Protein Cages: Methods and Protocols [1 ed.] 1493921304, 9781493921300

Citation preview

Methods in Molecular Biology 1252

Brendan P. Orner Editor

Protein Cages Methods and Protocols

METHODS

IN

MOLECULAR BIOLOGY

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

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

.

Protein Cages Methods and Protocols

Edited by

Brendan P. Orner Department of Chemistry, King’s College London, London, UK

Editor Brendan P. Orner Department of Chemistry King’s College London London, UK

ISSN 1064-3745 ISSN 1940-6029 (electronic) ISBN 978-1-4939-2130-0 ISBN 978-1-4939-2131-7 (eBook) DOI 10.1007/978-1-4939-2131-7 Springer New York Heidelberg Dordrecht London Library of Congress Control Number: 2014952733 © 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. Exempted from this legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the Publisher’s location, in its current version, and permission for use must always be obtained from Springer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law. 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. While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein. Printed on acid-free paper Humana Press is a brand of Springer Springer is part of Springer Science+Business Media (www.springer.com)

Preface In recent years, there has been a surge in research focused on proteins that form cage-like structures. While these multicomponent proteins have been of great interest to scientists studying protein folding and structural biology, they have also posed one of the next big challenges in protein design and engineering. Furthermore, because of their often nanoscaled size, they have caught the eye of materials scientists, and their interior cavities have been exploited by chemists and enzymologists to explore the possibility of enhancing reactivity by means of confinement. They also have been used to develop bioconjugation chemical reactions, some applications of which have been for the presentation of ligands multivalently and to attach directing functionality in order to control the delivery of drugs or imaging agents encapsulated inside the cage. Protein cages are currently inspiring diverse scientific disciplines and are therefore at the crossroads of extremely widely scoped research. This volume emphasizes the geographic and scientific diversity of the field by highlighting the expertise of researchers based on three continents who are developing new techniques to help understand protein cages and to apply them to a variety of technologies. Following a nanomaterials direction, the chapter written by Tsvetkova and Dragnea provides procedures for the assembly of virus capsids around nanoparticles, whereas Uchida et al. describe methods to generate iron particles inside ferritin and organic polymers inside capsids. In addition, Pulsipher and Dmochowski present procedures for growing gold nanoparticles inside ferritins and assembling the proteins around the particles, and Sana and Lim describe how to measure relaxivities of protein cage encased iron nanoparticles. Protein cages have been the focus of protein design endeavors, and Ardejani and Orner describe the use of simple computational strategies to engineer more stable ferritins. As the field matures, developing improved methods for the production of these proteins is especially important, and the chapter by Rurup et al. describes how to isolate encapsulin proteins, and Lassila details procedures for the purification of bacterial microcompartments. Investigation into biochemical and physical properties of protein cages is also a large slice of this field and the development of novel approaches is required due to the unique challenges associated with this class of proteins. Cornell and Orner describe a technique to identify specific oligomerization states of protein cages by designing binding sites for a fluorescent reagent at key protein–protein interfaces, whereas Grove et al. present techniques to determine the role structural metal has on the assembly of a miniferritin, and Zhang and Ardejani describe how to apply Differential Scanning Calorimetry to the assembly of ferritins. Van Rosmalen et al. review the use of Atomic Force Microscopy to probe the structure and mechanical properties of protein cages, and Gibbons et al. present a theoretical model to understand the deformation and indentation of hollow protein nanostructures. The chapters herein collectively provide an introduction to the rich world of protein cage research and specific techniques to understand and exploit this fascinating class of proteins. Moreover, it is hoped that this volume will help to inspire and further propel the current multidisciplinary enthusiasm in studying and discovering new applications for protein cages. London, UK

Brendan P. Orner

v

.

Contents Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v ix

1 Encapsulation of Nanoparticles in Virus Protein Shells . . . . . . . . . . . . . . . . . . . . . Irina B. Tsvetkova and Bogdan G. Dragnea 2 Use of Protein Cages as a Template for Confined Synthesis of Inorganic and Organic Nanoparticles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Masaki Uchida, Shefah Qazi, Ethan Edwards, and Trevor Douglas 3 Ferritin Encapsulation and Templated Synthesis of Inorganic Nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Katherine W. Pulsipher and Ivan J. Dmochowski 4 Determining the Relaxivity Values of Protein Cage-Templated Nanoparticles Using Magnetic Resonance Imaging . . . . . . . . . . . . . . . . . . . . . . . . Barindra Sana and Sierin Lim 5 Computationally Assisted Engineering of Protein Cages. . . . . . . . . . . . . . . . . . . . Maziar S. Ardejani and Brendan P. Orner 6 Recombinant Expression and Purification of “Virus-like” Bacterial Encapsulin Protein Cages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . W. Frederik Rurup, Jeroen J.L.M. Cornelissen, and Melissa S.T. Koay 7 Production of Bacterial Microcompartments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jonathan K. Lassila 8 Detection of Protein Cage Assembly with Bisarsenic Fluorescent Probes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thomas A. Cornell and Brendan P. Orner 9 Determining the Role of Metal Binding in Protein Cage Assembly . . . . . . . . . . Anne Grove, Ambuj K. Kushwaha, and Khoa H. Nguyen 10 Differential Scanning Calorimetry to Quantify the Stability of Protein Cages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yu Zhang and Maziar S. Ardejani 11 Material Properties of Viral Nanocages Explored by Atomic Force Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mariska G.M. van Rosmalen, Wouter H. Roos, and Gijs J.L. Wuite 12 Computational Mechanics of Viral Capsids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Melissa M. Gibbons, Luigi E. Perotti, and William S. Klug

1

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii

17

27

39 51

61 69

79 91

101

115 139 189

.

Contributors MAZIAR S. ARDEJANI  Department of Chemistry, King’s College London, London, UK JEROEN J.L.M. CORNELISSEN  Laboratory for Biomolecular Nanotechnology, MESA+ Institute for Nanotechnology, University of Twente, Enschede, The Netherlands THOMAS A. CORNELL  Department of Chemistry, King’s College London, London, UK IVAN J. DMOCHOWSKI  Department of Chemistry, University of Pennsylvania, Philadelphia, PA, USA TREVOR DOUGLAS  Department of Chemistry and Biochemistry, Center for Bio-Inspired Nanomaterials, Montana State University, Bozeman, MT, USA BOGDAN G. DRAGNEA  Department of Chemistry, Indiana University, Bloomington, IN, USA ETHAN EDWARDS  Department of Chemistry and Biochemistry, Center for Bio-Inspired Nanomaterials, Montana State University, Bozeman, MT, USA MELISSA M. GIBBONS  Department of Mechanical & Aerospace Engineering, University of California, Los Angeles, CA, USA ANNE GROVE  Department of Biological Sciences, Louisiana State University, Baton Rouge, LA, USA WILLIAM S. KLUG  Department of Mechanical & Aerospace Engineering, University of California, Los Angeles, CA, USA MELISSA S.T. KOAY  Laboratory for Biomolecular Nanotechnology, MESA+ Institute for Nanotechnology, University of Twente, Enschede, The Netherlands AMBUJ K. KUSHWAHA  Department of Biological Sciences, Louisiana State University, Baton Rouge, LA, USA JONATHAN K. LASSILA  DuPont Industrial Biosciences, Palo Alto, CA, USA SIERIN LIM  School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore, Singapore KHOA H. NGUYEN  Department of Biological Sciences, Louisiana State University, Baton Rouge, LA, USA BRENDAN P. ORNER  Department of Chemistry, King’s College London, London, UK LUIGI E. PEROTTI  Department of Mechanical & Aerospace Engineering, University of California, Los Angeles, CA, USA KATHERINE W. PULSIPHER  Department of Chemistry, University of Pennsylvania, Philadelphia, PA, USA SHEFAH QAZI  Department of Chemistry and Biochemistry, Center for Bio-Inspired Nanomaterials, Montana State University, Bozeman, MT, USA WOUTER H. ROOS  Natuur- en Sterrenkunde and LaserLab, Vrije Universiteit, Amsterdam, The Netherlands W. FREDERIK RURUP  Laboratory for Biomolecular Nanotechnology, MESA+ Institute for Nanotechnology, University of Twente, Enschede, The Netherlands BARINDRA SANA  School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore, Singapore IRINA B. TSVETKOVA  Department of Chemistry, Indiana University, Bloomington, IN, USA

ix

x

Contributors

MASAKI UCHIDA  Department of Chemistry and Biochemistry, Center for Bio-Inspired Nanomaterials, Montana State University, Bozeman, MT, USA MARISKA G.M. VAN ROSMALEN  Natuur- en Sterrenkunde and LaserLab, Vrije Universiteit, Amsterdam, The Netherlands GIJS J.L. WUITE  Natuur- en Sterrenkunde and LaserLab, Vrije Universiteit, Amsterdam, The Netherlands YU ZHANG  College of Chemical Engineering, Nanjing Forestry University, Nanjing, Jiangsu, China

Chapter 1 Encapsulation of Nanoparticles in Virus Protein Shells Irina B. Tsvetkova and Bogdan G. Dragnea Abstract The self-assembly of virus-like particles may lead to materials which combine the unique characteristics of viruses, such as precise size control and responsivity to environmental cues, with the properties of abiotic cargo. For a few different viruses, shell proteins are amenable to the in vitro encapsulation of non-genomic cargo in a regular protein cage. In this chapter we describe protocols of high-efficiency in vitro self-assembly around functionalized gold nanoparticles for three examples of icosahedral and non-icosahedral viral protein cages derived from a plant virus, an animal virus, and a human retrovirus. These protocols can be readily adapted with small modifications to work for a broad variety of inorganic and organic nanoparticles. Key words Virus-like particle, Gold nanoparticle, Templated self-assembly, Protein cage

1

Introduction Virus protein shells have elicited great interest in the past 10–15 years for features that allow nanoscale control of chemical properties through either genetic engineering or direct chemical manipulations [1]. A number of applications are currently pursued ranging from nanomedicine [2–4] to catalysis [5–7] and energy conversion technologies [8–11]. The encapsulation of genetic or nongenetic materials in virus capsids provides routes to engineer new materials [6, 12–16] and opportunities for gaining new insights in the virus assembly mechanisms [17–21]. Gold nanoparticles (GNPs) are particularly interesting targets for encapsulation due to relatively good size control and biocompatible surface chemistry which allows for conjugation with a variety of biomolecules [22] of interest in gene and drug delivery [23, 24] and diagnosis and detection [25, 26]. A distinctive optical property of GNPs is their surface plasmon polariton resonance at visible wavelengths which is a function of the nanoparticle size and where optical scattering and absorption are significantly enhanced [27]. The surface plasmon resonance makes it possible to discriminate between nanoparticles of different sizes spectroscopically and in situ by nonintrusive

Brendan P. Orner (ed.), Protein Cages: Methods and Protocols, Methods in Molecular Biology, vol. 1252, DOI 10.1007/978-1-4939-2131-7_1, © Springer Science+Business Media New York 2015

1

2

Irina B. Tsvetkova and Bogdan G. Dragnea

optical techniques relying on either direct absorption or scattering detection [28] or assisted by organic fluorophore tags [29]. Virus-like particles encapsulating multiple GNPs can be formed by covalent binding of small gold nanoparticles (8 nm) can be encapsulated through self-assembly from protein subunits and GNPs inside the cavity normally occupied by the viral genome [33–36]. Here we focus on the latter approach and discuss three examples of GNP encapsulation by the capsid of the brome mosaic virus (BMV), a simple nonenveloped icosahedral virus; the nucleocapsid of the Sindbis virus, an icosahedral enveloped animal virus; and the polyprotein shell of an enveloped human retrovirus, the type I human immunodeficiency virus (HIV). 1.1

General Features

1.2 Brome Mosaic Virus

To direct the assembly of capsid proteins around GNPs, it is necessary to functionalize the GNP surface in a way that promotes essential interactions which are reminiscent of those occurring between the natural cargo (nucleic acid) and the capsid protein. Some virus proteins will spontaneously self-assemble through nonspecific electrostatic interactions with cargo as it was shown by Bancroft and colleagues for small icosahedral plant viruses [37]. However, not all nonspecific, attractive interactions will result in a regular capsid. If interactions are too strong, kinetic bottlenecks will result. The assembly of other viruses requires specific interactions between specific nucleic acid sequences and coat proteins, which promote the growth of a functional proteinaceous cage around the nucleic acid cargo [34, 38, 39]. This protein cage carries crucial functions for many of the virus life stages. BMV is one of the first icosahedral plant viruses to be reconstituted in vitro from purified components [40]. The crystal structure of the BMV capsid was determined at 3.4 A˚ resolution. BMV’s capsid belongs to the widespread architecture of icosahedral shells, which have the biological advantage of the highest volume-to-surface ratio of any of the regular polyhedra. Icosahedral shells can only form from a number of subunits belonging to a subset of natural numbers consequence of the quasi-symmetry rules outlined in 1962 by Caspar and Klug [41]. BMV possesses a protein coat composed of 180 identical 20 kDa coat protein subunits [42]. The capsid protein N‐terminus points toward the interior of the capsid, is flexible, and is rich in basic residues. While the assembly of empty capsids is possible in certain conditions, the strong electrostatic attraction between negatively charged RNA and the Nterminus significantly speeds up assembly and stabilizes the final assembly result. Protein-protein interactions are responsible for the formation of a regular array and result from repulsive electrostatic interactions balanced by attractive hydrophobic interactions. The balance between attractive and repulsive interactions within

Virus-Like Nanoparticles

3

the protein shell is responsive to the chemical environment [43, 44]. Similar aspects of the nucleic acid-protein and proteinprotein interactions in many simple viruses [45] made BMV an excellent model system and, thus, a source of inspiration for future virus-based materials. Because BMV can be reconstituted in vitro from capsid proteins and viral and nonviral single-stranded nucleic acids as well as anionic polymers, it was long believed that BMV assembly is carried out by nonspecific electrostatic interactions between proteins and cargo [37]. However, the accumulating evidence of specific interactions suggests that assembly pathways in cells may be different than those involved in in vitro reconstitution [46–48]. In the later case, BMV can be dissociated into coat protein dimer subunits and RNAs by increasing solution pH above 7 and the ionic strength to above 1 M. Viral RNA is then separated from protein. By mixing negatively charged gold nanoparticle with proteins at high ionic strength and pH, when capsid proteins are stable as dimers in solution and upon reestablishing low pH and ionic strength, virus-like particles of sizes commensurate with the GNP diameter form spontaneously [49, 50]. 1.3 Alphavirus: Ross River Virus

The Ross River virus is a small, icosahedral, enveloped virus from the Alphavirus genus that can infect humans and animals [51]. Its structure is more complex than BMV, with two concentric protein shells interconnected across a lipid membrane and pseudo T ¼ 4 icosahedral symmetry for both shells [52]. The intact virion is 70 nm in diameter. Alphaviruses have an icosahedral nucleocapsid of 39 nm diameter which assembles from 240 identical protein subunits. The inner cavity encapsulating the genome has a diameter of 33 nm. To construct an alphavirus-like nanoparticle, it is crucial to preserve the native interactions between nucleocapsid proteins adsorbed on the surface of the nanoparticle in order to maintain the shell structure and ability to connect to glycoproteins [53]. Although alphavirus core-like particles (CLP) have been shown to assemble around negatively charged particles [38], functionalization with 48mer-ssDNA containing the encapsidation signal from genomic RNA increased the yield and morphological resemblance to wild-type nucleocapsids [54].

1.4 Human Immunodeficiency Virus (HIV)

The HIV-1 life cycle entails two distinct assembly states: the noninfectious immature state and the infectious, mature state. Immature HIV particles are spherical and formed from a polyprotein Gag shell encapsulating two RNA strands [55]. The structure of isolated Gag protein has been described at atomic-level resolution, but the quaternary structure of the immature particle remains unclear, mainly due to particle variability [55, 56]. Cryoelectron tomography studies of immature virions provided evidence of incomplete Gag shells with local hexagonal order and the absence of a regular

4

Irina B. Tsvetkova and Bogdan G. Dragnea

array of pentamers—a hallmark of icosahedral capsids [56, 57]. Obtaining structural information at higher resolution could be facilitated by Gag constructs that form more regular spherical assemblies [58]. It was proposed that nanoparticle-templated assembly could provide a venue toward this end [39]. The major structural element of HIV-1, the Gag polyprotein, contains four major domains: the matrix (MA), capsid (CA), nucleocapsid (NC), and C-terminal P6 [55]. The in vitro assembly of recombinant Gag proteins around nucleic acid results in virus-like particles (VLPs) that have similar structural properties with immature HIV-1 viruses isolated from cells [56, 59]. A well-studied assembly-competent recombinant Gag is the Δ16–99 mutant. It has a deletion of the N-terminal residues 16–99 and lacks the myristylated acid for improved solubility. This construct has been studied extensively for its ability to form Gag-VLPs in the presence of nucleic acid [55, 56, 59]. The major drawback for higher-resolution structural studies of the current in vitro assembly system is that the Gag-VLP assembly was inefficient and the variation in size and morphology precludes structural methods based on averaging among many particles [60]. To assemble Gag-VLPs by the nanoparticle-directed method, the presence of immobilized DNA on a spherical nanoparticle template is required [39]. GNP template Gag-VLPs display a much more uniform size than immature VLPs obtained from nucleic acid and protein. This allowed the use of single-particle reconstruction in the study of defects associated with the ground state of the assembly result [39]. In this chapter we describe the synthesis and functionalization of gold nanoparticle cores with different diameters, purification of capsid protein subunits, and assembly conditions for the three virus capsids mentioned above.

2

Materials Prepare and store all reagents at room temperature (unless indicated otherwise). Be aware of and follow all waste disposal regulations. All solutions were prepared using ultrapure water (18 MΩ cm at 25  C) and analytical grade reagents.

2.1 Buffers for BMV Protein Purification and Capsid Reassembly

Disassembly: 0.5 M CaCl2, 0.005 M Trizma HCl, pH 7.4. Desalting: 0.01 M Trizma, pH 7.4. TKM (protein storage): 1 M KCl, 0.01 M Tris-HCl, 0.005 M MgCl2, pH 7.4. TNKM (VLP reassembly): 0.05 M Trizma HCl, 0.05 M NaCl, 0.01 M KCl, 0.005 M MgCl2, pH 7.4. SAMA (virus storage): 0.05 M NaOAc, 0.008 M Mg(OAc)2, pH 4.6.

Virus-Like Nanoparticles

5

2.2 Buffers for RRV Protein Purification and Capsid Reassembly

Elution: 500 mM NaCl, 20 mM HEPES, 2 mM EDTA, pH 8.0.

2.3 Buffers for HIV Δ16–99 Gag Capsid Reassembly

HIS (high ionic strength): 50 mM HEPES, 0.5 M NaCl, pH 7.5.

2.4

60 nm citrate-coated gold nanoparticles (Ted Pella, Redding, CA) were stored at 4  C.

Other Reagents

HNE: 20 mM HEPES, 0.15 M NaCl, 0.1 mM EDTA, pH 7.5.

Storage buffer: 50 mM HEPES, 0.5 M NaCl, 4 mM DTT, pH 7.5. LIS (low ionic strength): 50 mM HEPES, 0.1 M NaCl, 1 mM DTT, pH 8.0.

Carboxyl-terminated tetraethylene glycol undecane thiol HS(CH2)11-(OCH2CH2)4-OCH2-COOH (ProChimia, Poland) was stored at 4  C in nitrogen. Lyophilized thiol-PEG-modified alphavirus-specific sequence HS-PEG-48mer (50 -HS-3[(CH2CH2O)6 phosphoramidite](CCGTTAATGCATGTCGAGATATAAAGCATAAGGGACATGCATTAACGG-30 ). Lyophilized thiol-modified oligonucleotides for Gag assembly HS-PEG-(TG)25 (50 -HS-3[(CH2CH2O)6 phosphoramidite](TG)25-30 ) (Integrated DNA Technologies Inc., Coralville, IA). NAP-25 columns (Sephadex G-25 DNA grade). Slide-A-Lyzer mini-dialysis units 20 k MWCO, 10 k MWCO. Float-A-Lyzer microdialysis unit 100 kDa. Nanosep centrifugal devices 300 K MWCO, 30 K MWCO. The recombinant protein, Δ16–99 Gag, was a courtesy from Dr. Alain Rein’s laboratory (NCI, Frederick, MD 21702).

3

Methods

3.1 Synthesis, Functionalization, and Purification of Gold Cores

Citrate-coated GNPs with an average diameter of 12 nm are prepared following the Slot and Geuze method with tannic acid added as an additional reducing agent [61]: 1. Add 1 ml of 1 %w/v tetrachloroauric acid monohydrate (HAuCl4H2O) to 79 ml of water and heat to 60  C. 2. At the same time mix 4 ml of 1 %w/v trisodium citrate dehydrate, 45 μL of 1 %w/v tannic acid, and 45 μL of 2.5 mM K2CO3 with 16 ml of water and also heat to 60  C. 3. Mix both solutions rapidly and bring to boiling while mixing. Upon this operation, a color is expected to change from dark blue-gray to purple to orange red in a few minutes.

6

Irina B. Tsvetkova and Bogdan G. Dragnea

4. After reaching the orange-red color, boil the mixture for another 10 min and cool down to room temperature. When correctly prepared, citrate-stabilized GNPs are stable upon storage and should remain suspended in solution for longer than 1 year. Citrate-coated GNPs with an average diameter of 19 nm are prepared following the Frens [62] method: 1. Heat 100 ml of 0.01 %w/v tetrachloroauric acid monohydrate solution until boiling in a 200 ml beaker while stirring vigorously. 2. Add 1.8 ml of 1 %w/v trisodium citrate dehydrate solution to boiling gold solution, while maintaining temperature and stirring rate. After addition, solution color should change from colorless to gray to dark blue to purple to red in 5–10 min. 3. Boil mixture for 30 min with taking care to replenish water volume as it evaporated by adding preheated water. 4. After 30 min, turn off the heat and cool down the solution to room temperature. GNPs are functionalized by mixing with carboxylate‐terminated thiolalkylated tetraethyleneglycol (TEG) ligand or with thiolmodified polyethylene glycol (PEG)-derived ssDNA. Since the thiol bond formed between the ligands and the metal surface is stronger than the bond formed with the initial citrate molecules, citrate is easily displaced in the presence of ligand excess: 1. Add 10 μL of TEG ligand to 100 mL of citrate-coated 12 nm GNP solution. An immediate darkening of the solution color should be observed indicating ligand substitution. Stir mixture overnight to guarantee maximum coverage. 2. Ligand excess is removed through pelleting down functionalized GNPs from solutions by centrifugation (parameters of centrifugation for different GNP sizes are in Table 1). Remove the supernatant and resuspend GNP pellet in water. Repeat centrifugation for three times. Resuspended the final pellet and store it in water. UV-Vis could be used to determine the concentration of GNP-TEG stock solution (see Note 1). Table 1 Centrifugation conditions for GNP purification from ligand excess GNP size (nm)

RCF, g

Time (min)

12

100,000

45

19

70,000

40

60

38,000

30

Virus-Like Nanoparticles

7

The functionalization of citrate-coated 19 and 60 nm gold particles with 50 SH-PEG-DNA is carried out with a slight modification of the Hurst et al. protocol [63]: 1. Prior to use, resuspend lyophilized HS-PEG-ssDNA in 100 μL of 0.1 M DTT, 0.18 M phosphate buffer, pH 8.0, and incubate the resultant mixture at room temperature for 1 h. Purify the ligand with reduced disulfide bonds using a NAP-5 column. 2. Add freshly prepared ligands to the gold nanoparticle solution in water at a number ratio of 3 equivalents HS-PEG-ssDNA to 0.152 nm2 GNP surface (explanation of calculation is in Note 2). 3. After 2 h incubation, adjust the mixture to 0.01 M in phosphate buffer pH 8 and 0.01 % in sodium dodecyl sulfate. 4. After the second 90 min incubation, bring the mixture to 0.05 M NaCl in phosphate buffer and then to 0.1 M, and then sonicate for 10 s. 5. Subsequently increase the ionic strength to 1 M at 0.1 M intervals every 15 min. Incubate the reaction mixture thereafter for 2 days at room temperature in order to ensure oligonucleotide saturation on the nanoparticle surface. 6. Remove ligand excess by pelleting down functionalized GNPs from solutions (parameters of centrifugation for different GNP sizes are in Table 1). Remove supernatant and resuspend GNPs in HIS buffer. Repeat wash steps for a total of five times. Resuspend the final pellet in water and store at 4  C. 3.2 BMV Protein Purification from WT-BMV

Briefly, wild-type BMV virus is harvested from Nicotiana benthamiana infected according to the Agrobacterium tumefaciens infiltration protocol [64], purified through a CsCl gradient, and frozen at 80  C for storage. To prepare BMV coat proteins, first thaw the virus solution and then perform the following steps of virus disassembly and protein purification at 4  C: 1. To dissociate BMV, dialyze 100–200 μL of 2 mg/ml virus solution overnight against disassembly buffer. 2. Remove RNAs containing sediment by centrifugation at 17,000  g for 30 min. 3. Dialyze overnight the supernatant containing dissociated coat proteins against a desalting buffer to remove calcium, then against the TKM buffer for storage purposes. 4. To determine the protein concentration, measure the absorbance at 280 nm and use the extinction coefficient 0.82 cm2/g [65]. The ratio of the absorbance at 260 nm to that at 280 nm should be 0.6 or less, which indicates complete removal of RNA (remaining RNA is less than 0.2 %) [66]. Before proceeding to

8

Irina B. Tsvetkova and Bogdan G. Dragnea

the following experiments, BMV protein integrity should be tested with MALDI‐TOF for the right molecular weight of ~20,300 amu. 3.3 RRV Capsid Protein Purification [67]

1. Express RRV CP in pET-29b DNA plasmid (Novagen, EMD Chemicals Inc., Gibbstown, NJ) in Rosetta2 cells until the OD600 is 0.4–0.6. At that time, induce cells with 1 mM ITPG (isopropyl β-D-1-thiogalactopyranoside) and grow at 37  C for 4 h before harvesting. 2. Lyse cells by French press and apply the clarified lysate to a HiTrap SP FF 1 mL column. Elute the protein with elution buffer. 3. Four peaks should be observed in the absorbance profile as a function of elution time. The last peak (#4) from the chromatogram contains the capsid protein monomer. An analysis of concentrated sample of peak #4 shows the presence of capsid proteins as well a truncated form of the protein. 4. Concentrate proteins by centrifugation with Nanosep centrifugal devices 30 K MWCO, and buffer exchange into HNE for storage. Use BSA Bradford assay to determine the concentration of the capsid protein.

