatment, off-target effects have been documented [61]. A variety of PKI inhibitor peptide analogs are commercially available that have a high affinity for PKA and are recognized to have exquisite specificity for PKA at lower concentrations. H89 is an isoquinoline-based small molecule that was derived from an earlier inhibitor, H8 [62]. While H8 targeted both PKA and PKG, H89 was found to be a potent inhibitor of PKA but also had weak antagonistic activity against several other kinases including PKG, PKC, casein kinases I and II, and CamKII [63]. H89 acts as a competitive inhibitor of ATP binding to occupy and prevent substrate phosphorylation. While H89 is an effective inhibitor of

Disruption of AKAP Complexes

145

PKA, numerous off-target effects have been documented including disruption of various intracellular signaling pathways and inhibition of a significant number of kinases, including some that were inhibited at greater levels than PKA [64]. Although H89 is among the most commonly used of all PKA inhibitors, caution should be used in interpretation of results due to its numerous off-target effects. Cyclic nucleotide analogs such as Rp-cAMPS (adenosine-3′,5′cyclic monophosphorothioate Rp-isomer) have also been used as an inhibitory agent for PKA. Rp-cAMPS is cell permeable and acts as an antagonist of cAMP to prevent activation of PKA by binding to the cAMP-binding sites on the regulatory subunits of PKA [65, 66]. This cAMP analog also demonstrates resistance to hydrolysis by phosphodiesterases. Although Rp-cAMPS has limited cell permeability, newer versions such as Rp-8-Br-cAMPS and Rp-8Cl-cAMPS are recognized to have improved permeability and greater potency [67]. Yet, since additional signaling elements bind cAMP, it is possible that these analogs may also have other cellular targets aside from PKA-R and can thereby cause unintended secondary effects. 1.5 Practical Considerations

2

There are clear advantages of screening peptide disruptors in vitro prior to their use for cellular analysis. This is particularly relevant since cross talk is extremely common in kinase signaling cascades and AKAP-specific signaling events are not fully elucidated. Strategies that specifically interrogate the physical interaction between AKAPs and the D/D binding groove of the R-subunits provide a clearer route to identify disruptors. These interactions can be further validated using competitive binding experiments using a known disruptor such as Ht-31 or RIAD to confirm binding to the same interaction surface. A common strategy used for screening is fluorescence polarization where binding of potential disruptors is measured in solution using an increasing concentration of disruptor. This will provide dissociation constant for each disruptor and is a critical first step before entering more complex, cell-based experiments. Outlined below is a protocol that we have used to develop and characterize these anchoring disruptor peptides.

Materials 1. Purified D/D domain of RI and RII [68, 69]. 2. Lysis Buffer: 20 mM Tris, pH 8; 100 mM NaCl, 0.1 mM PMSF. 3. Assay Buffer: 10 mM HEPES, pH 7.4; 150 mM NaCl, 3 mM EDTA, 0.005 % P20. 4. Fluorescently labeled peptide stocks in DMSO. 5. Black opaque low-binding plates (384 well). 6. Plate reader capable of FP.

146

3

Eileen J. Kennedy and John D. Scott

Methods 1. For each protein concentration to be tested, prepare an 80 μL solution of fluorescently labeled peptide (diluted in assay buffer; see Note 1) to a microcentrifuge tube at a concentration of 20 nM. The peptide should be clear in solution with no visible precipitate (see Note 2). Once the protein is added, the final concentration will be 10 nM. 2. In a separate microcentrifuge tube, create serial dilution stocks of RI or RII in assay buffer at a 2× concentration (see Note 3). The final protein concentration should be tested to as high a concentration as possible to reach a plateau of nonbinding. For the D/D domains of RI and RII, higher concentrations should reach 50–100 μM. The protein should be tested over a serial dilution range down to 0.1 nM or lower, depending on the affinity of the peptide. The protein should be serially diluted two- to tenfold over this concentration range. For each protein concentration, prepare 80 μL of stock solution. 3. Combine the peptide solution with each solution of the different protein concentrations (80 μL protein and 80 μL peptide) and gently invert multiple times to ensure proper mixing (see Note 4). Add 50 μL of the solution to each of three wells. 4. Repeat step 3 for each protein concentration tested. 5. As a negative control lacking protein, combine 80 μL peptide (20 nM) diluted in assay buffer with 80 μL assay buffer. Invert to mix and plate 50 μL per well into three wells. 6. As a positive control, a known AKAP inhibitor peptide can be used (see Note 5). Combine 80 μL of the control peptide (20 nM in assay buffer) with either RI or RII over the same concentration range (80 μL per concentration). For each concentration, plate 50 μL in triplicate. 7. Store the plate in the dark at room temperature for 30–60 min before reading (see Note 6). 8. Read the plate using absorbance/emission values that are suitable for the peptide label and obtain FP values for each well. 9. Convert polarization values to anisotropy to determine the relative KD values for each inhibitor peptide tested.

4

Notes 1. Many AKAP disruptor peptides are extremely hydrophobic. To facilitate solubility in buffer, concentrated peptide stock solutions are prepared in DMSO, often in the range of 1–10 mM. Tenfold serial dilutions of the peptide in buffer are performed to accurately reach a 10 nM final concentration.

Disruption of AKAP Complexes

147

2. If the peptide precipitates out of solution, sonication of the solution may promote solubility of the peptide. As an alternative, a minimal amount of peptide-solubilizing agent may need to be added to the media (e.g., DMSO). 3. To maintain protein stability in solution, the protein dilutions are prepared on ice using freshly thawed protein. 4. Inverting the solution should be performed with great care to minimize the introduction of bubbles in solution, which will interfere with fluorescence readings. If bubbles are present, they may be removed by low-speed centrifugation of the microplate. 5. A standard AKAP disruptor and its corresponding negative control that is commercially available are Ht31 and Ht31P. 6. Longer or shorter incubation times may be necessary for optimal results. As an initial trial, the plate can be analyzed every 30 min over a 2-h time course to identify an optimal time point for analysis.

Acknowledgements This work was supported, in whole or in part, by National Institutes of Health Grants 1K22CA154600 to EJK, and DK105542 and DK054441 to JDS. References 1. Scott JD, Pawson T (2009) Cell signaling in space and time: where proteins come together and when they’re apart. Science 326: 1220–1224. doi:10.1126/science.1175668 2. Carr DW, Stofko-Hahn RE, Fraser ID et al (1991) Interaction of the regulatory subunit (RII) of cAMP-dependent protein kinase with RII-anchoring proteins occurs through an amphipathic helix binding motif. J Biol Chem 266:14188–14192 3. Newlon MG, Roy M, Morikis D et al (2001) A novel mechanism of PKA anchoring revealed by solution structures of anchoring complexes. EMBO J 20:1651–1662. doi:10.1093/emboj/20.7.1651 4. Gold MG, Lygren B, Dokurno P et al (2006) Molecular basis of AKAP specificity for PKA regulatory subunits. Mol Cell 24:383–395. doi:10.1016/j.molcel.2006.09.006 5. Welch EJ, Jones BW, Scott JD (2010) Networking with AKAPs: context-dependent regulation of anchored enzymes. Mol Interv 10:86–97. doi:10.1124/mi.10.2.6

6. Skroblin P, Grossmann S, Schafer G et al (2010) Mechanisms of protein kinase A anchoring. Int Rev Cell Mol Biol 283:235– 330. doi:10.1016/S1937-6448(10)83005-9 7. Dessauer CW (2009) Adenylyl cyclase – A-kinase anchoring protein complexes: the next dimension in cAMP signaling. Mol Pharmacol 76:935–941. doi:10.1124/mol.109.059345 8. Sanderson JL, Dell’Acqua ML (2011) AKAP signaling complexes in regulation of excitatory synaptic plasticity. Neuroscientist 17:321–336. doi:10.1177/1073858410384740 9. Diviani D, Dodge-Kafka KL, Li J et al (2011) A-kinase anchoring proteins: scaffolding proteins in the heart. Am J Physiol Heart Circ Physiol 301:H1742–H1753. doi:10.1152/ ajpheart.00569.2011 10. Klauck TM, Faux MC, Labudda K et al (1996) Coordination of three signaling enzymes by AKAP79, a mammalian scaffold protein. Science 271:1589–1592 11. Coghlan VM, Hausken ZE, Scott JD (1995) Subcellular targeting of kinases and phosphatases

148

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

Eileen J. Kennedy and John D. Scott by association with bifunctional anchoring proteins. Biochem Soc Trans 23:592–596 Dodge KL, Khouangsathiene S, Kapiloff MS et al (2001) mAKAP assembles a protein kinase A/PDE4 phosphodiesterase cAMP signaling module. EMBO J 20:1921–1930. doi:10.1093/emboj/20.8.1921 Tasken KA, Collas P, Kemmner WA et al (2001) Phosphodiesterase 4D and protein kinase a type II constitute a signaling unit in the centrosomal area. J Biol Chem 276:21999– 22002. doi:10.1074/jbc.C000911200 Bauman AL, Soughayer J, Nguyen BT et al (2006) Dynamic regulation of cAMP synthesis through anchored PKA-adenylyl cyclase V/VI complexes. Mol Cell 23:925–931. doi:10.1016/j.molcel.2006.07.025 Taylor SS, Yang J, Wu J et al (2004) PKA: a portrait of protein kinase dynamics. Biochim Biophys Acta 1697:259–269. doi:10.1016/j. bbapap.2003.11.029 Herberg FW, Maleszka A, Eide T et al (2000) Analysis of A-kinase anchoring protein (AKAP) interaction with protein kinase A (PKA) regulatory subunits: PKA isoform specificity in AKAP binding. J Mol Biol 298:329–339. doi:10.1006/jmbi.2000.3662 Aye TT, Mohammed S, van den Toorn HW et al (2009) Selectivity in enrichment of cAMP-dependent protein kinase regulatory subunits type I and type II and their interactors using modified cAMP affinity resins. Mol Cell Proteomics 8:1016–1028. doi:10.1074/ mcp. M800226-MCP200 Lester LB, Coghlan VM, Nauert B et al (1996) Cloning and characterization of a novel A-kinase anchoring protein. AKAP 220, association with testicular peroxisomes. J Biol Chem 271:9460–9465 Trotter KW, Fraser ID, Scott GK et al (1999) Alternative splicing regulates the subcellular localization of A-kinase anchoring protein 18 isoforms. J Cell Biol 147:1481–1492 Huang LJ, Durick K, Weiner JA et al (1997) Identification of a novel protein kinase A anchoring protein that binds both type I and type II regulatory subunits. J Biol Chem 272:8057–8064 Wang Y, Ho TG, Bertinetti D et al (2014) Isoform-selective disruption of AKAPlocalized PKA using hydrocarbon stapled peptides. ACS Chem Biol 9:635–642. doi:10.1021/cb400900r Scott JD, Dessauer CW, Tasken K (2013) Creating order from chaos: cellular regulation by kinase anchoring. Annu Rev Pharmacol Toxicol 53:187–210. doi:10.1146/annurevpharmtox-011112-140204

23. Carr DW, Hausken ZE, Fraser ID et al (1992) Association of the type II cAMP-dependent protein kinase with a human thyroid RIIanchoring protein. Cloning and characterization of the RII-binding domain. J Biol Chem 267:13376–13382 24. Vijayaraghavan S, Goueli SA, Davey MP et al (1997) Protein kinase A-anchoring inhibitor peptides arrest mammalian sperm motility. J Biol Chem 272:4747–4752 25. Alto NM, Soderling SH, Hoshi N et al (2003) Bioinformatic design of A-kinase anchoring protein-in silico: a potent and selective peptide antagonist of type II protein kinase A anchoring. Proc Natl Acad Sci U S A 100:4445– 4450. doi:10.1073/pnas.0330734100 26. Faruque OM, Le-Nguyen D, Lajoix AD et al (2009) Cell-permeable peptide-based disruption of endogenous PKA-AKAP complexes: a tool for studying the molecular roles of AKAP-mediated PKA subcellular anchoring. Am J Physiol Cell Physiol 296:C306–C316. doi:10.1152/ajpcell.00216.2008 27. Hundsrucker C, Krause G, Beyermann M et al (2006) High-affinity AKAP7delta-protein kinase A interaction yields novel protein kinase A-anchoring disruptor peptides. Biochem J 396:297–306. doi:10.1042/BJ20051970 28. Christian F, Szaszak M, Friedl S et al (2011) Small molecule AKAP-protein kinase A (PKA) interaction disruptors that activate PKA interfere with compartmentalized cAMP signaling in cardiac myocytes. J Biol Chem 286:9079– 9096. doi:10.1074/jbc.M110.160614 29. Schafer G, Milic J, Eldahshan A et al (2013) Highly functionalized terpyridines as competitive inhibitors of AKAP-PKA interactions. Angew Chem Int Ed Engl 52:12187–12191. doi:10.1002/anie.201304686 30. Verdine GL, Hilinski GJ (2012) Stapled peptides for intracellular drug targets. Methods Enzymol 503:3–33. doi:10.1016/ B978-0-12-396962-0.00001-X 31. Gold MG, Fowler DM, Means CK et al (2013) Engineering A-kinase anchoring protein (AKAP)-selective regulatory subunits of protein kinase A (PKA) through structure-based phage selection. J Biol Chem 288:17111– 17121. doi:10.1074/jbc.M112.447326 32. Burns-Hamuro LL, Ma Y, Kammerer S et al (2003) Designing isoform-specific peptide disruptors of protein kinase A localization. Proc Natl Acad Sci U S A 100:4072–4077. doi:10.1073/pnas.2628038100 33. Carlson CR, Lygren B, Berge T et al (2006) Delineation of type I protein kinase A-selective signaling events using an RI anchoring disruptor. J Biol Chem 281:21535–21545

Disruption of AKAP Complexes 34. Torheim EA, Jarnaess E, Lygren B et al (2009) Design of proteolytically stable RI-anchoring disruptor peptidomimetics for in vivo studies of anchored type I protein kinase A-mediated signalling. Biochem J 424:69–78. doi:10.1042/ BJ20090933 35. Sarma GN, Kinderman FS, Kim C et al (2010) Structure of D-AKAP2:PKA RI complex: insights into AKAP specificity and selectivity. Structure 18:155–166 36. Jarnaess E, Ruppelt A, Stokka AJ et al (2008) Dual specificity A-kinase anchoring proteins (AKAPs) contain an additional binding region that enhances targeting of protein kinase A type I. J Biol Chem 283:33708–33718. doi:10.1074/jbc.M804807200 37. Bhat SV, Bajwa BS, Dornauer H, de Souza NJ (1977) Structure and stereochemistry of new labdane diterpenoids from Coleus forskohlii Briq. Tetrahedron Lett 19:1669–1672 38. Seamon KB, Padgett W, Daly JW (1981) Forskolin: unique diterpene activator of adenylate cyclase in membranes and in intact cells. Proc Natl Acad Sci U S A 78:3363–3367 39. Patel TB, Du Z, Pierre S et al (2001) Molecular biological approaches to unravel adenylyl cyclase signaling and function. Gene 269: 13–25 40. Premont RT, Matsuoka I, Mattei MG et al (1996) Identification and characterization of a widely expressed form of adenylyl cyclase. J Biol Chem 271:13900–13907 41. Dessauer CW, Scully TT, Gilman AG (1997) Interactions of forskolin and ATP with the cytosolic domains of mammalian adenylyl cyclase. J Biol Chem 272:22272–22277 42. Beavo JA, Rogers NL, Crofford OB et al (1970) Effects of xanthine derivatives on lipolysis and on adenosine 3′,5′-monophosphate phosphodiesterase activity. Mol Pharmacol 6:597–603 43. Omori K, Kotera J (2007) Overview of PDEs and their regulation. Circ Res 100:309–327. doi:10.1161/01.RES.0000256354.95791.f1 44. Robison GA, Butcher RW, Sutherland EW (1968) Cyclic AMP. Annu Rev Biochem 37:149–174. doi:10.1146/annurev. bi.37.070168.001053 45. Shear M, Insel PA, Melmon KL et al (1976) Agonist-specific refractoriness induced by isoproterenol. Studies with mutant cells. J Biol Chem 251:7572–7576 46. Sugimoto Y, Narumiya S (2007) Prostaglandin E receptors. J Biol Chem 282:11613–11617. doi:10.1074/jbc.R600038200 47. Zeng L, An S, Goetzl EJ (1998) EP4/EP2 receptor-specific prostaglandin E2 regulation

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

149

of interleukin-6 generation by human HSB.2 early T cells. J Pharmacol Exp Ther 286: 1420–1426 Schwede F, Maronde E, Genieser H et al (2000) Cyclic nucleotide analogs as biochemical tools and prospective drugs. Pharmacol Ther 87:199–226 Zorn M, Maronde E, Jastorff B et al (1993) Differential effects of two structurally related N6-substituted cAMP analogues on C6 glioma cells. Eur J Cell Biol 60:351–357 Beavo JA, Brunton LL (2002) Cyclic nucleotide research – still expanding after half a century. Nat Rev Mol Cell Biol 3:710–718. doi:10.1038/nrm911 Walsh DA, Ashby CD, Gonzalez C et al (1971) Krebs EG: purification and characterization of a protein inhibitor of adenosine 3′,5′-monophosphate-dependent protein kinases. J Biol Chem 246:1977–1985 Dalton GD, Dewey WL (2006) Protein kinase inhibitor peptide (PKI): a family of endogenous neuropeptides that modulate neuronal cAMP-dependent protein kinase function. Neuropeptides 40:23–34. doi:10.1016/j. npep.2005.10.002 Demaille JG, Peters KA, Fischer EH (1977) Isolation and properties of the rabbit skeletal muscle protein inhibitor of adenosine 3′,5′-monophosphate dependent protein kinases. Biochemistry 16:3080–3086 Scott JD, Fischer EH, Demaille JG et al (1985) Identification of an inhibitory region of the heat-stable protein inhibitor of the cAMPdependent protein kinase. Proc Natl Acad Sci U S A 82:4379–4383 Scott JD, Fischer EH, Takio K et al (1985) Amino acid sequence of the heat-stable inhibitor of the cAMP-dependent protein kinase from rabbit skeletal muscle. Proc Natl Acad Sci U S A 82:5732–5736 Cheng HC, van Patten SM, Smith AJ et al (1985) An active twenty-amino-acid-residue peptide derived from the inhibitor protein of the cyclic AMP-dependent protein kinase. Biochem J 231:655–661 Scott JD, Glaccum MB, Fischer EH et al (1986) Primary-structure requirements for inhibition by the heat-stable inhibitor of the cAMP-dependent protein kinase. Proc Natl Acad Sci U S A 83:1613–1616 Glass DB, Cheng HC, Kemp BE et al (1986) Differential and common recognition of the catalytic sites of the cGMP-dependent and cAMP-dependent protein kinases by inhibitory peptides derived from the heat-stable inhibitor protein. J Biol Chem 261:12166–12171

150

Eileen J. Kennedy and John D. Scott

59. Cheng HC, Kemp BE, Pearson RB et al (1986) A potent synthetic peptide inhibitor of the cAMP-dependent protein kinase. J Biol Chem 261:989–992 60. Reed J, De Ropp JS, Trewhella J et al (1989) Conformational analysis of PKI(5–22)amide, the active inhibitory fragment of the inhibitor protein of the cyclic AMP-dependent protein kinase. Biochem J 264:371–380 61. Glass DB, Feller MJ, Levin LR et al (1992) Structural basis for the low affinities of yeast cAMP-dependent and mammalian cGMPdependent protein kinases for protein kinase inhibitor peptides. Biochemistry 31: 1728–1734 62. Hidaka H, Inagaki M, Kawamoto S et al (1984) Isoquinolinesulfonamides, novel and potent inhibitors of cyclic nucleotide dependent protein kinase and protein kinase C. Biochemistry 23:5036–5041 63. Chijiwa T, Mishima A, Hagiwara M et al (1990) Inhibition of forskolin-induced neurite outgrowth and protein phosphorylation by a newly synthesized selective inhibitor of cyclic AMP-dependent protein kinase, N-[2-(pbromocinnamylamino)ethyl]-5isoquinolinesulfonamide (H-89), of PC12D pheochromocytoma cells. J Biol Chem 265: 5267–5272

64. Murray AJ (2008) Pharmacological PKA inhibition: all may not be what it seems. Sci Signal 1:re4. doi:10.1126/scisignal.122re4 65. Rothermel JD, Stec WJ, Baraniak J et al (1983) Inhibition of glycogenolysis in isolated rat hepatocytes by the Rp diastereomer of adenosine cyclic 3′,5′-phosphorothioate. J Biol Chem 258:12125–12128 66. Rothermel JD, Jastorff B, Botelho LH (1984) Inhibition of glucagon-induced glycogenolysis in isolated rat hepatocytes by the Rp diastereomer of adenosine cyclic 3′,5′-phosphorothioate. J Biol Chem 259:8151–8155 67. Gjertsen BT, Mellgren G, Otten A et al (1995) Novel (Rp)-cAMPS analogs as tools for inhibition of cAMP-kinase in cell culture. Basal cAMP-kinase activity modulates interleukin-1 beta action. J Biol Chem 270: 20599–20607 68. Kinderman FS, Kim C, von Daake S et al (2006) A dynamic mechanism for AKAP binding to RII isoforms of cAMP-dependent protein kinase. Mol Cell 24:397–408 69. Banky P, Newlon MG, Roy M et al (2000) Isoform-specific differences between the type Ialpha and IIalpha cyclic AMP-dependent protein kinase anchoring domains revealed by solution NMR. J Biol Chem 275:35146–35152. doi:10.1074/jbc.M003961200