3.4 Reassembly of BMV Protein Capsid Around 12 nm GNPs

There are three main intermolecular interactions in a virus whose relative magnitudes can be controlled by changing ionic strength and pH [43]: the hydrophobic interaction between protein subunits, the electrostatic repulsions between deprotonated carboxyl groups, and the electrostatic attractions between negatively charged cargo and positively charged flexible protein arm. It is worth noting that the presence of divalent ions like Mg2+ or Ca2+ is necessary for stabilizing virus capsids by interactions with carboxyl groups at subunit interface [42]. Based on these facts we used a two-step protocol [17, 68] to assemble BMV around GNPs. First, we mixed functionalized particles with proteins at high pH and high ionic strengths where attractive interactions between particle and capsomers are minimal and repulsive interaction between proteins overcomes hydrophobic attraction. Then, by lowering the ionic strength, proteins concentrate around negatively charged particles due to interactions with the positively charged protein tail. However, in these conditions the shell appearance in micrographs is mostly amorphous, without signs of capsomer organization. Lowering pH in the presence of divalent ions is a necessary subsequent step for proteins to anneal into the regular structure of the capsid: 1. TEG‐stabilized GNPs were dialyzed against TKM buffer (see buffer table) before they were mixed with BMV protein which was stored in TKM buffer.

Virus-Like Nanoparticles

9

2. A 0.75 equivalent of GNPs was used per 180 capsid proteins (corresponding to a T ¼ 3 structure). The volumes of freshly prepared proteins (within 1 or 2 days after the virus dissociation process) were adjusted to reach a final concentration of protein of 0.5 mg/mL in the dialysis mixture and then mixed with functionalized particles and a first dialysis performed against TNKM buffer. 3. The mixture was then dialyzed against the SAMA buffer. Assembled VLPs are stable in this buffer and can be frozen at 80  C for storage up to 6 months. The efficiency of incorporation was close to 90 % (see Subheading 3.7). 3.5 Assembly of Alphavirus Capsid Proteins with Functionalized GNPs

Ross River CLPs with GNP cores functionalized by HS-PEG-48mer were assembled in vitro following assembly with nucleic acid conditions previously established with slight modifications [67]. Previous work has shown that capsids do not assemble at ionic strength higher than 0.2 M. A phase diagram of the alphavirus capsid was constructed by varying ionic strength and pH (data not published), and optimal conditions of assembly were found at ionic strength between 0.05 and 0.75 M and pH between 7.5 and 8 (see Note 3). Moreover, it was found that charge density is crucial and complete capsids were produced in high yields (80 %) when templates with 8,000–12,000 charges were used: 1. Mix 2 μl of functionalized 19 nm GNP (stock concentration of 0.25 μM) in 0.005 % Tween 20 with 1 μl of 0.1 M NaCl to obtain 0.05 M NaCl in the reaction and vortex for a few seconds. 2. Add 4.2 μL of RRV CP (5 mg/ml in HNE buffer) to obtain the number ratio of 1 GNP to 960 CP (or GNP:CLP ¼ 1:4) and the final protein concentration in the mixture 3.0 mg/ml. 3. Mix assembly reaction for a few seconds and let it stand at room temperature for 30 min. After that, the Au-CLP solutions should be kept at 4  C. The efficiency of incorporation should be at least 85 % and the CLP diameter measured from negative-stained TEM images around 40 nm (37.5  3.2 nm in our case) which is close to the diameter of alphavirus nucleocapsid (see Subheading 3.7).

3.6 Assembly of Δ16–99 Gag Around DNA-Coated Gold Nanoparticles

Encapsulation of GNP by recombinant Gag was carried out by adapting the Datta et al. protocol [69]. Before assembly, GNPs functionalized with HS-PEG-TG25 should be dialyzed against HIS buffer at room temperature: 1. Mix corresponding amounts of concentrated recombinant Δ16–99 Gag in storage buffer and GNPs at room temperature to obtain a ratio of 15,000 equivalents of Gag to 1 equivalent of 60 nm GNP. 2. Dilute mixture with HIS buffer to a final protein concentration of 1 mg/ml.

10

Irina B. Tsvetkova and Bogdan G. Dragnea

3. Dialyze assembly reaction (20 k MWCO) overnight against LIS buffer at 4  C. 4. Next day change dialysis membrane to 100 k MWCO and dialyze overnight to remove unbound Gag. 5. Concentrate final sample (Nanosep, 300 k MWCO, 7,000  g, 4  C) for analysis and storage. VLP diameter should be around 100 nm and 80 % of gold particles are encapsulated. 3.7 VLP Characterization

The successful assembly of VLPs is routinely interrogated by negative-stain transmission electron microscopy (TEM). Micrographs are analyzed with the ImageJ Processing Toolkit to estimate NP and VLP diameters (Fig. 1a–d). The morphological features of protein shell such as defects or capsomer location could be visualized and measured from TEM images (Fig. 2). The yield of VLPs for each assembly (efficiency of encapsulation) is determined from the TEM images by dividing the number of complete VLPs over the total number of gold particles in the pictures (either involved in complete VLPs, partial VLPs, or free GNPs). Pictures should be taken from several different areas of the TEM grid, and a minimum of 200–300 particles are counted for each sample.

Fig. 1 Negative-stained TEM images of (a) citrate-coated GNP, (b) TEG-coated GNP, and (c) BMV VLPs. The white ring with a thickness of 1–1.5 nm around GNP in (b) represents the ligand layer. A dark gray ring can be observed between the ligand and the protein coat, which indicates enhanced stain penetration at that location. The outer light-gray layer with a thickness of 5–7 nm in (c) is the protein layer. This thickness value corresponds well to the known radial extent of capsomers. (d) Histograms of TEM-measured diameters for GNP cores and VLPs. (e) Hydrodynamic diameter distribution measured by dynamic light scattering

Virus-Like Nanoparticles

a

11

c Gray density profile

120 100 80 60 40 20 0

b

5 10 15 20 25 30 35 40 nm

200

Gray density profile

160 120

Au-Gag-VLP NC CA

80 DNA-PEG-Au

40 0 0

10

20

30

40 50 60 70 80 Particle cross-section (nm)

90

100 110 120

Fig. 2 Morphological features of single Au-Gag-VLPs. (a) Negative-stained TEM images of two individual 60 nm gold particles encapsulated in Gag shells. The presence of a scar defect (arrows) was observed in most particles. Scale bar is 50 nm. (b) Radial density distribution of a single Au-Gag VLP assembled on a 60 nm core. Stain density profile variations are consistent with the known sizes of different protein domains in Gag. (c) Density variation along the perimeter of a single Au-Gag VLP assembled on a 60 nm core [the density profile corresponds to the yellow circular arc of radius 48 nm depicted in (a)] (Reproduced from ref. 39 with permission from Elsevier)

Low-resolution 3D reconstructions can be obtained from negative-stained TEM images by classifying and averaging at least 2,000–3,000 particles in the EMAN software package [70]. However, in this case, one should be aware of potential artifacts due to staining and vacuum. 3D images were rendered using UCSF Chimera software [71]. Dynamic light scattering (DLS) measurements provide complementary information on particle hydrodynamic diameter distributions close to native conditions (Fig. 1e). Also, the formation of aggregates during assembly or storage can be easily monitored this way. Samples for DLS should be diluted and filtered with a 0.2 μm syringe filter.

12

4

Irina B. Tsvetkova and Bogdan G. Dragnea

Notes 1. Since gold colloid optical absorbance at a 400 nm wavelength depends only on the total concentration of gold atoms and is practically independent on the particle size (at least within the size range of 5–30 nm), it is convenient to determine the weight concentration of GNPs simply from measurements of optical absorbance. Accordingly, an extinction coefficient of 11.3 cm2/mg can be used for citrate-coated GNPs and 12.7 cm2/mg for TEG-functionalized GNPs (to account for an increase in scattering). For 60 nm GNP the extinction coefficient is provided by the manufacturer. 2. It was previously shown that bound thiols occupy 0.152 nm2 per group on the surface of a spherical GNP [72]. We use this value for our calculations for HS-PEG-DNA to gold ratio. A 3 excess of ligand for particle functionalization yielded a ratio of 5,000 ligand molecules/GNP for 19 nm and 50,000 ligand molecules/GNP for 60 nm. 3. Conditions of templated assembly should be optimized for any specific virus capsid. In some cases, virus assembly phase diagrams are available in the literature and can be easily adapted for cargo encapsulation [43, 44]. When assembly conditions are not well established, optimization studies should be carried out including the effects of ionic strength and pH variations, temperature, change of template size, surface functionalization, and ratio between assembly components [39].

Acknowledgments This work was supported by the US Department of Energy, Office of Science, Basic Energy Sciences, under Award # DE-SC0010507. References 1. Douglas T, Young M (2006) Viruses: making friends with old foes. Science 312:873–875 2. Steinmetz NF, Manchester M (2011) Viral nanoparticles: tools for materials science and biomedicine. Pan Stanford Publishing, Singapore 3. Plummer EM, Manchester M (2011) Viral nanoparticles and virus-like particles: platforms for contemporary vaccine design. WIREs Nanomed Nanobiotechnol 3:174–196 4. Lee LA, Wang Q (2006) Adaptations of nanoscale viruses and other protein cages for medical applications. Nanomedicine 2:137–149

5. Fiedler JD, Brown SD, Lau JL et al (2010) RNA-directed packaging of enzymes within virus-like particles. Angew Chem Int Ed 49:9648–9651 6. Comellas-Aragones M, Engelkamp H, Claessen VI et al (2007) A virus-based singleenzyme nanoreactor. Nat Nanotechnol 2:635–639 7. Patterson DP, Schwarz B, El-Boubbou K et al (2012) Virus-like particle nanoreactors: programmed encapsulation of the thermostable CelB glycosidase inside the P22 capsid. Soft Matter 8:10158–10166

Virus-Like Nanoparticles 8. Stephanopoulos N, Carrico ZM, Francis MB (2009) Nanoscale integration of sensitizing chromophores and porphyrins with bacteriophage MS2. Angew Chem Int Ed 48:9498–9502 9. Nam YS, Magyar AP, Lee D et al (2010) Biologically templated photocatalytic nanostructures for sustained light-driven water oxidation. Nat Nanotechnol 5:340–344 10. Nam KT, Kim D-W, Yoo PJ et al (2006) Virusenabled synthesis and assembly of nanowires for lithium ion battery electrodes. Science 312:885–888 11. Niu Z, Liu J, Lee LA et al (2007) Biological templated synthesis of water-soluble conductive polymeric nanowires. Nano Lett 7:3729–3733 12. DuFort CC, Dragnea B (2010) Bio-enabled synthesis of metamaterials. Annu Rev Phys Chem 61:323–344 13. Kostiainen MA, Hiekkataipale P, Laiho A et al (2013) Electrostatic assembly of binary nanoparticle superlattices using protein cages. Nat Nanotechnol 8:52 14. Carette N, Engelkamp H, Akpa E et al (2007) A virus-based biocatalyst. Nat Nanotechnol 2:226–229 15. Douglas T, Young M (1998) Host-guest encapsulation of materials by assembled virus protein cages. Nature 393:152–155 16. Jung B, Rao ALN, Anvari B (2011) Optical nano-constructs composed of genomedepleted brome mosaic virus doped with a near infrared chromophore for potential biomedical applications. ACS Nano 5:1243–1252 17. Tsvetkova I, Chen C, Rana S et al (2012) Pathway switching in templated virus-like particle assembly. Soft Matter 8:4571–4577 18. Cadena-Nava RD, Hu YF, Garmann RF et al (2011) Exploiting fluorescent polymers to probe the self-assembly of virus-like particles. J Phys Chem B 115:2386–2391 19. Hu Y, Zandi R, Anavitarte A et al (2008) Packaging of a polymer by a viral capsid: the interplay between polymer length and capsid size. Biophys J 94:1428–1436 20. Porterfield JZ, Dhason MS, Loeb DD et al (2010) Full-length hepatitis B virus core protein packages viral and heterologous RNA with similarly high levels of cooperativity. J Virol 84:7174–7184 21. Daniel M-C, Tsvetkova IB, Quinkert ZT et al (2010) Role of surface charge density in nanoparticle-templated assembly of bromovirus protein cages. ACS Nano 4:3853–3860 22. Mout R, Moyano DF, Rana S et al (2012) Surface functionalization of nanoparticles for nanomedicine. Chem Soc Rev 41:2539–2544

13

23. Thomas M, Klibanov AM (2003) Conjugation to gold nanoparticles enhances polyethylenimine’s transfer of plasmid DNA into mammalian cells. Proc Natl Acad Sci 100:9138–9143 24. Ghosh P, Han G, De M et al (2008) Gold nanoparticles in delivery applications. Adv Drug Deliv Rev 60:1307–1315 25. Cognet L, Tardin C, Boyer D et al (2003) Single metallic nanoparticle imaging for protein detection in cells. Proc Natl Acad Sci 100:11350–11355 26. Berciaud S, Cognet L, Blab GA et al (2004) Photothermal heterodyne imaging of individual nonfluorescent nanoclusters and nanocrystals. Phys Rev Lett 93:257402 27. Haiss W, Thanh NTK, Aveyard J et al (2007) Determination of size and concentration of gold nanoparticles from UV Vis spectra. Anal Chem 79:4215–4221 28. Boyer D, Tamarat P, Maali A et al (2002) Photothermal imaging of nanometer-sized metal particles among scatterers. Science 297:1160–1163 29. Capehart SL, Coyle MP, Glasgow JE et al (2013) Controlled integration of gold nanoparticles and organic fluorophores using synthetically modified MS2 viral capsids. J Am Chem Soc 135:3011–3016 30. Blum AS, Soto CM, Wilson CD et al (2004) Cowpea mosaic virus as a scaffold for 3-D patterning of gold nanoparticles. Nano Lett 4:867–870 31. Slocik JM, Naik RR, Stone MO et al (2005) Viral templates for gold nanoparticle synthesis. J Mater Chem 15:749–753 32. Radloff C, Vaia RA, Brunton J et al (2005) Metal nanoshell assembly on a virus bioscaffold. Nano Lett 5:1187–1191 33. Chen C, Daniel MC, Quinkert ZT et al (2006) Nanoparticle-templated assembly of viral protein cages. Nano Lett 6:611–615 34. Loo L, Guenther RH, Basnayake VR et al (2006) Controlled encapsidation of gold nanoparticles by a viral protein shell. J Am Chem Soc 128:4502–4503 35. Wang TJ, Zhang ZP, Gao D et al (2011) Encapsulation of gold nanoparticles by simian virus 40 capsids. Nanoscale 3:4275–4282 36. Aniagyei SE, Kennedy CJ, Stein B et al (2009) Synergistic effects of mutations and nanoparticle templating in the self-assembly of cowpea chlorotic mottle virus capsids. Nano Lett 9:393–398 37. Bancroft JB, Hiebert E, Bracker CE (1969) The effects of various polyanions on shell formation of some spherical viruses. Virology 39:924–930 38. Goicochea NL, De M, Rotello VM et al (2007) Core-like particles of an enveloped animal virus

14

Irina B. Tsvetkova and Bogdan G. Dragnea

can self-assemble efficiently on artificial templates. Nano Lett 7:2281–2290 39. Goicochea NL, Datta SAK, Ayaluru M et al (2011) Structure and stoichiometry of template-directed recombinant HIV-1 Gag particles. J Mol Biol 410:667–680 40. Hiebert E, Bancroft JB, Bracker CE (1968) The assembly in vitro of some small spherical viruses, hybrid viruses, and other nucleoproteins. Virology 34:492–508 41. Caspar D, Klug A (1962) Physical principles in the construction of regular viruses. Cold Spring Harb Symp Quant Biol 27:1–24 42. Lucas R, Larson S, McPherson A (2002) The crystallographic structure of brome mosaic virus. J Mol Biol 317:95–108 43. Cuillel M, Berthetcolominas C, Timmins PA et al (1987) Reassembly of Brome Mosaicvirus from dissociated virus - a neutronscattering study. Eur Biophys J 15:169–176 44. Lavelle L, Gingery M, Phillips M et al (2009) Phase diagram of self-assembled viral capsid protein polymorphs. J Phys Chem B 113:3813–3819 45. Bahadur RP, Rodier F, Janin J (2007) A dissection of the protein–protein interfaces in icosahedral virus capsids. J Mol Biol 367:574–590 46. Rao ALN (2006) Genome packaging by spherical plant RNA viruses. Annu Rev Phytopathol 44:61–87 47. Choi YG, Rao ALN (2003) Packaging of brome mosaic virus RNA3 is mediated through a bipartite signal. J Virol 77:9750–9757 48. Ni P, Wang Z, Ma X et al (2012) An examination of the electrostatic interactions between the N-terminal tail of the brome mosaic virus coat protein and encapsidated RNAs. J Mol Biol 419:284–300 49. Sun J, DuFort C, Daniel M-C et al (2007) Core-controlled polymorphism in virus-like particles. Proc Natl Acad Sci 104:1354–1359 50. Chen C, Kwak ES, Stein B et al (2005) Packaging of gold particles in viral capsids. J Nanosci Nanotechnol 5:2029–2033 51. Strauss JH, Strauss EG (1994) The alphaviruses: gene expression, replication, and evolution. Microbiol Rev 58:491–562 52. Cheng RH, Kuhn RJ, Olson NH et al (1995) Nucleocapsid and glycoprotein organization in an enveloped virus. Cell 80:621–630 53. Lopez S, Yao JS, Kuhn RJ et al (1994) Nucleocapsid-glycoprotein interactions required for assembly of alphaviruses. J Virol 68:1316–1323

54. Goicochea NL (2010) Nanoparticle-directed assembly of enveloped virus components and applications. Ph.D. Thesis, Indiana University 55. Ganser-Pornillos BK, Yeager M, Sundquist WI (2008) The structural biology of HIV assembly. Curr Opin Struct Biol 18:203–217 56. Briggs JAG, Riches JD, Glass B et al (2009) Structure and assembly of immature HIV. Proc Natl Acad Sci 106:11090–11095 57. Wright ER, Schooler JB, Ding HJ et al (2007) Electron cryotomography of immature HIV-1 virions reveals the structure of the CA and SP1 Gag shells. EMBO J 26:2218–2226 58. Briggs JAG, Kr€ausslich H-G (2011) The molecular architecture of HIV. J Mol Biol 410:491–500 59. Datta SK, Rein A (2009) Preparation of recombinant HIV-1 Gag protein and assembly of virus-like particles in vitro. In: Prasad V, Kalpana G (eds) HIV protocols, vol 485. Humana Press, New York, pp 197–208 60. Wilk T, Gross I, Gowen BE et al (2001) Organization of immature human immunodeficiency virus type 1. J Virol 75:759–771 61. Slot JW, Geuze HJ (1985) A new method of preparing gold probes for multiple-labeling cytochemistry. Eur J Cell Biol 38:87–93 62. Frens G (1973) Controlled nucleation for the regulation of the particle size in monodisperse gold suspensions. Nature Phys Sci 241:20–22 63. Hurst SJ, Lytton-Jean AKR, Mirkin CA (2006) Maximizing DNA loading on a range of gold nanoparticle sizes. Anal Chem 78:8313–8318 64. Gopinath K, Kao CC (2007) Replicationindependent long-distance trafficking by viral RNAs in Nicotiana benthamiana. Plant Cell 19:1179–1191 65. Cuillel M, Zulauf M, Jacrot B (1983) Selfassembly of brome mosaic virus protein into capsids: initial and final states of aggregation. J Mol Biol 164:589–603 66. Yamazaki H, Kaesberg P (1963) Degradation of bromegrass mosaic virus with calcium chloride and isolation of its protein and nucleic acid. J Mol Biol 7:760–762 67. Mukhopadhyay S, Chipman PR, Hong EM et al (2002) In vitro-assembled alphavirus core-like particles maintain a structure similar to that of nucleocapsid cores in mature virus. J Virol 76:11128–11132 68. Garmann RF, Comas-Garcia M, Gopal A et al (2014) The assembly pathway of an icosahedral single-stranded RNA virus depends on the

Virus-Like Nanoparticles strength of inter-subunit attractions. J Mol Biol 426:1050–1060 69. Datta SAK, Curtis JE, Ratcliff W et al (2007) Conformation of the HIV-1 Gag protein in solution. J Mol Biol 365:812–824 70. Ludtke S, Baldwin P, Chiu W (1999) EMAN: semiautomated software for high-resolution single-particle reconstructions. J Struct Biol 128:82–97

15

71. Pettersen EF, Goddard TD, Huang CC et al (2004) UCSF Chimera–a visualization system for exploratory research and analysis. J Comput Chem 25:1605–1612 72. Chen S, Kimura K (1999) Synthesis and characterization of carboxylate-modified gold nanoparticle powders dispersible in water. Langmuir 15:1075–1082

Chapter 2 Use of Protein Cages as a Template for Confined Synthesis of Inorganic and Organic Nanoparticles Masaki Uchida, Shefah Qazi, Ethan Edwards, and Trevor Douglas Abstract Protein cages are hollow spherical proteins assembled from a defined number of subunits. Because they are extremely homogeneous in size and structure, their interior cavities can serve as ideal templates to encapsulate and synthesize well-defined nanoparticles. Here, we describe the exemplary synthesis of a hard and a soft material in two representative protein cages, i.e., magnetite nanoparticles in ferritin and a poly(2-aminoethyl)methacrylate inside a viral capsid derived from the bacteriophage P22. Key words Protein cages, Ferritin, Viral capsid, Biomineralization, Atom transfer radical polymerization (ATRP)

1

Introduction Biology has provided inspiration for material scientists for years. One of these examples is the synthesis of nano-sized particles [1, 2], nucleated and constrained within a protein cage-like architecture. Nanoparticles have made a significant impact in material science for the past few decades, because they exhibit unique properties (e.g., magnetic, photonic, and catalytic) different from bulk materials. One of the critical challenges in the preparation of nanoparticles is the control of their size and morphology and, because the emergent properties of nanoparticles are intimately related to their dimension, the homogeneity of the size distribution is critical. Through millions of years of evolution, nature has developed nano-size cage-like protein assemblies that are designed to encapsulate guest molecules inside and protect them from the external environment. Ferritins that sequester iron as iron oxide particles [3] and viral capsids that encapsulate their nucleic acid genome [4] are examples of protein cages in this category. Material scientists have been inspired from the natural function of protein cages and exploited them as platforms for the development of a wide range of nanomaterials [4–7]. Cage-like protein structures are ideal

Brendan P. Orner (ed.), Protein Cages: Methods and Protocols, Methods in Molecular Biology, vol. 1252, DOI 10.1007/978-1-4939-2131-7_2, © Springer Science+Business Media New York 2015

17

18

Masaki Uchida et al.

Fig. 1 (a) Schematic illustration of magnetite (or maghemite) nanoparticle synthesis in HFn. (b) SEC profiles of HFn before and after mineralization reaction with 3,000Fe/cage. Co-elution profile of HFn (280 nm) and iron oxide (410 nm) after the mineralization reaction indicates the composite nature of the mineralized HFn

templates to synthesize nanoparticles because they have homogeneous size and their interior cavity can provide a confined environment for the selectively directed synthesis or for entrapment of guest molecules. In this chapter, we describe two examples of nanoparticle synthesis in protein cages, the synthesis of magnetite (or maghemite) nanoparticles (hard inorganic materials) in ferritin (Fig. 1a) and the synthesis of a 2-aminoethyl methacrylate (AEMA) polymer, via atom transfer radical polymerization (ATRP), in a viral capsid derived from the bacteriophage P22 (Fig. 2).

2

Materials All solutions should be prepared using ultrapure water (resistivity of 18.2 MΩ cm at 25  C) unless otherwise noted.

2.1 Synthesis of Magnetite Nanoparticle in Human Ferritin

1. The apo form of human heavy-chain ferritin (HFn) is heterologously expressed in, and purified from, E. coli (BL21 (DE3)). The cloning, expression, and purification procedure has been described in previous papers [8, 9]. The purified HFn can be

Use of Protein Cages as a Template for Confined Synthesis. . .

P22S39C

P22-int Initiator Attachment

19

P22-xAEMA ATRP

xAEMA

Bis-Acrylamide =

AEMA =

Fig. 2 Reaction scheme for the synthesis of P22 protein-polymer hybrid. P22 is first modified with a cysteine reactive ATRP initiator which is used as a macroinitiator to start the growth of an ATRP polymer on the inside of the virus-like particle (VLP)

stored at 80  C in a standard buffer such as 50 mM Tris, 100 mM NaCl, pH 7.5, until further use. 2. 10 reaction solution: 1 M NaCl. Weigh 58.44 g NaCl and add to 900 mL water. Dissolve the salt and add water up to a final volume of 1.0 L. Filter through 0.22 μm membrane filter. Store at room temperature. 3. Degassed water and 100 mM NaCl: Degas about 30 mL of ultrapure water and 10 mL of 100 mM NaCl (for use in Subheading 3.1, step 3) by bubbling nitrogen (or argon) gas for at least 30 min. Degas the solutions immediately before magnetite synthesis and keep them under a positive pressure of the inert gas. 4. Ammonium iron (II) sulfate hexahydrate (12.5 mM) solution in degassed water: Weigh 49.0 mg ammonium iron (II) sulfate hexahydrate (NH4)2Fe(SO4)2
6H2O and prepare a 10 mL solution using degassed water. Prepare the solution just prior to magnetite synthesis and maintain under a positive pressure of N2. 5. Hydrogen peroxide (4.17 mM) solution in degassed water: Dilute 4.7 μL of 30 % hydrogen peroxide solution (8.82 M) to 10 mL using degassed water. Prepare the solution just prior to magnetite synthesis and maintain under a positive pressure of N2. 6. NaOH (100 mM) solution. Weigh 4.00 g NaOH and add to about 900 mL water. Once the NaOH is well dissolved, add

20

Masaki Uchida et al.

water to a final volume of 1.0 L. Filter through 0.22 μm membrane filter. Store at room temperature. Degas the solution by gently bubbling nitrogen gas through the solution just before magnetite synthesis and maintain under a positive pressure of N2. 7. Sodium citrate (300 mM, pH 7.0) solution: Weigh 8.82 g of sodium citrate dihydrate and add to about 90 mL of water. Adjust the pH with 2 M HCl and make up to 100 mL with water. Filter through 0.22 μm membrane filter. Store at room temperature. 8. Dulbecco’s phosphate buffered saline (DPBS): Weigh 19.2 g of premixed chemical reagents of DPBS purchased from SigmaAldrich and dissolve into 1,900 mL of water. Adjust the pH to 7.4 and make up to a final volume of 2.0 L with water. Filter through 0.22 μm membrane filter. Store at room temperature. 2.2 Synthesis of Cross-Linked AEMA (x-AEMA) in P22 Viral Capsid

1. Phosphate buffered saline (PBS, 100 mM sodium phosphate, 50 mM sodium chloride): Three different PBS buffers, each having a different pH (i.e., 7.0, 7.6, and 8.0), are used during the synthesis. Weigh 31.2 g sodium phosphate monobasic dihydrate and 5.84 g sodium chloride and add to 1,900 mL water. Dissolve well and adjust to the target pH with 2 M NaOH. Make up to 2.0 L with water. Filter through 0.22 μm membrane filter. Store at room temperature. 2. 0.5 M guanidine-HCl in PBS, pH 7.0: Weigh 47.8 g of guanidine-HCl and add to about 900 mL of PBS, pH 7.0. Dissolve well and make up to 1.0 L with water. 3. ATRP polymer is initiated from a genetically introduced cysteine residue mutated from serine at position 39, which is on the interior of the capsid of wild-type P22 coat protein as described [10]. According to the available P22 structural models, this site is expected to be exposed to the interior cavity [11] and is therefore suitable for attachment of the ATRP initiator (see Note 1) [10, 12]. The procapsid form (PC) of P22S39C is heterologously expressed in and purified from E. coli (BL21 (DE3)). The cloning, expression, and purification procedure has been previously described [10, 13]. The PC of P22S39C is transformed to the expanded form (EX) as previously described [10] (see Note 2). 4. 2-Bromoisobutyryl aminoethyl maleimide: The ATRP initiator, 2-bromoisobutyryl aminoethyl maleimide, is synthesized by a modification of the procedures reported previously [14, 15]. Mix N-2-aminoethylmaleimide (250 mg, 0.98 mmol) with dry triethylamine (300 μl, 2.2 mmol) in 5 ml dry dichloromethane on ice. Add 2-bromo-2-methylpropionylbromide (200 μl, 1.6 mmol) to the reaction dropwise. Allow the

Use of Protein Cages as a Template for Confined Synthesis. . .

21

reaction to warm to room temperature and then extract three times with dichloromethane and water, followed by drying over anhydrous sodium sulfate. Purify the crude product using column chromatography (silica gel, 10 % ethyl acetate in dichloromethane). 5. Metal catalyst solution for ATRP: Degas ultrapure water in advance by gently bubbling argon gas through the solution for 30 min. Weigh 19.2 mg CuBr (0.134 mmol), 29.9 mg CuBr2 (0.134 mmol), and 83.5 mg 2,20 -bipyridine (0.535 mmol) into a 10 mL crimp-top vial, seal the vial, and degas with argon gas flow. After 30 min, add 10 mL of degassed water to the vial and sonicate for 5 min. Repeat as necessary until solution turns dark brown to black in color (see Note 3). The solution is prepared just prior to the polymerization experiment with P22. 6. High-salt buffer solution (250 mM sodium sulfate, 100 mM sodium phosphate, pH 8.0): Weigh 35.5 g sodium sulfate and 15.6 g sodium phosphate monobasic dihydrate and add to about 900 mL water. Dissolve well and adjust to the target pH with 2 M NaOH. Adjust the volume to 1.0 L with water. Filter through a 0.22 μm membrane filter. Store at room temperature.

3

Methods

3.1 Synthesis of Magnetite Nanoparticle in Human Ferritin

1. Make 2  1.0 L solutions of 0.1 M NaCl solution. Dialyze HFn into 1.0 L of 100 mM NaCl for buffer exchange. After 6 h, replace the dialysis solution with fresh 0.1 M NaCl solution and dialyze for an additional 12 h. Filter the HFn through a 0.1 μm syringe filter to remove dust and adjust the concentration to 2 mg/mL with 100 mM NaCl. 2. Insert a pH electrode connected to a pH auto titrator (Brinkmann 718 STAT Titrino) into the mouth of a jacketed (and temperature-controlled) reaction vessel pre-equilibrated at 65  C. Use a PTFE/silicone septum to seal the mouth of the vessel. 3. Add 7.5 mL of degassed 100 mM NaCl and 2.5 mL of the HFn solution prepared above to the vessel under an Ar atmosphere (i.e., final HFn concentration is 0.5 mg/mL) (see Notes 4 and 5). Stir the solution gently with a magnetic stirrer. 4. Load two disposable syringes with 12.5 mM ammonium iron (II) sulfate hexahydrate solution and 4.17 mM hydrogen peroxide solution, respectively. The volumes of the solutions needed depend on the size of the magnetite particle (i.e.,

22

Masaki Uchida et al.

iron-loading factor per HFn cage) (see Note 6). Insert the syringes, with injection tubes, into a syringe pump and insert the tubes into the reaction vessel. 5. Connect the inlet and outlet hoses of the jacketed reaction vessel to a water bath set at 65  C so that temperature of the vessel is brought up to and maintained at 65  C during the mineralization reaction by circulating water through the jacketed flask. 6. Raise the pH of the solution slowly to 8.5 with 100 mM NaOH using the auto titrator. When the pH reaches 8.5, start the simultaneous injection of ammonium iron (II) sulfate hexahydrate and hydrogen peroxide solutions at a constant rate of 100 Fe/(cage · min) (i.e., 79 μL/min) using the syringe pump. H+ generated during the reaction is titrated using 100 mM NaOH to maintain the pH at 8.5. The reaction is considered to be completed 5 min after the addition of ammonium iron (II) sulfate hexahydrate and hydrogen peroxide solutions necessary for the targeted iron-loading factor (see Note 7). 7. Add 200 μL of 300 mM sodium citrate solution to the completed reaction in order to chelate any free iron in the solution. Dialyze the mineralized HFn solution against 1.0 L of DPBS. 8. Purify the mineralized HFn away from any aggregated products by fast protein liquid chromatography (FPLC) equipped with a Superose 6 size-exclusion column (SEC) while detecting absorbance at 280 and 410 nm to monitor elution of protein and mineral, respectively. Typical SEC profiles of HFn before and after mineralization reaction are shown in Fig. 1b. 3.2 Synthesis of x-AEMA in P22 Viral Capsid

1. Dissolve 2-bromoisobutyryl aminoethyl maleimide (4.63 mg, 0.016 mmol) in DMSO (400 μL) to a concentration of 40 mM. 2. Add 322 μL of the initiator solution (prepared above) to 24 mL of P22S39C EX (5 mg/mL) in PBS, pH 7.6 (this corresponds to a fivefold excess per P22 subunit). Gently shake the mixture to react for three hours at room temperature. After 3 h, quench the reaction with DTT (322 μL, 40 mM in water). Pellet the P22S39C by ultracentrifugation at approximately 209,000  g (45,000 rpm) (F50L-8x39 rotor, Piramoon Technologies) for 50 min, to remove any excess of the soluble initiator and DTT, and then resuspend the pellet in 1.0 mL PBS, pH 7.6. 3. Dissolve 160 mg of 2-aminoethyl methacrylate (AEMA) and 40 mg of bis-acrylamide in 4 mL of PBS pH 8.0. Adjust back to pH 8.0 with 2 M NaOH and adjust the volume to 5.0 mL. Spin down the mixture for 1 min at 4,000  g to pellet and remove any aggregate.

Use of Protein Cages as a Template for Confined Synthesis. . .

23

4. Transfer 4 mL of the AEMA/Bis-acrylamide solution to a 10 mL crimp-top vial and add 2.2 mL of PBS, pH 8.0. Add 1.2 mL of 8.0 mg/mL P22S39C with the initiator molecule prepared in step 2 to this mixture. Seal the crimp-top vial, followed by a cycle of pumping with a vacuum and back filling with argon gas four times to deaerate the mixture (see Note 8). 5. Add the metal catalyst solution (0.6 mL) to the monomerP22S39C mixture anaerobically (under N2 or Ar) and cover the entire vial in foil, to avoid any photochemical reactions, and maintain the temperature at 23  C for the duration of the polymerization reaction (180 min). 6. Quench the reaction by exposure to air and spin for 5 min at 17,000  g to remove any aggregates formed during polymerization. Purify the P22S39C-polymer composite away from any unreacted monomer and the metal catalyst by pelleting the protein as described previously by ultracentrifugation and resuspend in 1 mL of PBS, pH 8.0. Add 2 mL of the highsalt buffer (250 mM sodium sulfate, 100 mM sodium phosphate, pH 8.0) and agitate gently for 3 h at 4  C to remove any AEMA and bis-acrylamide monomer which might be noncovalently associated with the P22. Pellet the protein again by ultracentrifugation and resuspend in 1.0 mL PBS, pH 8.0. Analyze the obtained material as previously described [10, 12].

4

Notes 1. The location of the initiator on a protein cage is critical to achieve confined polymerization inside of the cage cavity. 2. The scaffolding protein of P22S39C is removed using 0.5 M guanidine-HCl. Ultracentrifuge the P22S39C at approximately 209,000  g for 50 min to pellet the capsid. Resuspend the pellet in 0.5 M guanidine-HCl by rocking gently for 1–2 h at 4  C. Ultracentrifuge the protein again at approximately 209,000  g for 50 min [16]. Repeat the cycle of pelleting and resuspension four times. Dialyze the empty-shell (ES) form of P22 against PBS, pH 7.0, overnight. Heat-treat the ES form of P22 for 20 min at 65  C to transform the protein cage into expanded shell (EX) form. Purify the heat treated sample by ultracentrifugation pelleting as above, followed by resuspension into PBS, pH 7.6. Analyze the transformation of the P22 from PC form to EX form as previously described [10, 13, 17]. 3. The catalyst solution should be dark brown to black in color. Slight blue color suggests insufficient deaeration and could cause poor polymerization.

24

Masaki Uchida et al.

4. The typical HFn concentration we have described is 0.5 mg/mL, but protein concentrations in the range of 0.1–1 mg/mL work equally well for the synthesis. 5. A critical key to the successful synthesis of magnetite (or maghemite) is keeping the reaction vessel atmosphere anaerobic. Continuous gentle flow of Ar (or N2) to the reaction vessel through narrow-bore tubing (exterior diameter 1/16 in., interior diameter 0.02 in.) helps to maintain anaerobic conditions. 6. The iron loading in HFn can be successfully achieved up to 5,000 Fe per HFn cage. When 5 mg of HFn is used for the synthesis of 5,000 Fe per HFn cage, the volume of ammonium iron (II) sulfate hexahydrate solution (12.5 mM) is 3,951 μL and the volume of hydrogen peroxide solution (4.17 mM) is 3,951 μL as well. For lower loadings, the required volumes are scaled accordingly. 7. The color of the mineralized HFn solution should be dark brown if magnetite (or maghemite) has been successfully synthesized in HFn. A bright orange/brown color suggests contamination of the product with ferrihydrite (or other iron oxides) formation. Definitive analysis can be achieved with electron diffraction equipped with transmission electron microscope or X-ray diffraction. 8. Tap the vial during the vacuum pumping cycle to remove air bubbles.

Acknowledgment This work was supported with grants from the US Department of Energy, Office of Basic Energy Sciences, Division of Materials Science and Engineering DE-FG02-07ER46477 (for the inorganic nanoparticle work) and the National Institutes of Health NIAID R01AI104905 (for the polymer work). References 1. Douglas T, Young M (1998) Host-guest encapsulation of materials by assembled virus protein cages. Nature 393:152–155 2. Meldrum FC, Wade VJ, Nimmo DL, Heywood BR, Mann S (1991) Synthesis of inorganic nanophase materials in supramolecular protein cages. Nature 349:684–687 3. Harrison PM, Arosio P (1996) Ferritins: molecular properties, iron storage function and cellular regulation. Biochim Biophys Acta 1275:161–203

4. Douglas T, Young M (2006) Viruses: making friends with old foes. Science 312:873–875 5. Manchester M, Steinmetz NF (2009) Viruses and nanotechnology, vol 327, Current topics in microbiology and immunology. Springer, Berlin 6. Uchida M, Klem MT, Allen M, Flenniken ML, Gillitzer E, Varpness Z, Suci P, Young MJ, Douglas T (2007) Protein cage architecture: containers as templates for materials synthesis. Adv Mater 19:1025–1042

Use of Protein Cages as a Template for Confined Synthesis. . . 7. Witus LS, Francis MB (2011) Using synthetically modified proteins to make new materials. Acc Chem Res 44:774–783 8. Uchida M, Flenniken ML, Allen M, Willits DA, Crowley BE, Brumfield S, Willis AF, Jackiw L, Jutila M, Young MJ, Douglas T (2006) Targeting of cancer cells with ferrimagnetic ferritin cage nanoparticles. J Am Chem Soc 128:16626–16633 9. Uchida M, Terashima M, Cunningham CH, Suzuki Y, Willits DA, Willis AF, Yang PC, Tsao PS, McConnell MV, Young MJ, Douglas T (2008) A human ferritin iron oxide nanocomposite magnetic resonance contrast agent. Magn Reson Med 60:1073–1081 10. Lucon J, Qazi S, Uchida M, Bedwell GJ, LaFrance B, Prevelige PE Jr, Douglas T (2012) Use of the interior cavity of the P22 capsid for site-specific initiation of atomtransfer radical polymerization with highdensity cargo loading. Nat Chem 4:781–788 11. Chen D-H, Baker ML, Hryc CF, DiMaio F, Jakana J, Wu W, Dougherty M, HaasePettingell C, Schmid MF, Jiang W, Baker D, King JA, Chiu W (2011) Structural basis for scaffolding-mediated assembly and maturation of a dsDNA virus. Proc Natl Acad Sci U S A 108:1355–1360 12. Lucon J, Edwards E, Qazi S, Uchida M, Douglas T (2013) Atom transfer radical

25

polymerization on the interior of the P22 capsid and incorporation of photocatalytic monomer crosslinks. Eur Polym J 49:2976–2985 13. Kang S, Uchida M, O’Neil A, Li R, Prevelige PE, Douglas T (2010) Implementation of P22 viral capsids as nanoplatforms. Biomacromolecules 11:2804–2809 14. Mantovani G, Lecolley F, Tao L, Haddleton DM, Clerx J, Cornelissen JJ, Velonia K (2005) Design and synthesis of N-maleimidofunctionalized hydrophilic polymers via copper-mediated living radical polymerization: a suitable alternative to PEGylation chemistry. J Am Chem Soc 127:2966–2973 15. Heredia KL, Bontempo D, Ly T, Byers JT, Halstenberg S, Maynard HD (2005) In situ preparation of protein-“smart” polymer conjugates with retention of bioactivity. J Am Chem Soc 127:16955–16960 16. Prevelige PE, Thomas D, King J (1988) Scaffolding protein regulates the polymerization of P22 coat subunits into icosahedral shells in vitro. J Mol Biol 202:743–757 17. Teschke CM, McGough A, ThumanCommike PA (2003) Penton release from P22 heat-expanded capsids suggests importance of stabilizing penton-hexon interactions during capsid maturation. Biophys J 84:2585–2592

Chapter 3 Ferritin Encapsulation and Templated Synthesis of Inorganic Nanoparticles Katherine W. Pulsipher and Ivan J. Dmochowski Abstract Understanding how inorganic nanoparticles interact with proteins is paramount to their safe and effective use in vivo. Ordered protein-inorganic nanomaterial assemblies will also enable the creation of patterned structures with useful physical properties. Thermophilic ferritin (tF) from Archaeoglobus fulgidus has unique structural features and self-assembly properties that facilitate stable but also reversible interaction with gold nanoparticles (AuNPs). In this chapter we describe how to express and purify tF and induce its assembly around AuNPs. We also describe methods for characterizing the tF-AuNP complex as well as templating NP growth within the protein cavity. Key words Ferritin assembly, Protein self-assembly, Nanoparticle-protein interaction, Biomineralization, Gold nanoparticle, Ferritin, Nano-bio interface

1

Introduction Inorganic nanomaterials possess unique optical, electronic, and magnetic properties that are highly useful in bio- and medicinerelated applications such as imaging [1], photothermal therapy [2], and sensing [3]. However, there is little current understanding of how nanoparticles and proteins interact and affect one another. Our group has sought to further this basic understanding by characterizing the interactions of metal ions and nanoparticles with a variety of wild-type as well as computationally designed ferritin protein systems [4–8]. The native self-assembly processes of ferritins and viral capsids can be used to fill protein cages with nanomaterial. Researchers have taken advantage of the pH-mediated assembly/ disassembly of ferritin protein to encapsulate nanomaterials within the protein cavity [9–11]. This method relies on a high concentration of target compound and often suffers from low encapsulation yields and damaged protein structure due to the near-denaturing conditions required and repetitive passage through the pI of the protein. In contrast, viral capsid assembly around nanomaterials can

Brendan P. Orner (ed.), Protein Cages: Methods and Protocols, Methods in Molecular Biology, vol. 1252, DOI 10.1007/978-1-4939-2131-7_3, © Springer Science+Business Media New York 2015

27

28

Katherine W. Pulsipher and Ivan J. Dmochowski

be induced by the presence of the particles themselves. This has been shown with gold nanoparticles [12–18], quantum dots [19, 20], magnetic iron particles [21], and CoFe2O4 particles [20]. In similar fashion, our group has studied the assembly of a thermophilic ferritin protein around size-matched gold nanoparticles. Thermophilic ferritin (tF) from the archaeon Archaeoglobus fulgidus has a number of unique properties that aid its interaction with nanoparticles. Structurally, its 24 monomers are arranged in tetrahedral symmetry, forming a nearly spherical cage with four large (4.5 nm) triangular pores [22]. These pores allow the diffusion of reagents into and out of the protein cavity and yield relatively large, exposed surfaces for further nanoparticle functionalization. tF exists in low-salt conditions as a stable dimer of four-helix bundles and at higher ionic strength can assemble into its full 24mer cage, in contrast to the pH-mediated assembly of other ferritin proteins [23]. The tF self-assembly process also avoids the need for high cargo concentrations for serendipitous encapsulation upon pH reassembly. Instead, the presence of AuNP drives ferritin self-assembly resulting in NP encapsulation. Our group has sought to engineer a stable, native-like tF-AuNP assembly, where the wild-type stability and structure of the protein are preserved, and the NP surface is effectively passivated to prevent particle aggregation and improve biocompatibility. We found that a 48 h, room-temperature assembly process with gentle stirring leads to a stable tF-AuNP conjugate, where the AuNP is 5 nm in diameter and capped with bis-p-(sulfonatophenyl)phenyl phosphine (BSPP) ligand [8]. This protein-NP conjugate can be characterized by a number of techniques including native gel electrophoresis, transmission electron microscopy (TEM), and size-exclusion chromatography (SEC) as described in this chapter.

2

Materials Prepare all solutions using distilled water purified to at least 18 Ω. All solutions can be stored at room temp, unless otherwise noted. Storage of reagents should be done according to the manufacturer’s recommendation.

2.1 tF Transformation and Expression

1. pAF0834 plasmid containing tF gene (Dr. Eric Johnson, California Institute of Technology). Store at 20  C. 2. NZYM Plus Broth medium: 1.0 % casein digest (N-Z-Amine), 0.5 % NaCl, 0.5 % yeast extract, 0.4 % glucose, 12.5 mM MgSO4·7H2O, 12.5 mM MgCl2. 3. Agar plates: 37 g granulated Luria-Bertani (LB) agar, Miller medium (5 g yeast extract, 10 g peptone from casein, 10 g NaCl, 12 g agar) dissolved in 1 L H2O. Autoclave, let them

Ferritin Encapsulation and Templated Synthesis of Inorganic Nanoparticles

29

cool to ~60  C, add 50 mg ampicillin sodium salt, and pour into sterile plastic Petri dishes. After letting them solidify at room temp, store at 4  C. 4. BL21-CodonPlus competent cells. 5. LB-Miller medium: 5 g yeast extract, 10 g peptone from casein, 10 g NaCl dissolved in 1 L H2O. Autoclave to sterilize (see Note 1). 6. Terrific Broth (TB) medium: 12 g tryptone, 24 g yeast extract, 9.4 g K2HPO4, 2.2 g KH2PO4 dissolved in 1 L H2O. Autoclave to sterilize in either a 4 L non-baffled Erlenmeyer flask or a 2 L baffled flask. 7. Ampicillin sodium salt solution: 100 mg dissolved in 1 mL H2O. Store at 4  C. 8. Chloramphenicol solution: 35 mg dissolved in 1 mL H2O. Store at 20  C. 9. IPTG solution: 2.38 g dissolved in 10 mL H2O. Aliquot into 1 mL portions and store at 20  C. 2.2 tF Purification and Characterization

1. Protease inhibitor cocktail tablets. 2. Buffer A: 10 mM Tris–HCl, 10 mM NaCl, pH 8.4. Store at 4  C. 3. Sonicator. 4. Buffer B: 10 mM Tris–HCl, 2 M NaCl, pH 8.4. Store at 4  C. 5. Assembly buffer: 20 mM NaH2PO4, 2 mM EDTA, 2.5 M NaCl, pH 7.6. Store at 4  C. 6. High-salt buffer: 800 mM NaCl, 20 mM NaH2PO4, pH 7.6. Store at 4  C. 7. Low-salt buffer: 20 mM NaCl, 20 mM NaH2PO4, pH 7.6. Store at 4  C. 8. HiTrap QFF anion-exchange column, 5 mL. 9. Superdex 200 10/300 GL size-exclusion column. 10. 10 and 100 kDa centrifugal filter units. 11. 4–15 % Tris–HCl, Mini-PROTEAN TGX gel. 12. Bradford assay or Lowry assay kit. 13. Bovine gamma globulin (BGG)—2 mg/mL capsules.

2.3 tF Nanoparticle Assembly and Seeded Growth

1. Bis-p-(sulfonatophenyl)phenyl phosphine (BSPP). 2. 5 nm diameter citrate-capped AuNPs. 3. NaCl. 4. Methanol. 5. Low-salt buffer: 20 mM NaH2PO4, 20 mM NaCl, pH 7.6. Store at 4  C.

30

Katherine W. Pulsipher and Ivan J. Dmochowski

6. Gold(III) chloride: 100 mM solution in H2O. 7. Ascorbic acid: 100 mM solution in H2O (see Note 2). 8. 10DG desalting column. 2.4 tF-AuNP Characterization

1. Native gel buffer: 20 mM NaH2PO4, 5 mM NaCl, pH 7.6. Store at 4  C. 2. Agarose. 3. Glycerol sample loading buffer: 80 % v/v glycerol in H2O. 4. Coomassie dye staining solution: 0.5 g Coomassie dye, 500 mL ethanol, 75 mL glacial acetic acid, 425 mL H2O. 5. Gel destaining solution: 300 mL methanol, 100 mL glacial acetic acid, 600 mL H2O. 6. Carbon-coated copper TEM grid. 7. Uranyl acetate solution: 1 g dissolved in 100 mL H2O. 8. Superdex 200 10/300 GL column. 9. Low-salt buffer: 20 mM NaH2PO4, 20 mM NaCl, pH 7.6. Store at 4  C.

3

Methods

3.1 Day 1: tF Transformation

For all bacteria culture work, use a clean workspace: wipe down the benchtop with ethanol, keep everything near an open Bunsen burner flame, and sterilize all containers by opening near the flame. 1. Put plasmid, BL21-CodonPlus(DE3)-RIL or -RP cells, and 15 mL Falcon tubes (1 per plasmid transformation) in an ice bucket for 5–10 min to allow frozen plasmid and cell solutions to thaw. 2. Set a water bath to 42  C. 3. Near an open flame, put into each Falcon tube: 1 μL plasmid solution, 50 μL BL21-CodonPlus(DE3)-RIL or -RP cells. Let the solution sit for 30 min. 4. While waiting, sterilize a cell spreader by holding it in the flame for several seconds and then submerging it in 100 % ethanol. 5. After 30 min, add 450 μL NZYM Plus Broth to each Falcon tube, and put the tubes in the 42  C water bath for 45 s to heat shock the bacteria, inducing them to take up the plasmids. After heat shocking, put the tubes in an incubator set to 37  C 225 rpm for 1 h. 6. To plate the cells, pipet 200 μL of the incubated bacteria solution onto an agar plate containing ampicillin. Sterilize the spreader again by putting it in a flame. Once the spreader has cooled, use it to spread the bacteria solution into a thin layer on

Ferritin Encapsulation and Templated Synthesis of Inorganic Nanoparticles

31

the plate. Put the plate in a 37  C incubator for at least 12 h but no more than 24 h. After 12–24 h, the plates can be stored at 4  C for up to 2 weeks. 3.2 Day 2: Small Growth

1. Into two 50 mL Falcon tubes, place 10 mL of LB, 10 μL of ampicillin solution, 10 μL of chloramphenicol solution and, using a blue plastic sterile loop, a single colony from the plated bacteria. Choose a colony that is not overlapping or touching any others. 2. Put tubes into a shaker: 30  C, 220 rpm, 12–16 h. To avoid risk of overgrowth and resource exhaustion and subsequent cell death, do not let the bacteria grow more than 16 h.

3.3 Day 3: Large Growth

1. Centrifuge small growth Falcon tubes from incubator: 3,220  g, 4  C, 20 min. While centrifuging, add 100 mg ampicillin and 35 mg chloramphenicol to prepared TB solutions, near a Bunsen burner flame. 2. Discard the yellow supernatant in the Falcon tubes. Use a 10 mL serological pipet to add 10 mL TB/ampicillin/chloramphenicol solution to the tan bacteria pellet. Use the pipet to resuspend the pellet. Transfer the 10 mL TB/ampicillin/chloramphenicol + bacteria pellet back to 1 L TB solution. 3. Put the TB solutions in the shaker: 37  C, 225 rpm if using baffled 2 L flask, 180 rpm if using non-baffled 4 L flask, 3h. 4. After 3 h, check OD560 and OD600 of the cultures using a UVvisible (UV-Vis) spectrometer. Repeat every 30 min or so until OD560 and OD600 ~ 0.8. Induce protein growth by adding 1 mL IPTG solution to each flask. Incubate for 4 h. 5. After 4 h, centrifuge TB solutions: 2,430  g, 4  C, 15 min. Discard supernatant and store pellets at 20  C until ready to use. Frozen bacteria pellets can be stored for months at 20  C.