Chapter 12 Screening for Small Molecule Disruptors of AKAP–PKA Interactions Carolin Schächterle, Frank Christian, João Miguel Parente Fernandes, and Enno Klussmann Abstract Protein–protein interactions (PPIs) are highly specific and diverse. Their selective inhibition with peptides, peptidomimetics, or small molecules allows determination of functions of individual PPIs. Moreover, inhibition of disease-associated PPIs may lead to new concepts for the treatment of diseases with an unmet medical need. Protein kinase A (PKA) is an ubiquitously expressed protein kinase that controls a plethora of cellular functions. A-kinase anchoring proteins (AKAPs) are multivalent scaffolding proteins that directly interact with PKA. AKAPs spatially and temporally restrict PKA activity to defined cellular compartments and thereby contribute to the specificity of PKA signaling. However, it is largely unknown which of the plethora of PKA-dependent signaling events involve interactions of PKA with AKAPs. Moreover, AKAP– PKA interactions appear to play a role in a variety of cardiovascular, neuronal, and inflammatory diseases, but it is unclear whether these interactions are suitable drug targets. Here we describe an enzyme-linked immunosorbent assay (ELISA) for the screening of small molecule libraries for inhibitors of AKAP–PKA interactions. In addition, we describe a homogenous time-resolved fluorescence (HTRF) assay for use in secondary validation screens. Small molecule inhibitors are invaluable molecular tools for elucidating the functions of AKAP–PKA interactions and may eventually lead to new concepts for the treatment of diseases where AKAP–PKA interactions represent potential drug targets. Key words Protein kinase A (PKA), A-kinase anchoring protein (AKAP), Inhibitory peptides, Nonpeptidic helix mimetics, Small molecules, Homogenous time-resolved fluorescence (HTRF) assay, Enzyme-linked immunosorbent assay (ELISA)

1

Introduction Cells perceive changes of their environment mostly through sensing chemical cues such as hormones or neurotransmitters via receptors. Stimulation of receptors then often causes a rise in the level of second messengers that modulates intracellular signaling to enable cellular responses that are specific for each stimulus. The modulation of intracellular signaling depends on accurate protein– protein interactions (PPI) between proteins organized along signaling cascades. The signaling cascades involve, among others,

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_12, © Springer Science+Business Media New York 2015

151

152

Carolin Schächterle et al.

scaffolding proteins and protein kinases. Scaffolding proteins are multivalent protein binders that constitute platforms for the spatial and temporal coordination of signal propagation through bound proteins; protein kinases play crucial roles as they phosphorylate and thereby activate or inactivate their effectors. One of such kinases is protein kinase A (PKA). PKA is ubiquitously expressed and involved in a plethora of cellular processes. The tetrameric PKA holoenzyme consists of a homodimer of regulatory (RIα, RIβ, RIIα, or RIIβ) subunits and two catalytic (Cα, Cβ, or Cγ) subunits. The dimeric R subunits form an N-terminal hydrophobic pocket, the dimerization and docking (D/D) domain, which directly interacts with RII-binding domains (RBD) of a family of scaffolding proteins, the A-kinase anchoring proteins (AKAPs). AKAPs can also directly interact with PKA substrates and other signaling proteins like phosphodiesterases or protein phosphatases [1–3]. By virtue of their unique targeting domains, AKAPs direct their associated binding partners to defined cellular compartments and facilitate spatial and temporal coordination of cellular signaling events. Activation of seven transmembrane receptors inducing a rise in the level of the second messenger cAMP leads to activation of PKA; cAMP binds to the R subunits, causing a conformational change and the dissociation of the C subunits which phosphorylate substrates in close proximity. In addition to PKA’s activity itself, it is the local confinement of its activity achieved by its interaction with AKAPs which is crucial for many of its physiological functions. Examples are the control of cardiac myocyte contractility and arginine–vasopressin (AVP)-mediated water reabsorption in renal principal cells [1, 2, 4–6]. RII-binding domains of the approximately 50 AKAP family members are structurally conserved amphipathic α-helices formed by 14–25 amino acids. The α-helices dock with their hydrophobic phases into the D/D domains of the regulatory subunits of PKA. The conserved RII-binding domains have been used to identify novel AKAPs such as GSKIP [7] and were the basis for uncovering the involvement of AKAP–PKA interactions in defined processes. The latter has been achieved with peptides for disruption of the interactions. The peptides were developed from RII-binding domains of various AKAPs. The original peptide, Ht31, had been derived from the RII-binding domain of AKAP13 (AKAP-Lbc) [8, 9]. Nowadays, various peptides such as those derived from the RIIbinding domain of AKAP7 (AKAP18) [10] or in silico-designed ones such as AKAPIS are available [11]. They all bind to the D/D domain with nanomolar affinity and thereby competitively inhibit AKAP–PKA interactions [6]. The use of such peptides in vitro and in cell culture experiments revealed key roles of AKAP–PKA interactions in cardiac myocyte contractility, AVP-mediated water reabsorption, and many other processes [2]. However, due to the generally low membrane permeability and stability, peptides are not suitable for cell and animal studies and they are also considered

Interference with PKA Anchoring

153

difficult for the development towards drugs, although several peptide-based drugs have reached the market [12]. Nonpeptidic agents such as peptidomimetics and small molecules are promising alternatives to peptides. Such agents would constitute invaluable tools to elucidate functions of AKAP–PKA interactions in complex biological systems, in particular in animal experiments, where knockout or knockdown of an AKAP would interfere with all of its functions as all of its PPIs would be affected and where knock-in studies to interfere with defined PPIs are laborious and time and cost intensive. Since dysregulation of AKAPs and their interactions with PKA are associated with various cardiovascular diseases such as heart failure, inflammatory diseases such as chronic obstructive pulmonary disease (COPD), and neurological disorders such as schizophrenia, disruptors of AKAP–PKA interactions may be utilized for the validation of the interactions as drug targets [5, 13]. Using a rational approach, we have recently developed nonpeptidic terpyridine-based α-helix mimetics on the basis of an interaction model of a peptide derived from the RII-binding domain of AKAP18δ, AKAP18δ-L314E with the D/D domain of RIIα. The peptide binds RIIα subunits with a KD = 0.4 ± 0.3 nM [10]. The mimetics are designed to combine advantages of peptides, specificity and target selectivity, and of small molecules, increased membrane permeability and stability [14]. The terpyridine-based agents mimic the peptidic α-helix by presenting amino acid-derived side chains from a hydrophobic rodlike axis [14]. The rodlike axis mimicking the hydrophobic phase of the peptidic α-helix consists of pyridine, cyclopentyl, and benzyl rings interacting with the bottom of the D/D domain pocket; hydrophilic carboxyl groups enable the terpyridines to interact with hydrophilic residues at the rim of the D/D domain pocket and enhance binding affinity. KD values for the interaction of the terpyridines with the D/D domain of RIIα range from 30 to 148 μM; IC50 values for the inhibition of the interaction of AKAP18 with RIIα are between 38 and 138 μM [14]. This chapter describes a detailed protocol to screen small molecule libraries for inhibitors of AKAP–PKA interactions via enzyme-linked immunosorbent assay (ELISA) [15]. This primary screening identifies hits, which need to be validated via secondary approaches such as modified ELISAs or homogenous time-resolved fluorescence (HTRF) assays, which we will also describe. The ELISA, a solid phase assay, requires two recombinant proteins whose interaction is detected using a primary antibody directed against one of the proteins and a secondary antibody coupled to horseradish peroxidase. The enzyme converts a substrate whereby light is emitted. The light intensity can be quantitatively measured. An inhibitory compound reducing the interaction between the proteins decreases the signal. The HTRF assay is solution based and this method quantitatively analyzes PPIs via fluorescence resonance energy transfer

154

Carolin Schächterle et al.

(FRET) [14]; antibodies directed against each of the interacting proteins are used, one is labeled with a donor and the other with an acceptor chromophore. Excitation of the donor causes FRET to the acceptor when in close proximity, i.e., when interaction between the two binding partners occurs; emission of a fluorescent signal from the acceptor chromophore can be quantitatively detected [16]. Peptides or small molecules inhibiting the PPI reduce or abolish FRET. Being a ratiometric measurement, this technique is significantly more robust than other fluorescence-based methods. It is also more sensitive, thanks to the use of exceptionally long-lived donor chromophores. This allows for the introduction of a time delay between excitation and detection of emitted light; consequently, the typically short-lived fluorescence of most common contaminants is not measured, and the background signal is reduced. Each assay may be used as a primary or secondary screening system and allows the determination of IC50 values for identified hits. IC50 values are indicators of the inhibitory potency of a compound, defining the concentration of an inhibitor that decreases the interaction to 50 % of the maximum. Hits confirmed in secondary approaches can undergo a hit to lead development towards a drug candidate where selectivity, affinity, and pharmacological properties are optimized. Additional assays that may be adopted for high-throughput screening of small molecule libraries include, for example, fluorescence polarization-, AlphaScreen-, Biacore-, and cell-based automatic microscopy assays. Any screening assay requires adaptation to the particular high-throughput screening platform.

2 2.1

Materials ELISA

1. White 384 well microtiter plates. 2. Phosphate buffered saline (PBS): 1.4 M NaCl, 25 mM KCl, 60 mM Na2HPO4, and 15 mM KH2PO4. 3. Coating buffer: 0.5 mM PMSF (phenylmethylsulfonylfluoride), 1 mM DTT (dithiothreitol), and protease inhibitor cocktail in PBS. 4. Blocking buffer: 0.3 % dried skimmed milk, 0.05 % Tween-20, 0.5 mM PMSF, and protease inhibitor cocktail in PBS. 5. Washing buffer: 0.05 % Tween-20 in PBS. 6. RIIα (recombinant full length regulatory RIIα subunit [14]; 15 ng/well; see Note 1). 7. Recombinant full length GST-AKAP18δ (N-terminal GST-tag [14]; 80 ng/well; see Note 1). 8. Small molecule fragment library “FMP20000” (www.chembionet.info; www.fmp-berlin.de).

Interference with PKA Anchoring

155

9. DMSO (dimethylsulfoxide). 10. Compounds FMP-API-1 (=3,3′-diamino-4,4′-dihydroxydiphenylmethane), 990, 2348 (small molecule fragment library FMP20000). 11. Antibodies: Anti-A18δ3 (custom-made rabbit antibody directed against the epitope QGNPKRSKENRGDRND, amino acid residues 60–76 of rat AKAP18δ) and anti-RIIα mouse antibody (BD Biosciences, Heidelberg, Germany) (see Note 1). 12. Horseradish peroxidase-conjugated anti-rabbit secondary antibody. 13. Lumi-Light Western Blotting substrate solution (Roche, Grenzach-Wyhlen, Germany). 14. Genios Pro plate reader (Tecan Genios Pro; Tecan Austria GmbH, Grödig, Austria) with Magellan 5 software (see Note 2). 15. GraphPad Prism software. 16. Heraeus Megafuge 1.0; swing-out rotor for microtiter plates. 2.2

HTRF

1. Assay buffer 2×: 0.1 % BSA, 0.1 % Tween-20 in PBS, pH 7.4. 2. Recombinant GST-tagged AKAP18α-Δ-2-10 [14]. 3. Recombinant His-tagged DD domain of PKA-RIIα (amino acids 1–44) [14]. 4. Terbium-conjugated anti-GST antibody (Cisbio Bioassays, Codolet, France). 5. XL665-conjugated anti-His antibody (Cisbio Bioassays, Codolet, France). 6. AKAP18δ-L314E peptide 10 μM in DMSO. 7. AKAP18δ-PP peptide 10 μM in DMSO. 8. DMSO. 9. ProxiPlate-384 Plus Shallow Well Microplates (PerkinElmer, Waltham, USA). 10. Genios Pro plate reader (Tecan Genios Pro; Tecan Austria GmbH, Grödig, Austria) or another HTRF-certified plate reader.

3

Methods

3.1 ELISA for Screening for Small Molecule Inhibitors of AKAP– PKA Interactions

We established this protocol for screening of the small molecule library “FMP20000” (www.chembionet.info; www.fmp-berlin.de) to analyze the effects of 20,064 diverse, commercially available compounds (ChemDiv, San Diego, USA) on the interaction of AKAP18δ with RIIα. First, we illustrate an ELISA-based primary screening [15]; in a second protocol, we describe the validation of hits obtained in the primary screen. In this modified ELISA, compounds FMP-API-1 (=3,3′-diamino-4,4′-dihydroxydiphenylmethane) and

156

Carolin Schächterle et al.

990 were tested using a serial dilution. Compound 2348 was identified to have no influence on the AKAP18δ–RIIα interaction and was included as a negative control [15]. 3.1.1 Compound Screening Protocol

Recombinant GST-AKAP18δ and full length untagged RIIα were encoded in the vector pGEX4T3 and generated in E. coli (strain Rosetta DE3) [14]. A pipetting scheme and the assay layout for the screening are depicted in Fig. 1a. 1. The compounds of the library are stored on 384 well microtiter plates (10 mM stock solutions) at −20 °C. For thawing, the plates were incubated at 37 °C for 30 min, shortly centrifuged (see Notes 3 and 4), and kept at room temperature until usage. 2. Coat wells of the 384 well assay microtiter plates with 20 μl/well coating buffer containing 15 ng recombinant RIIα, centrifuge, and incubate 1–2 h at room temperature (see Notes 4 and 5). 3. Remove protein solution. 4. Block free binding sites of the plates by adding 95 μl blocking buffer to each well, centrifuge, and incubate for 1 h at 22 °C (see Note 4). 5. Wash wells three times with washing buffer using an automatic microtiter plate washer (see Note 6). 6. Empty plate and add 10 μl of blocking buffer to each well (see Note 7). 7. Transfer of 0.5 μl/well of each compound from storage plate to the screening plate. This was performed by using the Sciclone ALH 3000 Workstation (Caliper LifeSciences, Hopkinton, USA). 8. Add 10 μl of blocking buffer with 80 ng recombinant GSTAKAP18δ to the wells. This pipetting scheme (see Fig. 1a) assures accurate mixing of inhibitor and GST-AKAP18δ in the total volume of = 20.5 μl for proper incubation. The final concentration of the compounds is 244 μM. 9. Incubate 1 h at 22 °C. 10. Wash wells three times with washing buffer using an automatic microtiter plate washer (see Note 6). 11. Add 20 μl of anti-AKAP18δ antibody A18δ3 (1:1,000) in blocking buffer according to the pipetting scheme and incubate for 1 h at 22 °C (see Fig. 1a). 12. Wash wells three times with washing buffer using an automatic microtiter plate washer (see Note 6). 13. Add 20 μl of horseradish peroxidase-conjugated anti-rabbit secondary antibody, diluted 1:4,000 in blocking buffer, and incubate for 1 h at 22 °C (at least 15 min). 14. Wash wells three times with washing buffer using an automatic microtiter plate washer (see Note 6).

Interference with PKA Anchoring

157

Fig. 1 Exemplary ELISA design for the validation of identified hits. (a) Assay design and plate layout for library screening. (b) Assay design and plate layout for compound validation

15. Add 20 μl of Lumi-Light Western Blotting substrate solution and incubate for 5 min at 22 °C or 37 °C. 16. Measure light intensity. 17. Analyze data and calculate the IC50 as described below. 3.1.2

Data Analysis

1. Subtract averaged blank value from other averaged (two times quadruplicates) values. 2. Use GraphPads one site competition model with the correspondX - log IC 50 ) ing equation Y = Bottom + ( Top - Bottom ) / 1 + 10( . The lower asymptote corresponds to Bottom, which in the calculation settings was defined to be greater than 0.0 to exclude an incorrect fit leading to negative binding values.

(

)

3. Values are visualized in a graph and an IC50 value is calculated.

158

Carolin Schächterle et al. 150

990 FMP-API1 2348

binding [%]

125 100 75 50 25 0 -2

0

2

4

log(c(compound)) [µM] HN

FMP-API-1 HO

990 OH

HN 2

NH2 S

NH

log (IC50) = 1.4 IC50 = 25 M

2

log (IC50) = 1.1 IC50 = 13 M

Fig. 2 FMP-API-1 and compound 990 inhibit the interaction between AKAP18δ and RIIα subunits of PKA

3.1.3 Compound Validation Protocol

Figure 1b displays the layout of the 384 well microtiter plate. For instructions of the design of this assay, refer to Note 8. The hit validation protocol runs similarly as the screening protocol. It is designed to apply 1 μl/well of the particular inhibitor dilutions. Thus, serial dilutions of inhibitors are necessary (dilutions in DMSO in μM: 0.0025, 0.025, 0.25, 2.5, 25, 62.5, 125, 187.5, 250). Calculate the stock concentration according to the final incubation volume per well (21 μl). Instead of step 7 of the compound screening protocol, add 1 μl each of FMP-API-1, 990 and 2348 in the concentrations indicated in the pipetting scheme of Fig. 1b. The following steps are as mentioned in the compound screening protocol above. The inhibitory effect of compounds FMP-API-1 and 990 is shown in Fig. 2. The IC50 values of FMP-API-1 and compound 990 are 25 and 13 μM, respectively. Compound 2348 does not affect the interaction.

3.2 HTRF for the Identification of AKAP–PKA Interaction Inhibitors

The solution-based method HTRF can be used as a secondary screening for validation of inhibitory compounds. We established the method for characterizing the inhibitory effects of terpyridinebased α-helix mimetics on the interaction of AKAP18 with RIIα [14]. Exemplary, we describe here the protocol for the application of an AKAP18δ-derived peptide, AKAP18δ-L314E as inhibitor of AKAP–PKA interactions.

Interference with PKA Anchoring

159

The assay includes the D/D domain of RIIα and AKAP18α that lacks the 10 N-terminal amino acid residues, which comprise the membrane-targeting domain. The AKAP18α version was coupled to a glutathione S-transferase (GST) tag and the D/D domain to a His tag. The interaction of the D/D domain with AKAP18α was detected with antibodies directed against the tags. The antiGST antibody was terbium- and the anti-His antibody XL665conjugated facilitating FRET measurements. A range of other tags and anti-tag antibodies are available commercially; therefore, the choice of tags and antibodies is left to the user. Carry out all procedures at room temperature. 1. Prepare serial dilutions of the compound(s) to be tested; in this case, the AKAP18δ-L314E peptide and the inactive control peptide AKAP18δ-PP (see Note 9). 2. Dilute the recombinant proteins separately in assay buffer 2× to a final concentration of 50 nM each (see Notes 10 and 11). 3. Add the fluorophore-conjugated antibodies to the each of the protein solutions to a final concentration of 2 μg/ml. 4. Dispense 5 μl of donor protein solution (GST-AKAP18α and terbium-coupled antibody; final AKAP concentration 25 nM) to each well of a 384 well plate, except for wells containing the negative control. An example plate scheme is available in Fig. 3, including positive and negative controls (see Notes 12 and 13). 5. Dispense 0.2 μl of each dilution of the compound into the appropriate wells (see Note 14). 6. Dispense 5 μl of acceptor protein solution containing the D/D domain and the XL665-conjugated antibody into each well (see Note 15).

Fig. 3 Exemplary HTRF assay plate to assess the effect of the peptide AKAP18δ-L314E on the interaction of AKAP18 with RIIα subunits of PKA. AKAP18δ-L314E is an established inhibitor of AKAP–PKA interactions, whereas AKAP18δ-PP is a peptide without effect on AKAP–PKA interaction

160

Carolin Schächterle et al.

7. Shake at 500 × g for 15 s and briefly centrifuge the plate to make sure all the components are properly mixed and at the bottom of the wells. 8. Incubate for 1–2 h in the dark (see Note 16). 9. Measure fluorescence emission at 620 nm (donor) and 665 nm (acceptor) with a time delay of 150 μs after donor excitation at 320 nm in an HTRF-certified plate reader (see Notes 17 and 18). 10. Please (see Note 19 to 23) for general suggestions concerning optimization of the HTRF assay and validation of results obtained by HTRF. 3.2.1

Data Analysis

1. Calculate the FRET ratio by dividing the fluorescence intensity of the acceptor by the fluorescence intensity of the donor for each well. 2. Subtract the average FRET ratio of the negative control wells. 3. Normalize to the background-subtracted average FRET ratio of the wells containing positive control for a relative measure of the residual interaction. 4. Plot the obtained values in a logarithmic X-axis and calculate an IC50 value by performing a nonlinear regression as described in the Data Analysis section of the ELISA method. For this purpose, we use the GraphPad Prism version 5.0c for MacOS. A representative example of two distinct results is shown in Fig. 4, which show that the peptide AKAP18δL314E effectively inhibits the AKAP18–PKA interaction and that the AKAP18δ-PP peptide fails to do so.