3.4 Protein Extraction

1. Take bacteria pellets out of freezer to thaw slightly. To a plastic beaker add bacteria pellets and two protease inhibitor tablets (crushed up to dissolve). Weigh the bacteria pellets and add 2 mL buffer A to the beaker for every 1 g of bacteria. Stir gently to suspend pellets in the solution. 2. Set a water bath to 80  C. 3. Sonicate the resuspended bacteria on ice, 6 for 2 min (see Note 3). 4. Transfer cell suspension to 50 mL Falcon tubes and centrifuge: 3,220  g, 4  C, 20 min. Pour the supernatant into a 125 or 250 mL Erlenmeyer flask. 5. Heat the flask for 10 min at 80  C. The solution will go from brown to pale yellow and somewhat thick.

32

Katherine W. Pulsipher and Ivan J. Dmochowski

6. After 10 min, transfer the solution from the flask to 15 mL Falcon tubes. Centrifuge: 12,400  g, 4  C, 90 min. Transfer the clear, yellow supernatant to a clean 50 mL Falcon tube. Filter the solution using a 60 mL syringe and a 0.22 μm syringe filter. Store the filtered protein solution at 4  C until ready for purification. It is recommended to proceed to purification within a day or 2. 3.5 tF Purification: Anion-Exchange Chromatography

1. Equilibrate a HiTrap QFF anion-exchange column (flow rate ¼ 5 mL/min) with buffer A. Load the filtered protein solution onto the column via superloop and elute by running a linear gradient: 40 % buffer B over 20 min. 2. Collect protein-containing fractions eluted with buffer B (usually 4–30 % B) and concentrate using a 10 kDa centrifugal filter unit, down to a total volume of ~7 mL (see Note 4). Double the volume of the concentrated protein solution by adding an equal volume of assembly buffer. 3. Store the protein at 4  C for 2 days to allow the protein to assemble into its full 24mer cage.

3.6 tF Purification: Size-Exclusion Chromatography

1. Concentrate the protein in assembly buffer for loading onto the Superdex 200 10/300 GL column using a 100 kDa centrifugal filter unit. Run the Superdex 200 10/300 GL purification method. For best resolution, we recommend an injection volume of 500 μL or less. The protein in its 24mer form elutes at approximately 9–10 mL, depending on the specific column. Calibrating the column using a molecular weight calibration is recommended to know exactly which fractions should be collected. The MW of tF 24mer is approximately 487,500 Da. 2. After collecting all tF-containing fractions, concentrate them to approximately 0.5 mL total volume. Total protein yield varies from batch to batch, but a 5 mg yield from 2 L of bacteria is fairly typical. 3. The protein concentration can be estimated using the absorbance at 280 nm and an extinction coefficient of 1.67 mL/ (mg  cm). Once an approximate concentration is known, a more exact value can be found using a Bradford or Lowry assay, using BGG dilutions as standards. The protein can also be characterized by mass spectrometry (see Note 5) and CD spectroscopy to verify its MW and alpha-helical character, respectively.

3.7 AuNP Ligand Exchange

For tF assembly, we use 5 nm gold nanoparticles capped with BSPP ligand, but commercially available gold nanoparticles often come capped with citrate. The following is the procedure to replace citrate with BSPP on the particle surface.

Ferritin Encapsulation and Templated Synthesis of Inorganic Nanoparticles

33

1. Transfer 50 mL of commercial citrate-AuNP solution and 20 mg BSPP into a round-bottom flask with a stir bar. Stir overnight at room temp. 2. To remove citrate and excess BSPP, add solid NaCl with stirring until the solution turns dark purple as the nanoparticles precipitate. Centrifuge the solution: 1,810  g at 4  C, 30 min. 3. Discard the yellow supernatant and wash the purple precipitate with 0.5 mL H2O (see Note 6). Add 0.5 mL methanol to reprecipitate the particles. Centrifuge again:1,810  g, 4  C, 30 min. 4. Pour off the supernatant and resuspend the particles in 250–500 μL H2O, depending on desired concentration. 5. To calculate particle concentration, use the absorbance at 450 nm and the extinction coefficient determined by Haiss et al. [24]: 7.20  106 M 1 cm 1. 3.8 tF-AuNP Assembly

1. To assemble the protein around nanoparticles, it must be in dimer form. Calculate the volume of protein and nanoparticles needed to make 1 mL of 0.6 μM solution of tF 24mer and 5 nm AuNPs. Mix the purified high-salt protein solution with the appropriate amount of low-salt buffer to bring the final concentration to 0.6 μM when AuNPs are later added. Let the protein sit in low salt for at least 48 h at 4  C to ensure protein disassembly into dimers (see Note 7). 2. After 48 h, add the appropriate volume of AuNPs. Pipet up and down to mix, and put on a gel rocker or other sources of gentle agitation for 48 h at room temp.

3.9 tF-AuNP Assembly Characterization: Native Gel

1. To make the gel, microwave 500 mg agarose with 35 mL native gel buffer in a 250 mL Erlenmeyer flask for 1 min, stopping every 15 s or so to prevent the solution from boiling over. After 1 min, add another 35 mL of native gel buffer, and repeat microwaving. Allow the solution to cool to ~60  C, pour it into a gel mold and let it sit at room temp for approximately 1 h to set. 2. While the gel is setting, prepare the gel samples. Mix 20 μL of sample with 5 μL of 80 % glycerol in H2O solution. 3. Run the gel at 100 V for 15–18 min. Stain with Coomassie dye solution for 1 h and destain as long as necessary to see distinct blue protein bands (3 h to overnight). If the protein has successfully assembled around the particles, the blue protein band and red AuNP bands will overlap in the gel.

3.10 tF-AuNP Assembly Characterization: TEM

1. Dip a carbon-coated Cu grid in a drop of methanol on a piece of parafilm and let it dry to increase the hydrophilicity of the surface.

34

Katherine W. Pulsipher and Ivan J. Dmochowski

2. Pipet 5 μL of sample onto the grid and let it sit for 10 min. After 10 min, blot excess solution from the grid using the ripped edge of a filter paper. 3. Wash away excess buffer salts by floating the grid on top of a water droplet on a piece of parafilm and blotting the excess solution away with a filter paper, taking care to avoid drying the grid. Repeat for a total of three water droplets. 4. Stain the grid by pipetting 3 μL of 1 % uranyl acetate solution onto the grid for 30 s. Blot excess using filter paper, and let the grid dry for several minutes. 5. Image the grid and perform particle size analysis using ImageJ (version 1.41 h, National Institutes of Health) (see Note 8). 3.11 tF-AuNP Assembly Characterization: SEC

The tF-AuNP complex can be characterized by SEC to verify that the protein and NP coelute. This technique can also be used to calculate an equilibrium constant for the reaction (see Note 9): AuNP

12  tF 2mer

Ð 1  tF

AuNP

The equation used to determine the association constant is KA ¼

½tF24

AuNPŠ

½tF2 Š12

:

1. Inject 100 μL of tF-AuNP solution onto the column and run the Superdex 200 10/300 GL purification method. 2. To determine the concentrations of tF protein in dimer form and 24mer form, integrate the areas of the relevant 280 nm absorbance peaks for the FPLC traces (9–10 mL for 24mer, 14 mL for dimer, specific volume varies from column to column). 3. To account for absorbance of the AuNP at 280 nm, subtract 50 % from the integrated area of the 24mer peak. The dimer and 24mer concentrations can then be normalized: integrated area for tF2 peak total area corrected area for tF24 AuNP AuNP ¼ original ½tFŠ  total area where total area ¼ area of tF2 þ corrected area of tF24

concentration of tF2 ¼ original ½tFŠ  concentration of tF24

Using this method we have calculated an apparent association constant, KA, for tF-AuNP association to be (2.1  0.4)  1078 M 11 and KA for the native protein salt-mediated assembly to be (2.2  0.5)  1068 M 11, at 25  C.

Ferritin Encapsulation and Templated Synthesis of Inorganic Nanoparticles

3.12 AuNP Seeded Growth

35

Because the inner diameter of the protein cavity is 8 nm and the encapsulated AuNP has a diameter of 5 nm, it is possible to add gold ions and reducing agent to “grow” the particle to fill the cavity. 1. To 0.5 mL of 0.6 μM tF-AuNP, add 30 μL of 100 mM AuCl3 (see Note 10) and incubate at room temp for 2 h. The solution will turn a darker shade of red/purple but should remain clear with no observable precipitate. 2. After 2 h, desalt using a 10DG column. Equilibrate the column with low-salt buffer, add the sample to the column, and add (3 mL—sample volume) low-salt buffer to the column. Collect 0.5 mL fractions, checking each fraction for tF and AuNP content by measuring the absorbance of each sample at 280 and 520 nm. 3. Combine all tF-AuNP fractions into one vial and add 90 μL of 100 mM freshly prepared ascorbic acid. The solution will turn slightly more red and less purple. Let the solution incubate at room temp overnight. 4. The resulting solution can be characterized by UV-Vis spectroscopy to verify lack of formation of large aggregates (there should be little to no change in surface plasmon resonance wavelength), as well as small X-ray scattering (SAXS) or dynamic light scattering (DLS). It can also be characterized by TEM to demonstrate increase in particle diameter using the particle analysis tool in ImageJ (version 1.41 h, National Institutes of Health).

4

Notes 1. So long as careful flame sterilization is maintained, the covered LB solution will stay good for months at room temp. To be safe, the solution should be autoclaved again prior to use in the small growth. Dispose of the LB solution if it becomes cloudy, indicating microbial growth. 2. The ascorbic acid solution should be freshly prepared, as it oxidizes easily. When freshly prepared, the solution is clear and colorless. After several days the solution turns yellow and is unsuitable for use in the seeded growth reaction. 3. The bacteria solution should go from being highly viscous and tan to darker colored and much less viscous. If after six cycles of sonication the solution is still somewhat viscous, continue sonicating at 2 min intervals. Sonication has at times taken ten cycles to reach the nonviscous stage. 4. To avoid protein precipitation during concentration, centrifuge for short periods of time (~10 min), and in between each round of centrifugation resuspend highly concentrated protein located at the bottom of the centrifugal filter unit near the filters.

36

Katherine W. Pulsipher and Ivan J. Dmochowski

5. Using matrix-assisted laser desorption ionization (MALDI) time-of-flight (TOF) mass spectrometry as a characterization technique usually leads to observation of only the monomer and dimer peaks, rather than the full 487,500 Da 24mer protein. We have used both sinapinic acid and Super-DHB as matrices with success. 6. Sometimes the particles do not precipitate after centrifuging. If this happens, add more NaCl and centrifuge a second time with the same conditions. 7. To confirm tF disassembly, Trp fluorescence can be measured by exciting at 295 nm. The disassembled dimer protein will have a peak at approximately 336 nm, compared to 24mer at 330 nm. 8. Properly assembled conjugates will appear in micrographs as a dark AuNP circle within a white protein “halo.” Be careful with the defocus settings used when imaging the sample by TEM, as significant underfocus can lead to a white ring artifact—a Fresnel fringe. At high magnification, a defocus of approximately 1.5 μm is usually appropriate. Imaging areas that are wellstained increases the likelihood of observing a true protein halo. 9. There is no evidence of other types of assemblies (monomer, 12mer, etc.) in the FPLC traces; thus, we calculate the KA for native protein assuming an equilibrium only between dimer and 24mer. 10. We have used both HAuCl4 and AuCl3 as the gold salt and have found less particle growth for HAuCl4 compared to AuCl3, presumably because it is less favorable for negatively charged AuCl4 to congregate within the also negatively charged tF cavity compared to neutral AuCl3. Using HAuCl4 as the gold salt necessitates a second round of gold ion addition, incubation, desalting, and reducing agent addition to see complete filling of the protein cavity.

Acknowledgments The authors thank Eric Johnson for providing the Archaeoglobus fulgidus ferritin gene in the pAF0834 plasmid and Jeffery Saven for use of instruments. We also thank Joe Swift and Jasmina CheungLau for doing the foundational work of this project. This work was supported by NSF CHE 0548188 and DMR-0520020. References 1. Michalet X, Pinaud FF, Bentolila L et al (2005) Quantum dots for live cells, in vivo imaging, and diagnostics. Science 307:538–544 2. Huang XH, El-Sayed IH, Qian W et al (2006) Cancer cell imaging and photothermal therapy

in the near-infrared region by using gold nanorods. J Am Chem Soc 128:2115–2120 3. Alivisatos P (2004) The use of nanocrystals in biological detection. Nat Biotechnol 22:47–52

Ferritin Encapsulation and Templated Synthesis of Inorganic Nanoparticles 4. Swift J, Wehbi WA, Kelly BD et al (2006) Design of functional ferritin-like proteins with hydrophobic cavities. J Am Chem Soc 128:6611–6619 5. Zhang L, Swift J, Butts CA et al (2007) Structure and activity of apoferritin-stabilized gold nanoparticles. J Inorg Biochem 101:1719–1729 6. Butts CA, Swift J, Kang S-G et al (2008) Directing noble metal ion chemistry within a designed ferritin protein. Biochemistry 47:12729–12739 7. Swift J, Butts CA, Cheung-Lau J et al (2009) Efficient self-assembly of Archaeoglobus fulgidus ferritin around metallic cores. Langmuir 25:5219–5225 8. Cheung-Lau JC, Liu D, Pulsipher KW et al (2013) Engineering a well-ordered, functional protein-gold nanoparticle assembly. J Inorg Biochem http://dx.doi.org/10.1016/j. jinorgbio.2013.10.003 9. Liu X, Wei W, Yuan Q et al (2012) ApoferritinCeO2 nano-truffle that has excellent artificial redox enzyme activity. Chem Commun 48:3155–3157 10. Hennequin B, Turyanska L, Ben T et al (2008) Aqueous near-infrared fluorescent composites based on apoferritin-encapsulated PbS quantum dots. Adv Mater 20:3592–3596 11. Zheng B, Zettsu N, Fukuta M et al (2011) Versatile protein-based bifunctional nanosystems (encapsulation and directed assembly): selective nanoscale positioning of gold nanoparticle-viral protein hybrids. Chem Phys Lett 506:76–80 12. Sun J, DuFort C, Daniel M-C et al (2007) Corecontrolled polymorphism in virus-like particles. Proc Natl Acad Sci U S A 104:1354–1359 13. Aniagyei SE, Kennedy CJ, Stein B et al (2009) Synergistic effects of mutations and nanoparticle templating in the self-assembly of cowpea chlorotic mottle virus capsids. Nano Lett 9:393–398

37

14. Capehart SL, Coyle MP, Glasgow JE et al (2013) Controlled integration of gold nanoparticles and organic fluorophores using synthetically modified MS2 viral capsids. J Am Chem Soc 135:3011–3016 15. Dragnea B, Chen C, Kwak E-S et al (2003) Gold nanoparticles as spectroscopic enhancers for in vitro studies on single viruses. J Am Chem Soc 125:6374–6375 16. Loo L, Guenther RH, Basnayake VR et al (2006) Controlled encapsidation of gold nanoparticles by a viral protein shell. J Am Chem Soc 128:4502–4503 17. Chen C, Daniel M-C, Quinkert ZT et al (2006) Nanoparticle-templated assembly of viral protein cages. Nano Lett 6:611–615 18. Daniel M-C, Tsvetkova IB, Quinkert ZT et al (2010) Role of surface charge density in nanoparticle-templated assembly of bromovirus protein cages. ACS Nano 4:3853–3860 19. Dixit SK, Goicochea NL, Daniel M-C et al (2006) Quantum dot encapsulation in viral capsids. Nano Lett 6:1993–1999 20. Loo L, Guenther RH, Lommel SA et al (2007) Encapsidation of nanoparticles by red clover necrotic mosaic virus. J Am Chem Soc 129:11111–11117 21. Huang X, Bronstein LM, Retrum J et al (2007) Self-assembled virus-like particles with magnetic cores. Nano Lett 7:2407–2416 22. Johnson E, Cascio D, Sawaya MR et al (2005) Crystal structures of a tetrahedral open pore ferritin from the hyperthermophilic archaeon Archaeoglobus fulgidus. Structure 13:637–648 23. Kim M, Rho Y, Jin KS et al (2011) pHdependent structures of ferritin and apoferritin in solution: disassembly and reassembly. Biomacromolecules 12:1629–1640 24. Haiss W, Thanh NTK, Aveyard J et al (2007) Determination of size and concentration of gold nanoparticles from UV-vis spectra. Anal Chem 79:4215–4221

Chapter 4 Determining the Relaxivity Values of Protein Cage-Templated Nanoparticles Using Magnetic Resonance Imaging Barindra Sana and Sierin Lim Abstract The application of magnetic resonance imaging (MRI) is often limited by low magnetic relaxivity of currently used contrast agents. This problem can be addressed by developing more sensitive contrast agents by synthesizing new types of metal complex or metallic nanoparticles. Protein cage has been used as a template in biological synthesis of magnetic nanoparticles. The magnetic nanoparticle-protein cage composites have been reported to have high magnetic relaxivity, which implies their potential application as an MRI contrast agent. The magnetic relaxivity is determined by measuring longitudinal and transverse magnetic relaxivities of the potential agent. The commonly performed techniques are field-cycling NMR relaxometry (also known as variable field relaxometry or nuclear magnetic relaxation dispersion (NMRD) profiling) and in vitro or in vivo MRI relaxometry. Here, we describe techniques for the synthesis of nanoparticle-protein cage composite and determination of their magnetic relaxivities by in vitro MR image acquisition and data processing. In this method, longitudinal and transverse relaxivities are calculated by measuring relaxation rates of water hydrogen nuclei at different nanoparticle-protein cage composite concentrations. Key words Magnetic relaxivity, Ferritin, Nanoparticle, MRI, Imaging, Protein cage

1

Introduction Magnetic resonance imaging (MRI) is a noninvasive diagnostic tool for assessing the anatomy and function of soft tissues. MRI measures the relaxation of hydrogen nuclei of tissue water and the technique typically relies on high water content of soft tissues. Contrast agents shorten proton relaxation time and improve contrast difference within an MR image [1]. Several limitations of currently available MRI contrast media can be addressed by developing highly sensitive contrast agents [2–4]. The development of more sensitive contrast agents may enable the detection of smaller tissue lesions beyond the limitation of currently available contrast media and may expand the regular application of MRI technique.

Brendan P. Orner (ed.), Protein Cages: Methods and Protocols, Methods in Molecular Biology, vol. 1252, DOI 10.1007/978-1-4939-2131-7_4, © Springer Science+Business Media New York 2015

39

40

Barindra Sana and Sierin Lim

Potential contrast enhancement (in MR images) by magnetic compounds can be assessed from their magnetic relaxivity, i.e., the increase of relaxation rate (of the surrounding water proton spins) by 1 mM contrast agent. Magnetic relaxation is a complex phenomenon by which nuclear magnetization developed in a nonequilibrium state returns to the equilibrium or steady-state condition. The shift from nonequilibrium to steady-state is not instantaneous and the time required for this shift is known as relaxation time. There are two types of magnetic relaxations: (1) the relaxation parallel to the external magnetic field known as longitudinal relaxation (T1 relaxation or spin-lattice relaxation) and (2) the relaxation perpendicular to the external field known as transverse relaxation (T2 relaxation or spin-spin relaxation). The corresponding longitudinal and transverse relaxation times are known as T1 and T2, respectively. Longitudinal and transverse magnetic relaxations are used in two distinct types of MR imaging, i.e., T1-weighted imaging and T2-weighted imaging. These two types of imaging techniques differ in the sequence of repetition time (TR) or echo time (TE). The T1-weighted images are acquired by changing the TR values, while T2-weighted images by changing the TE values. T1-weighted imaging differentiates fat and water by showing water as the darker and fat as the brighter phase. In contrast, T2weighted imaging may not differentiate between fat and water as they may both appear bright. Typically two different types of contrast agents are used in T1- and T2-weighted imaging; they are called T1 and T2 contrast agents and their relaxivities are expressed as R1 and R2, respectively. T1 contrast agents (also known as bright contrast agent) enhance contrast by increasing signal intensity and appear bright on MR images, while T2 contrast agents (also known as dark contrast agent) work by decreasing signal intensity and the contrast agents appear darker on the MR images. Contrast agents improve proton relaxivity of water molecule by several possible molecular mechanisms including increasing the number of innersphere water molecule, optimizing the water exchange rate, and decreasing the molecular tumbling rate. However, the science behind contrast development is extremely complex and performance of a contrast agent depends on chemical composition and physical properties of the material. The search for ultrasensitive contrast agents has drawn remarkable attention in recent years. Various types of chemical complexes and nanoparticles are being synthesized with the objective of developing ultrasensitive MRI contrast agents. Complexes and chelates of gadolinium, iron, and manganese are approved contrast agents for clinical MRI [5–7]. Magnetic nanoparticles and metal-protein conjugates were reported to have high magnetic relaxivity [8–12]. Complexation or chelation with large molecules increases the rotational correlation time (decreases tumbling rate) of small paramagnetic ions (fast tumbling molecule), which enhances magnetic

Magnetic Relaxivity of Metal-Protein Nanocomposites

41

relaxivity of the small molecules [12]. Molecular tumbling rate of small molecules decreases in the same way by conjugation with bulky protein molecules. In addition, hydration water molecules associated with the protein structure may also contribute in water exchangeability and thus affect magnetic relaxivity of metal-protein conjugates [15, 20]. Ferritin is a large protein cage (480 kDa) with inherent capability of metallic nanoparticle synthesis within its cavity [13, 14]. Gadolinium, iron, and manganese nanoparticles have been synthesized using ferritin as the template and characterized as potential MRI contrast agents [15–20]. The metal nanoparticles synthesized within the ferritin cage showed extremely narrow size distribution due to uniform size of the template. The proteinaceous nature of ferritin allows easy chemical or genetic modification to facilitate encapsulation of molecular cargo within the protein cage, targeted delivery, and targeted accumulation. The following protocols describe the synthesis of iron nanoparticle by mineralization of the ferritin cage, characterization of the metalprotein nanocomposite, and evaluation of its magnetic relaxivity by analyzing MR images. The same protocol can be used for determining magnetic relaxivities of other chemicals, metal complexes, chelates, and nanoparticles.

2

Materials Prepare all solutions using deionized water and analytical grade chemicals unless otherwise mentioned.

2.1

Protein

The following protocols are optimized for engineered Archaeoglobus fulgidus ferritin (AfFtn-AA) as the template for nanoparticle synthesis. A. fulgidus is a hyperthermophilic archaeon. Hence, AfFtn-AA derived from it is thermostable. The recombinant production and preparation of the ferritin has been previously described [19].

2.2

Working Buffer

25 mM HEPES containing 50 mM NaCl, pH 7.5. Dissolve 5.96 g HEPES powder and 2.92 g NaCl in 1 l water and adjust the pH to 7.5 using 4 M NaOH solution. For chromatography steps, filter the buffer with 0.2 μm membrane.

2.3

Chemicals

1. Prepare 100 mM stock of ferrous solution by dissolving 0.0278 g FeSO4·7H2O crystal in 1 ml 0.1 % HCl and 100 mM stock of manganese solution by dissolving 0.0198 g MnCl2·4H2O crystal in 1 ml water (see Note 1). 2. Prepare 1.2 % agarose (biotechnology grade) solution by boiling a mixture of 1.2 g agarose powder in 100 ml working buffer. 3. BCA protein assay kit (Thermo Scientific). 4. Gel filtration calibration kit (GE Healthcare).

42

2.4

Barindra Sana and Sierin Lim

Equipment

1. Centrifuge, Sorvall Legend XTR (Thermo Scientific). 2. Inductively coupled plasma mass spectrometer (ICP-MS) (PerkinElmer). 3. Spectrophotometer, UV-1600PC (VWR). 4. AKTA purifier fast protein liquid chromatography (FPLC) (GE Healthcare). 5. Transmission electron microscope (TEM) (JEOL, JEM-1400). 6. Vibrating sample magnetometer (VSM), 1.5 T (Lakeshore, 7300 series). 7. MRI scanner, 3.0 T, MAGNETOM Verio (Siemens).

2.5 Other Consumables

1. Amicon Ultra centrifugal filter device, 100 kDa molecular weight cutoff (MWCO) (Millipore). 2. PD-10 column (GE Healthcare). 3. Superdex 200 10/300 GL column (GE Healthcare). 4. Drop dialysis membrane (Millipore). 5. Carbon-coated copper grid, 300 mesh (Electron Microscopy Sciences).

3

Methods Since the protein is thermostable, all experiments are performed at room temperature unless otherwise mentioned.

3.1 Nanoparticle Synthesis in Ferritin Cage (Mineralization/ Loading)

1. Aliquot 0.1 mg/ml (~0.2 μM 24-mer) ferritin solution in 100 mM HEPES buffer containing 50 mM NaCl (pH 7.5) (see Note 2). 2. Add the metal (stock solution) in the protein solution at 1:1,000 (v/v) ratio to achieve 0.1 mM metal concentration (i.e., 480 metal ion/24mer) and mix immediately (see Note 3). 3. Repeat the addition (of 0.1 mM metal stock at 1:1,000 (v/v) ratio) at 10-min interval to achieve the desired metal concentration. The stepwise addition allows for higher final metal concentration and higher metal to protein ratio to be achieved. 4. Incubate at room temperature for 1 h followed by overnight incubation at 4  C. 5. Centrifuge the mixture at 14,000  g for 15 min to remove any particulates. 6. Concentrate the supernatant using 100 kDa MWCO Amicon Ultra centrifugal filter device (see Note 4). 7. Remove the unbound metal (if any) by filtration through PD10 desalting column.

Magnetic Relaxivity of Metal-Protein Nanocomposites

43

8. Measure the final metal concentration of the preparation using ICP-MS (see Note 5). ICP-MS samples were prepared by mixing 1 ml of mineralized ferritin with 1 ml of 63 % HNO3 and 1 ml 35 % H2O2, and the mixture was heated at 80  C for 30 min. The sample was cooled down, the volume adjusted to 10 ml with DI, and filtered. 9. Determine the final protein concentration with BCA protein assay kit following the manufacturer’s protocol (see Note 6). 10. Calculate the amount of metal encapsulated in each protein from the molar ratio of the metal to ferritin in the final preparation. 3.2 Confirmation of Iron Binding and Molecular Size of Ferritin

Confirm the iron binding from the co-elution of iron and protein in the size-exclusion chromatography (SEC) experiment. The molecular weight (MW) of the protein cage can be estimated from the elution volume. 1. Attach the Superdex 200 10/300 GL column to the FPLC system and equilibrate with two column volumes (vc) (48 ml) of the working buffer at 0.5 ml/min. Use the same buffer as the running buffer. 2. Inject 100 μl of 0.1 mg/ml of each protein standard from the gel filtration calibration kit dissolved in the working buffer using a 100-μl injection loop (see Note 7). Elute the sample at 0.5 ml/min. Monitor the elution of each standard protein at 280 nm. Calculate the elution volume (ve) by subtracting the volume at injection from the volume at elution. 3. Determine the void volume (vo) of the column by injecting blue dextran solution in the same column and record its elution volume. 4. Generate a standard curve by plotting K av ¼ vvec vvoo vs. Log10(MW) of each protein standard in the calibration kit and obtain the equation from linear fitting of the standard curve. 5. Inject 100 μl of 0.1 mg/ml protein sample into the column following step 2. 6. Monitor the absorbance of the protein and the iron nanoparticle at 280 nm and 310 nm, respectively. 7. Record the elution volumes of the protein and iron nanoparticle. The same elution volume will indicate iron binding to or core formation within the protein cage. The unbound iron will elute in larger elution volume than the protein molecules. 8. Calculate the molecular weight of the ferritin cage by plotting its Kav in the equation mentioned in step 4. Compare the measured molecular weight to the theoretical molecular weight of the 24-meric ferritin cage (~500 kDa).

44

Barindra Sana and Sierin Lim

3.3 Confirmation of Nanoparticle Formation

Confirm the presence of iron nanoparticle within the protein cage by imaging under TEM. 1. Prepare 0.2 mg/ml mineralized ferritin solution and remove the salts by performing drop dialysis into water using “V” series membrane following manufacturer’s protocol. 2. Soak the carbon-coated copper grid (Electron Microscopy Sciences) in the protein solution by floating the grid on a drop of solution with the carbon-coated surface facing the solution. Allow the protein sample to adhere to the grid for ~2 min and air dry the grid for ~10 min. 3. Negative staining of the protein cage (see Note 8). Prepare a 1.5 % uranyl acetate (see Note 9) solution in water and stain the air-dried grid by floating it on a drop of this solution with the protein-adhered surface facing the uranyl acetate solution. Allow to stain for ~2 min and air dry the grid for ~15 min. Store the grid in a desiccator for at least 12 h before observing under TEM. 4. Examine the grid under a TEM with 25,000–30,000 times magnification. The white shell (~13 nm) confirms the formation of the ferritin cage. 5. Fix a sample without uranyl acetate staining (step 3) on the TEM grid. Examining this sample under the TEM will show the metal nanoparticles without the protein cage.

3.4 Magnetic Characterization of the Mineralized Ferritin

Study the magnetic property of the nanoparticles by VSM. 1. Buffer exchange about 10 mg of the mineralized ferritin with water using a PD-10 desalting column and remove the water by lyophilization. 2. Place the dried sample in the VSM sample holder and measure the magnetization at room temperature under the field strength of 1.2 to +1.2 T using a 1.5 T 7300 series VSM (Lakeshore). 3. Measure the magnetization of the empty sample holder at the same condition under the same field strength. Subtract the magnetization of the holder from the magnetization of the sample (in holder) to obtain the actual magnetization of the sample. 4. Plot the field strength against magnetization to obtain a hysteresis loop (of magnetization) of the sample. The shape of the hysteresis loop indicates the magnetic nature (paramagnetic, superparamagnetic, etc.) of the compound. 5. For quantitative characterization of magnetic properties, measure the magnetic parameters including saturated magnetization (Ms), retentivity (Mr), and coercivity (Hci). 6. Unloaded ferritin may serve as a control.

Magnetic Relaxivity of Metal-Protein Nanocomposites

3.5 Fixing the Nanoparticles in Agarose Matrix for Imaging

45

Fix the nanoparticles in different dilution within 0.8 % (final concentration) agarose in a 24-well tissue culture plate (Fig. 1) using the following protocol: 1. Dilute the nanoparticle-protein composite stock using the working buffer to get final protein concentrations of 2.4, 1.2, 0.6, 0.3, 0.15, and 0.075 mg/ml by serial dilution. Make 1 ml of each dilution (in duplicate) directly in the wells of a 24-well tissue culture plate. Usually a blank containing media is taken as “0” sample concentration. “No iron” or “no ferritin” control experiments can also be prepared for comparison. 2. Prepare 1.2 % agarose solution in the working buffer and cool down to about 70–80  C. Keep the agarose solution on a hot plate at 80  C (see Note 10). 3. Add 2 ml agarose solutions in 1 ml of samples within the 24well tissue culture plate and mix immediately by repeated pipetting (see Note 11). Mixing should be done gently to avoid formation of air bubble. In the final preparation, the protein concentration would be 0.8, 0.4, 0.2, 0.1, 0.05, and 0.025 mg/ml (see Note 12). 4. Allow the sample to solidify (Fig. 1) and keep at 4  C until imaging is done on the next day.

3.6

Imaging

Scan sequences will require optimization depending on the materials tested. 1. Place the tissue culture plates (with the samples) horizontally in a knee coil and secure the plate by fixing with tape to prevent any movement during imaging.

Fig. 1 Different concentrations of samples fixed in 0.8 % agarose in 24-well tissue culture plate. Sample 1 to sample 4 contains four types of nanoparticleprotein cage composite synthesized by encapsulating 1200, 2400, 4800, and 7200 iron per protein cage, respectively; all wells in a column for sample 1 to sample 4 have identical iron concentration but different protein concentrations

46

Barindra Sana and Sierin Lim

2. Place the coil within the magnetic field of the MAGNETOM Verio 3.0 T whole body scanner (Siemens). Make sure that the sample is located at the center of the magnetic field. 3. Scan the plate at horizontal cross section (coronal/frontal plane) with slice thickness of 3.0–6.0 mm. 4. For T1-weighted scanning, acquire the images at variable inversion time (TI) of 30, 50, 100, 200, 400, 500, 600, 800, 100, 1,200, 1,800, 2,000, and 2,500 ms. Set the other parameters as follows: repetition time (TR) ¼ 2,800 ms, echo time (TE) ¼ 8.5 ms, matrix ¼ 512 512, slice thickness ¼ 6.0 mm, number of averages ¼ 1, and field of view ¼ 27 cm. 5. Capture the T2-weighted images at variable echo times (TE) of 10, 15, 20, 40, 60, 80, 100, 150, and 200 ms. Other parameters are set as TR ¼ 1,000 ms, matrix ¼ 512 512, slice thickness ¼ 6.0 mm, number of averages ¼ 1, and field of view ¼ 27 cm. 3.7 Relaxivity Calculation

1. Measure the mean signal intensity of all MR images taken at various inversion times (TI) and echo times (TE), using ImageJ software (NIH). Measurement should be done around the center of images (Fig. 2) (see Note 13). 2. Plot the mean signal intensities of T1-weighted images of each dilution against the corresponding inversion times of the imaging. Fit the curve to the “inversion recovery equation” [y ¼ a b  e cx, where x is inversion time (TI), y is mean signal intensity (I), a and b are constants, and c is longitudinal relaxation rate (1/T1)] using the “curve-fitting tool” in MATLAB software (MathWorks) (Fig. 3a). Get the 1/T1 value of each dilution from the fittings.

Fig. 2 Typical T1-weighted MR images of the samples showing the area of intensity measurement (red circle)

Magnetic Relaxivity of Metal-Protein Nanocomposites

47

Fig. 3 Typical curve fitting of time vs. signal intensity plot for calculation of T1 (A) and T2 (B) values of different concentrations of a sample. The different colors indicate different concentrations of the same samples

Fig. 4 Relaxivity calculation by plotting relaxation rate against metal concentration

3. Plot the mean signal intensities of T2-weighted images of each dilution against the corresponding echo times of the imaging. Fit the curve to the “monoexponential decay equation” [y ¼ a  e bx + c, where x is echo time (TE), y is mean signal intensity (I), a and c are constants, and b is transverse relaxation rate (1/T2)] using the “curve-fitting tool” in MATLAB software (Fig. 3b). Get the 1/T2 value of each dilution from the fittings. 4. Plot the longitudinal or transverse relaxation rates (1/T1 or 1/T2) against the molar concentrations of the metal in different dilutions. Fit the curve to the linear equation [y ¼ mx + c, where y is relaxation rate (1/T1 or 1/T2) in s 1, m is gradient of the linear fit, x is metal concentrations in mM, and c is constant]. The gradients of the linear fits give the respective longitudinal or transverse relaxivity (R1 or R2) of the sample in mM 1 s 1 (Fig. 4).

48

4

Barindra Sana and Sierin Lim

Notes 1. Several other metals can be encapsulated within the ferritin cage following the protocols reported previously [21, 22]. 2. Low ferritin concentration is used to avoid iron precipitation at higher local concentration outside the protein cage, which may happen during iron loading in concentrated ferritin solution. 3. Immediate mixing is important after the addition of metal solution. Otherwise, local iron precipitation may be observed due to autoxidation leading to bulk precipitation in subsequent steps. 4. Concentrating with Amicon Ultra centrifugal filter also removes unbound metals. 5. Usually ICP-MS measures the metal concentration in ppm. The value can be converted to molar concentration by dividing with the atomic weight of the metal. 6. BCA protein assay method is sensitive to iron; the presence of free iron may interfere with protein quantification. Any trace of unbound iron should be removed before performing the BCA assay. 7. Protein standard can be injected individually or as a mixture. For mixture, it is recommended that the mixture contains standards whose elution volumes are ~2 ml apart. 8. The electrons during TEM imaging destroy the protein. To visualize the protein cage, staining with uranyl acetate or phosphotungstic acid is required. The heavy atoms fill the voids of the protein and preserve the protein structures upon electron bombardment in the TEM. The staining allows visualization of the voids, hence the term negative stain. Nanoparticles are visible without any staining. 9. Uranyl acetate is radioactive. Take proper measure for its handling and disposal. Phosphotungstic acid is an alternative staining reagent. It is not radioactive but uranyl acetate shows better staining. 10. High temperature of the agarose solution prevents fast solidification of agarose during mixing with the protein-nanoparticle solution (at room temperature), which otherwise may cause inefficient mixing and inhomogeneity in the final preparation. Using high concentration of agarose also may cause fast solidification and inhomogeneity in the final preparation. 11. It is possible to prepare the sample in any other vials (such as test tubes or microcentrifuge tubes) which can be arranged in different ways to fit the bore size of the MRI scanner. 12. The concentrations of the test materials should be adjusted depending on their contrast enhancement property and the strength of magnetic field.

Magnetic Relaxivity of Metal-Protein Nanocomposites

49

13. It is highly recommended to avoid the extreme edges of the images that incorporate some parts of the plate image, particularly during intensity measurement of T2-weighted images. The extreme edges may appear very dark and it is difficult to distinguish the edge of sample from the plate on the images.

Acknowledgment The authors thank Dr. Cher Heng Tan at Tan Tock Seng Hospital, Singapore, for technical advices on magnetic resonance imaging. The work is supported by Singapore National Medical Research Council New Investigator Grant (NMRC/NIG/1073/2012) References 1. Caravan P, Ellison JJ, McMurry TJ, Lauffer RL (1999) Gadolinium(III) chelates as MRI contrast agents: structure, dynamics, and applications. Chem Rev 99:2293–2352 2. Bulte JW, Kraitchman DL (2004) Iron oxide MR contrast agents for molecular and cellular imaging. NMR Biomed 17:484–499 3. Lu J, Ma S, Sun J, Xia C, Liu C, Wang Z, Zhao X, Gao F, Gong Q, Song B, Shuai X, Ai H, Gu Z (2009) Manganese ferrite nanoparticle micellar nanocomposites as MRI contrast agent for liver imaging. Biomaterials 30:2919–2928 4. Weissleder R, Mahmood U (2001) Molecular imaging. Radiology 219:316–333 5. Kubicek V, Toth E (2009) Design and function of metal complexes as contrast agents in MRI. Adv Inorg Chem 61:63–129 6. Reimer P, Balzer T (2003) Ferucarbotran (Resovist): a new clinically approved RESspecific contrast agent for contrast-enhanced MRI of the liver: properties, clinical development, and applications. Eur Radiol 3:1266–1276 7. Bleicher AG, Kanal E (2008) Assessment of adverse reaction rates to a newly approved MRI contrast agent: review of 23,553 administrations of gadobenate dimeglumine. Am J Roentgenol 191:W307–W311 8. Wang YXJ (2011) Super paramagnetic iron oxide based MRI contrast agents: current status of clinical applications. Quant Imaging Med Surg 1:35–40 9. Yalappu MM, Othman SF, Curtis ET, Gupta BK, Jaggi M, Chauhan SC (2010) Multifunctional magnetic nanoparticles for magnetic resonance imaging and cancer therapy. Biomaterials 32:1890–1905

10. Huang J, Zhong X, Wang L, Yang L, Mao H (2012) Improving the magnetic resonance imaging contrast and detection methods with engineered magnetic nanoparticles. Theranostics 2:86–102 11. Koylu MZ, Asubay S, Yilmaz A (2009) Determination of proton relaxivities of Mn(II), Cu (II) and Cr(III) added to solutions of serum proteins. Molecules 14:1537–1545 12. Yang JJ, Yang J, Wei L, Zurkiya O, Yang W, Li S, Zou J, Maniccia AL, Mao H, Zhao F, Malchow R, Zhao S, Johnson J, Hu X, Krogstad E, Liu ZR (2008) Rational design of proteinbased MRI contrast agents. J Am Chem Soc 130:9260–9267 13. Galvez N, Fernandez B, Valero E, Sanchez P, Cuesta R, Dominguez-Vera JM (2008) Apoferritin as a nanoreactor for preparing metallic nanoparticles. Comp Rendus Chim 11:1207–1212 14. Yoshizawa K, Iwahori K, Sugimoto K, Yamashita I (2006) Fabrication of gold sulfide nanoparticles using the protein cage of apoferritin. Chem Lett 35:1192–1193 15. Aime S, Frullano L, Crich SG (2002) Compartmentalization of a gadolinium complex in the apoferritin cavity: a route to obtain high relaxivity contrast agents for magnetic resonance imaging. Angew Chem Int Ed 41:1017–1019 16. Crich SG, Bussolati B, Tei L, Grange C, Esposito G, Lanzardo S, Camussi G, Aime S (2006) Magnetic resonance visualization of tumor angiogenesis by targeting neural cell adhesion molecules with the highly sensitive gadolinium-loaded apoferritin probe. Cancer Res 66:9196–9201 17. Sanchez P, Valero E, Galvez N, DominguezVera JM, Marinone M, Poletti G, Corti M,

50

Barindra Sana and Sierin Lim

Lascialfari A (2009) MRI relaxation properties of water-soluble apoferritin-encapsulated gadolinium oxide-hydroxide nanoparticles. Dalton Trans 5:800–804 18. Uchida M, Terashima M, Cunningham CH, Suzuki Y, Willitis DA, Yang PC, Tsao PS, McConnell MV, Young MJ, Douglas T (2008) A human ferritin iron oxide nanocomposite magnetic resonance contrast agent. Magn Reson Med 60:1073–1081 19. Sana B, Johnson E, Sheah K, Poh CL, Lim S (2010) Iron based ferritin nanocore as a contrast agent. Biointerphases 5:AF48–AF52

20. Sana B, Poh CL, Lim S (2012) A manganeseferritin nanocomposites as an ultrasensitive T2 contrast agent. Chem Commun 48:862–864 21. Sana B, Calista M, Lim S (2012) Protein cage assisted metal-protein nanocomposite synthesis: optimization of loading conditions. AIP Conf Proc 1502:82–96 22. Qiu H, Dong X, Sana B, Peng T, Parapelle D, Chen P, Lim S (2013) Ferritin-templated synthesis and self-assembly of Pt nanoparticles on monolithic porous graphene network for electrocatalysis in fuel cell. ACS Appl Mater Interfaces 5:782–787

Chapter 5 Computationally Assisted Engineering of Protein Cages Maziar S. Ardejani and Brendan P. Orner Abstract A hybrid computational method incorporating topographic analysis of protein surfaces and free-energy calculations of protein-protein interactions in protein nanocages is described. This design strategy can be used to engineer protein cages for enhanced structural stability and assembly. Key words Protein engineering, Computational design, Protein-protein interactions, Ferritins

1

Introduction Many self-assembling protein cages have evolved to selectively isolate biological material, such as mineralized iron in the case of the ferritins or genetic material within virus capsids, and release it on demand. Due to these encapsulation properties in addition to their nanoscale, protein cages are being harnessed increasingly for a wide range of nanobiotechnological applications from drug delivery to nanomaterials and catalysis [1]. However these applications are often limited by the non-ideal physical properties of these protein complexes. Therefore, it may be desirable to reengineer protein cages for enhanced utility. However, complete de novo rational design of complex folded proteins is extremely difficult [2], and this challenge is compounded for self-assembling nanoscaled protein quaternary structure [3]. Thus, the development and implementation of tools for the computationally assisted manipulation of protein-protein interactions involved in self-assembly, especially those generating nanostructured protein complexes, are of great recent interest [4]. Here, we describe a hybrid computational method incorporating topographic analysis of protein surfaces and free-energy calculations of protein-protein interactions to enhance the structural energetics of a protein nanocage. Protein nanostructure is often assembled from multiple protein chains and is therefore controlled by different protein-protein interfaces, the number and nature of which depend on

Brendan P. Orner (ed.), Protein Cages: Methods and Protocols, Methods in Molecular Biology, vol. 1252, DOI 10.1007/978-1-4939-2131-7_5, © Springer Science+Business Media New York 2015

51

52

Maziar S. Ardejani and Brendan P. Orner

stoichiometry, symmetry, and similarities between the monomers. Thus, to stabilize oligomeric protein complexes, an obvious strategy would be to manipulate interactions at the protein-protein interfaces directly [5]. However, the direct strategy ignores the fact that the hierarchical interdependence of the interactions made at these interfaces often complicates the redesign process through positive and negative cooperativity [4]. In other words, unfavorable structural deviations caused by small atomic clashes at lowerstoichiometry interfaces can be amplified through the structural hierarchy, resulting in complete disruption of higher-order assembly even if the specific redesigned interface is enhanced [6]. Thus, current protein design methodology is limited when applied to engineering nanostructures, and it rarely can be applied to multisubunit, self-assembling systems [5]. Hence, improving the accuracy and predictive power of the currently available computational methodology is desirable. One strategy to do this could be the development of new hybrid computational approaches. Proteins typically fold into well-packed three-dimensional structures. However, some oligomeric proteins, for various functional and non-functional reasons, possess protein-protein interfaces interrupted by solvent-accessible “concavities” which are further characterized by the number of openings they provide to bulk solvent: pockets (one opening) and pores (two openings). For example, some ferritins have concavities near their twofold, threefold, fourfold, and “B-site” interfaces (see Fig. 2 for a graphical definition of these interfaces). Some of these concavities are thought to be involved in the traffic of chemical species required for iron biomineralization inside the cage [7, 8]. In most protein design strategies, solvent-accessible concavities are rarely distinguished from other “packing imperfections” although in a few examples, with monomeric proteins, cavity-filling or cavity-making mutations have been used to probe folding [9–17]. The structural analysis program CASTp differentiates packing imperfections at protein-protein interactions based on their solvent accessibility [18]. The structural distinctness of these imperfections, which can be analytically detected by CASTp, has been exploited to design new molecular interactions through the optimization of side-chain packing in protein-protein interfacial concavities [6, 19]. In this chapter, we describe a method that we have successfully employed to stabilize [6] and enhance the self-assembling properties [19] of a protein nanocage, E. coli bacterioferritin. Our previous analyses have shown that most stabilizing mutations were associated with interfacial concavities; thus we reasoned that an efficient computational search could be achieved by limiting our focus to optimizing pocket-associated residues rather than the entire interface. In addition, we thought that the assembly state would be more tolerant to mutations within concavities because

Thermostabilization of Protein Cages

53

there is more volume within which to generate new interactions [6]. Moreover, this method finds mutations that either block the interfacial pores or reduce the volume of the pockets, each of which effects is respectively suitable for making the cage impermeable and/or enhancing its stability. The procedure described here is general and should be easily ported to other ferritins and protein cages. Our approach has been applied to single-point mutations but could be adapted to larger, more extensive studies. Furthermore, our suggestions regarding filtering out potential problematic residues and the number of structures to carry forward are based on developing a manageable project for a single PhD student in a modestly funded lab.

2

Materials PDB file for a high-resolution structure of the protein cage [20] (downloadable at: http://www.rcsb.org/pdb/). Structure visualization program (e.g., UCSF Chimera [21], downloadable at: http://www.cgl.ucsf.edu/chimera/). FoldX program [22] (downloadable at: http://foldx.crg.es/). CASTp [23] (web server accessible at http://cast.engr.uic.edu/). BLAST [24] (user interface: http://blast.ncbi.nlm.nih.gov/). Sequence alignment program (for example, ClustalX [25] downloadable at: http://www.clustal.org/).

3

Methods This approach to engineer enhanced protein cage stability and selfassembly is based on a hybrid computational method incorporating topographic analysis of protein surfaces and free-energy calculations of protein-protein interactions. A summary of this approach is shown in the workflow diagram (see Fig. 1). This approach requires a protein data bank file (.pdb) containing the high-resolution structure of the protein cage of interest. These can be obtained from the RCSB Protein Data Bank. To prepare the structure for the calculations, it first must be simplified through the generation of an oligomeric subcomplex focusing on the protein-protein interaction of interest. The energy of this subcomplex is then computationally minimized to generate a structure with slightly altered atomic coordinates of the side chains. The virtual screening for stabilizing mutations is initiated by the identification of pockets at the oligomeric interfaces through surface topographic analysis of the minimized subcomplexes. Saturation mutagenesis is next performed through a series of free-energy calculations. Because these are low-level calculations, it is essential to

54

Maziar S. Ardejani and Brendan P. Orner

Fig. 1 A minimal computational workflow to design stabilized protein cages. Three stages of our strategy consist of: (1) preparation of the initial structure file into a minimized subcomplex, (2) virtual screening for stabilizing mutations using a minimal computation approach employing a hybrid of surface topographic analysis of protein-protein interfaces and free-energy simulation of their interactions, and (3) virtual control experiments to filter out potentially unreliable hits

perform replicates and virtual control experiments to improve the success rate. Therefore, interaction network and phylogenetic analyses are then performed on the predicted stabilizing mutations to ensure they do not impose any potentially destabilizing effects. Only variants that pass these controls are taken on for cloning, expression, and in vitro analysis. 3.1 Preparation of the Cage Subcomplex Structures

1. Retrieve the atomic coordinates of the protein cage either generated de novo or downloaded from the RCSB Protein Data Bank (see Notes 1 and 2). 2. Use the visualization program to extract the coordinates of an oligomeric subcomplex of the cage from the PDB file. A subcomplex oligomer (e.g., dimer, trimer, etc.) is created through deletion of all except the protein chains that comprise the protein-protein interface of interest. As most protein cages are symmetric, this simplification will reduce the computation time while causing little loss in accuracy (see Note 3). 3. Use the hRepairPDBi command in the FoldX software to perform an initial minimization of the subcomplex (at 298 K, pH 7, 0.05 M ionic strength; see Note 4).

Thermostabilization of Protein Cages

3.2 Screening for Stabilizing Mutation by Virtual Saturation Mutagenesis of Pocket-Forming Residues

55

1. Submit the repaired subcomplex to CASTp [23] for detection and analysis of the pockets. Visualize the CASTp results using the visualization program. The CASTp calculations will detect many pockets and pores in the structure of each subcomplex. Rank them based on the molecular volume and visually confirm that these concavities are in the protein-protein interface of interest and not involved with protein monomers located outside the subcomplex (see Note 5). 2. Using the visualization program, export the list of residues involved in the selected pocket (e.g., use the hwrite listi command in Chimera). 3. Virtually mutate all of the residues involved in the interfacial concavity to all other amino acids using the hBuildModeli command of FoldX. This will generate an output file whose name begins with “Average_” Save this file for further analysis (see Note 6). 4. Open the resulting “Average_” output file using a spreadsheet editor and sort the data in ascending order by “total energy” to rank the mutations based on their degree of stabilization. Select the five most stabilizing mutations for further analysis (see Note 7).

3.3 Structural Analysis as Virtual Control Experiments to Filter Out Potential False Positives

1. Generate the minimized structure of the subcomplex for each of the selected mutants using the hBuildModeli command in FoldX and setting the “OutPDB” value to “true” 2. Submit each of these minimized structures to CASTp to analyze the structural effect of the mutation on the targeted concavity. This analysis will determine if the mutation indeed reduces the volume of the concavity as expected (see Note 8). 3. Use the hPrintNetworksi command in FoldX to extract the interaction network information from the minimized subcomplex structures of WT and the mutants. Identify mutated residues and compare the WT with the mutants to determine if new interactions across the interface were generated (see Note 9). 4. Obtain the amino acid sequence of homologous proteins from BLAST and perform an alignment to make sure that the mutated residues are not highly conserved (see Note 10). 5. Select three mutants that pass the control filters in the last three steps and assess them in vitro for stability.

4

Notes 1. If downloading the structure from the PDB, be sure to select the “biological assembly” as this will usually be the entire oligomeric cage as opposed to a single monomer.