4

Notes One of the first ELISA experiments should determine the capacity of the microtiter plates for binding either regulatory PKA subunits or the AKAP. Our initial experiments revealed a better output for plate-bound RIIα compared to plate-bound GST-AKAP18δ. Also, initial experiments are required to determine optimal concentrations of each of the recombinant proteins. It is possible to preincubate multiple microtiter plates with RIIα and to store these plates for a few days at 4 °C. The collection of a number of plates simplifies the library screening. A further setup showed that blocking buffer containing skimmed milk powder (0.3 %) caused less unspecific luminescence compared to blocking buffer containing BSA. 1. Tags on recombinant proteins are only required if antibodies directed against the proteins of interest are not available or not suitable for the ELISA. If both proteins are tagged, the tags

Interference with PKA Anchoring

161

AKAP18δ-L314E 1,2 1,0

relative activity

0,8 0,6 IC50 = 9,62nM 0,4 0,2 0,0 -0,2 0,01

0,1

1

10

100

1000

10000

100

1000

10000

peptide (nM)

AKAP18δ-PP 1,2 1,0

relative activity

0,8 0,6 0,4 0,2 0,0 -0,2 0,01

0,1

1

10 peptide (nM)

Fig. 4 Representative results showing effects of peptides that either inhibit (a) or do not affect (b) the interaction of AKAP18 with RIIα subunits of PKA

must be different to assure detection of each of the proteins with different anti-tag-antibodies and to avoid artificial dimer formation as is sometimes observed with GST or MBP tags. Full-length untagged RII can be recombinantly generated and excellent antibodies are available. Therefore, only a tag on an AKAP may be required. In this case it should be verified that it does not disturb the AKAP–PKA interaction. One advantage of tags is the possibility of increased binding of tagged

162

Carolin Schächterle et al.

recombinant protein to the surface of microtiter plates that are specially coated in high density with a ligand. For example, precoating the plate with glutathione facilitates the binding of GST-tagged recombinant proteins. 2. Choosing the 384 well microtiter plates depends on the imaging system. The detection with fluorescent or chemiluminescent signals requires black or white plates, respectively. The use of tetramethylbenzidine (TMB) requires transparent microtiter plates (e.g., Corning, 384 well, high binding, polystyrene) with flat bottom. Measuring ODs at 450 nm should be performed with an appropriate reader (Genios Pro plate reader, for instance, Tecan Austria GmbH). 3. The use of an ultrasonic bath may dissolve potential solid particles in the compound stock solutions. The plates are treated for 1 min in the ultrasonic bath and subsequently incubated for 5 min at 30 °C to allow plates to dry on the surface. Finally, the protective foil can be removed. 4. The centrifugation steps during the assay (500 × g, 1 min) are to ensure concentration of the sample at the well bottoms. 5. In this protocol we mention an incubation for one to two hours at 22 °C for the first recombinant protein, RIIα. An overnight incubation at 4 °C to split the procedure into 2 days would work as well. A prolonged incubation of compound, recombinant AKAP, and recombinant regulatory PKA subunit is not recommended as proteins may not be stable and small molecules may precipitate upon longer incubation. Peptides do not precipitate even during longer incubation. 6. A microtiter plate washer (like BioTek, ELx405 Select CW) simplifies the washing procedure as it discards and adds the washing buffer automatically. Washing with a multichannel pipette works comparably but takes much longer. Of note, the washing buffer should not spill over from one well into another to prevent mixing of differently loaded wells, and the wells should be washed without bubbles and foam development to assure accurate washing for diminishing the background signal. Consider the maximum volume fitting in 384 well microtiter plate wells (≈110 μl). 7. Emptying the wells has to be carried out through vigorous pouring out and completed with softly bashing on tissue sheets to assure complete removal of the liquid from the plate. 8. To compensate for pipetting errors, this assay should comprise twofold quadruplicates. Further, it should include several controls to verify coating of the plate and protein–protein binding. The negative control (blank) displays the unspecific binding of the AKAP-directed antibody to wells lacking recombinant GST-AKAP18δ. The coating test reveals the

Interference with PKA Anchoring

163

binding efficiency of RIIα to the wells, as these wells are loaded with recombinant RIIα that are detected with anti-RIIα antibodies. The third control is to assess the influence of the added compound on the plate-bound recombinant RIIα. To evaluate this, serial compound dilutions are added to RIIα bound to the surface of the wells. The subsequently added anti-RIIα antibody will show whether the compounds negatively affect plate-bound RIIα. Finally, the influence of the compounds on the RIIα–AKAP18δ interaction is measured by incubating RIIα and GST-AKAP18δ in the presence of the serially diluted hits (see Fig. 1b). 9. Since the volume of compound added to the mix is typically very small, it is dispensed with an automatic liquid handling system. Therefore, we prepare the serial dilutions in a plate with the same format as the assay plate and include the same volume of DMSO in the wells to which no compound is added. 10. Make sure that the proteins are concentrated enough so that their addition will not significantly affect the concentration of the assay buffer; alternatively, adjust the concentration of the buffer used so that the final concentration is as desired. 11. The buffer indicated in the materials section is a good starting point, but, depending on the proteins in the assay, the addition of cofactors that can influence the interaction often increases the range of the signal and thus assay sensitivity. Generally, cofactors such as Mg2+ or Zn2+ affect protein–protein interactions and may be included to improve interactions in ELISA assays. 12. When setting up a new HTRF assay, it is important to optimize the concentrations of both recombinant proteins, which is easily done by cross-titrating them in one plate. This can also be useful after the assay has been established, for instance, when changing to a new batch of recombinant protein, which sometimes causes a drop in the signal range. It is important to have in mind that too little protein results in a smaller signal range and that in the presence of too much protein, the assay is not in a dynamic range, i.e., it is in saturation. 13. The FRET ratios determined using an HTRF assay are not sufficient by themselves for quantification of the interaction between two proteins. It is therefore essential in any HTRF experiment to include adequate controls to determine the range of the signal. When assessing the inhibitory potency of small molecules towards a PPI, typical controls are: (a) 100 % control—both proteins and both fluorophoreconjugated antibodies in the absence of compounds (this will generate the maximum FRET ratio)

164

Carolin Schächterle et al.

(b) 0 % control—no recombinant proteins (or only acceptor) and both fluorophore-conjugated antibodies in the absence of compounds (this will generate the lowest FRET ratio, corresponding to the background of the assay) Optimization of an HTRF assay is generally aimed at increasing the difference between these two controls; the higher the 100 % control is compa\red to the 0 % control, the more sensitive the assay will be. 14. This protocol was established in a format compatible with high-throughput screening. Where the volumes to be dispensed onto the plate were too small to be accurately measured with an automatic pipette, as is the case with the addition of the peptide AKAP18δ-L314E, a Freedom EVO automated liquid handling platform (Tecan AG, Männedorf, Switzerland) was used. 15. The pipetting schemes should avoid carry-over effects; always pipette from the wells containing lower concentration of compounds or proteins to the wells containing higher concentration, and use a new pipette tip when changing compounds or adding protein to wells containing a different compound. 16. When setting up a new HTRF assay, it is reasonable to perform a time course experiment to determine optimal measurement time. Incubation periods are usually around 1–2 h. 17. Determining the optimal time delay between excitation and detection can help improve sensitivity. A compromise must be reached to exclude as much background fluorescence due to contaminants in the mixture (generally short-lived fluorescence) as possible, while at the same time allowing for detection of high signals from the HTRF fluorophores, which have a long fluorescence lifetime. 18. Plate readers for HTRF measurements should fulfill two fundamental requirements: they should be able to excite and detect fluorescence emission at the appropriate wavelengths of course (this depends on the fluorophores used) and they should allow for the introduction of a time delay between the excitation and the detection of emitted fluorescence. Several plate readers compatible with HTRF are certified by Cisbio Bioassays (Codolet, France) and bear an “HTRF compatibility” sticker. 19. After setting up the conditions for an HTRF assay, it should be validated by using known inhibitors of the interaction. If none is available, recombinant untagged donor or acceptor proteins can be used; these should decrease the FRET ratio by competing with the equivalent tagged protein for binding to the interaction partner without generating a FRET signal.

Interference with PKA Anchoring

165

20. Especially when using HTRF for high-throughput screening of small molecule inhibitors of PPIs, positive hits should be validated in at least one counter-assay, to make sure that the observed effect is due to inhibition of the target interaction and not due to interference with the photophysical properties of the fluorophores. Two elegant HTRF-based counter-assays are described by Benicchi et al. [17]. 21. Great care should be taken when preparing the recombinant proteins. These should be as free of other contaminating proteins as possible, and measures should be taken to avoid proteolytic degradation and oxidation (keep the proteins at 4 °C; include protease inhibitors and reducing agents such as DTT or β-mercaptoethanol in the protein solution). If there is the possibility of checking beforehand whether the proteins are functional, this would increase confidence in the method. Aliquoting them also helps preventing their inactivation during storage. 22. If the compounds to be tested are dissolved in DMSO, it is advisable to include the same concentration of DMSO in the control wells. 23. The best results are generally obtained with white plates, although black plates are also suitable. Glass-bottom plates should be avoided. Assay volumes should match well volumes as close as possible to allow for optimal detection.

Acknowledgements We thank Sylvia Niquet for technical assistance. This work was supported by the Deutsche Forschungsgemeinschaft (DFG KL1415/4-2), the Else Kröner-Fresenius-Stiftung (2013_A145), and the German-Israeli Foundation (I-1210-286.13/2012). References 1. Scott JD, Dessauer CW, Tasken K (2013) Creating order from chaos: cellular regulation by kinase anchoring. Annu Rev Pharmacol Toxicol 53:187–210 2. Skroblin P, Grossmann S, Schafer G et al (2010) Mechanisms of protein kinase a anchoring. Int Rev Cell Mol Biol 283:235–330 3. Szaszak M, Christian F, Rosenthal W et al (2008) Compartmentalized cAMP signalling in regulated exocytic processes in nonneuronal cells. Cell Signal 20:590–601 4. Diviani D, Dodge-Kafka KL, Li J et al (2011) A-kinase anchoring proteins: scaffolding proteins in the heart. Am J Physiol Heart Circ Physiol 301:H1742–H1753

5. Esseltine JL, Scott JD (2013) AKAP signaling complexes: pointing towards the next generation of therapeutic targets? Trends Pharmacol Sci 34:648–655 6. Hundsrucker C, Klussmann E (2008) Direct AKAP-mediated protein-protein interactions as potential drug targets. Handb Exp Pharmacol 186:483–503 7. Hundsrucker C, Skroblin P, Christian F et al (2010) Glycogen synthase kinase 3beta interaction protein functions as an A-kinase anchoring protein. J Biol Chem 285:5507–5521 8. Carr DW, Hausken ZE, Fraser ID et al (1992) Association of the type II cAMP-dependent protein kinase with a human thyroid RII-anchoring

166

9.

10.

11.

12.

Carolin Schächterle et al. protein. Cloning and characterization of the RII-binding domain. J Biol Chem 267: 13376–13382 Carr DW, Stofko-Hahn RE, Fraser ID et al (1991) Interaction of the regulatory subunit (RII) of cAMP-dependent protein kinase with RII-anchoring proteins occurs through an amphipathic helix binding motif. J Biol Chem 266:14188–14192 Hundsrucker C, Krause G, Beyermann M et al (2006) High-affinity AKAP7delta-protein kinase A interaction yields novel protein kinase A-anchoring disruptor peptides. Biochem J 396:297–306 Alto NM, Soderling SH, Hoshi N et al (2003) Bioinformatic design of A-kinase anchoring protein-in silico: a potent and selective peptide antagonist of type II protein kinase A anchoring. Proc Natl Acad Sci U S A 100:4445–4450 Craik DJ, Fairlie DP, Liras S et al (2013) The future of peptide-based drugs. Chem Biol Drug Des 81:136–147

13. Troger J, Moutty MC, Skroblin P et al (2012) A-kinase anchoring proteins as potential drug targets. Br J Pharmacol 166:420–433 14. Schafer G, Milic J, Eldahshan A et al (2013) Highly functionalized terpyridines as competitive inhibitors of AKAP-PKA interactions. Angew Chem Int Ed Engl 52: 12187–12191 15. Christian F, Szaszak M, Friedl S et al (2011) Small molecule AKAP-protein kinase A (PKA) interaction disruptors that activate PKA interfere with compartmentalized cAMP signaling in cardiac myocytes. J Biol Chem 286: 9079–9096 16. Herman B, Krishnan RV, Centonze VE (2004) Microscopic analysis of fluorescence resonance energy transfer (FRET). Methods Mol Biol 261:351–370 17. Benicchi T, Iozzi S, Svahn A et al (2012) A homogeneous HTRF assay for the identification of inhibitors of the TWEAK-Fn14 protein interaction. J Biomol Screen 17:933–945

Chapter 13 Structure-Based Bacteriophage Screening for AKAP-Selective PKA Regulatory Subunit Variants Ryan Walker-Gray and Matthew G. Gold Abstract cAMP-dependent protein kinase (PKA) is tethered at different subcellular locations by A-kinase anchoring proteins (AKAPs). AKAPs present amphipathic helices that bind to the docking and dimerization (D/D) domain of PKA regulatory subunits. Peptide disruptors derived from AKAP anchoring helices are powerful tools for determining whether PKA anchoring is important in different biological processes. Focusing on the reciprocal side of the AKAP-PKA interface can enable development of tools for determining the roles of individual AKAPs. Accordingly, here we describe a bacteriophage screening procedure for identifying variants of PKA regulatory subunit D/D domains that bind selectively to individual AKAPs. This procedure can be adapted for engineering specificity into other shared protein interfaces. Key words PKA, AKAP, cAMP, Phage display, Protein engineering, Directed evolution

1

Introduction It is important to understand which A-kinase anchoring proteins (AKAPs) coordinate cAMP-dependent protein kinase (PKA) to perform specific physiological functions since targeting individual AKAP-PKA complexes is an emerging avenue for disease intervention [1]. Biotechnological innovations for manipulating the location of enzymes in time and space support research in this vein [2]. Knowledge of the three-dimensional molecular basis of AKAPPKA interactions has previously enabled the design of peptide disruptors for investigating PKA anchoring. All AKAPs contain an amphipathic helix that presents a series of aliphatic amino acids on one face that interact with a hydrophobic groove on the docking and dimerization domain (D/D) of type II (RII) PKA regulatory subunits [3, 4] and type I (RI) PKA regulatory subunits in some cases [5]. Peptide disruptors were originally derived from native anchoring helices [6]. Synthetic sequences have latterly been engineered for higher AKAP affinity [7] and selectivity for either RI [8] or RII regulatory subunits [3]. Focusing on the reciprocal side of

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_13, © Springer Science+Business Media New York 2015

167

168

Ryan Walker-Gray and Matthew G. Gold

the interface (the regulatory subunit D/D domain) is an avenue for developing tools for distinguishing the contributions of different AKAPs in targeting PKA. Isolated D/D variant domains with AKAP selectivity may be applied to drive PKA away from individual AKAPs. Conversely, full-length regulatory subunits incorporating D/D domain variants with AKAP selectivity can be applied in combination with general anchoring peptide disruptors such as Ht31 to drive PKA selectively to an individual AKAP [9]. This latter capability can potentially be combined with intracellular fluorescence resonance energy transfer (FRET) reporters for cAMP and local PKA activity that are described elsewhere in this volume. Inspection of AKAP-PKA crystal structures indicates positions in the RII D/D domain where amino acid substitutions could impart AKAP selectivity (Fig. 1). AKAPs exhibit higher variability (lighter shading in Fig. 1) in their anchoring helices at positions that can potentially interact with Ile3, Ile5, Thr10, and Gln14 on the surface of RII (Fig. 1). Therefore, substitutions at positions 3, 5, 10, and 14 in RII have the potential to impart binding preference for specific AKAPs. This protocol has been written to explain how to identify an RII variant bearing substitutions at these four positions that possesses selectivity for a generic target AKAP. This protocol may also serve as a template for identifying AKAP-selective RI variants or RII variants with substitutions at additional positions. We detail how to generate a phage library presenting variants of the type II regulatory subunit D/D domain (Subheading 3.1), how to express and purify an AKAP bait protein for phage selection (Subheading 3.2), how to enrich phage bearing AKAPselective D/D variants (Subheading 3.3), and how to sequence clonal phage plaques grown on LB plates (Subheading 3.4).

Fig. 1 RII docking and dimerization (D/D) domain amino acids that contact high variability positions in AKAP anchoring helices. Sequence alignment of AKAP anchoring helices (left) and location of low (dark) and high (light) variability positions in the anchoring helix in relation to amino acids on the surface of the RII D/D (right)

Engineering AKAP-Selective PKA Variants

2

169

Materials Unless otherwise stated, prepare solutions using ultrapure water, e.g., with resistivity >18 MΩ after purification with a Milli-Q system.

2.1 DNA Amplification, Modification, and Sequencing

1. Oligonucleotide primers for amplification of D/D variant sequences (see Note 1): EcoRI_Linker_DD_F: 5′-GGTTCATGTGCTCAGAATTCTGGTTCTGGTT CTTCTGGTGGTTCTGGT-3′ DD_stop_HindIII_R: 5′-CCGTGACACAGCAGAAGCTTTCATTACTAAGC ACGAGCTTCACGCAGACGGGTGA AGTATTCAACAGCAAATTCAACCAGGTCCGGCGG CTGCTG-3′ DD_Template: 5′-GTTCTTCTGGTGGTTCTGGTATGTCTCAC NNK CAGNNKCCGCCGGGTCTGNNKGA ACTGCTG NNK GGTTACACCGTTGAAGTTCTGCGT CAGCAGCCGCCGG-3′ (where N = A + C + G + T, K = G + T, see Note 2). 2. Tris-EDTA buffer: 10 mM Tris–HCl pH 8.0, 1 mM EDTA. 3. DNA modifying enzymes (New England Biolabs): Exonuclease I (ExoI)/Shrimp Alkaline Phosphatase (SAP) mixture (make stock mixture containing each enzyme at 1 U/μL); BamHI, EcoRI, Hind III, and NotI restriction enzymes. 4. DNA polymerases: Platinum Taq DNA polymerase (Life Technologies); GoTaq DNA polymerase (Promega). 5. Qiaquick PCR purification and gel extraction kits (Qiagen). 6. BigDye Terminator v3.1 Cycle Sequencing Kit (Life Technologies). 7. Thin-walled 0.2 mL PCR tubes 8. Apparatus for imaging DNA using visible light: 4 % agarose gel with thick wells containing 1 μL/100 mL SYBR Safe DNA gel stain (Life Technologies), UV light box equipped with a VisiBlue UV/Blue converter plate (UVP), SYBR Safe Imager viewing glasses (Life Technologies). 9. Small volume (90 % purity and dissolve stocks according to Table 1 prior to storage of aliquots at −20 °C. 11. An aspirator for removing supernatant from 1.5 mL tubes, e.g., the Safe Aspiration Station (Gilson). 12. Temperature-controlled shaker for 1.5 mL Eppendorf tubes, e.g., the Thermomixer Comfort (Eppendorf).

3

Methods

3.1 Construction of RII D/D Variant Bacteriophage Library

1. Dissolve oligonucleotide primers in Tris-EDTA buffer at 100 μM for long-term storage at −80 °C. Dilute primers into molecular biology-grade water for 10 μM (EcoR1_Linker_ DD_F & DD_stop_HindIII_R) and 50 ng/μL (DD_ Template) working solutions stored at −20 °C.

172

Ryan Walker-Gray and Matthew G. Gold

Fig. 2 Preparation of RII D/D variant library DNA for ligation into T7 vector arms. (a) T7 bacteriophage vector map showing method for inserting a linker and the RII D/D between T7 vector arms. Diversity is introduced at D/D codons 3, 5, 10, and 14 by performing PCR with a central primer containing wobble positions: N = A + C + G + T, K = G + T. (b) Insert DNA, following digestion with EcoRI and HindIII, is extracted from a 4 % agarose gel stained with SYBR Safe on a UV light box equipped with a Visi-Blue UV/Blue converter plate. DNA bands are visualized through SYBR Safe Imager viewing glasses

2. Set up a 100 μL PCR reaction mixture comprising 74.5 μL molecular biology-grade water, 4 μL of 10 μM EcoR1_Linker_ DD_F primer, 4 μL of 10 μM DD_stop_HindIII_R primer, 1 μL DD_Template primer (50 ng/μL), 2 μL dNTPs (10 mM mixture), 10 μL High Fidelity PCR Buffer (10× stock), 4 μL magnesium sulfate (50 mM stock), and finally 0.5 μL Platinum Taq DNA Polymerase High Fidelity. 3. Amplify RII variant insert sequences (Fig. 2a, see Note 3) by equally dividing the reaction mixture into two thin-walled PCR tubes and incubating in a thermal cycler according to the following program: 94 °C for 2 min; 28 cycles of [94 °C for 30 s, 50 °C for 30 s, 68 °C for 1 min]; 68 °C for 1 min; hold at 4 °C. Pool reactions and purify amplified DNA using the Qiagen PCR purification kit. Elute with 36.3 μL molecular biology-grade water into a 1.5 mL tube. 4. Add 4.3 μL NEBuffer 2.1 (10×), 0.43 μL BSA (100×), 1 μL EcoRI (10 U/μL), and 1 μL HindIII (10 U/μL) and incubate for 6 h in a 37 °C water bath. Remove, add 8.6 μL 6× loading dye, and load onto a 4 % agarose gel containing SYBR Safe reagent. Run for 90 min at 90 V before cutting out the DNA band at ~190 bp with a steel razor, using the Visi-Blue converter plate and SYBR Safe Imager viewing glasses to visualize the DNA band (Fig. 2b—see Note 4). Extract the DNA using the Qiagen DNA gel extraction kit. Elute DNA with 30 μL molecular biology-grade water and measure DNA concentration