56

Maziar S. Ardejani and Brendan P. Orner

2. Having a high-quality structure of the cage (resolution below ˚ ) is essential for successful analysis. Structures with a 2.2 A resolution below 2.8 A˚ can also be considered, but the computational results will be less reliable, and the software may have problems with some high-energy side-chain conformations. 3. A typical protein cage is composed of tens of thousands of atoms. Performing molecular-modeling calculations for such a large system requires tremendous amounts of computational power and time, making the task unfeasible. Our approach to break down large cage complexes into smaller oligomeric subcomplexes reduces the computational demand. For example, in the octahedral, homo-24-meric cage of a maxiferritin (see Fig. 2), each monomer interacts with six surrounding monomers through different types of interfaces, some of which are symmetry related. That is, in each ferritin cage, there are 6 tetramerization sites surrounding the fourfold axes, 8 trimerization sites on the threefold axes, 12 dimerizing interactions at the twofold axes of symmetry, and 24 “B-site interfaces”. A combination of these interfaces constitutes the complete set of protein-protein contacts made by a monomer in the cage oligomerization state (e.g., the black monomer in Fig. 2).

Fig. 2 Four different types of representative subcomplex oligomers into which a typical maxiferritin nanocage with octahedral, 432 symmetry can be broken down to reduce computational demand when analyzing their protein-protein interfaces. Each monomer (in black) interacts with six surrounding monomers through different types of interfaces, some of which are symmetry related. Clockwise, starting on the left: in each ferritin cage, there are 12 dimerizing interactions at the twofold axes of symmetry, 6 tetramerization sites surrounding the fourfold axes, 24 “B-site” interfaces, and 8 trimerization sites on the threefold axes. In order to ensure that interfacial concavities are analyzed accurately, some of these subcomplexes may be symmetry redundant. For example, in the maxiferritin, we define the “B-site interface” for the calculations although formally it is defined by combinations of the other interfaces (see Note 3)

Thermostabilization of Protein Cages

57

4. A prerequisite for the computational analysis is to have a good structure, that is, a protein with no van der Waals clashes or no residues with high-energy rotamers. Also it is necessary to make sure that there are no missing atoms or any gaps in the backbone. To fulfill these requirements the hRepairPDBi command in FoldX is used to prepare PDB files for subsequent calculations. During this process, residues that have unfavorable torsion angles, van der Waals clashes, etc. are identified and optimized to find new energy minima. In the FoldX protocol, the side chains are rotated, while the backbones are fixed, and buffer salts and unwanted ligands are removed. These “repaired” PDB files are then used to perform surface topography analysis using CASTp in the next step. 5. Concavities detected by CASTp on the edges of a subcomplex are less useful because the effect of the missing monomers is neglected during their detection. With that said, this emphasizes the importance of analyzing a number different protein-protein interactions through the selection of different subcomplexes. Thus, some of these subcomplexes may be symmetry redundant. For example, in the maxiferritin, we define a “B-site interface” for the calculations although formally combinations of the other interfaces define it. 6. To save hard drive space, set the “OutPDB” value to “false” in the virtual screening step. Later in the control step, we will set this value to “true” in order to generate PDB files for the minimized structure of only the selected mutants. 7. It has been previously shown that methionine side chains in a protein can accommodate many different geometries and thus provide a considerable degree of conformational flexibility resulting in the lack of specificity for methionine-rich protein surfaces [26, 27]. Moreover, the oxidation of methionine, which occurs in a wide variety of conditions, has been shown to cause structural instability and changes in the aggregation states of proteins [28]. Thus, we generally remove Met from the list of viable stabilizing designs. In addition, mutations to proline can also be problematic due to loss of a backbone hydrogen bond and backbone flexibility effects. Because FoldX assumes a rigid backbone, we typically remove proline mutations from candidate designs. 8. In the E. coli bacterioferritin cage for instance, there is a large concavity (see Fig. 3) at the twofold dimeric interface which contracts upon computationally predicted replacement of Asn23 with hydrophobic amino acids. We have shown in vitro that these mutations enhance the thermal stability of the ferritin [6]. 9. If the WT residue that was mutated has a large number of contacting atoms (more than 20), it is considered “highly buried.”

58

Maziar S. Ardejani and Brendan P. Orner

Fig. 3 Computationally predicted mutation of Asn23 to hydrophobic residues causes the large structural concavity (depicted as a blue surface) at the twofold dimeric interface of E. coli bacterioferritin to contract. The figure is adopted from Ardejani et al. [6] with permission

This may indicate that it is making structurally important interactions (as opposed to interactions with bound solvent within the concavity). In these cases, we often choose to avoid mutating these residues. This consideration is especially critical if the residue is involved in hydrogen bonding. We have noticed that the energetic penalty implemented in the FoldX energy function for removing a hydrogen bond can be understated compared to the other energy terms. Thus, the software may often falsely predict that a mutation, which removes a WT hydrogen bond, to have a stabilizing effect on the complex. Therefore, special care should be taken when residues involved in hydrogen bonds are mutated especially if the hydrogen bonds are part of a large interaction network. 10. A high level of conservation can give an indication that the residue is structurally or functionally important. Thus, avoiding mutation of these might be recommended, depending on the goals of the engineering exercise. References 1. Uchida M, Klem MT, Allen M, Suci P, Flenniken M, Gillitzer E, Varpness Z, Liepold LO, Young M, Douglas T (2007) Biological containers: protein cages as multifunctional nanoplatforms. Adv Mater 19:1025–1042 2. Dahiyat BI, Mayo SL (1997) De novo protein design: fully automated sequence selection. Science 278:82–87 3. Ardejani MS, Orner BP (2013) Obey the peptide assembly rules. Science 340:561–562

4. Grigoryan G, Kim YH, Acharya R, Axelrod K, Jain RM, Willis L, Drndic M, Kikkawa JM, DeGrado WF (2011) Computational design of virus-like protein assemblies on carbon nanotube surfaces. Science 332:1071–1076 5. King NP, Sheffler W, Sawaya MR, Vollmar BS, Sumida JP, Andre´ I, Gonen T, Yeates TO, Baker D (2012) Computational design of selfassembling protein nanomaterials with atomic level accuracy. Science 336:1171–1174

Thermostabilization of Protein Cages 6. Ardejani MS, Li NX, Orner BP (2011) Stabilization of a protein nanocage through the plugging of a protein–protein interfacial water pocket. Biochemistry 50:4029–4037 7. Yao H, Wang Y, Lovell S, Kumar R, Ruvinsky AM, Battaile KP, Vakser IA, Rivera M (2012) The structure of the BfrB–Bfd complex reveals protein–protein interactions enabling iron release from bacterioferritin. J Am Chem Soc 134(32):13470–13481 8. Tosha T, Ng H-L, Bhattasali O, Alber T, Theil EC (2010) Moving metal ions through ferritin protein nanocages from three-fold pores to catalytic sites. J Am Chem Soc 132:14562–14569 9. Mendel D, Ellman JA, Chang Z, Veenstra DL, Kollman PA, Schultz PG (1992) Probing protein stability with unnatural amino acids. Science 256:1798–1802 10. Karpusas M, Baase WA, Matsumura M, Matthews BW (1989) Hydrophobic packing in T4 lysozyme probed by cavity-filling mutants. Proc Natl Acad Sci U S A 86:8237–8241 11. Saito M, Kono H, Morii H, Uedaira H, Tahirov TH, Ogata K, Sarai A (2000) Cavityfilling mutations enhance protein stability by lowering the free energy of native state. J Phys Chem B 104:3705–3711 12. Ishikawa K, Nakamura H, Morikawa K, Kanaya S (1993) Stabilization of Escherichia coli ribonuclease HI by cavity-filling mutations within a hydrophobic core. Biochemistry 32:6171–6178 13. Eijsink VGH, Dijkstra BW, Vriend G, van der Zee JR, Vettman OR, van der Vinne B, van den Burg B, Kempe S, Venema G (1992) The effect of cavity-filling mutations on the thermostability of Bacillus stearothermophilus neutral protease. Protein Eng 5:421–426 14. Akasako A, Haruki M, Oobatake M, Kanaya S (1997) Conformational stabilities of Escherichia coli RNase HI variants with a series of amino acid substitutions at a cavity within the hydrophobic core. J Biol Chem 272:18686–18693 15. Kono H, Saito M, Sarai A (2000) Stability analysis for the cavity-filling mutations of the Myb DNA-binding domain utilizing freeenergy calculations. Proteins 38:197–209 16. Ohmura T, Ueda T, Ootsuka K, Saito M, Imoto T (2001) Stabilization of hen egg white lysozyme by a cavity-filling mutation. Protein Sci 10:313–320

59

17. Tanaka M, Chon H, Angkawidjaja C, Koga Y, Takano K, Kanaya S (2010) Protein core adaptability: crystal structures of the cavity-filling variants of Escherichia coli RNase HI. Protein Pept Lett 17:1163–1169 18. Shortle D, Stites WE, Meeker AK (1990) Contributions of the large hydrophobic amino acids to the stability of staphylococcal nuclease. Biochemistry 29:8033–8041 19. Ardejani MS, Chok XL, Foo CJ, Orner BP (2013) Complete shift of ferritin oligomerization toward nanocage assembly via engineered protein-protein interactions. Chem Commun 49:3528–3530 20. Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, Shindyalov IN, Bourne PE (2000) The Protein Data Bank. Nucleic Acids Res 28:235–242 21. Eric FP, Thomas DG, Conrad CH, Gregory SC, Daniel MG, Elaine CM, Thomas EF (2004) UCSF chimera - a visualization system for exploratory research and analysis. J Comput Chem 25:1605–1612 22. Schymkowitz J, Borg J, Stricher F, Nys R, Rousseau F, Serrano L (2005) The FoldX web server: an online force field. Nucleic Acids Res 33:W382–W388 23. Binkowski TA, Naghibzadeh S, Liang J (2003) CASTp: Computed Atlas of Surface Topography of proteins. Nucleic Acids Res 31:3352–3355 24. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment search tool. J Mol Biol 215:403–410 25. Larkin MA, Blackshields G, Brown NP, Chenna R, McGettigan PA, McWilliam H, Valentin F, Wallace IM, Wilm A, Lopez R, Thompson JD, Gibson TJ, Higgins DG (2007) Clustal W and Clustal X version 2.0. Bioinformatics 23:2947–2948 26. O’Neil KT, DeGrado WF (1990) How calmodulin binds its targets: sequence independent recognition of amphiphilic [alpha]-helices. Trends Biochem Sci 15:59–64 27. Gellman SH (2002) On the role of methionine residues in the sequence-independent recognition of nonpolar protein surfaces. Biochemistry 30:6633–6636 28. Hu D, Qin Z, Xue B, Fink AL, Uversky VN (2008) Effect of methionine oxidation on the structural properties, conformational stability, and aggregation of immunoglobulin light chain LEN. Biochemistry 47:8665–8677

Chapter 6 Recombinant Expression and Purification of “Virus-like” Bacterial Encapsulin Protein Cages W. Frederik Rurup, Jeroen J.L.M. Cornelissen, and Melissa S.T. Koay Abstract Ultracentrifugation, particularly the use of sucrose or cesium chloride density gradients, is a highly reliable and efficient technique for the purification of virus-like particles and protein cages. Since virus-like particles and protein cages have a unique size compared to cellular macromolecules and organelles, the rate of migration can be used as a tool for purification. Here we describe a detailed protocol for the purification of recently discovered virus-like assemblies called bacterial encapsulins from Thermotoga maritima and Brevibacterium linens. Key words Virus-like assemblies, Protein cages, Virus purification, Bacterial encapsulins, Nanotechnology

1

Introduction Protein cages such as ferritins [1–9], viruses [5, 10–12], bacteriophages, bacterial encapsulins [13, 14], and the recently discovered bacterial microcompartments [6–8, 14–19] are hollow protein-based assemblies of various sizes that are found in nature and play a crucial role in biological catalysis, mineralization, molecular storage, detoxification, and gene delivery. Icosahedral capsids are assembled according to the Caspar-Klug quasi-equivalence theory in which 60 N subunits (where N is the triangulation (T) number) are symmetrically arranged as pentamers and hexamers to form the final icosahedron [20]. The smallest icosahedral virus is composed of 60 protein subunits arranged as 12 pentamers to form a T ¼ 1 capsid assembly. Owing to their diverse role in nature, viruses and virus-like assemblies are increasingly used in materials science, engineering, and nanotechnology, as tools and building blocks for controlled catalysis, nanoparticle synthesis, and electronics and as molecular cargo delivery systems [4, 10, 21–24]. In the past, prokaryotes were thought to lack subcellular organization and compartmentalization; however, the recent discovery of bacterial

Brendan P. Orner (ed.), Protein Cages: Methods and Protocols, Methods in Molecular Biology, vol. 1252, DOI 10.1007/978-1-4939-2131-7_6, © Springer Science+Business Media New York 2015

61

62

W. Frederik Rurup et al.

Fig. 1 Electron density surface mapping of the encapsulin from (left) Thermatoga maritima and (right) Brevibacterium linens. The native cargo proteins, ferritin-like protein and dye-decolorizing peroxidase, respectively, are shown as insets

encapsulins and microcompartments, or so-called primitive organelles, has since provided first evidence that subcellular organization is important for even simple organisms [6–8]. For example, the 24 nm bacterial encapsulins from Thermotoga maritima and Brevibacterium linens are similar in morphology and size diameter to plant-based viruses (Fig. 1) [14]. The interior cavity of the B. linens bacterial encapsulin contains a dimer of trimers of a dyedecolorizing peroxidase (DyP), whereas the bacterial encapsulins from T. maritima contain a dimer of pentamers of a ferritin-like protein (Flp) and are involved in bacterial catalysis and/or mineralization, respectively. However, although morphologically similar to small icosahedral plant-based viruses, both structural and gene homology studies suggest that encapsulins are of non-viral origin, due to the absence of any genes of viral origin within the vicinity of the encapsulin gene [14, 25]. Bacterial encapsulins are much more stable against temperature, pH, ionic strength, and chaotropes than their virus-based counterparts and therefore exhibit enormous potential for future applications of nanotechnology. This chapter provides a detailed description for the recombinant expression and purification of bacterial encapsulins in Escherichia coli (Fig. 2) [3]. For both non-viral and viral assemblies, the use of ultracentrifugation has significantly improved the purification and isolation of such assemblies [3]. Ultracentrifugation takes advantage of the relative size to density migration of virus-like particles and protein cages through a gradient of viscous liquid, typically sucrose or cesium chloride [3]. Since virus-like particles and protein cages have a unique size compared to cellular macromolecules and organelles, the rate of migration can be used as a tool for purification.

Purification of Virus-like Assemblies

63

Fig. 2 Schematic diagram for the recombinant expression and purification of bacterial encapsulins

2

Materials Antibiotics: Stock 100 mg/mL ampicillin sterilized by 0.2 μm filtration in ultrapure sterile H2O. Stock 34 mg/mL chloramphenicol prepared in pure ethanol. Bacterial growth medium: Autoclave 10 g bacterotryptone, 5 g yeast extract, 5 g NaCl made up to 1 L deionized water at 120  C for 20 min. Encapsulin buffer: 20 mM Tris-Cl, 150 mM NH4Cl, 20 mM MgCl2, 1 mM β-mercaptoethanol, pH 7.5. Specialist equipment: Thermo Scientific Sorvall WX80 ultracentrifuge T 865 rotor (30 mL polycarbonate tubes) and Surespin 630/36 rotor (38.5 mL polyclear tubes).

3

Methods 1. Inoculate 5–7 mL of sterile LB medium with E. coli Rosetta (DE3) cells containing pET21a plasmid vector expressing either B. linens or T. maritima (see Note 1). Add ampicillin (stock 100 mg/mL, 5–7 mL) and chloramphenicol (stock 34 mg/mL, 5–7 mL) and grow 18 h (overnight) at 37  C with continuous shaking. 2. Prepare 0.5 L of sterile LB medium in 2 L culture flasks for bacterial expression the next step. 3. The next morning, add the overnight cultures to inoculate 0.5 L sterile LB medium containing ampicillin (stock 100 mg/mL, 50 mL) and chloramphenicol (stock 34 mg/mL, 50 mL) and grow at 37  C with continuous shaking until OD600 reaches 0.6–0.8. 4. Allow the bacterial cells to cool slowly to 22  C with continuous shaking (see Note 2).

64

W. Frederik Rurup et al.

5. Once the bacterial cells are cooled, add isopropyl β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 0.1 mM and grow 18–22 h (overnight) at 22  C with continuous shaking. 6. The following day, pellet the bacterial cells by centrifugation at 10,000  g for 15 min at 4  C. 7. Resuspend the bacterial cells in 10 mL encapsulin buffer (see Note 3). 8. Transfer the resuspended cells to a 50 mL Falcon tube. 9. Keeping the cells cooled in an ice box, insert an ultrasonic probe into the suspended bacterial cell mixture and lyse the bacterial cells at full amplitude with full power for 2 min. 10. Add DNAse and RNAse (10 μL of each nuclease) to the lysed cells and incubate for 2 h at 4  C with gentle mixing. 11. Transfer the lysed cells into ultracentrifuge tubes, weigh, and tare to g ; for ^λN  0 < λN gN þ 2 N AL ΠN ¼ (49) 1 2 > > : λN ; for ^λN > 0; 2kN ^ with λN ¼ λN þ kN gN . For tangential contact we define a tangential gap function g T as the displacement of the node in the plane tangent to the hemisphere at the contact point, and a tangential Lagrange multiplier λT representing the friction force. The tangential contact potential is defined as: 8 kT > > < λT
g T þ g T
g T ; for jj^λT jj  μ^pN 2 ΠAL (50) ¼ T 1 > > : ðjjλT jj μ^pN Þ2 ; for jj^λT jj > μ^pN : 2kT For the state of no contact (^pN > 0), we have ΠAL ¼ T

1 jjλT jj2 ; 2kT

8^λT :

(51)

These augmented energy terms penalize the gap functions quadratically, which effectively leads to a linear constitutive response of the contact interface. The contact forces from the penalty terms are however supplemented directly by the Lagrange multiplier terms. In order to compute the Lagrange multipliers, we adopt the Uzawa algorithm, which updates the multipliers iteratively, holding them constant during each iteration. Based on the forces calculated from the penalty terms, the update from contact iteration i to i + 1 is

Computational Mechanics of Viral Capsids i λiþ1 N ¼ λN þ kN gN ;

λiþ1 ¼ λiT þ kT g T : T

159

(52)

To accelerate convergence of the Uzawa update, we also increment the values of the penalty parameters. Initially, the penalty parameters are chosen to be comparable with the stiffness of the shell, and subsequently they are increased geometrically at each contact iteration: iþ1 i ¼ 10  kN ; kN

3

kTiþ1 ¼ 10  kTi :

(53)

Some Case Studies and Examples In this section we apply the proposed modeling framework to analyze the mechanical response of viral capsids to external forces. In the examples presented here we highlight the capabilities and key practical aspects (e.g., choice of parameters governing the mesh generation) of our modeling framework. While a large number of viruses have been indented, here we select as our primary test cases the cowpea chlorotic mottle virus (CCMV), the hepatitis B virus (HBV), and the ϕ29 bacteriophage. We will also extend the application of the proposed modeling framework to the analysis of non-viral macromolecules and present the indentation of microtubules. These several applications will highlight the high flexibility and robustness of the method. In particular, the study of microtubules indentation is included to illustrate how the presented approach is applicable to general geometries and protein structures, which could also be significantly different from viral shells. We will conclude describing several applications of the thin-shell model, which offers complementary deep insights in the mechanical response of viral capsids to indentation. We note that, although the results presented in the following are all related to the study of macromolecules indentation, the same numerical models could be used to study the response of viral capsids to tensile forces, providing more insights into the capsids assembly process. However, indentation experiments are more common due to the simplicity of contact loading without binding the shell to the tip or substrate, and therefore are better suited to illustrate the new methods presented herein. In interpreting the following results and applications, it is useful to note a clear distinction between the proposed and other methods. In several two- and three-dimensional approaches, simplified continuum models have been used to study viral nanoindentation [5–7, 9–11, 42–44]. The similarity between all of these models is the use of both geometric and constitutive homogeneity. Although these geometrically homogeneous elastic models are able to capture the linear response of the capsids as observed experimentally, the influence of geometric heterogeneity on the mechanical

160

Melissa M. Gibbons et al.

response is not clear. On the contrary the proposed 3-D modeling framework captures the capsid geometric heterogeneity, which may become especially important at the large indentations corresponding to the capsid failure. 3.1

CCMV

The cowpea chlorotic mottle virus (CCMV) is a roughly spherical plant virus with a diameter of  28 nm, and triangulation number T ¼ 3 as determined by the Caspar–Klug classification scheme for spherical viruses [39]. Having outlined the procedures for creating finite element meshes of macromolecules from both atomic and electron density data (Subheading 2.1), it now becomes important to assess the effects of each of the parameters in the model, such as the structural and mesh resolution, to determine what features of the topography are necessary to capture a heterogeneous mechanical response. Given the large amount of information available about the structural response of CCMV to nanoindentation (e.g. [6, 7, 29, 42]), this viral capsid serves as an excellent basis for parametric studies. The atomic-level structure of the native CCMV capsid has been solved with X-ray crystallography, and the atomic coordinates of the capsid are available from VIPERdb (PDB-ID 1CWP [45]). This kind of structural input, as described above, allows for the most flexibility in terms of the final mesh produced. In order to determine the relative importance of mesh resolution δ and structural resolution σ (Subheading 2.1.1) on the capsid mechanical response, parametric studies on their effects and comparison to a spherical model are performed. This will allow for the determination of the coarsest heterogeneous mesh that can be used while still capturing mechanical details not present in the homogeneous spherical models. Figure 6 shows a grid of nine surface meshes that will be inputs to the series of parametric analyses, with one axis showing three values of the standard deviation σ, and the other axis showing three values of the grid spacing δ. Columns represent meshes with constant underlying structural resolution, while the mesh varies vertically from coarse to fine. Horizontal rows represent meshes with constant grid spacing, and these meshes tend to have approximately the same number of degrees of freedom. The diagonal represents ideal combinations of δ and σ. As the standard deviation σ decreases from 12 to 6 A˚, the level of structural detail in the underlying global density field increases and a much higher mesh resolution is necessary to capture the details of the topography. Meshes above the diagonal are created with grid spacings δ that are too large to capture the underlying amount of details, while meshes below the diagonal contain a larger number of degrees of freedom without capturing significantly more structural details. The colored rectangles in Fig. 6 surround the sets of meshes that will be used in the parametric studies to determine which meshes include the most important structural details. The input parameters and mesh details for all nine meshes are given in Table 1.

Fig. 6 Nine possible combinations of grid spacing δ and decay parameter σ produce meshes that range from low-resolution, coarse representation of the capsid, to high-resolution, fine representation. As δ decreases, the fineness of the mesh increases, as shown on the ordinate axis, while decreasing the value of σ increases the structural resolution, as shown on the abscissa. The blue rectangle on the diagonal encloses meshes produced by combinations of δ and σ that produce high-quality meshes with enough mesh detail to capture the nonuniform topography at the given structural resolution Table 1 Mesh parameters and statistics for the meshes of Fig. 6. Grid spacing δ, decay parameter σ, radius r, and thickness t are given in A˚ Inputs

Finite element mesh

σ

δ

Isoval.

ravg

tavg

Nodes

Elements

12

8

120

120

27.3

25,946

129,016

12

6

120

120

27.1

54,946

281,286

12

4

120

120.2

27.1

149,460

788,528

9

8

50

120.2

27.3

28,144

140,760

9

6

50

120.4

27.0

59,313

303,576

9

4

50

120.8

26.9

161,007

847,701

6

8

13.5

120.5

27.2

31,294

156,863

6

6

13.5

121

27.0

69,041

354,478

6

4

13.5

121.1

26.5

195,700

1,038,171

Melissa M. Gibbons et al.