Engineering AKAP-Selective PKA Variants

173

using a NanoDrop 1000 (see Note 5). Dilute the insert DNA if necessary to 0.66 pmol/μL (~80 ng/μL). 5. Set up the following 5 μL ligation reaction in a 1.5 mL tube: 1.5 μL insert DNA (1 pmol), 1 μL T7Select 10-3b vector arms (0.5 μg), 0.5 μL 10× ligase buffer, 0.5 μL 10 mM ATP, 0.5 μL 100 mM DTT, and finally 1 μL T4 DNA ligase (see Note 6). Incubate overnight at 16 °C in the Thermomixer without shaking. In addition, inoculate 30 mL LB supplemented with 50 μg/mL carbenicillin with BLT5403 cells from the glycerol stock. Grow overnight at 37 °C with shaking to produce a saturated starter culture. 6. On the following morning, thaw one 25 μL T7Select packaging extract aliquot on ice. Add 5 μL ligation reaction and gently mix with a pipette tip. Incubate at 22 °C for 2 h. Simultaneously, transfer 500 μL of the BLT5403 starter culture into 50 mL M9LB and 50 mL LB (both supplemented with 50 μg/mL carbenicillin), and shake the cultures for approximately 2 h at 37 °C until OD600nm reaches ~0.8, at which point transfer the cultures to 4 °C. Terminate packaging by adding 270 μL sterile LB. 7. Set aside 5 μL of the packaging reaction for titering. Add the remaining 295 μL to the BLT5403 cells grown in 50 mL LB (OD600nm ~ 0.8) for amplification. Once the BLT5403 cells have cleared after 3–4 h, spin for 10 min at 8,000 × g, collect the supernatant, and store the amplified library at 4 °C in a Falcon tube wrapped in aluminum foil. 8. Take 2 μL from the remaining packaging reaction and dilute into 198 μL sterile LB (10−2 dilution). Set up a series of tenfold dilutions by serially diluting 100 μL into 900 μL sterile LB to establish a series ranging from 10−2 to 10−6 times the concentration of the original packaging reaction. 9. Boil 100 mL of top agarose in a 250 mL screw cap bottle in a microwave and then cool and hold at 50 °C in a water bath (see Note 7). 10. Pipette 250 μL of the BLT5403 cell culture grown in M9LB (OD600nm ~ 0.8) into four sterile 14 mL round-bottom tubes. Add 100 μL of either the 10−3, 10−4, 10−5, or 10−6 dilution to each tube. 11. Add 3 mL of top agar to each 14 mL tube, and immediately invert each tube onto a labeled pre-warmed (37 °C) LB + 100 μg/mL ampicillin plate while gently rolling the plate to ensure even and complete coverage with the top agar mixture. Incubate at 37 °C for 3–4 h (see Note 8). 12. Count plaques on a plate bearing between 10 and 100 plaques. The number of packaged phage in the amplified

174

Ryan Walker-Gray and Matthew G. Gold

library = 2.95 × (number of plaques on the plate)/(phage concentration in the diluent added to the plate relative to the original packaging reaction). For example, 80 plaques on the 10−5 dilution plate = 2.95 × 80/10−5 = 2.4 × 107 packaged phage. A titer greater than 1 × 107 indicates that all possible combinations of amino acids at RII positions 3, 5, 10, and 14 are represented in the library (see Note 9). 3.2 Expression and Purification of Bait AKAP Fragment

1. Perform secondary structure prediction of the primary sequence of the target AKAP, using Jpred 3 [10], in a 100amino acid region centered on its amphipathic anchoring helix. 2. Design primers for synthetically amplifying the coding sequence of a sixty-amino acid region in the target AKAP that does not interrupt predicted secondary structure elements and positions the start of the anchoring helix at least twenty amino acids from the N-terminus. Use Jcat codon optimization software [11] to optimize the coding sequence for bacterial expression. Incorporate 5′-BamHI and 3′-NotI restriction sites in the construct to enable ligation into pGEX6P1 (see Note 10). 3. PCR-amplify the target AKAP fragment coding sequence and clone into pGEX6P1 using BamHI and NotI restriction sites. Verify that the correct sequence has been inserted into pGEX6P1 before proceeding. 4. Transform the pGEX6P1-AKAP fragment vector into BL21 DE3 Star cells (see Note 11), plating onto LB/ampicillin plates. Inoculate 50 mL LB containing 100 μg/mL ampicillin with bacteria picked from a single colony and shake at 37 °C overnight. On the following morning, transfer 6 mL of the saturated starter culture into 2 × 600 mL of the same medium, and shake at 37 °C until the culture reaches an OD600nm = 0.5. Add 300 μL of 1 M IPTG, shake at 37 °C for a further 3 h, pellet, and flash freeze in liquid nitrogen prior to storage at −80 °C. 5. Perform all protein purification steps on ice. Thaw pelleted bacteria, and resuspend in 100 mL lysis buffer. Clarify the lysate by 30 min centrifugation at 20,000 × g following sonication (3 × 5 s). 6. Collect the supernatant and incubate with 0.5 mL Glutathione Sepharose 4B in a glass chromatography column for 1 h. Wash the beads with 3 × 10 mL wash buffer. 7. Elute GST-AKAP fusion protein with 4 × 0.5 mL wash buffer supplemented with 10 mM L-glutathione. 8. Equilibrate ~100 mL Superdex 200 gel filtration column with gel filtration buffer using an FPLC. Inject 2 mL GST-AKAP fusion protein sample, and collect 2 mL fractions (Fig. 3a).

Engineering AKAP-Selective PKA Variants

175

Fig. 3 Purification of GST-AKAP fragment fusion protein. (a) Example elution of a purified GST-AKAP fragment (GST-AKAP-Lbc 1,186–1,246 is shown) from an 80 mL Superdex 200 gel filtration column. ( b ) Coomassie-stained 4–12 % SDS-PAGE gel. Protein collected in fractions C-E serves as the bait in phage selection in this case

9. Assess the purity of GST-AKAP fusion protein in different fractions using Coomassie staining following SDS-PAGE (Fig. 3b). Pool peak fractions and concentrate using Amicon Ultra-15 Centrifugal Filter Units. Calculate protein concentration from absorbance at 280 nm using a NanoDrop 1000, aliquot, and flash freeze in liquid nitrogen for storage at −80 °C. 3.3 Enrichment of Phage Bearing RII D/D Variants with Selectivity for the Target AKAP

1. Make up a competitor AKAP anchoring helix mixture (see Note 12) by diluting every AKAP peptide listed in Table 1, with the exception of the target AKAP, in phage wash buffer at a concentration of 1.5 mg/mL (0.1 μg/μL for each peptide). Aliquot and store at −20 °C. 2. Prior to each round of selection, mix 0.5 μL glutathione magnetic bead slurry with 0.1 μg of the GST-AKAP fragment fusion protein in 200 μL phage wash buffer in a 1.5 mL tube, and shake in the Thermomixer at a mixing frequency of 1,250 rpm for 1 h at 16 °C. 3. Insert the 1.5 mL tube in the DynaMag-2 and aspirate the supernatant. Add 500 μL phage wash buffer, 400 μL competitor peptide mix, and 100 μL amplified phage library (~1 × 1010 phage). Shake at a mixing frequency of 1,250 rpm overnight in the Thermomixer at 16 °C. In addition, inoculate 30 mL LB supplemented with 50 μg/mL carbenicillin with BLT5403 cells from the glycerol stock. Grow overnight at 37 °C with shaking to produce a saturated starter culture. 4. Perform the following wash procedure six times: Insert the 1.5 mL tube in the DynaMag-2 (see Note 13); aspirate the supernatant; add 1 mL phage wash buffer; vortex briefly; apply a pulse of rotation in a benchtop microfuge to remove solution from the lid of the 1.5 mL tube.

176

Ryan Walker-Gray and Matthew G. Gold

Fig. 4 Procedure for enriching phage bearing AKAP-selective RII D/D variants. Prior to each round of selection, magnetic glutathione beads are charged with a GST-AKAP fusion protein corresponding to the target AKAP. The charged beads are incubated with the D/D-phage library in the presence of free competitor AKAP helices. Unbound phages are washed away, and subsequently bound phages are released by incubation with PreScission protease that bisects the GST-AKAP fusion protein. Released phages are amplified and serve as the input in the next round of selection. Released D/D-phage may also be sequenced (see Fig. 5)

5. Resuspend the magnetic beads in 100 μL phage wash buffer supplemented with 0.5 μg PreScission protease and incubate in the Thermomixer shaking at a mixing frequency of 1,250 rpm for 3 h at 25 °C (see Note 14). Simultaneously, transfer 500 μL of the BLT5403 starter culture into 50 mL LB supplemented with 50 μg/mL carbenicillin, and shake for approximately 2 h at 37 °C until OD600nm reaches ~0.8, at which point transfer the culture to 4 °C. 6. Insert the 1.5 mL tube in the DynaMag-2. Collect the 100 μL supernatant. Amplify the eluted phage by adding 95 μL of the eluted phage to 3 mL BLT5403 in a 14 mL round-bottom tube and incubating for 3–4 h until the cells have cleared. Store the remaining 5 μL at 4 °C for plaque sequencing if appropriate. Spin for 10 min at 8,000 × g, collect the supernatant, and store these amplified phage wrapped in aluminum foil at 4 °C. 7. Repeat steps 2–6 (summarized in Fig. 4) until the library converges onto a predominant D/D variant, determined by plaque sequencing once every three rounds of selection.

Engineering AKAP-Selective PKA Variants

3.4 Plaque Sequencing

177

1. Set up a plaque assay (see Subheading 3.1, steps 8–11) using 10−3 to 10−6 dilutions of the eluted phage following a round of selection. 2. Prepare 80 μL of polymerase chain reaction mixture comprising 45.9 μL molecular biology-grade water, 1.6 μL 10 mM dNTPs, 3.2 μL 50 mM MgCl2, 6.4 μL 5 μM T7Select UP primer, 6.4 μL 5 μM T7Select DOWN primer, 16 μL GoTaq Flexi DNA Polymerase 5× colorless buffer, and 0.5 μL GoTaq Flexi DNA Polymerase. Dispense 10 μL fractions into 8 × 0.2 mL thin-walled tubes. 3. Insert a pipette tip precisely into an individual plaque, taken from the top agar of a plaque assay dilution presenting approximately 20 spherical well-separated plaques (Fig. 5a), and transfer phage into the PCR reaction by pipetting the solution twice up and down the tip. 4. Incubate the samples in a thermocycler using the following program: 95 °C for 2 min; 30 cycles of [95 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min 30 s]; 72 °C for 5 min; hold at 4 °C. 5. Add 1 μL of ExoI/SAP mixture (see Note 15) to eight new 0.2 mL tubes, then transfer in 8 × 2.5 μL of the completed PCR reactions from the previous step using a multichannel pipette. Incubate at 37 °C for 15 min, then heat inactivate the enzymes by 15 min incubation at 80 °C. Add 2.5 μL of molecular biology-grade water to each tube. 6. Prepare 63.75 μL of 4/3× BigDye sequencing reaction mix by combining 34 μL molecular biology-grade water, 8.5 μL BigDye terminator v3.1, 17 μL BigDye sequencing buffer, and 4.25 μL 5 μM T7Select DOWN primer. Dispense 7.5 μL fractions

Fig. 5 Plaque sequencing. (a) Transfer of phage into PCR reaction. (b) Removal of dNTP and single-stranded DNA following PCR. (c) BigDye terminator reaction

178

Ryan Walker-Gray and Matthew G. Gold

into eight new 0.2 mL thin-walled PCR tubes. Transfer in 8 × 2.5 μL from the diluted ExoI/SAP-treated sample from the previous step using a multichannel pipette. Incubate the samples in a thermocycler according to the following program: 94 °C for 2 min; 25 cycles of [96 °C for 10 s, 50 °C for 5 s, 60 °C for 3 min]; hold at 4 °C. 7. Transfer each BigDye v3.1 sequencing reaction to 0.5 mL tubes containing 10 μL water, 2 μL 3 M sodium acetate pH 5.2, and 50 μL ethanol. Incubate the samples on ice for 15 min and then spin in a prechilled centrifuge at 4 °C for 15 min at 14,000 × g. 8. Aspirate the supernatant and wash the pelleted DNA with 100 μL chilled 70 % ethanol. Mix the tubes by inversion, and spin for 5 min at 14,000 × g. Aspirate the supernatant and allow tubes to air-dry at room temperature (see Note 16). Submit the samples to a sequencing facility that uses, e.g., an Applied Biosystems 3100 DNA Analyzer. 9. Compare the primary amino acid sequence of the RII variant presented by phage in each plaque. If more than four of the eight sequenced plaques code for the same primary sequence, proceed with validation and application experiments using this RII variant (see Note 17). If the sequences have not converged, continue with phage selection for three further rounds (Subheading 3.3) before repeating plaque sequencing.

4

Notes 1. Order custom oligonucleotide primers (e.g., with TriLink BioTechnologies), requesting synthesis on the micromolar scale with primers purified by making a tight cut on a gel following polyacrylamide gel electrophoresis. 2. The base triplet (A/C/G/T)(A/C/G/T)(G/T) represents all 20 amino acids with only 32 DNA codon variants. 3. These primers amplify the linker sequence SGSGSSGGSG followed by the first 45 amino acids of PKA RIIα. An N-terminal EcoRI site and C-terminal HindIII enable the sequence to be ligated after residue Asn351 of the T7 capsid protein. There is scope for modifying the input library by performing this initial step with different primers. For example, extra or different variable codons could be targeted within RII, or a type I regulatory subunit D/D library could be amplified. 4. UV transilluminators generate light with a wavelength between 254 and 365 nm that causes DNA damage. The Visi-Blue converter plate converts light within this UV band to a spectrum between 400 and 500 nm. The converter can be

Engineering AKAP-Selective PKA Variants

179

used to visualize SYBR Safe-stained DNA with Safe Imager viewing glasses, which results in improved ligation efficiency. 5. The purified DNA insert should exhibit a 260/280 nm absorbance ratio of ≥1.8 and a 260/230 nm absorbance ratio of ≥2. A lower ratio at 260/230 nm indicates carry-over of contaminants from the gel extraction process. We find that including an additional wash step with PB buffer during extraction using the Qiagen gel extraction kit markedly improves DNA purity. 6. Although the T7Select manual recommends vector to insert ratios of between 1:1 and 1:3, we find that the higher ratio of 1:5 is optimal. 7. If no bacteria are visible on the LB plates 3–4 h after the plaque assay, it is possible that the top agar was not cooled sufficiently, resulting in the death of the BLT5403 cells. Phage contamination of the BLT5403 cells prior to the plaque assay leads to the same result. 8. It is important to rapidly add the top agar and plate out once the phage dilutions have been added to the BLT5403 cell aliquots. Delays will allow phage to amplify prior to plating leading to erroneous phage titers. 9. There are 324 = 1,048,576 total DNA variants, thus the fractional library coverage = 1 – (1,048,575/1,048,576)n, where n is the packaging number. For a typical packaging efficiency of 8 × 107 inserts, it is therefore very likely that all possible DNA variants are represented in the library. 10. pGEX6P1 is a vector for bacterial expression of proteins with glutathione S-transferase (GST) and the PreScission protease recognition sequence fused to their N-terminus. These elements enable transgenic sequences to be immobilized to Glutathione Sepharose and glutathione magnetic beads via GST and released into solution when desired by addition of PreScission protease. 11. BL21 Star strains typically enable higher protein yields than BL21 strains due to the increased stability of mRNA due to a mutation in the RNaseE gene. 12. The concentration of the competitor peptides may be varied. Higher competitor peptide concentration will enrich phage more on the basis of their relative affinity rather than their absolute affinity for the target AKAP and vice versa. Since new AKAPs are still being identified, future screens may incorporate additional AKAP competitors. Alternatively, the competitor list may be limited to the AKAPs present in a given tissue with the intention of developing an AKAP-selective RII variant to be applied in a specific tissue.

180

Ryan Walker-Gray and Matthew G. Gold

13. The most effective method for aspiration is to slide a P10 pipette tip, attached to a vacuum, immediately to the bottom of the 1.5 mL tube down the side that is opposite the magnet. This approach limits unwanted aspiration of beads along with buffer. 14. PreScission™ protease (GE Healthcare) cleaves at the sequence LEVLFQ/GP. Since PreScission™ protease contains a noncleavable GST-tag, it is retained on the glutathione magnetic beads following release of the AKAP fragment and associated phage. 15. ExoI digests single-stranded DNA, and SAP dephosphorylates dNTPs. These molecules would otherwise carry over from the PCR reaction and interfere with the BigDye reaction. 16. Allow the samples to air-dry slowly at room temperature, since drying in a heating block can lead to salt crystal formation that interferes with sequencing. 17. The AKAP selectivity of identified RII variant sequences may be validated by approaches including pull-down and competition binding assays using purified proteins and live-cell imaging experiments [9]. References 1. Gold MG, Gonen T, Scott JD (2013) Local cAMP signaling in disease at a glance. J Cell Sci 126:4537–4543 2. Scott JD, Pawson T (2009) Cell signaling in space and time: where proteins come together and when they’re apart. Science 326: 1220–1224 3. Gold MG, Lygren B, Dokurno P et al (2006) Molecular basis of AKAP specificity for PKA regulatory subunits. Mol Cell 24: 383–395 4. Kinderman FS, Kim C, von Daake S et al (2006) A dynamic mechanism for AKAP binding to RII isoforms of cAMP-dependent protein kinase. Mol Cell 24:397–408 5. Means CK, Lygren B, Langeberg LK et al (2011) An entirely specific type I A-kinase anchoring protein that can sequester two molecules of protein kinase A at mitochondria. Proc Natl Acad Sci U S A 108:E1227–E1235 6. Rosenmund C, Carr DW, Bergeson SE et al (1994) Anchoring of protein kinase A is required for modulation of AMPA/kainate

7.

8.

9.

10.

11.

receptors on hippocampal neurons. Nature 368:853–856 Alto NM, Soderling SH, Hoshi N et al (2003) Bioinformatic design of A-kinase anchoring protein-in silico: a potent and selective peptide antagonist of type II protein kinase A anchoring. Proc Natl Acad Sci U S A 100:4445–4450 Carlson CR, Lygren B, Berge T et al (2006) Delineation of type I protein kinase A-selective signaling events using an RI anchoring disruptor. J Biol Chem 281:21535–21545 Gold MG, Fowler DM, Means CK et al (2013) Engineering A-kinase anchoring protein (AKAP)-selective regulatory subunits of protein kinase A (PKA) through structure-based phage selection. J Biol Chem 288:17111–17121 Cole C, Barber JD, Barton GJ (2008) The Jpred 3 secondary structure prediction server. Nucleic Acids Res 36:W197–W201 Grote A, Hiller K, Scheer M et al (2005) JCat: a novel tool to adapt codon usage of a target gene to its potential expression host. Nucleic Acids Res 33:W526–W531

Chapter 14 A Yeast-Based High-Throughput Screen for Modulators of Phosphodiesterase Activity Ana Santos de Medeiros and Charles S. Hoffman Abstract Cell-based high-throughput screens (HTSs) targeting heterologously expressed proteins in yeast identify compounds that often display relevant biological activity when tested in cell culture. We developed a fission yeast-based HTS to detect small-molecule inhibitors of mammalian cyclic nucleotide phosphodiesterases (PDEs). These screens are carried out in Schizosaccharomyces pombe using a PKA-repressed fbp1-ura4 reporter whose expression due to low PKA activity prevents cells from growing in medium containing the pyrimidine analog 5-fluoro orotic acid (5FOA). We describe here the steps required to construct strains for screening and to optimize conditions for successful screens. Key words Cyclic nucleotide phosphodiesterase, Fission yeast, Schizosaccharomyces pombe, fbp1, High-throughput screen, Inhibitors

1

Introduction Glucose signaling in the fission yeast Schizosaccharomyces pombe acts via a cAMP pathway to regulate PKA [1]. The fbp1 gene is transcriptionally repressed by PKA activity. Cells with low PKA activity show a 200-fold increase in fbp1 expression relative to those with high PKA activity [2, 3]. The identification of genes required for glucose detection and PKA activation relied on the use of a translational fusion of the fbp1 promoter and a portion of the open reading frame to the S. pombe ura4 gene, whose expression is required for growth on medium lacking uracil, but is toxic when cells are grown in medium containing the pyrimidine analog 5-fluoro-orotic acid (5FOA) [2]. This allowed for the isolation of strains that are defective in glucose signaling by their ability to form colonies on glucose-rich solid medium lacking uracil. It also allowed for the cloning of genes that restored glucose repression by their ability to confer 5FOA-resistant (5FOAR) growth [1]. HTSs for small-molecule inhibitors of heterologously expressed PDEs also make use of the fbp1-ura4 reporter. By establishing conditions in

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_14, © Springer Science+Business Media New York 2015

181

182

Ana Santos de Medeiros and Charles S. Hoffman

which the expressed PDE is responsible for the low PKA, 5FOAsensitive (5FOAS) phenotype, one can screen for compounds that inhibit the PDE to restore 5FOAR growth [4]. As exogenous cGMP can be taken up by S. pombe to activate PKA [5], this screen can detect inhibitors of both cAMP- and cGMP-hydrolyzing PDEs and has led to the identification of PDE4, PDE7, PDE8, and PDE11 inhibitors that show biological activity in cell culture [4, 6–8]. This cell-based assay allows for inexpensive HTSs to detect and characterize small-molecule inhibitors of cloned PDEs expressed in S. pombe.

2

Materials All media lacking 5FOA or cyclic nucleotides can be stored at room temperature, while media containing 5FOA or cyclic nucleotides should be stored at 4 °C. When preparing commercially formulated growth media such as Edinburgh Minimal Medium (EMM), note whether or not autoclaving is an acceptable form of sterilization. Some formulations have heat-labile components and require filter sterilization. Prepare cyclic nucleotide-containing solutions at room temperature and use fresh, although short-term storage at 4 °C is acceptable. Filter sterilize solutions using a syringe filter-driven unit for small volumes or a bottle-top filter for large volumes.

2.1

Media

1. 5FOA medium: For liquid medium, combine 950 mL distilled H2O with 80 g glucose, 1.45 g yeast nitrogen base w/o amino acids and w/o (NH4)2SO4, 5 g ammonium sulfate, 0.4 g 5FOA, 2 g SC-uracil dropout mix (see Note 1), and 50 mg uracil. Dissolve by stirring with low heat. Filter sterilize. For solid medium, make a 2× concentrated nutrient solution in half the volume. After filter sterilization, combine with 490 mL water plus 20 g Bacto agar that has been autoclaved in a 2 L flask (see Note 2). 2. EMM complete medium: For liquid medium, combine 970 mL distilled H2O with 12.33 g EMM w/o dextrose and 30 g glucose. Add 150 mg leucine and 75 mg each of adenine, histidine, lysine, and uracil. Dissolve by mixing under low heat and filter sterilize. For solid medium, make a 2× concentrated nutrient solution in half the volume. After filter sterilization, combine with 490 mL water plus 20 g Bacto agar that has been autoclaved in a 2 L flask. 3. Cyclic nucleotides: Add 32.9 mg cAMP (for 10 mM stock) or 18.6 mg of cGMP (for 5 mM stock) per 10 mL growth medium. Mix and make sure cyclic nucleotide is fully dissolved before filter sterilizing (see Note 3). 4. Small-molecule PDE modulators: Bring water in a beaker to a boil. Place Eppendorf tube containing medium into water for

Yeast-based HTS for PDE Modulators

183

30 s to 1 min before adding compounds that are dissolved in DMSO (generally 20–100 mM stock solutions). Mix well, but avoid vortexing as some compounds will come out of solution (see Note 4). 2.2

Equipment

1. 5FOA-based growth screens for PDE inhibitors use 384-well clear sterile dishes such as the Corning 3680 assay plate. 2. Multichannel pipettes (16 channels) that hold up to 50 μL are useful in delivering medium and cells to wells of a microtiter dish. 3. When large numbers of wells need to be filled with medium, consider using a liquid handler such as the Thermo Multidrop 384 or Wellmate microplate dispenser.

3

Methods These HTSs are not limited to use with the 15 mammalian PDE genes and three Trypanosome PDE genes that we have introduced into our strain collection. Construction of strains expressing new PDE genes is relatively straightforward.

3.1 Construct Yeast Strains Expressing PDE of Interest

1. PCR amplify the PDE gene of interest using 80-mer oligonucleotides that contain about 20 bases of sequence at the 3′ ends to target amplification of the PDE of interest and 60 bases of sequence at the 5′ ends to target insertion into the locus to be used for expression. We can provide S. pombe expression vectors that use promoters of varying strength such as the strong adh1 and nmt1 promoters, the moderately active nmt41 and tif471 promoters, or the weakly active cgs2 and lys7 promoters. We have also targeted insertion directly into the S. pombe cgs2 PDE gene locus [4] (see Note 5). 2. PCR products carrying the PDE gene of interest are introduced into expression vectors by gap-repair transformation [9] in which the PCR product and a linearized vector are cotransformed into a host S. pombe strain that lacks PDE activity. By using a strain of the h90 homothallic mating type, one can detect PDE activity as a function of increased mating. 3. Select for transformants based on the selectable marker in the cloning vector (e.g., transformants carrying LEU2-marked plasmids are selected for on EMM medium lacking leucine). Screen for increased mating by inverting the plate of colonies over finely crushed iodine to produce a vapor that stains asci that form in mating-competent colonies (see Note 6). 4. Rescue candidate plasmid from yeast to E. coli [10] in order to purify plasmid and confirm by DNA sequence analysis that it carries the desired PDE gene.

184

Ana Santos de Medeiros and Charles S. Hoffman

5. Linearize the plasmid in a region that is homologous to a portion of the S. pombe genome and transform a host strain so as to generate a transformant carrying a stably integrated plasmid (see Note 7). 6. Construct strains by genetic crosses and tetrad dissection to combine the PDE of interest with the fbp-ura4 reporter and other mutations that will facilitate the screen (see Note 8). The desired level of PKA activity in the host strain will be low when preparing for a PDE inhibitor screen (see Note 9). 7. Pilot the 5FOA assay to determine the optimal conditions for the HTS as described in Subheading 3.2. 3.2 Pilot 5FOA Assays 3.2.1 Strains That Lack Adenylyl Cyclase Activity

This procedure will identify the optimal concentration of cAMP or cGMP for use in a HTS. This concentration will cause a small increase in the OD600 of a culture growing in a microtiter dish well. For example, one should identify conditions in which the culture reaches an OD600 of 0.07 in the absence of added cyclic nucleotide and 0.1–0.2 in the presence of cyclic nucleotide. Under such conditions, inhibition of the expressed PDE will likely produce an OD600 of ~1.2 (Fig. 1).

Fig. 1 5FOA growth response to exogenous cAMP in strains lacking adenylyl cyclase activity. Strains expressing wild-type human PDE7B1 and two mutant alleles of PDE7B1 were subjected to a 5FOA growth assay in medium containing various concentrations of cAMP. As seen by their relative sensitivity to cAMP, PDE7B1-D7 is less active than wild-type PDE7B1, while PDE7B1-G21 is more active than wild-type PDE7B1. Strains that lack PDE activity reach an OD600 of 1.2 in the presence of ~0.025 mM cAMP [5]

Yeast-based HTS for PDE Modulators

185

1. Transfer freshly growing cells from solid medium to 5 mL EMM complete liquid medium in an 18 × 150 mm culture tube and grow overnight at 30 °C with shaking. 2. Count cells using a hemacytometer and dilute into fresh medium that will allow for repression of the fbp1-ura4 reporter and will achieve a cell concentration of 1 × 107 cells/mL the following day (see Notes 10 and 11). 3. Pipet 25 μL of 5FOA medium into wells in columns 2–16 of a 384-well microtiter dish. 4. Pipet 75 μL of 5FOA medium containing a cyclic nucleotide (10 mM cAMP or 5 mM cGMP) into wells in column 1 of a 384-well microtiter dish. 5. Using a multichannel pipette, transfer 50 μL from column 1 to column 2 and proceed to perform 2/3 serial dilutions in wells of columns 2–14 (remove 50 μL of medium from column 14 after mixing and discard to leave 25 μL in the well). Do not add cyclic nucleotides to columns 15 and 16 that serve as negative controls (see Note 12). 6. Pellet cells by centrifugation at ~1,000 × g for 5 min and resuspend in 5FOA medium lacking cyclic nucleotides to a density of 3 × 105 cells/mL (see Note 13). 7. Pipet 25 μL cells into each well. Avoid making bubbles in the medium that would affect the OD600 reading (see Note 14). 8. Incubate microtiter dishes for 48 h at 30 °C in a sealed container with wet paper towels to reduce evaporation in the wells. Sandwich the dishes containing cells between two blank microtiter dishes to reduce evaporation of condensation on the lids of the experimental dishes. 9. After incubation, vortex using a microtiter plate vortex to resuspend cells (alternatively, cells can be resuspended by pipetting). Determine the optical densities in the wells using a plate reader (see Note 15). 3.2.2 Cells That Express a Functional Adenylyl Cyclase Gene

Strains that express the git2 adenylyl cyclase gene, but are 5FOAS due to a combination of PDE activity and mutations in genes involved in adenylyl cyclase activation do not require the addition of cyclic nucleotides to the 5FOA growth medium. However, cyclic nucleotides or a PDE inhibitor is still required in the EMM medium used to grow the cultures prior to the screen so that the fbp1-ura4 reporter is repressed at the start of the HTS. A slightly modified protocol is needed to optimize the 5FOA assay conditions to be used in such HTSs. 1. Follow steps 1 and 2 as described in Subheading 3.2.1. Use either a cyclic nucleotide or an inhibitor of the target PDE to activate PKA during the growth of the cultures.

186

Ana Santos de Medeiros and Charles S. Hoffman

2. Pipet 25 μL of 5FOA medium containing either (1) no cyclic nucleotide (negative control), (2) 10 mM cAMP or 5 mM cGMP (cyclic nucleotide positive control), or (3) 40 μM of a known PDE inhibitor in 0.5 for an effective screen. It is calculated by subtracting three times the sum of the standard deviations of positive (σp) and negative (σn) controls, divided by the absolute value of the difference between means of the positive (μp) and negative (μn) controls, from 1: Z¢ factor = 1 -

3( s p + s n ) mp - mn

17. Adjusting the starting concentration of cells in a 5FOA assay is the best way to optimize a HTS. If the starting density is too low, the positive controls may not grow to saturation. If the starting density is too high, the negative controls may have a relatively high OD600 and a large standard deviation, both of which reduce the Z′ factor of the assay.

190

Ana Santos de Medeiros and Charles S. Hoffman

18. A Z score is a statistical assessment of the significance of an individual data point in a HTS. The Z score is determined by taking the experimental sample value and subtracting the mean of the negative controls. This is then divided by the standard deviation of the negative controls. Candidate hits should have Z scores of at least 6 to be considered statistically significant. Depending upon the quality of the HTS, one may choose a much larger cutoff when identifying the most promising candidates.

Acknowledgments This work was supported by NIH grant GM079662, the Peter Rieser Lectureship Fund, and a grant from Boston College to C.S.H. References 1. Hoffman CS (2005) Glucose sensing via the protein kinase A pathway in Schizosaccharomyces pombe. Biochem Soc Trans 33:257–260 2. Hoffman CS, Winston F (1990) Isolation and characterization of mutants constitutive for expression of the fbp1 gene of Schizosaccharomyces pombe. Genetics 124:807–816 3. Hoffman CS, Winston F (1991) Glucose repression of transcription of the Schizosaccharomyces pombe fbp1 gene occurs by a cAMP signaling pathway. Genes Dev 5:561–571 4. Ivey FD, Wang L, Demirbas D et al (2008) Development of a fission yeast-based highthroughput screen to identify chemical regulators of cAMP phosphodiesterases. J Biomol Screen 13:62–71 5. Demirbas D, Ceyhan O, Wyman AR et al (2011) Use of a Schizosaccharomyces pombe PKArepressible reporter to study cGMP metabolising phosphodiesterases. Cell Signal 23:594–601 6. Alaamery MA, Wyman AR, Ivey FD et al (2010) New classes of PDE7 inhibitors identified by a fission yeast-based HTS. J Biomol Screen 15:359–367

7. Ceyhan O, Birsoy K, Hoffman CS (2012) Identification of biologically active PDE11selective inhibitors using a yeast-based highthroughput screen. Chem Biol 19:155–163 8. Demirbas D, Wyman AR, Shimizu-Albergine M et al (2013) A yeast-based chemical screen identifies a PDE inhibitor that elevates steroidogenesis in mouse leydig cells via PDE8 and PDE4 inhibition. PLoS One 8:e71279 9. Ivey FD, Taglia FX, Yang F et al (2010) Activated alleles of the Schizosaccharomyces pombe gpa2+ Gα gene identify residues involved in GDP-GTP exchange. Eukaryot Cell 9: 626–633 10. Hoffman CS, Winston F (1987) A ten-minute DNA preparation from yeast efficiently releases autonomous plasmids for transformation of Escherichia coli. Gene 57:267–272 11. Wang L, Griffiths K Jr, Zhang YZ et al (2005) Schizosaccharomyces pombe adenylate cyclase suppressor mutations suggest a role for cAMP phosphodiesterase regulation in feedback control of glucose/cAMP signaling. Genetics 171:1523–1533

Chapter 15 Separation of PKA and PKG Signaling Nodes by Chemical Proteomics Eleonora Corradini, Albert J.R. Heck, and Arjen Scholten Abstract The chemically quite similar cyclic nucleotides cAMP and cGMP are two second messengers that activate the homologous cAMP- and cGMP-dependent protein kinases (PKA and PKG, respectively). To gain specificity in space and time in vivo, PKA is compartmentalized by the interaction of its regulatory subunits with A-kinase-anchoring proteins (AKAPs), which often form the core of larger signaling protein machineries. In a similar manner, PKG is also found to be compartmentalized close to specific, local pools of cGMP through interaction with G-kinase-anchoring proteins (GKAPs), although the extent and mechanisms mediating these interactions are only marginally understood. In affinity-based chemical proteomics strategies, small molecules are immobilized on solid supports in order to enrich for specific target proteins. We have shown the utility of immobilized cAMP and cGMP to enrich for PKA and PKG and their associated proteins. Unfortunately, both PKA and PKG are enriched in the pull downs with both immobilized compounds. Although this proved sufficient to identify novel AKAPs, the lower abundance of PKG has seriously hampered the enrichment and identification of novel GKAPs. Here we present an improved chemical proteomics method involving in-solution competition with low doses of different free cyclic nucleotides to segregate the cAMP/PKA- and cGMP/PKG-based signaling nodes, allowing the purification and subsequent identification of new scaffold proteins for PKG. Key words Chemical proteomics, cAMP, cGMP, PKA, AKAP, PKG, GKAP, Mass spectrometry

1

Introduction The cyclic nucleotides adenosine- and guanosine 3′,5′-cyclic monophosphate (cAMP and cGMP) are intracellular second messengers produced in response to hormone action. They play important roles in many cellular processes. Moreover, they control key physiological processes in the cardiovascular system such as cardiac contraction, vasorelaxation, and platelet signaling [1–5]. The major intracellular target of cAMP is the cAMP-dependent protein kinase (PKA), a heterotetramer composed of a regulatory subunit dimer (from one of four possible isoforms, PKA-RIα, RIβ, RIIα, and RIIβ) and two catalytic subunits (from one of three possible

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_15, © Springer Science+Business Media New York 2015

191

192

Eleonora Corradini et al.

isoforms, PKA-Cα, Cβ, Cγ) [6]. Compartmentalization of PKA’s actions are controlled through colocalization with its substrates via binding of the PKA-R dimer to distinct scaffold proteins named A-kinase-anchoring proteins (AKAPs) [7, 8]. In parallel, cGMPmediated signaling occurs mainly via the cGMP-dependent protein kinases (PKGIα, PKGIβ, and PKGII), which are PKA’s closest homologs [9]. The domain architecture of PKG is very similar to PKA, except that PKG is a homodimer with the two monomers held together by a leucine zipper at the N-terminus, while the PKA-R dimer consists of a four helix bundle. The N-terminus is different in the two cytosolic isoforms of PKGI (α and β) and binds to different PKG-anchoring proteins (GKAPs) to gain specificity in space and time, similarly to PKA and AKAPs [10, 11]. Both PKA and PKG are expressed in low abundance when compared to typical housekeeping proteins. Therefore, in order to study them, we need to isolate the kinases and their interacting scaffolds using specific enrichment techniques, such as the chemical proteomics approaches reported by us previously [12, 13]. The use of immobilized cyclic nucleotides as baits for PKA and its interactors has recently led to the discovery of several new AKAPs [13, 14]. The enrichment of the even less abundant GKAPs has proven more challenging, requiring the development of more specific methodology. In vitro, both cAMP and cGMP can activate PKG and PKA, respectively, albeit at a concentration around a ~100-fold higher than the Ka reported for their “own” kinase [15]. In vivo, strict compartmentalization of the kinases and their activating cyclic nucleotide likely prevents cross-reactivity. Nevertheless, in our chemical proteomics approach, we use a very high local concentration of either cAMP or cGMP on the surface of our affinity beads, leading to the simultaneous enrichment of both PKA and PKG with either resin. To dissect PKA- and PKG-driven local signaling complexes, here we present an improved chemical proteomics method based on a cAMP-resin and in-solution competition with a low dose of free cyclic cAMP or cGMP (Fig. 1). The chosen doses of free cAMP and cGMP are such that they are high enough to fully occupy their target kinase, but low enough that they do not bind their off-target kinase. We coupled this approach with quantitative, label-free liquid chromatography–high resolution mass spectrometry (LC-MS/MS) to separate and characterize cAMP from cGMP-directed signaling nodes. As a proof of principle, we were able to isolate PKA and many known AKAPs, from PKG when spiking competing concentrations of free cGMP and of free cAMP in the lysate. At the same time, we were able to isolate PKG and IRAG (one of the few known GKAPs) and, more importantly, several novel candidate GKAPs.

Proteomics of PKA and PKG Signaling

193

Fig. 1 Experimental setup. (a) Schematic representation of the chemical proteomics pull down using insolution competition with free cyclic nucleotides for the identification of novel possible GKAPs. (b) Coomassie stained SDS page of the pull downs performed in Hek293 cell line. A concentration of 10 μM of cAMP or cGMP is sufficient to compete PKA or PKG off the beads, and to leave the other kinase and its interactors on the beads

2

Materials

2.1 Cell and Tissue Lysis

1. Protease inhibitor cocktail (Complete mini, Roche Applied Science). 2. Phosphatase inhibitor cocktail (PhoSTOP, Roche Applied Science). 3. Phosphate-buffered saline (PBS), 10× stock: 1.37 M NaCl, 27 mM KCl, 100 mM KH2PO4. Adjust to pH 7.4 with HCl if necessary. 4. PBS, working solution: dilute one part of PBS 10× with nine parts of Milli-Q water. 5. Lysis buffer: PBS buffer, 0.1 % Tween 20, protease inhibitor cocktail (one tablet per 10 ml of buffer) and phosphatase inhibitor cocktail (one tablet per 10 ml of buffer) (see Note 1). 6. Cell line of human origin (e.g., HeLa, Hek293, Jurkat). 7. Tissues of mammalian origin (e.g., mouse, rat).

194

Eleonora Corradini et al.

2.2 cAMP-Affinity Pull Down

1. 8AHA-cAMP agarose beads (Biolog, CAT. NO. A 028). Prepare a 25 % bead slurry in PBS containing 0.1 % NaN3. Store at 4 °C (see Note 2). 2. Stock solutions: 100 mM adenosine 5′-diphosphate (ADP) sodium salt, 100 mM guanosine 5′-diphosphate (GDP) sodium salt (Sigma-Aldrich), 50 mM cAMP and 50 mM cGMP (Biolog) and store at 4 °C (see Note 3). 3. Washing buffers: 10 mM ADP and GDP in lysis buffer; 10 mM ADP, 10 mM GDP and 5 μM cAMP in lysis buffer; 10 mM ADP, 10 mM GDP and 5 μM cGMP in lysis buffer; PBS. Prepare from the stock solutions by diluting with lysis buffer. 4. Micro Bio-Spin™ Chromatography Columns (Bio-Rad).

2.3

SDS-PAGE

1. 4–12 % Bis–Tris gel (Bio-Rad). 2. XT-Mops running buffer (Bio-Rad). 3. Fixing solution: methanol:acetic acid: Milli-Q water, 50:10:40. 4. GelCode Blue Stain Reagent (Thermo Scientific).

2.4

In-Gel Digestion

1. 50 mM ammonium bicarbonate (ABC) in Milli-Q water, pH 8.5. 2. Acetonitrile (ACN). 3. 6.5 mM dithiothreitol (DTT) solution and 50 mM iodoacetamide (IAA) solution in 50 mM ABC for single use. 4. Trypsin (Sequencing grade, Roche Applied Science): 3 ng/μl in cold ABC.

2.5 LC–MS/MS Analysis

1. Suitable high-resolution mass spectrometer, e.g., Q-Exactive equipped with an electrospray ion source (Thermo, San Jose, CA) coupled online to a nano-UHPLC system, e.g., Proxeon Easy-nLC 1000 (Thermo Scientific). 2. Double fritted trapping column, 2 cm, 100 μm inner diameter, packed with 3 μm C18 resin (Dr Maisch Reprosil). 3. Analytical column, 50 cm, 75 μm inner diameter packed with 1.8 μm C18 resin (Agilent Zorbax SB-C18). 4. Distally coated fused-silica emitter (360 μm outer diameter, 20 μm inner diameter, 10 μm tip inner diameter). 5. HPLC solvent A: 0.1 % formic acid in Milli-Q water. 6. HPLC solvent B: 0.1 % formic acid in ACN.

2.6

Data Analysis

1. Proteome discoverer (version 1.3, Thermo Scientific, Bremen, Germany). 2. Mascot (version 2.3.02 Matrix Science, London, UK).

Proteomics of PKA and PKG Signaling

3

195

Methods

3.1 Preparation of Cell and Tissue Lysates for cAMPAffinity Pull Down with In-Solution Cyclic Nucleotide Competition 3.1.1 Cell Lysis

1. Harvest the cells at a density of 1–2 × 106 cells/ml by centrifugation (5 min, 190 × g ). Discard the supernatant and wash the pellet with PBS at 37 °C. Centrifuge again and discard the supernatant. The yield in protein content can vary depending on the cell line (e.g., 1–10 × 106 HeLa cells yield ~1 mg total protein). 2. Resuspend the pellet in 1 ml ice-cold lysis buffer. Transfer the cell suspension to a prechilled (ice) glass dounce homogenizer. 3. Lyse the cells using dounce homogenization on ice for 1–2 min (~20–30 strokes). Transfer the lysate to a prechilled Eppendorf tube. 4. Centrifuge the lysate at 4 °C at 20,000 × g for 10 min. Carefully transfer the soluble fraction to a new prechilled 15 ml falcon tube . Resuspend the pellet in 500 μl of lysis buffer and repeat the dounce homogenization in step 3. 5. Repeat centrifugation in step 4. Combine the supernatants from steps 3 and 4 (see Note 4). 6. Measure the protein concentration of the soluble fraction with an appropriate assay (e.g., Bradford, BCA, etc.).

3.1.2 Tissue Lysis

1. Prechill a steel mortar, pestle, and spoon in liquid nitrogen. 2. Pulverize the frozen tissue in the cold steel mortar with the cold pestle. 3. Use the cold spoon to transfer the pulverized tissue to a prechilled Eppendorf tube. 4. Add 1 ml of ice-cold lysis buffer and leave it on ice for 15 min. 5. Centrifuge the lysate at 4 °C at 20,000 × g. Carefully transfer the soluble fraction to a new prechilled falcon tube. The insoluble pellet can be washed several times as described in steps 3 and 4 of Subheading 3.1.1. 6. Measure the protein concentration of the soluble fraction with an appropriate assay (e.g., Bradford, BCA, etc.). The lysate is now ready for the pull-down assay (see Note 5).

3.2 Free Cyclic Nucleotide In-Solution Competition of cAMPAffinity Pull Down

1. Prior to the pull-down assay, add ADP and GDP solutions (100 mM) to the lysate to achieve a final concentration of 10 mM in a final protein concentration of 2 mg/ml cell lysate. 2. Divide the sample in three equal fractions containing at least 5 mg total protein each. 3. Add cAMP to a final concentration of 10 μM to one sample, cGMP with a final concentration of 10 μM to a second sample, use a third sample as control, and incubate at 4 °C under agitation for 30 min (see Note 6).