σ=1.2 nm σ=0.9 nm

1

σ=0.6 nm Contact Force (nN)

162

0.8

0.6

0.4

0.2

0 0

1

2

3

4

5

6

Indentation (nm)

Fig. 7 Force-indentation curves are shown for the three meshes shown in the horizontal green rectangle of Fig. 6. The three meshes are created with equal grid resolution, using δ ¼ 8A˚, and varying values of σ. All indentations are performed on the twofold symmetry orientation

The first parametric study is performed to determine the relative importance of the underlying structural resolution at a constant value of the mesh resolution. The horizontal green box in Fig. 6 surrounds three meshes created with the same value of grid ˚ , at varying levels of structural resolution. While the spacing, δ ¼ 8A three meshes have approximately the same number of degrees of ˚ mesh has about 31,000 nodes and the freedom (the σ ¼ 6A ˚ σ ¼ 12A mesh has about 26,000 nodes), it is clear that increasing the resolution of the underlying structural details causes the final ˚ to incorporate a higher number of details; this mesh with σ ¼ 6A becomes strikingly clear in the much higher thickness variation over the surface. It is noted that the average thickness is equal for all three meshes. It appears that the underlying structural resolution has little to no impact on the final results since Fig. 7 shows that the three force-indentation curves are coincident. This indicates that a coarse mesh with a higher structural resolution does not capture any additional feature that alter the response, despite the apparently more detailed surface. These results also indicate that decreasing σ is unnecessary unless δ is also decreased to capture the details. In the next parametric study, indentations are performed on three meshes created at a constant value of σ with varying mesh resolution as determined by the grid spacing δ. These meshes will have the exact same underlying structural resolution, and therefore the surfaces are nearly equivalent. The three meshes to be indented are surrounded by the vertical red box in Fig. 6, corresponding to

Computational Mechanics of Viral Capsids

163

Grid=0.8 nm 1

Grid=0.6 nm

Contact Force (nN)

Grid=0.4 nm 0.8

0.6

0.4

0.2

0 0

1

2

3

4

5

6

Indentation (nm)

Fig. 8 The validity of using coarse finite element meshes is studied by creating meshes of the CCMV capsid with varying fineness, and comparing their response. Force-indentation curves are shown for the three meshes in the red vertical rectangle of Fig. 6. The three meshes are created with equal structural resolution, using σ ¼ 12 A˚, and varying values of the grid spacing. All indentations are performed on the twofold symmetry orientation

˚ down to 4 A˚. Figure 8 shows that σ ¼ 12 A˚ and δ varied from 8 A the three force-indentation curves are very close to one another, although not coincident like the curves at a constant grid spacing, discussed above. The decrease in grid spacing allows for the capture of slightly more detail in the surface, as evidenced by the surface ˚ has both a higher meshes. For instance, the mesh at δ ¼ 4 A maximum outer radius and lower minimum radius while a smaller ˚ . These differences average thickness than the mesh at δ ¼ 8 A contribute to the slight softening evident as the grid spacing decreases. Additionally, the coarser meshes have a larger discretization error, while the finer meshes are much closer to convergence. However, the overall responses of the coarse and fine meshes agree quite well, and it is therefore unnecessary to use an overly fine mesh. Four-noded tetrahedral elements are known to be artificially stiff, so in order to assess the quantitative effects of this stiffening, a quadratic ten-noded tetrahedral mesh of the coarse representation ˚ , δ ¼ 8 A˚) is created. Both meshes are created from the (σ ¼ 12 A same surface triangulation, and therefore have the exact same number of elements, while the quadratic mesh has, as expected, significantly more nodes. As Fig. 9 shows, the difference in the mechanical response obtained using quadratic versus linear tetrahedral elements is not large. At an indentation of 3 nm, the linear

Melissa M. Gibbons et al. 0.45 Linear, σ=1.2 nm, δ=0.8 nm 0.4

Quadratic, σ=1.2 nm, δ=0.4 nm

0.35 Contact Force (nN)

164

0.3 0.25 0.2 0.15 0.1 0.05 0

0

0.5

1

1.5

2

2.5

3

Indentation (nm)

Fig. 9 Linear versus quadratic finite element meshes. Force-indentation curves are shown for the coarsest mesh in Fig. 6. The meshes are both created from the same surface triangulation. Both indentations are performed on the twofold symmetry orientation

mesh is approximately 11 % stiffer than the quadratic mesh. While this might affect the material property estimation, the effect is not profound enough to justify the computational cost of quadratic meshes. Additionally, it is argued that while the material property estimation would change slightly, these estimates are affected just as strongly by a number of other model parameters. One remaining question is the importance of the structural details that have been smoothed out by using a large value of σ (12 A˚) or a large value of δ (8 A˚). To address this question, three meshes of CCMV are created with varying values of σ and δ. In this latter parametric study, the grid spacing δ is adapted to fully capture the structural details present at higher structural resolution σ. The three meshes used for this study are surrounded by the blue diagonal box in Fig. 6. It is clear that the surface of the fine mesh is able to capture detailed topographic features, such as the placement of individual protein subunits within the pentamers and hexamers. The variation in the thickness over the surface is also more pronounced. But decreasing δ in order to capture the structural details causes the number of nodes to increase sharply; the coarsest mesh ˚ , σ ¼ 12 A˚) has 25,946 nodes, while the finest mesh (δ ¼ 8 A ˚ ˚ ) has 195,700 nodes, a factor of over seven in (δ ¼ 4 A, σ ¼ 6 A the number of degrees of freedom. Notably, the finest mesh has still fewer nodes than atoms (see Table 1). However, the resulting forceindentation curves shown in Fig. 10 present only moderate variations, with the finer meshes again producing softer curves. The

Computational Mechanics of Viral Capsids

165

σ=1.2 nm, Grid=0.8 nm 1

σ=0.9 nm, Grid=0.6 nm

Contact Force (nN)

σ=0.6 nm, Grid=0.4 nm 0.8

0.6

0.4

0.2

0 0

1

2

3

4

5

6

Indentation (nm)

Fig. 10 In order to determine the importance of the amount of structural details in the model, meshes with varying levels of structural and mesh resolution are indented (the meshes enclosed by the diagonal blue box in Fig. 6). Note that the fineness of the mesh increases as the structural resolution increases so as to accurately capture the increasing details. Force-indentation curves are shown for the three meshes. All indentations are performed on the twofold symmetry orientation

softening is slightly more amplified than in Fig. 8, which may be attributed to the slight decrease in average thickness as the fineness increases. This is a consequence of the manual choice of isovalues, which lead to some small but noticeable imprecision in the final mesh dimensions. This result is consistent with those shown in [42]. With these results in hand, it appears clear that, within the range of structural resolutions that capture a fair amount of topographic details while still having fewer degrees of freedom than the atomic structure, the coarsest mesh does indeed capture the heterogeneous response of the capsid when indented by an AFM tip. This is reasonable, as the mechanical probe itself (the AFM tip) is fairly coarse and is the means by which the global response of the capsid is measured. While σ can be reduced to represent the van der Waals radius of an atom, the corresponding δ needed to capture such a structure would produce a mesh with far more nodes than atoms, and would become computationally infeasible to indent. Additionally, given the results above, it is not clear what additional information could be gleaned from such a fine model. However, a mesh finer than the coarsest model presented in Fig. 6 can become important in capturing the virus mechanical response if indentation of the capsid is performed using AFM tips with radius of a few

Melissa M. Gibbons et al. 2 Two−fold

1.8

Three−fold

Contact Force (nN)

166

1.6

Five−fold

1.4

Sphere

1.2 1 0.8 0.6 0.4 0.2 0 0

2

4

6

8

10

12

Indentation (nm)

Fig. 11 Force-indentation curves for a spherical model, the solid black line, and the low-resolution nonuniform model from the upper left-hand corner of Fig. 6. The low-resolution nonuniform model was indented on all three icosahedral symmetry orientations

nanometer or less [46, 47]. With the advances in AFM tip manufacturing and indentation techniques, we expect that a larger amount of topographical details should be included to capture more accurately the interaction between the nanometer size AFM tip and the virus shell. Although it is now clear that among heterogeneous meshes, coarse, low-resolution models accurately capture the mechanical response to nanoindentation, it is not clear how much the response varies from that of a homogeneous spherical model. That is, if meshes from both low and high structural resolution produce force-indentation curves that are nearly equal, would the response of a spherical model match as well? For indentations up to approximately the value of the thickness of the CCMV capsid, the spherical and low-resolution nonuniform models produce equivalent results along all three icosahedral symmetry orientations. This can be seen in Fig. 11, where the force-indentation curve for the spherical model is plotted along with the force-indentation curves for the nonuniform model (for comparison, indentation is performed on all three icosahedral symmetry orientations). The spherical model was given a thickness and radius equal to the average thickness and average radius of the low-resolution nonuniform model seen in the upper left-hand corner of Fig. 6. However, after about 3 nm indentation, which is near the value of the thickness of the capsid, the response of the spherical model remains linear (which is similar to

Computational Mechanics of Viral Capsids

167

the results of Gibbons and Klug [42]), while the force-indentation curves of the nonuniform model begin to diverge due to variances in the contact geometry, which are explained more fully in [29]. These details are not adequately captured by a spherical model, which can neither distinguish among indentations along different icosahedral symmetry orientations. 3.2

HBV

Hepatitis B is a DNA human virus that, like CCMV, can assemble into icosahedral shells from protein components, producing structures that are indistinguishable from the intact virus [48]. The more common form of the capsid is a T ¼ 4 structure composed of 240 protein subunits, but a T ¼ 3 structure with 180 protein subunits can also be formed [49]. Nanoindentation experiments by AFM have been performed on both the T ¼ 3 and T ¼ 4 form of the capsid [15]. It was found that the T ¼ 3 capsid responds quite linearly up to indentations equaling its outer radius. The T ¼ 4 capsid has instead a pronounced nonlinear response, stiffening at early indentations and then softening as indentation approaches half the total height of the capsid. The crystal structure of Hepatitis B in the T ¼ 4 form has been solved [50], and can be downloaded from the VIPERdb database [4] (PDB-ID 1QGT). The less common T ¼ 3 form of the capsid has also been crystallized, and a model was received directly from A.C. Steven. Hepatitis B is a challenging structure to mesh, given that there are porous regions at multiple points on the surface and the dimer of the capsid forms a four-helix bundle, giving rise to “spikes” on the surface of the capsid [50]. These two factors combine to create a large amount of thickness variation across the outer surface akin to a series of valleys and peaks, which, for instance, is much more dramatic than in the CCMV capsid. In order to produce a coarse mesh, the decay parameter σ is chosen ˚ , such that the structural resolution is fairly low. In equal to 16 A contrast, to somewhat accurately capture the highly variegated surface, a higher value of grid spacing is necessary. For the T ¼ 3 mesh, the grid spacing is 6 A˚, while for the T ¼ 4 mesh the grid ˚ . The isovalue is chosen such that the thickness and spacing is 7. 4 A the size of the pores of the T ¼ 4 match the published values of Wynne et al. [50]. The only feature lacking in the final finite element mesh is the accurate detail of the “spikes,” which are to a great degree smoothed out at the chosen value of σ. Since the same protein subunits are used to create the T ¼ 3 structure, it is argued that while the radius will decrease, the average thickness of the T ¼ 3 capsid should equal that of the T ¼ 4 capsid. Therefore, the isovalue chosen for the T ¼ 3 mesh is such that the thicknesses of the two meshes are equivalent. The structural data and final mesh statistics for both models are given in Table 2. The finite element meshes are shown in Fig. 12, where the size differences between the T ¼ 3 and T ¼ 4 structures are apparent.

168

Melissa M. Gibbons et al.

Table 2 Data on the atomic structures and tetrahedral meshes for the T ¼ 3 and T ¼ 4 Hepatitis B capsids. davg for T ¼ 3 structure not published, other values determined directly from data. Note that the atomic structures do not contain hydrogen atoms. Diameters d, grid spacing δ, and decay parameter σ are given in A˚ Atomic structure

Finite element mesh

Capsid

Atoms

davg

dmin

dmax

δ

σ

Nodes

Elements

T¼3

205,080

N/A

220

308

6

16

45,769

221,089

T¼4

273,600

291

253

352

7.4

16

36,985

173,621

Fig. 12 The finite element meshes for the Hepatitis B capsid are shown for the T ¼ 3 (left) and T ¼ 4 (right) structures. The shape of an icosahedron is overlaid to highlight the differences in the contact at the various symmetry orientations. Both capsids are shown at the same scale, highlighting the size differences

The shape of an icosahedron is overlaid onto the meshes to show the differences in the contact geometry at the various symmetry orientation sites. For example, the twofold symmetry orientation, which is located at the center of any edge in the icosahedron, can be most easily seen in the figure at the vertical edge in the center of each structure. The twofold symmetry orientation for the T ¼ 3 capsid falls between neighboring sets of hexamers on either side, which form the threefold symmetry orientation sites (at the center of the triangular faces), and neighboring pentamers above and

Computational Mechanics of Viral Capsids

b

1 Two−fold

Contact Force (nN)

0.9

1 Two−fold

0.9

Three−fold

0.8

Five−fold

0.8

0.7

Average

0.7

Contact Force (nN)

a

0.6 0.5 0.4

T=3

0.3

169

Three−fold Five−fold Average

0.6 0.5 0.4

T=4

0.3

0.2

0.2

0.1

0.1 0

0 0

0.1

0.2

0.3

0.4

0.5

0.6

Relative Indentation (δ/R)

0.7

0.8

0.9

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Relative Indentation (δ/R)

Fig. 13 Simulated force versus indentation curves for the T ¼ 3 (a) and T ¼ 4 (b) HBV structures. Individual curves are shown for each symmetry orientation, and their weighted average is plotted as a solid black line in each graph. Indentation is plotted relative to the average outer radius of the meshes

below in the figure, which form the fivefold symmetry orientation sites (at the vertices of the icosahedron). The twofold symmetry orientation for the T ¼ 4 capsid is located such that indentation contact is made on a hexamer. The variations in the contact geometry cause the force-indentation curves to be quite distinct. Nonlinearities due to stiffening and softening are more pronounced and emerge earlier than in the indentation of CCMV capsid, but are caused by the same key factors [29]. Specifically, the stiffening events are related to discrete increases in the contact area, which allow for the capsid to more evenly distribute the load, while the softening events are related to a transition from bendingstretching to pure bending, which is an easier mechanism of deformation. Because the contact geometries are not equivalent between the T ¼ 3 and T ¼ 4 structures, the contact force curves for the three symmetry orientations do not follow the same trends. For example, at low values of indentation, the threefold orientation of the T ¼ 3 structure is the stiffest, while the twofold orientation is the softest (Fig. 13a). This trend is exactly reversed in the indentation of the T ¼ 4 structure (Fig. 13b). These initial responses are directly linked to the initial contact area; the higher the initial contact area, the stiffer the initial response since the capsid is able to distribute the loading throughout its entire volume more efficiently. In the case of the T ¼ 3 capsid, the initial contact area of the twofold orientation rests on a single spike, the initial contact area of the fivefold orientation rests on the five spikes composing a pentamer, and the initial contact area of the threefold orientation rests on the six spikes composing a hexamer. These differences can be seen in the top view of the contact force areas shown in Fig. 14a at indentations of

170

Melissa M. Gibbons et al.

Fig. 14 Top views of the contact area (showing the magnitude of the nodal contact force, in piconewtons) for (a) the T ¼ 3 capsid and (b) the T ¼ 4 capsid at several points during the indentation

δ=R ¼ 0:2. Due to the similar amount of initial contact area, it would be expected that the force-indentation curves of the threeand fivefold symmetry orientations would be quite close, and indeed, they are initially. The same observations hold true for the initial indentation on the T ¼ 4 capsid, where the initial contact area of the two- and fivefold symmetry orientations are both the highest (indentation begins on a hexamer and pentamer, respectively) and closest in value, as seen at an indentation of δ=R ¼ 0:2 in Fig. 14b. As expected, the initial contact force curves for the twoand fivefold symmetry orientations are quite near to each other, and noticeably stiffer than the initial force response for the threefold symmetry orientation, where initial contact occurs on three spikes from the three surrounding hexamers.

Computational Mechanics of Viral Capsids

171

Fig. 15 Cut side views of the von Mises stress (shown in MPa) at several points during the indentation, illustrating the transition to a bending dominated motion. This is highlighted in the upper capsomers, which noticeably rotate while undergoing very little deformation. The through-thickness state of stress is also indicative of bending, with nearly zero stress in the center of the capsid. (a) T ¼ 3. (b) T ¼ 4

At higher values of indentation, discrete increases in the contact area cause stiffening of the response, and transition from bendingstretching to bending dominated deformation causes softening of the response. These trends can be seen for both the T ¼ 3 and T ¼ 4 structures in Figs. 14 and 15. Since the general trends hold true for both structures and all symmetry orientations, only indentation of the T ¼ 3 capsid on its twofold symmetry orientation will be analyzed in detail. As discussed above, initial indentation on the

172

Melissa M. Gibbons et al.

twofold symmetry orientation of the T ¼ 3 capsid is quite soft due to the contact area being limited to a single spike. At a relative indentation value of δ=R ¼ 0:25, the upper portion of the capsid has deformed such that the two nearest spikes from the adjacent pentamers come into contact with the AFM tip, causing a rise in the force response. At much later points in the indentation, after δ=R ¼ 0:7, the curve is again observed to stiffen, which can be attributed to contact being made with four additional spikes from the adjacent hexamers. This final contact geometry is seen in Fig. 14a for δ=R ¼ 0:8. Preceding the final increase in contact area, the contact force curve is observed to be undergoing a distinct softening. As seen in the cut view of the capsid von Mises stress in Fig. 15a, by an indentation of δ=R ¼ 0:7, the upper portion of the capsid has become quite flat, and bending in this regime is a much softer mode of deformation. While the upper capsomers have rotated substantially, the deformation is mostly still localized to the center of the capsid. The black curves in Fig. 13 represent the weighted average of the individual symmetry orientation curves. In an icosahedral structure, there are 12 fivefold orientation sites, 20 threefold orientation sites, and 30 twofold orientation sites. Thus each curve cannot be given equal weight. Experimentally, the height profiles indicate which orientation is being indented, and the number of indentations on each orientation roughly follow the distribution of 12–20–30 stated above [6]. The weighted average curves for the T ¼ 3 and T ¼ 4 structures are directly compared in Fig. 16. The calibrated value of the Young’s modulus of the Hepatitis B capsid is E ¼ 250 MPa. This value causes both the T ¼ 3 and T ¼ 4 curves in Fig. 16 to match the experimental response quite well, which is a validating result as the material in both capsids is the same, and should arguably possess the same material properties. The simulated force-indentation curves manage to capture striking similarities with the experiments. Namely, the T ¼ 3 response, while not as linear as the experiments, is less nonlinear and slightly softer than the T ¼ 4 response. The T ¼ 4 response is nearly identical to the experimental response, with an initial stiffening followed by softening later in the indentation. 3.3 Bacteriophage ϕ29

ϕ29 is a bacteriophage with a double-stranded DNA genome, which is about 19,000 base pairs long. The ϕ29 capsid is comprised of 235 protein subunits, arranged into a prolate shell with ten hexamers in a central cylindrical region and two icosahedral end caps having a total of eleven pentamers and twenty hexamers, with T ¼ 3 quasi-symmetry [51]. ϕ29 has a complex assembly pathway that involves an intermediate structure, termed the prohead, which matures while the DNA is packed into the capsid by a head–tail connector. The head tail connector, in combination with a cyclic hexamer of prohead RNA (pRNA), forms a motor situated at the

Computational Mechanics of Viral Capsids

173

1.2 T=3 T=4

Contact Force (nN)

1

0.8

0.6

0.4

0.2

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Relative Indentation (δ/R)

Fig. 16 Weighted average force-indentation curves for both the T ¼ 3 and T ¼ 4 capsids. Direct comparison is possible due to plotting the indentation relative to the radius of the structure, which is different for the two capsids. Calibration has been performed on these curves with E ¼ 250 MPa such that the simulated response quantitatively matches the experimental response

apex of the capsid where the twelfth pentamer would be. Both the prohead and mature capsids are made of the same subunits and are formed into the same elongated icosahedral structure. The main difference between the prohead and the mature capsid lies in the prohead missing the tail complex, which is made of a lower collar, distal tail knob, and several appendages. The tail allows the doublestranded DNA to pass through during DNA release [51]. The minor outer diameter (while on its side) of the capsid is approximately 45 nm, and the major outer diameter without the tail complex is approximately 54 nm. The average thickness of the capsid is  1. 6 nm. The structure of the ϕ29 prohead and mature capsids has been solved using cryo-EM techniques [52], and serves as the basis for highlighting the use of the meshing method from electron density data (Subheading 2.1.2). The empty ϕ29 prohead was the subject of an AFM nanoindentation study by Ivanovska et al. [5], and it was found that the capsid, when indented on its side, responds linearly at indentations up to 30 % of its height. The linearity limit occurs at indentations near 12 nm, with a corresponding contact force of approximately 2. 8 nN. Between 12 and 15 nm of indentation, significant nonlinearities start to emerge and the response begins to soften. After this point, failure events occur, resulting in sudden drops in the contact force. Lowforce imaging reveals fracturing of the structure, indicating a material failure.

174

Melissa M. Gibbons et al.

The structural information of ϕ29 is known only from cryo-EM, and the density maps for several variants are available from EMBL [2]. The finite element mesh shown in Fig. 17 was ˚ resolution model of the empty, fiberless created using the 12. 7 A ϕ29 prohead, EMBL Accession code 1117 [52]. This is the same initial electron density data that was used to illustrate the meshing procedure, and as described in Subheading 2.1.2, the structural resolution of the data cannot be altered, leaving only the grid spacing and isovalue selections to the user. The initial grid spacing is 4. 27 A˚, which would lead to a mesh far too fine for the purpose of indentation. Therefore the grid is first smoothed using the Mapman smoothing filter [26] and then down-sampled by taking only every third grid point as input of the isosurfacing algorithm, effec˚ . By taking every third, tively increasing the grid spacing to 12. 8 A as opposed to every other grid point as in the example shown in Subheading 2.1.2, more of the surface detail is lost and the resulting surface mesh is noticeably coarser. The resulting surface mesh is finally smoothed using the VTK smoothing algorithm [25]. The final three-dimensional tetrahedral mesh is shown in Fig. 17, whereas the details of the atomic structure and final mesh are given in Table 3.

Fig. 17 Finite element mesh of the ϕ29 capsid. The top portion of the figure shows an outer view with highlighted elements edges, and the lower portion of the figure shows a cut view of the solid mesh (edges not shown)

Computational Mechanics of Viral Capsids

175

Table 3 Data on atomic structure and tetrahedral mesh of the ϕ29 capsid. Note, number of atoms and σ are N/A, as structural data are from electron density. Atomic data from Tao et al. [51]. Diameters d, thickness t, and grid spacing δ are given in A˚ Atomic structure

Finite element mesh

Capsid

Atoms

1 davg

2 davg

t

δ

σ

Nodes

Elements

ϕ29

N/A

524

436

16

12.8

N/A

36,645

179,529

Fig. 18 von Mises stress in the deformed ϕ29 mesh at 5 nm indentation: cut view showing the bendingdominated deformation mode (a), and top view highlighting the contact indentation forces (b)

The von Mises stress in the deformed capsid is shown at 5 nm indentation in Fig. 18a. Due to the larger size of the ϕ29 capsid, the indentation does not look as dramatic as it does for the smaller CCMV and Hepatitis B capsids. At 5 nm indentation, the deformation is localized to the bending of the upper portion of the capsid, and to a lesser degree, bending near the bottom portion, and is not transferred to the side regions as evidenced by the lack of stress in those areas. The top view of the contact force profile for the ϕ29 capsid at 5 nm indentation is shown in Fig. 18b. Due to the undulations in the surface, the actual contact area is quite small, and two central contact regions bear the brunt of the load. The resulting force-indentation curve, Fig. 19, shows that the response is linear, at least up to indentations of about three times the thickness. This is expected for several reasons. As evidenced by the contact force profile, up to the indentations shown here, there are no large or discrete changes in the contact area that would allow for better force distribution and lead to an increase in the response

Melissa M. Gibbons et al. 1.2

1

Contact Force (nN)

176

0.8

0.6

0.4

0.2

0 0

1

2

3

4

5

Indentation (nm)

Fig. 19 Contact force curve shown up to 5 nm indentation. The response is strikingly linear, and can be calibrated to match experimental data. The curve shown has been plotted using E ¼ 400 MPa

stiffness (as was seen, for instance, in all CCMV and Hepatitis B capsids). Moreover, the initial configuration of the capsid near the point of contact with the AFM is essentially flat. This is unlike the spherical capsids, which transition from curved to flat during the indentation, resulting in an apparent softening of the response. As can be seen in the view of the von Mises stress, Fig. 18a, the state of stress is most certainly bending dominated. These factors eliminate the two types of events that caused the spherical capsids to experience discrete stiffening and softening events in their force response. Given the simulated force-indentation curve, the Young’s modulus can be tuned such that the simulated contact force matches its experimentally measured counterpart in the linear regime. The estimated Young’s modulus is E ¼ 400 MPa, and the scaled contact force curve is shown in Fig. 19. The Young’s modulus value computed here is lower than the value estimated by Ivanovska et al. [5], which was E ¼ 1. 8 GPa, and lower than the estimate provided in [42], which was E ¼ 4. 5 GPa. However, the previous estimates are based on simplified modeling techniques, such as the use of point loads, and more importantly, homogeneous structures, which tend to cause an overestimation of the material properties. On the other hand, a likely factor in the apparent underestimation of the Young’s modulus in the foregoing is that the model used has a slightly larger than reported average thickness. As thicker capsids respond more stiffly, the resulting calibrated Young’s modulus is lower.

Computational Mechanics of Viral Capsids

3.4

Microtubules

177

Before turning our attention to the analyses based on the thin-shell model of viral capsids, we apply the 3-D bulk continuum formulation to study the nanoindentation of microtubules. The following analyses illustrate how the method applied so far to protein cages could be used seamlessly to study different geometries and protein structures. A microtubule is a non-viral macromolecular assembly that makes up one of the three main types of cytoskeletal filaments, the other two being actin and intermediate filaments [53]. Microtubules are assembled from multiple copies of the tubulin dimer, which polymerize end to end, creating a long protofilament. Typically, thirteen protofilaments are assembled into a hollow cylindrical structure with a diameter of about 25 nm, and microtubules are anywhere from 200 nm to several micrometers in length. As microtubules are one of the most important structural components of the cell, their response to radial indentation by AFM has been measured [8]. It was found that the microtubule responded linearly up to indentations equal to 15 % of their height, or  4 nm, with contact forces  0.3 nN. Indentations beyond this point produced sharp drops in the force response, and damage of the microtubule was observed. In order to create a three-dimensional finite element mesh of a microtubule, the crystal structure of the tubulin dimer (PDB-ID 1TUB [54]) was downloaded from the PDB. Given the dimer and several geometric parameters such as radius, number of protofilaments per microtubule, end to end distance between dimers, and the pitch between neighboring protofilaments, a microtubule can be generated. The number of protofilaments in a standard microtubule in vivo is thirteen, and in this configuration the protofilaments are paraxial [55]. There is a 9.46 A˚ pitch between neighboring protofilaments, which leads to a three monomer offset between the first and the last protofilament, and each monomer is approximately 4. 1 nm in length, leading to an end to end distance of 8. 2 nm. The microtubule is generated with 25 dimers per protofilament, making it approximately 200 nm long. This creates a structure that contains over two million atoms, resulting it the largest macromolecule meshed using the techniques developed in this work. Due to the large number of atoms and the desire for a relatively coarse mesh, the structural and mesh resolution are set to ˚ . Despite the low-resolution be low, with σ ¼ 15 A˚ and δ ¼ 10 A nature of the global density field and fairly large grid spacing, the structure is quite large, producing a fine mesh. However, the number of degrees of freedom in the three-dimensional mesh is still approximately 18 times fewer than the number of degrees of freedom in the atomic structure. The final tetrahedral mesh is shown in Fig. 20 and all the structural and mesh details are given in Table 4.

178

Melissa M. Gibbons et al.