196

Eleonora Corradini et al.

4. Transfer 600 μl of bead slurry (150 μl of dry beads) in a new Eppendorf tube and wash it with 500 μl of cold lysis buffer (see Note 7). 5. Remove the supernatant from the beads and repeat step 4 twice. 6. After the last wash, do not remove the supernatant and divide the bead slurries into 3 equal portions of 50 μl of dry beads. Remove the supernatants. 7. Add the three protein mixtures to the three equal portions of beads to yield a protein:beads ratio of 5 mg protein:50 μl of dry beads and incubate at 4 °C under agitation for 2 h. 8. Transfer the beads/lysate mixtures onto three Micro BioSpin™ Chromatography Columns, and let all the supernatants flow through by gravity. 9. Wash the three different pull downs with 1 ml of the appropriate ice-cold washing buffer (containing either free cAMP, or free cGMP or normal washing buffer for the control, see Note 8). 10. Repeat the washes in step 9 three times. 11. Wash each column three times with 1 ml ice-cold PBS. 12. Using the liquid volume of the last PBS wash, transfer the beads to a new Eppendorf tube. Spin the beads down at 1,000 rpm for 1 min and remove the supernatant (see Note 9). Elute the proteins by adding 50 μl of SDS loading buffer to the beads and boiling the samples at 95 °C for 10 min. The denatured protein samples are now ready for SDS-PAGE gel loading (see Note 10). 3.3 SDSPolyacrylamide Gel Electrophoresis

1. Set up a 4–12 % Bis–Tris SDS-PAGE, load 40 μl of each of the three samples and separate the proteins by applying 200 V across the gel for 1 h. Leave one empty lane in between each sample. 2. Remove the gel from the cassette, wash the gel with Milli-Q water, and fix the proteins on the gel using fixing solution. 3. Rinse the gel with Milli-Q water and stain the gel with GelCode Blue Stain Reagent for 1 h. 4. Remove the staining solution and rinse the gel with Milli-Q water for at least 4 h.

3.4

In-Gel Digestion

1. Different bands corresponding to different molecular weights are excised from the three gel lanes in an identical way and subjected to in-gel digestion. Each excised band is carefully chopped into 1 mm cubes, placed in 1.5 ml Eppendorf, and washed with Milli-Q water (see Note 11).

Proteomics of PKA and PKG Signaling

197

2. Discard the supernatant and add 100 μl of ACN, vortex the samples and leave them at room temperature until they shrink (~15 min). 3. Remove the ACN. Reduce the protein disulfide bonds with 100 μl of 6.5 mM DTT, at 60 °C for 1 h. Remove the excess of DTT, add 100 μl of ACN to dehydrate the gel pieces, and leave them at room temperature for 15 min. 4. Remove the ACN and alkylate the previously reduced cysteines with 50 mM IAA, at room temperature in the dark (see Note 12). 5. Remove the supernatant and dehydrate the gel pieces with 100 μl of ACN and leave them at room temperature for 15 min. Remove the ACN and rehydrate the gel with 100 μl of ABC. 6. Repeat step 5 and dehydrate the samples with 100 μl of ACN (see Note 13). 7. Add 20 μl of trypsin (3 ng/μl in ABC) to the samples and leave them on ice for 90 min. 8. Remove the excess of trypsin and add 50 μl of ABC to the samples and digest them overnight at 37 °C. 9. Extract the peptides from the gel pieces twice with 100 μl ACN for 15 min at room temperature. Combine the supernatants and discard the gel pieces. Dry the combined supernatants under vacuum. 10. Resuspend the samples with 30 μl of 10 % formic acid, 5 % DMSO (see Note 14). The samples are ready for LC–MS/MS analysis. 3.5 LC–MS/MS Analysis

1. Separate the peptides on a nanoflow ultra high pressure liquid chromatography setup (Proxeon Easy-nLC 1000) coupled with a Q-Exactive mass spectrometer. 2. A volume of 15 μl of the resuspended samples is used for the subsequent analysis. 3. The injected peptides are trapped with a double-fritted trapping column at a pressure of 800 bar with 100 % solvent A, and elution of the peptides is achieved with a gradient of 7–30 % solvent B in 35 min at a flow rate of 150 nl/min, in a total analysis time of 50 min. Nanospray is achieved using a goldcoated fused silica capillary biased to 1.7 kV. 4. The Q-Exactive mass spectrometer is operating in a datadependent mode. Full-scan MS spectra are acquired in the Orbitrap with a resolution of 35,000 after accumulation to a target value of 3,000,000. The ten most intense peaks are fragmented using normalized collision energy of 25 % after accumulation to a target value of 50,000 at a resolution of 17,500.

198

3.6

Eleonora Corradini et al.

Data Analysis

1. Create the peak list for each raw data file recorded by the mass spectrometer using Proteome Discoverer, using a standardized workflow. 2. The Mascot search is performed using a species appropriate database and settings in line with the used LC–MS setup. 3. Peak lists are searched against a concatenated forward-decoy Swissprot database with the following criteria: carbamidomethylation on cysteine residues as fixed modification and methionine oxidation as variable modifications. After identification the data are further filtered with the following criteria: mass deviation of ±10 ppm, Mascot Ion Score of at least 20, a minimum of six amino acid residues per peptide, and position rank 1. In order to identify new anchoring protein candidates, it is possible to use spectral count quantitation [12]. First calculate the enrichment ratio by dividing the Peptide Spectral Matches (PSMs) of the proteins identified in the pull down supplemented with cGMP by the PSMs observed of the same protein in the pull down supplemented with cAMP (Fig. 2) (see Note 15). Use the pull down where no cyclic nucleotides were supplemented as positive control. To define an appropriate cutoff it is possible to use the lowest PSMs ratio of the known AKAPs and GKAPs.

4

Notes 1. PBS buffer with 0.1 % Tween 20 can be prepared 24 h in advance and stored at 4 °C as the buffer should be cold when used. Lysis buffer should be prepared for single use by adding protease and phosphatase inhibitor cocktails only prior to use, in order to keep the efficiency of the inhibitors. 2. 0.1 % NaN3 is necessary to prevent bacteria and fungi grow in the bead slurry. 3. cAMP and cGMP stock solutions should be prepared only prior to use in order to keep their cyclic structures. 4. In order to remove the DNA from the lysate, filtering the lysate through a 0.45 μm filter is recommended. 5. Before pull down, dilute the lysate to a final protein concentration of 2 mg/ml. 6. Supplement the lysate prior to the pull down as well as the washing buffer of the first three washes after the pull down with 10 mM ADP and GDP and 10 μM cAMP or cGMP in order to reduce the interaction both with noncyclic nucleotidebinding proteins and remove completely PKA or PKG from

Proteomics of PKA and PKG Signaling

199

Fig. 2 Ratio vs. spectral counts plot. Ratioplot as illustrative example of the identification of the AKAPs and GKAPs in rat lung. The protein ratios are calculated by dividing the PSMs obtained for each protein in the pull down supplemented with free cGMP, by the PSMs obtained in the pull down supplemented with free cAMP. The ratios are plotted against the sum of the PSMs from both the pull downs. Upon competition with free cGMP (right panel), the plot clearly shows the enrichment of the PKA regulatory subunits and the known AKAPs. In contrast, when competed with free cAMP the pull down shows the opposite trend (left panel), with the enrichment of PKG, the known GKAP, IRAG, and new possible GKAP candidates

the beads. We recommend at least 1 ml of washing buffer per 50–100 μl of beads for each washing step. 7. Cut off the end of a 1 ml pipette tip when handling cAMPagarose beads to avoid disruption of the beads. To make sure that the measure of the volume of the beads is correct, compare the volume of the beads with the same volume of water in an Eppendorf tube. 8. In order not to have an exchange in the equilibrium, with consequent loss of the proteins bound to the beads, wash the beads with ice-cold washing buffer. 9. To remove the liquid as much as possible, without pipetting the beads we suggest to use thin-bore gel-loader tip. 10. An alternative to the in-gel digestion is in-solution digestion. In this case, after step 11 of Subheading 3.2, proteins are

200

Eleonora Corradini et al.

eluted by adding 100 μl 8 M urea in 50 mM ABC. Similar to the in-gel digestion the proteins are reduced and subsequently alkylated by adding 2 mM DTT (at 56 °C for 30 min) and 4 mM IAA (at room temperature for 20 min in the dark). Proteolysis is achieved by incubating the proteins with LysC (1:100 w/w enzyme:substrate ratio) at 37 °C for 4 h. The LysC digested solution is then diluted with ABC to a final concentration of 2 M urea and a second digestion step is performed with trypsin (1:100 enzyme:substrate ratio) at 37 °C overnight. For a single cAMP pull down with 5 mg protein in competition with free cyclic nucleotide, a bound fraction of ~5–10 μg is expected. 11. An alternative to the digestion of the whole lane is the selection and subsequent digestion of desired bands. 12. Reduction and alkylation of the cysteine residues is necessary for optimal unfolding to render all lysine and argninine residues accessible for the proteolytic enzymes. Also the formation of disulfide bridges in the digested peptide mixture is prevented in this way. 13. The use of ACN and ABC is required to destain the gel and remove the excess of DTT and IAA. 14. Centrifuge the resuspended samples before LC–MS/MS analysis to avoid small pieces of gel in the peptide mixture that can cause blockage in the LC system. 15. Three biological replicates should be analyzed in order to perform a statistical analysis such as t-test.

Acknowledgments This work was supported by the Netherlands Proteomics Center, by the PRIME-XS project, grant agreement number 262067, funded by the European Union 7th Framework Programme. References 1. Hofmann F, Bernhard D, Lukowski R et al (2009) cGMP regulated protein kinases (cGK). Handb Exp Pharmacol 191:137–162 2. Skalhegg BS, Tasken K (2000) Specificity in the cAMP/PKA signaling pathway. Differential expression, regulation, and subcellular localization of subunits of PKA. Front Biosci 5:D678–D693 3. Zagotta WN, Siegelbaum SA (1996) Structure and function of cyclic nucleotide-gated channels. Annu Rev Neurosci 19:235–263 4. Wegener JW, Nawrath H, Wolfsgruber W et al (2002) cGMP-dependent protein kinase I medi-

ates the negative inotropic effect of cGMP in the murine myocardium. Circ Res 90:18–20 5. Moncada S, Gryglewski R, Bunting S et al (1976) An enzyme isolated from arteries transforms prostaglandin endoperoxides to an unstable substance that inhibits platelet aggregation. Nature 263:663–665 6. Taylor SS, Kim C, Vigil D et al (2005) Dynamics of signaling by PKA. Biochim Biophys Acta 1754:25–37 7. Wong W, Scott JD (2004) AKAP signalling complexes: focal points in space and time. Nat Rev Mol Cell Biol 5:959–970

Proteomics of PKA and PKG Signaling 8. Scholten A, Aye TT, Heck AJ (2008) A multiangular mass spectrometric view at cyclic nucleotide dependent protein kinases: in vivo characterization and structure/function relationships. Mass Spectrom Rev 27:331–353 9. Corbin JD, Francis SH (1999) Cyclic GMP phosphodiesterase-5: target of sildenafil. J Biol Chem 274:13729–13732 10. Schlossmann J, Ammendola A, Ashman K et al (2000) Regulation of intracellular calcium by a signalling complex of IRAG, IP3 receptor and cGMP kinase Ibeta. Nature 404:197–201 11. Sharma AK, Zhou GP, Kupferman J et al (2008) Probing the interaction between the coiled coil leucine zipper of cGMP-dependent protein kinase Ialpha and the C terminus of the myosin binding subunit of the myosin light chain phosphatase. J Biol Chem 283: 32860–32869

201

12. Scholten A, Poh MK, van Veen TA et al (2006) Analysis of the cGMP/cAMP interactome using a chemical proteomics approach in mammalian heart tissue validates sphingosine kinase type 1-interacting protein as a genuine and highly abundant AKAP. J Proteome Res 5: 1435–1447 13. Kovanich D, van der Heyden MA, Aye TT et al (2010) Sphingosine kinase interacting protein is an A-kinase anchoring protein specific for type I cAMP-dependent protein kinase. Chembiochem 11:963–971 14. Burgers PP, Ma Y, Margarucci L et al (2012) A small novel A-kinase anchoring protein (AKAP) that localizes specifically protein kinase A-regulatory subunit I (PKA-RI) to the plasma membrane. J Biol Chem 287:43789–43797 15. Poppe H, Rybalkin SD, Rehmann H et al (2008) Cyclic nucleotide analogs as probes of signaling pathways. Nat Methods 5:277–278

Chapter 16 Development of Computational Models of cAMP Signaling Susana R. Neves-Zaph and Roy S. Song Abstract Despite the growing evidence defining the cAMP signaling network as a master regulator of cellular function in a number of tissues, regulatory feedback loops, signal compartmentalization, as well as cross-talk with other signaling pathways make understanding the emergent properties of cAMP cellular action a daunting task. Dynamical models of signaling that combine quantitative rigor with molecular details can contribute valuable mechanistic insight into the complexity of intracellular cAMP signaling by complementing and guiding experimental efforts. In this chapter, we review the development of cAMP computational models. We describe how features of the cAMP network can be represented and review the types of experimental data useful in modeling cAMP signaling. We also compile a list of published cAMP models that can aid in the development of novel dynamical models of cAMP signaling. Key words Computational modeling, Systems biology, cAMP, Protein kinase A, Phosphodiesterases, Adenylyl cyclase

1

Introduction Cyclic adenosine 3′, 5′-monophosphate (cAMP), the classic example of an intracellular second messenger, plays a central role in a diverse range of cellular processes such as secretion, cardiac contraction, metabolism, and synaptic plasticity [1–3]. Typically, activation of a G-protein-coupled receptor by binding a cytokine, hormone, or neurotransmitter stimulates cAMP production by increasing membrane-bound adenylyl cyclase (AC) activity [4]. Production of cAMP leads to the activation of the cAMP effectors protein kinase A, the GEF EPAC, and cyclic nucleotide-gated channels [5–7]. The balance between the AC production activities and phosphodiesterases (PDE), the enzymes that degrade cAMP into 5′AMP, maintains cAMP accumulation within an optimal range thereby determining the responsiveness to downstream signaling [8, 9]. The ubiquitous functional role of cAMP throughout the body is in sharp contrast with the high degree of specificity observed in cellular cAMP signaling pathways. In the same cell, exposure to

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7_16, © Springer Science+Business Media New York 2015

203

204

Susana R. Neves-Zaph and Roy S. Song

different hormones can elicit seemingly similar temporal changes in cAMP levels that result in distinct cellular outcomes [10]. This specificity of signaling, termed compartmentalization, is the subject of intense study, and we are just starting to appreciate how the time course, the amplitude, and, most notably, the location of cAMP signals are essential determinants of cellular output [11]. Several features of this versatile signaling cascade impart it with context-specific properties that promote compartmentalization such as: (1) the segregation of membrane production and cytosolic degradation activities allows cellular shape to play a role in cAMP accumulation; (2) AKAPs, scaffolds that spatially cluster specific ACs, PDEs, and cAMP effectors to subcellular locations, and direct downstream signaling; (3) a multitude of AC and PDE types display complex allosteric and posttranslational regulation [12–14]. These features give rise to a complexity in cAMP signaling that goes beyond the ability of intuition to fully understand. Therefore, the cAMP signaling system requires not just quantitative experimental methods, but also complementary mathematical/computational modeling to interpret the data. As key signaling components are identified, their activities and locations are experimentally determined, and novel features are discovered, computational studies are needed to assemble all this information into a coherent view of the contribution of cAMP to cellular output [15]. In this chapter, we will describe the sequential steps to develop a cAMP computational model and the type of experimental data useful in the construction of such models. We also provide a general overview of published cAMP models and their representation of the cAMP signaling. Our goal is that this chapter will be a useful resource to those interested in developing models of cAMP signaling.

2

Methods

2.1 Development of a cAMP Signaling Model

The goal of using computational models to understand signaling is to create a simplified quantitative representation that can guide experiments and lead to novel insight. Computational models of cAMP can be classified based on their purpose into two groups: (1) proposed mechanism models or (2) predictive models. Proposed mechanism models probe a possible detailed molecular mechanism that could give rise to the experimental observation of interest. This group of models is usually the ending figure in a mostly experimental study, and is used to theoretically explore, given the current knowledge, the presented signaling mechanism. Proposed mechanism models can also be used to investigate factors that could affect the signaling of interest that are outside of possible experimentation. An example is Calabiro et al., [16] where the authors tested, using a cAMP signaling model, whether the

cAMP Signaling Models

205

internalization of active G-protein-coupled receptors to endosomes could give rise to cAMP time courses similar to those observed experimentally, and probed how changing the endosomal compartment size, an impossible experimental test, could affect the accumulation of cAMP. Predictive models, on the other hand, are those that recapitulate many experimental observations about different aspects of signaling cascades and, by integrating these observations, provide nonobvious testable hypothesis that can lead to novel biological insight. For instance, Song et al., [17] used a computational model of cAMP and BDNF signaling to explore the regulation of PDE4 by ERK resulting in the control of AMPAR trafficking. Predictive models can also be built by assembling different sets of protein interactions or a variety of competing intracellular mechanisms to explain an observed experimental observation. Regardless of their ultimate purpose, both types of models are useful in guiding experimental questions, as they complement experimental efforts and lead to a formalism of the experimental data. The development of computational models can be divided into three stages: assembling of the model, constraining, and validating of the model. Assembly of the model entails compiling all the reactions that will be included so that the scope and goal of the model is defined. Constraining of the model involves collecting biological data needed to parametize the model. Validating the model consists of using experimental data not used during the constraining process, to substantiate the topology and scope of the model.

2.2 Assembling cAMP Signaling Model

Typically, modeling efforts are started after some experimental data has already been obtained, and based on these data a model is developed consisting of signaling reactions that can give rise to the experimental observation of interest. That entails drawing a diagram depicting the connectivity of all the proteins and small molecules involved, with each reaction represented with an arrow (for production or stimulatory reactions) or plunger (for degradation or inhibitory reactions). The next step is to describe those arrows and plungers in terms of equations derived from established chemical theory, usually in the form of mass action law for binding reactions or Michaelis–Menten approximations for enzymatic reactions. During this stage of model development it is necessary to keep in mind the level of detail needed, as it will determine the usefulness of a model [18]. One must decide, based on the scope of the model, what are the biologically relevant details to include and what can be ignored: a model that contains all known intricate details of the cellular event of interest vs. a model that recapitulates biological phenomena using minimal number of signaling reactions. A too simplistic representation of the signaling will limit the questions that can be explored and runs the risk of producing trivial

206

Susana R. Neves-Zaph and Roy S. Song

model predictions. However, a complex model, with excessive number of reactions and components, will increase the number of parameters needed and may be difficult to constrain (see Subheading 2.3 below). As a rule of thumb, it is recommended to always start with the simplest representation of the signaling that is compatible with the experimental observation of interest. In some models, details may be omitted or abstracted without loss of validity. For instance, AC activation may be represented as a function relating the EC50 of a receptor ligand of interest with cAMP levels, instead of including all the reactions leading from G-protein and cyclase activation [19]. However, if receptor or AC regulation is a key feature of the signaling pathway, this simplification is inadequate and an explicit representation of receptor activation should be included. The appropriate mathematical form should be applied depending on the goal and scope of the modeling effort. For a relatively simple pathway with few components, an analytical model can be constructed by using a simple set of descriptive equations that can highlight the relationship between parameter and component. For instance, Chen et al., [20] using an analytical solution showed how the interplay between dendritic diameter/length and PDEs activities controls cAMP accumulation and promotes compartmentalization in neurons. However, when models contain a significant number of components and reactions, such as the vast majority of published models of cAMP signaling, the use of coupled ordinary differential equation (ODE) based techniques is preferred. ODE models describe the changes in concentration of model components over time. The majority of these models are deterministic, meaning that they do not account for fluctuations in concentrations or rates due to biological noise. The published stochastic cAMP models that do take into account these noise fluctuations have dealt with cAMP signaling in very small volumes, such as in a dendritic spine, where variability in the small number of signaling molecules can give rise to a diversity of outputs [21, 22]. For those models where the location of signaling must be specified, one can use ODE models with compartments to define either membrane-delimited organelles (such as PKA signaling in the nucleus [23]) or subcellular functional locations (such as cAMP signaling in caveolae-rich membrane domains [24]). Each compartment may contain a specific subset of proteins within its specified domain, such as the study by Iancu et al., where cAMP signaling from caveolar vs. noncaveolar membrane domains was compared, and each domain contained a specific set of AC and PDEs [24]. The use of multicompartment ODE models can abstract spatial information; however, they require implicit information on the translocation rates of components between compartments, which may or may not be available and also assumes that the concentration of each component is homogenous within its respective compartment, which may not be the case.

cAMP Signaling Models

207

A more detailed approach to represent space is to use partial differential equation-based modeling, where one can track the changes in concentration of model components across time and space. Idealized or realistic cellular geometries derived from microscopy images can be included in these types of models, complementing imaging studies of cAMP signaling. However, these types of models require additional parameters (diffusion coefficients, concentration densities at the membrane, etc.) and tend to be computationally expensive. Partial differential equation-based cAMP models have been useful in identifying the key role that cell shape plays in cAMP accumulation, especially in highly polarized cells such as neurons [13]. There are also models that take into account stochastic diffusion/reaction events as those by the Blackwell lab, where the interplay of Ca++ and cAMP signaling leading to synaptic PKA activation was explored [25]. This study highlighted the importance of PKA colocalization with AC, not PKA targets, for efficient downstream signaling. There are many tools available to simulate cAMP signaling. MATLAB and Python are commonly used software tools for simulation. There are also a number of specialized software tools available to simulate cellular signaling depending on the scope and goal of the model (deterministic or stochastic, ODE based or partial differential equation based). A summary of available free-standing modeling platforms is presented in Table 1. For more information on modeling platforms, please refer to [26, 27].

Table 1 List of currently available simulating platforms

Simulating platforms

Ordinary differential algorithm

Partial differential algorithm

CellDesigner

*

COPASI

*

*

JDesigner/Jarnac

*

*

JSim

*

Kinetikit

*

Stochastic algorithm

* *

MCell

*

Moose

*

*

NeuroRD

*

*

Smoldyn VCell

* *

*

*

208

Susana R. Neves-Zaph and Roy S. Song

2.2.1 Representations of cAMP Signaling Components

It is useful to examine how other models have represented key events in cAMP signaling and the range of parameter values used. Table 2 shows a list of published cAMP models and their respective representations of the key components of cAMP signaling, which can be used to guide future model development. For the most part, these models contain no “black boxes,” i.e., relationships between components lacking molecular mechanism. Black boxes are at times used in models when there are gaps on mechanistic knowledge of the signaling event, or a lack of parameters values, and the right input–output behavior function is then used in place of the appropriate molecular reactions. An example would be to use a function that relates cAMP levels to PKA activation, without listing all of the binding reactions. While this approach reduces the effort needed to construct models of signaling and may provide information about general dynamic properties of the system, it cannot be used to investigate the behavior of specific molecular mechanisms and frequently fails to provide insight into the complexity of the resultant model.

Adenylyl Cyclases and Phosphodiesterases

There are ten genes encoding for ACs. AC1-9 gives rise to membrane-bound AC and AC10 codes for a soluble AC activated by intracellular bicarbonate levels [28, 29]. Membrane ACs display complex regulation by G-proteins, Ca++, and posttranslational modifications [30]. The AC enzymatic reaction of converting ATP to cAMP can be described as: kf

k cat

AC + ATP  AC.ATP  cAMP + AC kb

where AC.ATP is the enzyme:substrate intermediate and is assumed that the reaction converting cAMP back to the AC.ATP intermediate is negligible. In order to write the equations that describe the change over time for each of the molecular species (AC, ATP, AC.ATP, and cAMP), one must consider all the reactions that produce and consume each species: d [ AC] dt d [ ATP ] dt d [ AC.ATP ] dt d [ cAMP ] dt

= -k f [ AC][ ATP ] + kb [ AC.ATP ] = -k f [ AC][ ATP ] + kb [ AC.A ATP ] + kcat [ AC] = k f [ AC][ ATP ] - kb [ AC.ATP ] - kcat [ AC] = kcat [ AC.ATP ]

ATP levels are buffered, meaning that there is an endless supply mimicking the conditions of a healthy cell. The concentration of

X

X

X

X

cAMP input

X

X

Adrenal tissues

Neuron

HEK293

HEK293

Myocytes

Neuron

Neuron

Neuron

Neuron

HEK293

Cardiac myocytes

Neuron

Cardiac myocytes

B-Cells

HEK293

Neuron

Dempsher et al., [49]

Bhalla et al., [31]

Rich et al., [50]

Rich et al., [51]

Saucerman et al., [52]

Bhalla et al., [21]

Bhalla et al., [22]

Ajay et al., [53]

Hayer et al., [54]

Rich et al., [55]

Saucerman et al., [56]

Fernandez et al., [57]

Iancu et al., [24]

Fridlyand et al., [58]

Violin et al., [59]

Neves et al., [13]

X

cAMP input

X

X

X

X

X

Generic AC

Cell type

Model

X

AC1

X

AC2

X

AC4

Table 2 List of selected cAMP models and their representation of key components

X

AC5

X

AC6

X

AC7

AC8

X

X

X

X

X

X

X

X

X

X

X

X

Generic PDE

X

X

PDE1

X

PDE2

X

PDE3

X

X

X

PDE10

(continued)

PDE4

X

X

X

X

X

X

X

Neuron

Neuron

HEK293

B-Cells

Ventricular myocyte

Epithelial cell X

X

Thyroid follicles

Neuron

B-cells

Endothelial cells

HEK293

Neuron

Neuron

Podocytes

HEK293

Neuron

Calebiro et al., [16]

Nakano et al., [61]

Kim et al., [38]

Oliveira et al., [62]

Ni et al.,[37]

Heijman et al.,[63]

Xie et al., [64]

Kim et al., [65]

Fridlyand et al., [66]

Feinstein et al., [67]

Sample et al., [23]

Oliveira et al., [25]

Song et al., [17]

Azeloglu et al., [68]

Agarwal et al., [69]

Lindskog et al., [70]

X

X

cAMP input

Neuron

Cygnar et al., [60]

Generic AC

Cell type

Model

Table 2 (continued)

X

X

AC1

AC2

X

AC4

X

X

X

AC5

X

X

AC6

X

AC7

X

X

AC8

X

X

X

Generic PDE

X

X

X

X

X

X

X

X

PDE1

X

X

X

PDE2

X

X

X

PDE3

X

X

X

X

X

X

X

X

X

X

X

PDE4

X

PDE10

cAMP Signaling Models

211

cellular ATP is in the millimolar range, several orders of magnitude in excess of the concentration of AC. If the concentration of AC is not changing rapidly, the concentration of AC.ATP intermediate is at steady state and under these conditions it is possible to use the Michaelis-Menten approximation to describe the production of cAMP as: d [ cAMP ] dt

=

kcat [ AC][ ATP ]

[ ATP ] + kM

where kM is defined as: kM =

kb + kcat kf

Enzymatic studies will usually provide values for kM and Vmax (Vmax = kcat*[total enzyme]). Rarely are kf and kb values specified, however one can approximate the value for kb as 4kcat and calculate kf from the kM definition [31]. Most models will have a description of AC basal activity, with appropriate kinetics parameters, and a separate description of the activated AC so there will be two AC production activities. PDEs are also enzymes and can be described in the same manner as AC. The PDE superfamily consists of 11 families each with multiple genes giving rising to a number of isoforms due to splicing or alternative use of starting sites. Each PDE family is classified based on their affinities for cAMP or cGMP and are grouped as cAMP specific (PDE4, PDE7, and PDE8); cGMP specific (PDE5, PDE6, and PDE9); or dual specificity (PDE1, PDE2, PDE3, PDE10, and PDE11) [8]. The PDE superfamily displays complex regulation consisting of allosteric modulation and competitive inhibition by cyclic nucleotides, and posttranslational modifications by a number of kinases [32]. PKA

PKA, the ubiquitous cAMP effector, is a serine/threonine kinase that regulates the activity of a number of substrates, ranging from transcription factors to ionic channels [33]. In resting conditions, PKA is an inactive tetramer consisting of two regulatory subunits (R subunits) bound to two catalytic subunits (C subunits). Each regulatory subunit contains two cAMP-binding sites that upon increases in cAMP, act in a cooperative manner to induce a conformational change and release the two active catalytic subunits [34]. The initial concentration of the PKA holoenzyme ranges from mid-nanomolar values to 1 μM. The threshold for PKA holoenzyme activation depends on the identity of the R subunits, as those holoenzymes expressing RI type have a higher affinity for cAMP, and thus have a lower activation threshold than those containing RII [35].

212

Susana R. Neves-Zaph and Roy S. Song

Bhalla and Iyengar [31] developed one of the first computational models describing PKA activation based on the experimental data from [36]. PKA activation can be described by the binding of cAMP molecules to the regulatory subunits of the PKA holoenzyme (R2C2) in four consecutive binding events with differing KD to reflect its cooperative nature. This is followed by two reactions describing the release of the now-active catalytic subunits (C). R 2C2 + cAMP  cAMP.R 2C2 cAMP.R 2C2 + cAMP  cAMP 2.R 2C2 cAMP 2.R 2C2 + cAMP  cAMP3.R 2C2 cAMP3.R 2C2 + cAMP  cAMP 4.R 2C2 cAMP 4.R 2C2  cAMP 4.R 2C + C cAMP 4.R 2C  cAMP 4.R 2 + C PKA activation has also been described as two independent binding events and a single release reaction mechanism or further abstracted as a single binding reaction involving 4 cAMP molecules as R 2C2 + 4cAMP « PKA * [37, 38]. 2.3 Constraining and Validating a cAMP Signaling Model

When developing a cAMP model, it is helpful to identify and study in isolation a small subset of the cAMP signaling network. This modular format is guided by the availability of experimental data that can be used to compare and constrain the resulting simulations. For instance, there is data available for the activation of PKA in vitro, making it easy to first implement the reactions of the PKA activation module in isolation, making sure that varying cAMP levels result in the appropriate PKA activation. Once this module is behaving as expected, other modules may be connected such as cAMP production and degradation, PKA phosphorylation of targets, etc.

2.3.1 Experimental Data Suitable for cAMP Computational Models

Once the connectivity of the model has been established, one must collect experimental data to constrain and validate the model from the published literature. Most of the key players of the canonical cAMP pathway were discovered, purified, and biochemically characterized, over 30 years ago providing a wealth of quantitative data useful in the construction of computational models. The connectivity of cAMP pathway is relatively well established, and parameter values such as cellular concentrations (from tissue purifications), binding affinities, and reaction rates (from in vitro assays) are available. However, some of these parameter values come from measurements done in vitro conditions and may not accurately reflect the intracellular environment. For unknown parameters, biologically relevant placeholder values may be used and further constrained using additional experimental data. For instance, the parameters describing the phosphorylation of a substrate of PKA

cAMP Signaling Models

213

may be also used to describe other PKA phosphorylation events if experimental data is not available. The most valuable data to constrain models are measurements of concentrations or reaction rates as a function of time (time courses) or concentration of a ligand (dose responses). These types of data are called input:output relationships and are helpful to subdivide the model into modules by defining an input and output (e.g., Forskolin [input] vs. cAMP dose response [output]). Time courses with many data points are particularly useful, as they provide more information about the true dynamics of the signaling pathway. Förster resonance energy transfer (FRET) imaging with genetically encoded unimolecular biosensors allows the monitoring of cAMP activities at a temporal resolution unmatched by conventional biochemical methods [39–42]. FRET is the process by which a donor fluorophore transfers energy to a longer wavelengthshifted acceptor fluorophore and depends on the distance between the two fluorophores and their spectral overlap. cAMP FRET biosensors consist of two fluorescent proteins flanking a cAMP-binding domain that upon binding alters its conformation and changes the distance between the two fluorophores, resulting in a change in FRET. The cAMP-binding domain from EPAC is part of most widely used cAMP biosensors. Several versions of EPAC-based biosensors have been developed, with differing affinities and kinetics of cAMP sensing, and allow one to optimize the detection of cAMP to each cell type [43, 44]. Moreover, genetically encoded cAMP FRET biosensors can be expressed with targeting domains to specific membrane compartments (such as lipid rafts) or even AKAP scaffolding domains [45]. This allows the monitoring of the cAMP or PKA signaling activity in multiple compartments (e.g., the plasma membrane or the nucleus). For methods on FRET imaging using cAMP biosensors, please refer to ref. 46. FRET-imaging techniques allow the visualization of real-time changes of cAMP levels in intact cells. In most studies, cAMP FRET is presented as fold changes with relative units. And if one models the reaction involving the biosensor with cAMP, one can relate simulated intracellular concentrations of cAMP to the change in fluorescence in FRET experiments data. However, using fold change data to develop computational models of cAMP could possibly lead to a variety of model configurations or parameters to describe the same behavior, given that the basal cAMP concentration is not known. This is especially problematic when trying to construct models with detailed AC and PDE representations, as the concentrations of these may not be known. One way around this is to calibrate the FRET measurement in terms of cAMP concentration. In vitro characterizations for most cAMP biosensors have been published, which consist of mixing purified biosensor with known amounts of cAMP while monitoring FRET changes to relate FRET ratios to cAMP levels. The caveat of this approach is

214

Susana R. Neves-Zaph and Roy S. Song

that: (1) it depends on in vitro measurements; (2) the FRET measurement values will be specific to the optics of the microscope/spectrophotometer used. Preferably one can perform in-cell calibrations of cAMP biosensors by inhibiting adenylyl cyclase activity so that cAMP levels decrease to minimal levels to establish the minimum limit of the cAMP biosensor. This is followed by addition of cell-permeable cAMP PDE-resistant analogs of known concentration and monitoring the resulting FRET changes [47, 46]. Alternatively, in-cell calibrations can be done in cells expressing cAMP biosensors permeabilized with saponin beta-escin and then exposed to different levels of cAMP analogs [48] or by using a micro-infusion technique where the cell expressing the FRET reporter is challenged with known concentrations of cAMP via a patch pipette (described elsewhere in this volume). By calibrating the FRET response, a cAMP dose–response curve can be assembled, and the level of intracellular cAMP can be extrapolated from the change in FRET for direct comparison with the model.

3

Conclusions Computational modeling is a powerful tool for understanding the temporal and spatial regulation of cAMP signaling. Computational models of cAMP signaling can provide information that may not be accessible to direct experimental measure, propose causative relationships between unexpected cellular components, and guide future experiments into emergent system properties. The advent of exciting live-cell imaging technologies that provide fine temporal and spatial resolution to cAMP signaling will further facilitate the development of predictive models. The approach of combining high content imaging and computational models enhances the predictive power of an exclusively computational approach and the mechanistic understanding of a purely experimental effort.

4

Notes The Bhalla lab maintains a very useful annotated database of signaling reactions and their kinetic parameters (http://doqcs. ncbs.res.in/). There are number of repository sites with published signaling models such as Biomodels (http://www.ebi.ac.uk/biomodelsmain/) and Vcell (http://vcell.org/vcell_models/published_ models.html?current=five).

cAMP Signaling Models

215

References 1. Seino S, Shibasaki T (2005) PKA-dependent and PKA-independent pathways for cAMPregulated exocytosis. Physiol Rev 85: 1303–1342 2. Zaccolo M (2009) cAMP signal transduction in the heart: understanding spatial control for the development of novel therapeutic strategies. Br J Pharmacol 158:50–60 3. Daniel PB, Walker WH, Habener JF (1998) Cyclic AMP signaling and gene regulation. Annu Rev Nutr 18:353–383 4. Sunahara RK, Dessauer CW, Gilman AG (1996) Complexity and diversity of mammalian adenylyl cyclases. Annu Rev Pharmacol Toxicol 36:461–480 5. Kaupp UB, Seifert R (2002) Cyclic nucleotidegated ion channels. Physiol Rev 82:769–824 6. Gloerich M, Bos JL (2010) Epac: defining a new mechanism for cAMP action. Annu Rev Pharmacol Toxicol 50:355–375 7. Tasken K, Aandahl EM (2004) Localized effects of cAMP mediated by distinct routes of protein kinase A. Physiol Rev 84:137–167 8. Bender AT, Beavo JA (2006) Cyclic nucleotide phosphodiesterases: molecular regulation to clinical use. Pharmacol Rev 58:488–520 9. Conti M, Mika D, Richter W (2014) Cyclic AMP compartments and signaling specificity: role of cyclic nucleotide phosphodiesterases. J Gen Physiol 143:29–38 10. Steinberg SF, Brunton LL (2001) Compartmentation of G protein-coupled signaling pathways in cardiac myocytes. Annu Rev Pharmacol Toxicol 41:751–773 11. Karpen JW (2014) Perspectives on: cyclic nucleotide microdomains and signaling specificity. J Gen Physiol 143:5–7 12. Kapiloff MS, Rigatti M, Dodge-Kafka KL (2014) Architectural and functional roles of A kinase-anchoring proteins in cAMP microdomains. J Gen Physiol 143:9–15 13. Neves SR, Tsokas P, Sarkar A et al (2008) Cell shape and negative links in regulatory motifs together control spatial information flow in signaling networks. Cell 133:666–680 14. Rich TC, Webb KJ, Leavesley SJ (2014) Can we decipher the information content contained within cyclic nucleotide signals? J Gen Physiol 143:17–27 15. Saucerman JJ, Greenwald EC, PolanowskaGrabowska R (2014) Mechanisms of cyclic AMP compartmentation revealed by computational models. J Gen Physiol 143:39–48 16. Calebiro D, Nikolaev VO, Gagliani MC et al (2009) Persistent cAMP-signals triggered by

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

internalized G-protein-coupled receptors. PLoS Biol 7:e1000172 Song RS, Massenburg B, Wenderski W et al (2013) ERK regulation of phosphodiesterase 4 enhances dopamine-stimulated AMPA receptor membrane insertion. Proc Natl Acad Sci U S A 110:15437–15442 Aldridge BB, Burke JM, Lauffenburger DA et al (2006) Physicochemical modelling of cell signalling pathways. Nat Cell Biol 8:1195–1203 Greenwald EC, Polanowska-Grabowska RK, Saucerman JJ (2014) Integrating fluorescent biosensor data using computational models. Methods Mol Biol 1071:227–248 Chen W, Levine H, Rappel WJ (2009) Compartmentalization of second messengers in neurons: a mathematical analysis. Phys Rev E Stat Nonlin Soft Matter Phys 80:041901 Bhalla US (2004) Signaling in small subcellular volumes. I. Stochastic and diffusion effects on individual pathways. Biophys J 87:733–744 Bhalla US (2004) Signaling in small subcellular volumes. II. Stochastic and diffusion effects on synaptic network properties. Biophys J 87:745–753 Sample V, DiPilato LM, Yang JH et al (2012) Regulation of nuclear PKA revealed by spatiotemporal manipulation of cyclic AMP. Nat Chem Biol 8:375–382 Iancu RV, Jones SW, Harvey RD (2007) Compartmentation of cAMP signaling in cardiac myocytes: a computational study. Biophys J 92:3317–3331 Oliveira RF, Kim M, Blackwell KT (2012) Subcellular location of PKA controls striatal plasticity: stochastic simulations in spiny dendrites. PLoS Comput Biol 8:e1002383 Alves R, Antunes F, Salvador A (2006) Tools for kinetic modeling of biochemical networks. Nat Biotechnol 24:667–672 Blackwell KT (2013) Approaches and tools for modeling signaling pathways and calcium dynamics in neurons. J Neurosci Methods 220:131–140 Tresguerres M, Levin LR, Buck J (2011) Intracellular cAMP signaling by soluble adenylyl cyclase. Kidney Int 79:1277–1288 Sadana R, Dessauer CW (2009) Physiological roles for G protein-regulated adenylyl cyclase isoforms: insights from knockout and overexpression studies. Neurosignals 17:5–22 Hanoune J, Defer N (2001) Regulation and role of adenylyl cyclase isoforms. Annu Rev Pharmacol Toxicol 41:145–174

216

Susana R. Neves-Zaph and Roy S. Song

31. Bhalla US, Iyengar R (1999) Emergent properties of networks of biological signaling pathways. Science 283:381–387 32. Omori K, Kotera J (2007) Overview of PDEs and their regulation. Circ Res 100:309–327 33. Taylor SS, Zhang P, Steichen JM et al (2013) PKA: lessons learned after twenty years. Biochim Biophys Acta 1834:1271–1278 34. Taylor SS, Buechler JA, Yonemoto W (1990) cAMP-dependent protein kinase: framework for a diverse family of regulatory enzymes. Annu Rev Biochem 59:971–1005 35. Dostmann WR, Taylor SS (1991) Identifying the molecular switches that determine whether (Rp)-cAMPS functions as an antagonist or an agonist in the activation of cAMP-dependent protein kinase I. Biochemistry 30:8710–8716 36. Ogreid D, Doskeland SO (1983) Cyclic nucleotides modulate the release of [3H] adenosine cyclic 3′,5′-phosphate bound to the regulatory moiety of protein kinase I by the catalytic subunit of the kinase. Biochemistry 22:1686–1696 37. Ni Q, Ganesan A, Aye-Han NN et al (2011) Signaling diversity of PKA achieved via a Ca2+cAMP-PKA oscillatory circuit. Nat Chem Biol 7:34–40 38. Kim M, Huang T, Abel T et al (2010) Temporal sensitivity of protein kinase a activation in late-phase long term potentiation. PLoS Comput Biol 6:e1000691 39. Zhang J, Allen MD (2007) FRET-based biosensors for protein kinases: illuminating the kinome. Mol BioSyst 3:759–765 40. Harvey CD, Yasuda R, Zhong H et al (2008) The spread of Ras activity triggered by activation of a single dendritic spine. Science 321:136–140 41. Lam AJ, St-Pierre F, Gong Y et al (2012) Improving FRET dynamic range with bright green and red fluorescent proteins. Nat Methods 9:1005–1012 42. Komatsu N, Aoki K, Yamada M et al (2011) Development of an optimized backbone of FRET biosensors for kinases and GTPases. Mol Biol Cell 22:4647–4656 43. Klarenbeek JB, Goedhart J, Hink MA et al (2011) A mTurquoise-based cAMP sensor for both FLIM and ratiometric read-out has improved dynamic range. PLoS One 6:e19170 44. Sprenger JU, Nikolaev VO (2013) Biophysical techniques for detection of cAMP and cGMP in living cells. Int J Mol Sci 14:8025–8046 45. Stangherlin A, Koschinski A, Terrin A et al (2014) Analysis of compartmentalized cAMP: a method to compare signals from differently targeted FRET reporters. Methods Mol Biol 1071:59–71

46. Borner S, Schwede F, Schlipp A et al (2011) FRET measurements of intracellular cAMP concentrations and cAMP analog permeability in intact cells. Nat Protoc 6:427–438 47. Mironov SL, Skorova E, Taschenberger G et al (2009) Imaging cytoplasmic cAMP in mouse brainstem neurons. BMC Neurosci 10:29 48. Nausch LW, Ledoux J, Bonev AD et al (2008) Differential patterning of cGMP in vascular smooth muscle cells revealed by single GFPlinked biosensors. Proc Natl Acad Sci U S A 105:365–370 49. Dempsher DP, Gann DS, Phair RD (1984) A mechanistic model of ACTH-stimulated cortisol secretion. Am J Physiol 246:R587–R596 50. Rich TC, Fagan KA, Nakata H et al (2000) Cyclic nucleotide-gated channels colocalize with adenylyl cyclase in regions of restricted cAMP diffusion. J Gen Physiol 116:147–161 51. Rich TC, Fagan KA, Tse TE et al (2001) A uniform extracellular stimulus triggers distinct cAMP signals in different compartments of a simple cell. Proc Natl Acad Sci U S A 98: 13049–13054 52. Saucerman JJ, Brunton LL, Michailova AP et al (2003) Modeling beta-adrenergic control of cardiac myocyte contractility in silico. J Biol Chem 278:47997–48003 53. Ajay SM, Bhalla US (2004) A role for ERKII in synaptic pattern selectivity on the time-scale of minutes. Eur J Neurosci 20:2671–2680 54. Hayer A, Bhalla US (2005) Molecular switches at the synapse emerge from receptor and kinase traffic. PLoS Comput Biol 1:137–154 55. Rich TC, Xin W, Mehats C et al (2007) Cellular mechanisms underlying prostaglandin-induced transient cAMP signals near the plasma membrane of HEK-293 cells. Am J Physiol Cell Physiol 292:C319–C331 56. Saucerman JJ, Zhang J, Martin JC et al (2006) Systems analysis of PKA-mediated phosphorylation gradients in live cardiac myocytes. Proc Natl Acad Sci U S A 103:12923–12928 57. Fernandez E, Schiappa R, Girault JA et al (2006) DARPP-32 is a robust integrator of dopamine and glutamate signals. PLoS Comput Biol 2:e176 58. Fridlyand LE, Harbeck MC, Roe MW et al (2007) Regulation of cAMP dynamics by Ca2+ and G protein-coupled receptors in the pancreatic beta-cell: a computational approach. Am J Physiol Cell Physiol 293:C1924–C1933 59. Violin JD, DiPilato LM, Yildirim N et al (2008) beta2-adrenergic receptor signaling and desensitization elucidated by quantitative modeling of real time cAMP dynamics. J Biol Chem 283:2949–2961

cAMP Signaling Models 60. Cygnar KD, Zhao H (2009) Phosphodiesterase 1C is dispensable for rapid response termination of olfactory sensory neurons. Nat Neurosci 12:454–462 61. Nakano T, Doi T, Yoshimoto J et al (2010) A kinetic model of dopamine- and calciumdependent striatal synaptic plasticity. PLoS Comput Biol 6:e1000670 62. Oliveira RF, Terrin A, Di Benedetto G et al (2010) The role of type 4 phosphodiesterases in generating microdomains of cAMP: large scale stochastic simulations. PLoS One 5:e11725 63. Heijman J, Volders PG, Westra RL et al (2011) Local control of beta-adrenergic stimulation: effects on ventricular myocyte electrophysiology and Ca(2+)-transient. J Mol Cell Cardiol 50:863–871 64. Xie M, Rich TC, Scheitrum C et al (2011) Inactivation of multidrug resistance proteins disrupts both cellular extrusion and intracellular degradation of cAMP. Mol Pharmacol 80:281–293 65. Kim M, Park AJ, Havekes R et al (2011) Colocalization of protein kinase A with adeny-

66.

67.

68.

69.

70.

217

lyl cyclase enhances protein kinase A activity during induction of long-lasting long-termpotentiation. PLoS Comput Biol 7:e1002084 Fridlyand LE, Philipson LH (2011) Coupling of metabolic, second messenger pathways and insulin granule dynamics in pancreatic betacells: a computational analysis. Prog Biophys Mol Biol 107:293–303 Feinstein WP, Zhu B, Leavesley SJ et al (2012) Assessment of cellular mechanisms contributing to cAMP compartmentalization in pulmonary microvascular endothelial cells. Am J Physiol Cell Physiol 302:C839–C852 Azeloglu EU, Hardy SV, Eungdamrong NJ et al (2014) Interconnected network motifs control podocyte morphology and kidney function. Sci Signal 7:ra12 Agarwal SR, Yang PC, Rice M et al (2014) Role of membrane microdomains in compartmentation of cAMP signaling. PLoS One 9:e95835 Lindskog M, Kim M, Wikstrom MA et al (2006) Transient calcium and dopamine increase PKA activity and DARPP-32 phosphorylation. PLoS Comput Biol 2:e119

INDEX A

C

Adenosine-3',5'-cyclic monophosphorothioate Rp-isomer (Rp-cAMPS) .......................................... 139, 145 Adenovirus ........................ 72, 73, 77, 78, 104, 109, 112, 113 Adenylyl cyclase gene ................................................................. 7, 76, 110, 113, 138, 185–187 AKAPs. See A-kinase anchoring proteins (AKAPs) A-kinase anchoring proteins (AKAPs) AKAP18 ...................... 140, 141, 152–155, 158–161, 171 AKAP18α ........................................................... 159, 171 AKAP18δ .................................................. 140, 153–156, 158–160, 162–164 AKAP-PKA disruptors .......................139–142, 151–165, 167, 168 inhibitor ................................................ 138, 153–160 interaction...................................... 86, 137–139, 141, 151–165, 167 amphipathic helix ........................137, 139, 140, 142, 167 competitor peptides .................................... 171, 175, 179 disruption ................................................... 137–147, 152 peptide disruptors ........................139, 140, 145, 167, 168 signaling complex ...............................................137–147 Allosteric mechanism ...................................................46, 54 Aorta cannulation .................................................................111 ATP ......................................... 13, 38, 90, 144, 170, 208, 211 Automated image analysis process......................................62

Ca2+ intracellular measurement ................74, 76, 78–80, 85, 86 CAAX motif........................................................... 87, 88, 95 cAMP concentration ............................13, 14, 18, 26–29, 31–39, 51, 53, 59, 60, 66–69, 72, 76, 77, 79, 80, 82, 86, 87, 90, 91, 97–99, 104, 110, 134, 138, 142, 143, 184, 186, 192, 193, 195, 207, 212, 214 dynamics ..............................25, 46, 57, 85–100, 132, 213 exchange protein activated directly by cAMP (EPAC) binding domain.................................................14, 15 cAMP-resin ..........................................................192 dose response curve............................... 26, 27, 31–35 N-terminal DEP.....................................................15 oscillations ........................................................90, 91 pools .................................................................28, 35 Rap1 .................................................................15, 44 REM-domain .........................................................15 intracellular levels ...................................... 13–23, 26, 27, 29, 33, 35, 71, 74, 85, 110, 121, 138, 142, 143, 191, 203, 205, 212–214 measurement................................................ 4, 35, 59–61, 63–68, 72–76, 80–82, 85, 90, 91, 98–100, 213 synthesis ...............................................................13, 187 Cardiomyocytes adult murine .......................................................104, 114 autofluorescence........................................ 18, 20, 21, 114 contractile function .............................................104, 112 culture ......................................................... 104, 112, 113 isolation ...................................................... 106, 110–112 mouse .................................................................103–114 primary ...............................................................103–114 viral transduction ................................................103–114 Cervical dislocation ..................................................106, 111 cGMP reporters........................................................................62 Chemical proteomics ................................................191–200 Cloning strategy ...............................................................120–123 CNBD. See Cyclic nucleotides binding domain (CNBD) Coarse-grain .......................................................................46 Collagen ..................................................................... 96, 106 Collagenase ......................................................................106

B Bacteriophage screening amplification .......................................................170–171 bacteriophage construction .................................170–171 glutathione magnetic beads ............................... 170, 175, 176, 179, 180 PreScission protease.................................... 160, 176, 179 selection ..................................................... 120, 141, 168, 170–171, 176–178 β2-adrenergic receptor ......................................................143 Bioinformatics .................................................... 48, 140, 142 BL21 strains .....................................................................179 BLUF-type photoreceptor domains .................................131 Bovine serum albumin ................................ 96, 106, 112, 120 bPac .................................................................. 131, 132, 134

Manuela Zaccolo (ed.), cAMP Signaling: Methods and Protocols, Methods in Molecular Biology, vol. 1294, DOI 10.1007/978-1-4939-2537-7, © Springer Science+Business Media New York 2015

219

CAMP SIGNALING: METHODS AND PROTOCOLS 220 Index

Compartmentalization ............................2, 85, 192, 204, 206 Computational modeling..........................................204, 214 construction ........................................................204, 212 Confocal microscopy .............................................. 87, 88, 98 Cotransfection ..............................................................11, 97 C-terminal polybasic stretch.........................................86–87 Cyclic nucleotide-gated channel........................ 1, 71, 72, 74, 76, 78–82, 86, 203 Cyclic nucleotides binding domain (CNBD) .............. 43–46, 50, 51, 53–56 Cytochalasin D ..................................................... 9, 112, 113

D DB-cAMP. See Dibutyryl cyclic adenosine monophosphate (DB-cAMP) Dibutyryl cyclic adenosine monophosphate (DB-cAMP) ........................................... 143, 144 Disruptor peptide ..................................... 139–141, 145, 146 AKAP-in silico ...........................................................140 Ht31 ............................................139–141, 147, 152, 168 DNA amplification ...............................................................169 enzymes .......................120–122, 124–126, 137, 167, 169 microinjection of DNA ...........25, 86, 117, 123, 126–127 modification ...............................................................169 primers.................................119–120, 123, 169, 174, 178 sequencing ..................................................................169

E Edinburgh Minimal Medium (EMM).............................182 Electrophysiological measurements ........................ 71, 72, 80 ELISA, SeeEnzyme–linked immunosorbent assay (ELISA) EMM. See Edinburgh Minimal Medium (EMM) Enzyme-linked immunosorbent assay (ELISA)...... 132, 134, 135, 153–158, 160, 163 E prostanoids (EP) ...........................................................144 euPacα ................................................................ 131, 132, 135 euPacβ ................................................................ 131, 132, 135

F fbp1 ................................................................... 181, 185, 189 Fibronecting .......................................................................96 Fission yeast .....................................................................181 Flavins ........................................................................ 20, 133 Fluorescence acceptor ..............................................2, 7–10, 14, 15, 19, 22, 26, 42, 47, 49, 59, 61, 62, 64, 65, 67, 104, 118, 154, 159, 160, 164, 213 bleaching .............................................. 10, 14, 15, 18, 21 CFP .........................3, 5, 9, 15, 16, 21, 30–32, 51, 54–56, 60–62, 64, 65, 67, 86–92, 94, 95, 98–100, 110, 118 donor ............... 2, 7–10, 14, 15, 19, 22, 26, 42, 47, 49, 51, 59, 61, 62, 64, 66, 67, 118, 154, 159, 160, 164, 213 FD-FLIM ............................................ 14, 15, 18, 20, 21

fluorophore ................................. 2, 3, 5, 7–10, 14, 15, 18, 21, 26, 27, 36, 38, 59, 64, 91, 110, 111, 113, 114, 118, 120, 159, 163–165, 213 Förster resonance energy transfer (FRET) AKAR4 .......................................................... 2–6, 11 cyclic nucleotide-gated (CNG) channels probes .................................................71 efficiency .......................... 2, 26, 42, 47, 59, 60, 66–69 EPAC-based sensor ......................................110, 114 EpacH90 ...................................................... 2, 3, 5, 6 FRET reporter calibration ................ 25–39, 213, 214 genetically enginnered ..............................................2 limitations ................................................... 27, 60, 72 PKA-based sensor .................................... 2, 7, 10, 54 probe ...................... 2, 3, 10, 59, 61, 67, 68, 71, 86, 92 ratio changes ......................................... 26–28, 90, 99 reporters reporters: design................................................47–50 reporters: development ............2, 3, 25, 36, 41–56, 59 reporters: geometrical considerations ................42, 48 sensors ..................................2–5, 7–9, 13–23, 25–27, 35, 36, 38, 39, 42, 50, 57, 86, 95, 97, 103–114, 117–128, 133, 213, 214 signal-to-noise .........................15, 39, 60, 98, 99, 127 spatiotemporal resolution..............................103, 214 targeted sensors............................................. 3, 26, 36 unimolecular sensor ............................ 50, 53–56, 213 frequency-domain (FD).................................... 14, 17–19 imaging ...........13–23, 27, 41, 60, 62–64, 67, 76, 118, 213 lifetime................................................ 13–23, 59, 67, 164 life time imaging (FLIM) .......................................14–22 live cell .............................................60, 95, 103, 180, 214 mCherry ......................................4–8, 11, 91, 95, 96, 100 MTurquoise2, 15, 18, 19 phototoxicity ...........................................................15, 98 time correlated single photon counting (TCSPC)........14 Venus ........................................................................3, 15 YFP ..........................................3–5, 9, 30–32, 38, 51–54, 60–65, 87, 89–92, 94, 96, 98–100, 110, 114, 118 Fluorescence polarization (FP) ................................. 145, 154 Forskolin............................................... 5, 7, 9, 10, 18, 19, 21, 26, 27, 34, 35, 64, 68, 72, 110, 113, 142, 143, 213 Fura-2 ............................................................... 72–74, 77–80 Fura red ............................................................ 91, 92, 94, 99

G GEF-catalytic domain ........................................................15 Genetic crosses .................................................................184 Genotyping ...................................................... 120, 123, 127 GKAP identification.......................................................193, 199 IRAG ................................................................. 192, 199 GLP-1. See Glucagon-like peptide-1 (GLP-1) Glucagon ......................................................................90, 91

CAMP SIGNALING: METHODS AND PROTOCOLS

Index Glucagon-like peptide-1 (GLP-1) ...............................88–90 Gluco-incretin hormones ...................................................88

H H89 .....................................................5, 7, 11, 139, 144, 145 Helix mimetics .........................................................153, 158 Heparin ............................................................ 105, 106, 111 Homogenous time-resolved fluorescence assay (HTRF) ...................153, 155, 158–160, 163–165 HTRF. See Homogenous time-resolved fluorescence assay (HTRF)

I IBMX ................................................... 5, 7, 9, 10, 18, 19, 21, 26, 64, 68, 69, 72, 74, 88, 110, 132–135, 143 Image cytometry ...........................................................59, 61 Image segmentation ............................................... 62, 63, 65 In-gel digestion ............................................... 194, 196–197, 199, 200 Inhibitory peptides ........................................... 153, 154, 158 In-silico approach .........................................................41–57 In-solution competition............................ 192, 193, 195–196 Insulin-secreting β-cells..........................................88–90, 96 Inter-chromophore distance ....................... 47, 50, 52, 55, 56 Ionic currents measurement............................................... 71–76, 79–82 Isoproterenol ................................................... 17, 18, 68, 74, 119, 121, 122, 143, 144

K K-Ras .................................................................................87

L Label-free liquid chromatography-high resolution mass spectrometry (LC-MS/MS).......... 192, 194, 197, 200 Lambert instruments FD-FLIM........................................15 Laminin ...............................................96, 104, 106, 109, 113 Langendorff apparatus...................................... 105, 107, 111 LC-MS/MS. See Label-free liquid chromatography-high resolution mass spectrometry (LC-MS/MS) Liberase ............................................................................111 Light-induced cyclase.......................................................132 Linker peptides.......................................................47–48, 58 Lipofectamine .......................................5, 6, 92, 93, 119, 121 L-type Ca2+ channel ...........................................................80

M Mass action law for binding reactions ..............................205 Mass spectrometry............................................................192 Michaelis–Menten approximations ..........................205, 211 Microdomains ............................................................7–8, 25 Microinfusion .........................................................27–29, 31

221

Microscopy ............................... 59, 60, 62–64, 67, 68, 86–89, 92–93, 98, 100, 106, 114, 118, 154, 187, 189, 207 Modified Eagle's medium (MEM) with Hank's salts ......106 Modulation-optimum ........................................................20 Monoclonal cell line .....................................................30, 36 mRNA ..............................................................................179

N NaN3 ........................................................................ 194, 198 Neighboring cells..................................................................3 NIST-traceable light source..........................................61, 62 Nystatin .................................................................. 73, 75, 82

O optogenetics......................................................................134

P PAC. See Photoactive adenylyl cyclase (PAC) Patch-clamp ...........................................................25–39, 86 PDE. See Phosphodiesterases (PDE) Peptide Spectral Matches (PSMs) ............................ 198, 199 Peptidomimetics ...............................................................153 Perfusion system ........................................... 6–10, 74, 76, 97 PGE2. See Prostaglandin E2 (PGE2) pGEX6P1................................................................. 174, 179 pH fluorescent protein ........................................................72 intracellular ................................................. 29–31, 36, 37 Phage display library ......................................................... 168, 171–176 selection ....................... 141, 168, 171–172, 175, 177, 178 Phosphatase .........................................80, 138, 152, 193, 198 Phosphodiesterases (PDE) activity ...................................35, 143, 183–185, 187, 188 gene ............................................................................183 inhibitors .................................................... 181–183, 186 mammalian ...........................................................13, 183 modulator ...........................................................181–190 trypanosome ...............................................................183 Photoactive adenylyl cyclase (PAC) dark activity ........................................................132–135 Photoisomerization ............................................................10 PKA. See Protein kinase-A (PKA) PKG domain........................................................................192 PKI. See Protein kinase inhibitor (PKI) Plasma membrane .....................25, 72, 82, 85–100, 138, 213 Plasmid purification .........................................................126, 183 Porphyrins ........................................................................133 Predictive model ............................................... 204, 205, 214 Proposed mechanism models............................................204 Prostaglandin E2 (PGE2) .......................... 68, 114, 143, 144 Protease enzyme ...............................................................106

CAMP SIGNALING: METHODS AND PROTOCOLS 222 Index

Protein engineering ...........................................................42, 141 purification chromatography columns............................. 170, 174, 192, 194, 196, 197 E. coli TOP10 cells................................................170 filtration buffer .............................................170, 174 Glutathione Sepharose ......................... 170, 174, 179 G-proteins ............................................................208 IPTG solution ..............................................170, 174 LB................................................. 170, 173–176, 179 lysis buffer.............. 127, 128, 170, 174, 193–195, 198 protein-protein interactions ....................... 13, 14, 59, 137, 151, 154, 163, 165 wash buffer ........................................... 170, 174–176 Protein kinase-A (PKA) activation ............................................... 1, 145, 152, 181, 187, 207, 208, 211, 212 activity ....................................................2–5, 10, 80, 138, 144, 152, 168, 181, 184, 187, 188 co-localization ....................................................192, 207 docking and dimerization domain ..............................167 holoenzyme ...........................44, 137, 138, 152, 211, 212 interaction....................................152, 153, 160, 167, 198 phosphorylation ................................ 2, 13, 144, 212–213 regulatory subunit ...................................... 44, 45, 51, 52, 86, 88, 95, 97, 98, 137, 138, 145, 152, 167–180, 191, 199, 211, 212 RI .....................17, 25, 135, 137–142, 145, 146, 168, 211 RII ..................................................25, 86, 137–142, 145, 146, 152, 153, 161, 167, 168, 171, 175, 178, 211 Protein kinase inhibitor (PKI) ............................ 11, 139, 144 Proteolysis ........................................................................200 PSMs. See Peptide Spectral Matches (PSMs) Pyrimidine analog 5-fluoro orotic acid medium ...............181

S Schizosaccharomyces pombe .................................................. 181 Second messengers ............................................. 13, 151, 191 Sensitized emission (SE) ....................................................14 Signaling cascades ............................................ 145, 151, 205 Small molecules .........................140, 153, 154, 162, 163, 205 SNARF-1 .....................................................................29, 37 Spectrum ............................... 26, 42, 63, 64, 69, 99, 114, 178 STAD. See Stapled AKAP disruptor (STAD) Stapled AKAP disruptor (STAD) ....................................141 Statistic assessment graphpad prism ...................................................155, 160 ImageJ .............................................11, 18, 62, 64, 68, 96 KaleidaGraph .................................................................8 Mascot ................................................................ 194, 198 Matlab ........................................................19, 61, 62, 64, 73, 78, 207 Metafluor...................................................... 9, 10, 38, 96 OriginPro .......................................................................8 Proteome Discoverer ..........................................194, 198 Z' factor ...................................................... 186, 189, 190 Subcellular compartments ..............................................2, 25 Sub-plasma membrane ...............................................85–100

T

Q

Terpyridin ................................................................. 153, 158 Tetrad dissection...............................................................184 TIRF. See Total internal reflection fluorescence (TIRF) Total internal reflection fluorescence (TIRF) .............. 87, 89, 90, 92, 93, 95, 96, 98, 100 Transfection .............................. 3–6, 8–11, 16–17, 19, 20, 29, 30, 35, 37, 77, 80, 92–94, 96, 97, 99, 119, 121, 142 Transgenic mice ........................................................117–128 Transgenic tools........................................................131–132 Translocation ...................................................... 85–100, 206 Tyrode buffer solution ..............................................106, 109

Quantum yield..............................................................15, 42

V

R

Valinomycin ............................................................29–31, 37 Vannas scissors..........................................................107, 111

Rational design .......................................................47–50, 57 Reactive oxygen species (ROS)...........................................21 Refractive index ............................................................42, 60 RIAD. See RI-selective peptide (RIAD) RI-selective peptide (RIAD) ............................ 141, 142, 145 Rp-cAMPS. See Adenosine-3',5'-cyclic monophosphorothioate Rp-isomer (Rp-cAMPS) Rselect.................................................................................. 141

W Whole-cell measurement...........................................................81–82

Y Yeast-based high-troughput screen...........................181–190