Fig. 20 The finite element mesh for the microtubule is shown, without edges. The microtubule is created from 13 protofilaments, each containing 25 tubulin dimers Table 4 Data on atomic structure and the tetrahedral mesh of the microtubule. Diameters d, length l, grid spacing δ, and decay parameter σ are given in A˚ Atomic structure

Finite element mesh

Macromolecule

Atoms

out davg

in davg

l

δ

σ

Nodes

Elements

Microtubule

2,138,175

250

190

2,000

10

15

115,617

566,145

Indentation of the microtubule produces a linear response up to 6 nm, as shown in the force versus indentation curve reported in Fig. 21. It is noted that experimentally failure occurs soon after a  4 nm indentation. The contact force has been calibrated to match the experimental data described above, yielding a Young’s modulus E ¼ 140 MPa. Based on experimental measurements of the flexural rigidity of microtubules, Schaap et al. [8] estimate that the Young’s modulus of microtubules is between 0.04 and 1. 2 GPa. While the value estimated here is quite low, it does fall within this range. The stress state and displacements at 3 nm indentation corresponding to E ¼ 140 MPa are shown in Fig. 22. In agreement with the loading conditions, the ends are free of stress, and the displacement field of the mesh shows that only the center region is constrained, while the ends actually displace upward more than the plate displacement of 3 nm, indicating movement away from the plate. At 3 nm indentation, the stress in the microtubule is quite

Computational Mechanics of Viral Capsids

179

0.5 0.45

Contact Force (nN)

0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 0

1

2

3

4

5

6

Indentation (nm)

Fig. 21 The contact force curve is shown up to 6 nm indentation. The response is strikingly linear, and can be calibrated to match experimental data. The plotted curve has been scaled using E ¼ 140 MPa

Fig. 22 The microtubule is displayed at 3 nm indentation. (a) von Mises stress and displacement fields. (b) Two-dimensional cut through the center of the microtubule highlighting the internal state of stress

low everywhere except at the points of contact, and in the side of the microtubule directly below the AFM tip. The highest stress is directly beneath the area of contact with the AFM tip, which is where the microtubule fails experimentally. A two-dimensional slice of the cross-section in the center of the microtubule, Fig. 22b, shows the internal von Mises stress state. While the stress in the spherical viruses remains concentrated in the upper and lower portions of the capsid, at least at small indentations, the microtubule experiences noticeable amounts of stress in the middle of the

180

Melissa M. Gibbons et al.

cross-section. However, this state of stress is limited to the portion of the capsid directly underneath the AFM tip, due to the ends remaining free of any constraints. 3.5 Thin-Shell Inversion Buckling

The key mechanical feature of icosahedral shells formed by the Caspar–Klug construction (Subheading 2.3.4) is the compressive prestress locked into the fivefold disclinations at the icosahedral vertices. Lidmar et al. [38] showed that these stresses can incite an Euler-type buckling instability in flat sheets when the dimensionless Fo¨ppl–von Ka´rma´n number γ ¼ Y R2 =κ exceeds a critical value of γ  150. The interpretation is that when bending is more costly than stretching (κ large, Y and R small), a flat geometry is preferred at the expense of disclination stresses, but when stretching dominates, the shell will prefer to bend out of the plane to relax the disclination stresses. In closed shells this buckling mechanism drives a transition from spherical (bending dominated) shapes to faceted (stretching dominated) shapes as γ increases. Here we may interpret the background curvature of the closed shell as providing a natural bias that breaks symmetry and drives the icosahedral vertices to “pop outward” relative to a spherical state. In this way of thinking, it seems plausible that a force in the opposite direction might cause the vertices of a buckled icosahedral shell to invert, and “pop inward.” We have explored this inversion buckling hypothesis in the context of AFM nanoindentation experiments [7, 15]. Here we provide some generic results to illustrate the use of the methods described in Subheading 2.3. All simulations of capsid deformation begin by discretizing an icosahedron with the triangular subdivision finite elements (Fig. 23a) described in Subheading 2.3.3. Initially the capsid shape is computed by minimizing the combined bending and stretching potential energies reported in Subheading 2.3.2. In this initialization phase no external loads are applied to the capsid. As shown in Fig. 23, the computational surface generated by the

2

Fig. 23 Reference icosahedron (a) and relaxed shells for FvK number γ ¼ YRκ below (b) and above (c) the buckling transition

Computational Mechanics of Viral Capsids

181

Fig. 24 Deformed shapes of an icosahedral shell indented along the fivefold symmetry axis. (a)–(d) full view; (e)–(h) cut-away view. (a),(e) ζ=R ¼ 0; (b),(f) and (c),(g) immediately before and immediately after buckling at ζ/R  0. 6; (d),(h) ζ=R ¼ 1

subdivision surface elements has global C1 smoothness, such that the sharp icosahedral edges and vertices are slightly rounded, with a radius of curvature which diminishes with the refinement of the mesh. The resulting relaxed capsid shapes exhibit local concentrations of strain energy around the icosahedral vertices, indicating that the capsids are effectively “prestressed” even before being subjected to any indentation force. Indentation of the prestressed capsid shells is simulated by quasi-static compression between two rigid, frictionless surfaces (Subheading 2.4). Initialized with the relaxed shapes described above, simulations proceed by moving the rigid surfaces incrementally closer to each other (Fig. 24). At each increment, the energy of the model is minimized using the LBFGS-B method and the static equilibrium of the capsid consistent with contact inequality constraints is computed. The total contact force on each rigid surface is computed by summing the contact force multipliers. The contact force F as a function of the change in distance ζ between the plates gives an informative characterization of the overall structural mechanics pffiffiffiffiffiffiffiof the capsid. Forces are normalized by the characteristic value Y κ. Displacements are normalized by the average radius of the shell R. The effect of disclination stresses on indentation mechanics is illustrated by comparing the force-indentation response of faceted icosahedral shells to a perfectly spherical shells of the same dimensions and moduli (Fig. 25). The two curves for sub-critical γ ¼ 100 are smooth, monotonically increasing, and nearly coincident for the spherical and icosahedral capsids. For smaller indentations,

182

Melissa M. Gibbons et al.

Fig. 25 Normalized force versus indentation curves for spherical and icosahedral shells at small (γ ¼ 100 < γ B  150) and large (γ ¼ 900 > γ B  150) FvK number. The curves corresponding to a low value of γ are monotonic and smooth whereas the curves corresponding to an elevated FvK number show nonlinearities and abrupt drops due to buckling phenomena

ζ < R/2, the force curves for the large γ spherical and icosahedral shells are also nearly coincident. In this range all curves are approximately linear, consistent with small-strain linearized elasticity theory [56], which predicts a linear force-indentation dependence for a spherical shell with scaling pffiffiffiffiffiffiffi κY ζ: F / R In contrast with the response of the sub-critical γ ¼ 100 shells, the roughly linear response of the γ ¼ 900 spherical shell ends part way through the indentation with an apparently discontinuous drop in force. The faceted γ ¼ 900 icosahedral shell produces a similar force drop at approximately the same indentation as the spherical capsid. However, the force drop in the faceted capsid is preceded by a nonlinear softening. The shapes of the high-γ shells immediately before and immediately after the force vs indentation discontinuity show an abrupt detachment and separation of the shell from the indenter (Fig. 24 f and g), demonstrating that physically these drops are the signature of inversion buckling events. Examining force-indentation response over larger indentation ranges (Fig. 26), we see that also spherical shells with γ ¼ 100 will eventually buckle, even if only at extreme indentations (ζ/R  1. 8) corresponding to the shell being compressed to

Computational Mechanics of Viral Capsids

183

Fig. 26 Normalized force-indentation curves for icosahedral shells with varied FvK number γ ¼ YR 2 =κ indented along the fivefold symmetry axis by a spherical AFM tip of radius R equal to that of the shell

Fig. 27 Critical values of force and indentation at which inversion buckling is triggered leading to abrupt drops in force (Fig. 26)

nearly its full diameter. As γ is further increased above the buckling threshold γ B  150, the critical values of force Fcrit and indentation ζcrit at which a vertex inverts under indentation steadily decrease, eventually reaching what looks like a plateau around pffiffiffiffiffiffiffi Fcrit = Y κ  1:0 and ζ critR  0.5 (Fig. 27).

184

Melissa M. Gibbons et al.

The analyses presented highlight that the Fo¨ppl–von Ka´rma´n number, which was previously shown to control faceting, also affects the stability of the viral capsids under load. In particular, capsids with sub critical FvK number do not experience buckling during indentation (at least up to extremely large indentation values, ζ/R  1.8) whereas capsids with large FvK numbers exhibit a buckling mechanism marked by the inversion of the conical regions surrounding icosahedral vertices. Moreover, these results suggest an additional way in which the fivefold disclinations located at icosahedral vertices play the role of structural defects since it reduces the critical buckling force, and thus diminishes the strength of the shells against geometric failure. This is reminiscent of the effects of geometrical imperfections in macroscale thin-shell structures, which when introduced in the manufacturing process are known to trigger buckling at loads much smaller than theoretical critical loads (often by orders of magnitude) [57].

4

Conclusions The study of the mechanical behavior of viral capsids is fundamental to understand the significant deformation and structural changes viruses undergo during assembly, maturation, and infective stage. Several numerical models (e.g., NMA, MD, EN) have been proposed to study the mechanics of viral capsids in several loading and deformation scenarios. These methods typically capture the geometry of viruses to a very high level of details but, on another side, are very expensive from a computational point of view and, in general, do not offer a clear structural/mechanical characterization of the viral capsid response. In the analysis of macromolecules, it remains a challenge to develop a computational model capable to capture the necessary level of details, preserve numerical efficiency, and predict the mechanical response of viral capsids subjected to different loading scenarios. A solution to this challenge is given by models based on continuum mechanics. Continuum mechanics has been used in numerous occasions to study the large-scale deformations and response to loading of viral capsids and, in general, of macromolecules. Findings and predictions discovered using continuum mechanics models have been supported and confirmed by experiments showing the validity of these models, previously restricted to the study of largescale (  1 nm) systems. The several examples provided in this chapter support the usefulness of the continuum mechanics approach, which seamlessly allows the study of different macromolecules. Additionally, the comparison between experimental and computed results reported here further proves the validity of this approach to study the mechanics of viral capsids. These results

Computational Mechanics of Viral Capsids

185

support the idea that global shape and structure have more effect on the global response than modeling the fine-scale details of the atomic interactions. In this chapter, we have outlined all the steps necessary to construct a finite element model based on continuum mechanics and to employ it in the analysis of viral capsids. We started by showing two different techniques to construct geometric models based on publicly available X-ray crystallography and cryo-electron microscopy experimental data. The approaches described allow to incorporate several levels of geometric heterogeneity in the model. Subsequently we described the other components necessary to build a continuum mechanics model for the analysis of macromolecules: the kinematic assumptions, the material laws, and the numerical discretization procedures. The proposed method has then been applied to study and simulate the indentation of several macromolecules, for which numerous experimental data are available in the literature for validation. The examples presented showed the importance of including nonuniform topography in the model in order to accurately capture the nonlinear mechanical response of the viral capsids. The response of the viral capsids to indentation depends on the contact area between the indenter and the capsid itself and how this area changes during indentation due to a nonuniform topography. The effect of geometric heterogeneity may become even more important in the large and nonlinear deformation regimes corresponding to capsid failure and, by analogy, disassembly. While geometric heterogeneity is important—a spherical model is inadequate—the coarser mesh among the ones investigated (Fig. 6) is sufficient to capture the experimental results considered. However, a more detailed description of the capsid geometry may become necessary depending on the loading scenario, e.g. if the indentation probe is made of carbon nanotubes. We concluded the modeling and examples sections by describing the analyses of viral capsids using thin-shell models. The key mechanical feature of these models is the compressive prestress locked into the fivefold disclinations at the icosahedral vertices, which act as structural defects and reduce the critical load at which the capsid buckles under external loading. We have shown the flexibility of the models based on continuum mechanics and pointed out in several instances how this robust approach may be easily applied to different geometries and loading conditions (e.g., the response to tensile forces is modeled without any changes or addition to the model). Moreover, the extension of this framework to study filled capsids is also straightforward and several approaches presented in the literature can be easily incorporated. A simple choice would be to model the pressure exerted from the DNA on the viral shell as reported in several studies (e.g., [12, 58, 59]). Another approach readily implementable in the current framework consists in increasing the capsid

186

Melissa M. Gibbons et al.

thickness at the sites where the DNA binds to the capsid as proposed in [10]. A more modeling intensive approach consists in including a homogeneous incompressible fluid inside the viral capsid and in modeling the viral capsid using porous elasticity. In this latter setting, the fluid inside the capsid is allowed to flow through the capsid pores during indentation [60]. All the aforementioned options are easy extensions of the proposed continuum mechanics approach, which, once more, is adaptable to other scenarios than the ones presented in this chapter. References 1. Berman HM, Henrick K, Nakamura H (2003) Announcing the worldwide protein data bank. Nat Struct Biol 10(12):980. http://www.rcsb. org/pdb/home/home.do 2. Brooksbank C, Camon E, Harris MA, Magrane M, Martin MJ, Mulder N, O’Donovan C, Parkinson H, Tuli MA, Apweiler R, Birney E, Brazma A, Henrick K, Lopez R, Stoesser G, Stoehr P, Cameron G (2003) The european bioinformatics institute’s data resources. Nucleic Acids Res 31:43–50. http://www.ebi. ac.uk 3. Baker TS, Olson NH, Fuller SD (1999) Adding the third dimension to virus life cycles: three-dimensional reconstruction of icosahedral viruses from cryo-electron micrographs. Microbiol Mol Biol Rev 63:862–922 4. Shepherd CM, Borelli IA, Lander G, Natarajan P, Siddavanahalli V, Bajaj C, Johnson JE, Brooks III CL, Reddy VS (2006) Viperdb: a relational database for structural virology. Nucleic Acids Res 34(Database Issue): http://viperdb.scripps.edu/ D386–D389. index.php 5. Ivanovska IL, de Pablo PJ, Ibarra B, Sgalari G, MacKintosh FC, Carrascosa JL, Schmidt CF, Wuite GJL (2004) Bacteriophage capsids: tough nanoshells with complex elastic properties. Proc Natl Acad Sci USA 101:6700–6705 6. Michel J-P, Ivanovska IL, Gibbons MM, Klug WS, Knobler CM, Schmidt CF, Wuite GJL (2006) Nanoindentation studies of full and empty viral capsids and the effects of capsid protein mutations on elasticity and strength. Proc Natl Acad Sci USA 103(16):6184–6189 7. Klug WS, Bruinsma RF, Michel J-P, Knobler CM, Ivanovska IL, Schmidt CF, Wuite GJL (2006) Failure of viral shells. Phys Rev Lett 97(22):228101 8. Schaap IAT, Carrasco C, Pablo PJ de, MacKintosh FC, Schmidt CF (2006) Elastic response, buckling, and instability of

microtubules under radial indentation. Biophys J 91(4):1521–1531 9. Kol N, Gladnikoff M, Barlam D, Shneck RZ, Rein A, Rousso I (2006) Mechanical properties of murine leukemia virus particles: effect of maturation. Biophys J 91(2):767–774 10. Carrasco C, Carreira A, Schaap IAT, Serena PA, Gomez-Herrero J, Mateu MG, Pablo PJ de (2006) DNA-mediated anisotropic mechanical reinforcement of a virus. Proc Natl Acad Sci 103(37):13706–13711 11. Kol N, Shi Y, Tsvitov M, Barlam D, Shneck RZ, Kay MS, Rousso I (2007) A stiffness switch in human immunodeficiency virus. Biophys J 92:1777–1783 12. Ivanovska IL, Wuite GJL, Jonsson B, Evilevitch A (2007) Internal DNA pressure modifies stability of WT phage. Proc Natl Acad Sci 104 (23):9603–9608 13. Schmatulla A, Maghelli N, Marti O (2007) Micromechanical properties of tobacco mosaic viruses. J Microsc (Oxford) 225:264–268 14. Carrasco C, Castellanos M, Pablo PJ de, Mateu MG (2008) Manipulation of the mechanical properties of a virus by protein engineering. Proc Natl Acad Sci 105(11):4150–4155 15. Roos WH, Gibbons MM, Arkhipovv, Uetrecht C, Watts NR, Wingfield PT, Steven AC, Heck AJR, Schulten K, Klug WS et al (2010) Squeezing protein shells: how continuum elastic models, molecular dynamics simulations, and experiments coalesce at the nanoscale. Biophys J 99(4):1175–1181 16. Gibbons MM, Klug WS (2007) Mechanical modeling of viral capsids. J Mater Sci 42:8995–9004 17. Lu M, Ma J (2005) The role of shape in determining molecular motions. Biophys J 89:2395–2401 18. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, Ferrin TE (2004) UCSF chimera: a visualization

Computational Mechanics of Viral Capsids system for exploratory research and analysis. J Comput Chem 25(13):1605–1612 19. Humphrey W, Dalke A, Schulten K (1996) VMD: visual molecular dynamics. J Mol Graph 14:33–38 20. Zhang Y, Bajaj C, Sohn B (2005) 3D finite element meshing from imaging data. Comput Methods Appl Mech Eng 194:5083–5106 21. Zhang Y, Bajaj C (2006) Adaptive and quality quadrilateral/hexahedral meshing from volumetric data. Comput Methods Appl Mech Eng 195:942–960 22. Bathe M (2008) A finite element framework for computation of protein normal modes and mechanical response. Proteins: Struct Funct Bioinf 70:1595–1609 23. Lorensen WE, Cline HE (1987) Marching cubes: a high resolution 3D surface construction algorithm. Comput Graph 21(4):163–169 24. Buell WR, Bush BA (1973) Mesh generation— survey. J Eng Ind Trans ASME 95:332–338 25. Schroeder W, Martin K, Lorensen B (2004) The visualization toolkit. Kitware, New York. http://www.vtk.org 26. Kleywegt GJ (1999) Experimental assessment of differences between related protein crystal structures. Acta Crystallogr Sect D: Biol Crystallogr 55:1878–1884. http://xray.bmc. uu.se/usf/ 27. Gurtin ME (1981) An introduction to continuum mechanics. Academic Press, New York 28. Holzapfel GA (2001) Nonlinear solid mechanics: a continuum approach for engineering. Wiley, New York 29. Gibbons MM, Klug WS (2008) Influence of nonuniform geometry on nanoindentation of viral capsids. Biophys J 95:3640–3649 30. Sokolnikoff IS (1964) Tensor analysis: theory and applications to geometry and mechanics of continua. Wiley, New York 31. Carmo MP Do (1976) Differential geometry of curves and surfaces, vol 2. Prentice-Hall, Englewood Cliffs 32. Wempner G, Talaslidis D (2002) Mechanics of solids and shells: theories and approximations. CRC press, Boca Raton 33. Strang WG, Fix GJ (1973) Analysis of the finite element method. Prentice-Hall, Englewood Cliffs 34. Cirak F, Ortiz M, Schroder P (2000) Subdivision surfaces: a new paradigm for thin-shell finite-element analysis. Int J Numer Methods Eng 47(12):2039–2072

187

35. Cirak F, Ortiz M (2001) Fully C1-conforming subdivision elements for finite deformation thin-shell analysis. Int J Numer Methods Eng 51(7):813–833 36. Cirak F, Scott MJ, Antonsson EK, Ortiz M, Schro¨der P (2002) Integrated modeling, finite-element analysis, and engineering design for thin-shell structures using subdivision. Comput Aided Des 34(2):137–148 37. Cirak F, Ortiz M, Pandolfi A (2005) A cohesive approach to thin-shell fracture and fragmentation. Comput Methods Appl Mech Eng 194 (21):2604–2618 38. Lidmar J, Mirny L, Nelson DR (2003) Virus shapes and buckling transitions in spherical shells. Phys Rev E 68(5):051910 39. Caspar DL, Klug A (1962) Physical principles in the construction of regular viruses. Cold Spring Harb Symp Quant Biol 27:1–24 40. Zhu C, Byrd RH, Lu P, Nocedal J (1997) Algorithm 778: L-BFGS-B: fortran subroutines for large-scale bound-constrained optimization. ACM Trans Math Softw 23 (4):550–560 41. Wriggers P (2006) Computational contact mechanics. Springer, New York 42. Gibbons MM, Klug WS (2007) Nonlinear finite-element analysis of nanoindentation of viral capsids. Phys Rev E 75(3):031901 43. Vliegenthart GA, Gompper G (2006) Mechanical deformation of spherical viruses with icosahedral symmetry. Biophys J 91(3):834–841 44. Buenemann M, Lenz P (2007) Mechanical limits of viral capsids. Proc Natl Acad Sci 104:9925–9930 45. Speir JA, Munshi S, Wand GJ, Baker TS, Johnson JE (1995) Structures of the native and swollen forms of cowpea chlorotic mottle virus determined by X-ray crystallography and cryoelectron microscopy. Structure 3(1):63–78 46. Cheung CL, Hafner JH, Lieber CM (2000) Carbon nanotube atomic force microscopy tips: direct growth by chemical vapor deposition and application to high-resolution imaging. Proc Natl Acad Sci 97(8):3809–3813 47. Woolley AT, Cheung CL, Hafner JH, Lieber CM (2000) Structural biology with carbon nanotube AFM probes. Chem Biol 7(11): R193–R204 48. Cohen BJ, Richmond JE (1982) Electron microscopy of hepatitis B core antigen synthesized in E. coli. Nature 296:677–678

188

Melissa M. Gibbons et al.

49. Crowther RA, Kiselev NA, Bottcher B, Berriman JA, Borisova GP, Ose V, Pumpens P (1994) 3-dimensional structure of hepatitis B virus core particles determined by electron cryomicroscopy. Cell 77:943–950 50. Wynne SA, Crowther RA, Leslie AGW (1999) The crystal structure of the human hepatitis B virus capsid. Mol Cell 3:771–780 51. Tao YZ, Olson NH, Xu W, Anderson DL, Rossmann MG, Baker TS (1998) Assembly of a tailed bacterial virus and its genome release studied in three dimensions. Cell 95 (3):431–437 52. Morais MC, Choi KH, Koti JS, Chipman PR, Anderson DL, Rossmann MG (2005) Conservation of the capsid structure in tailed dsDNA bacteriophages: the pseudoatomic structure of phi 29. Mol Cell 18:149–159 53. Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter P (2002) Molecular biology of the cell. Garland Science, New York 54. Nogales E, Wolf SG, Downing KH (1998) Structure of the alpha beta tubulin dimer

by electron crystallography. Nature 391:199–203 55. Ray S, Meyho¨fer E, Milligan RA, Howard J (1993) Kinesin follows the microtubule’s protofilament axis. J Cell Biol 121:1083–1093 56. Landau LD, Lifshitz EM (1986) Theory of elasticity. Pergamon Press, Oxford 57. Calladine CR (1998) Theory of shell structures. Cambridge University Press, New York 58. Roos WH, Ivanovska IL, Evilevitch A, Wuite GJL (2007) Viral capsids: mechanical characteristics, genome packaging and delivery mechanisms. Cell Mol Life Sci 64(12):1484–1497 59. Evilevitch A, Roos WH, Ivanovska IL, Jeembaeva M, Jo¨nsson B, Wuite GJL (2011) Effects of salts on internal DNA pressure and mechanical properties of phage capsids. J Mol Biol 405 (1):18–23 60. Ahadi A, Johansson D, Evilevitch A (2013) Modeling and simulation of the mechanical response from nanoindentation test of DNAfilled viral capsids. J Biol Phy 39(2):183–199

INDEX A ABAQUS software ............................................... 147, 156 AEMA. See 2-aminoethyl methacrylate (AEMA) AFM. See Atomic force microscopy (AFM) Alphavirus ..................................................................3, 5, 9 2-Aminoethyl methacrylate (AEMA) ............... 18, 20–23 Archaeoglobus fulgidus ..............................................28, 41 Atomic force microscopy (AFM) .............. 115–134, 140, 157, 165–167, 172, 173, 176, 177, 179, 180, 183 Atom transfer radical polymerization (ATRP) ........18–21 ATRP. See Atom transfer radical polymerization (ATRP) AuNP. See Gold nanoparticle (AuNP)

B Bacterial microcompartment .......................61–62, 69–76 Bacterioferritin (BFR)....... 52, 57, 58, 84, 103, 106, 111 Bateriophage φ .. 29, 126, 128, 145, 146, 159, 172–176 Bateriophage P22......................................................18–23 BCM. See Bacterial microcompartment (BCM) BFR. See Bacterioferritin (BFR) Biomineralization ............................................................ 52 Bipartite binding site ................................................80, 82 Bipyridyl................................................. 91, 93, 94, 96–98 BMV. See Brome mosaic virus (BMV) Brevibacterium linens ...................................................... 62 Brome mosaic virus (BMV)............................... 2–4, 7–10

C Capsid-genome interactions ................................ 127–129 Caspar–Klug construction of closed spherical viruses........................................................ 155–156 CASTp ..........................................................52, 53, 55–57 CCMV. See Cowpea chlorotic mottle virus (CCMV) Colorimetric assay for detection of metal...................... 91 Computational design ...................................27, 140, 184 Constitutive Modeling.................................147–148, 185 Continuum mechanics ............... 140, 141, 147, 184–186 Contrast agent...........................................................39–41 Cowpea chlorotic mottle virus (CCMV)........... 126, 143, 159–167, 169, 175, 176

D Density field.........................................142–144, 160, 177 Differential scanning calorimetry (DSC)............ 101–112

DLS. See Dynamic light scattering (DLS) DNA binding protein from starved cells (Dps) .... 80, 82, 91, 94–96, 98 DSC. See Differential scanning calorimetry (DSC) Dye-decolorizing peroxidase (DyP).........................62, 65 Dynamic light scattering (DLS) ........................ 10, 11, 35 DyP. See Dye-decolorizing peroxidase (DyP)

E Empty shell (ES) form ........................................... 23, 130 Encapsulation .......................1–12, 27–36, 41, 51, 69, 72 Encapsulin ........................................................ 61–66, 134 Equilibrium constant ...................................................... 34 ES form. See Empty shell (ES) form Expanded shell (EX) form ................................. 20, 22, 23

F Ferritin ........................ 17–20, 27–36, 41–45, 48, 51–53, 55–57, 61, 80, 84, 103, 105, 109, 110, 112 Ferritin-like protein (Flp) .........................................62, 65 Finite element mesh................................... 141–146, 156, 160, 161, 163, 164, 167, 168, 174, 175, 177, 178 FlAsH-EDT2...................................................... 81–84, 87 Flp. See Ferritin-like protein (Flp) Fluorescence oligomerzation assay ................................ 82 Fluorescent reporter of protein unfolding .................... 92 FoldX .........................................................................53–58 Force indentation ............. 125, 126, 129–131, 162–167, 169, 170, 172, 173, 175, 176, 181–183 Force spectroscopy............................................... 120–121 Free energy calculation .............................................51, 53

G Gel filtration ...........................41, 43, 92, 94, 96–98, 104 Gold nanoparticle (GNP)....................... 1–10, 12, 28, 32

H Haliangium ochraceum bacterial microcompartment................................. 73, 75, 76 Halothiobavillus neapolitanus carboxysome .................. 73 HBV. See Hepatitis B virus (HBV) Heat capacity ........................................................ 101, 102 Hepatitis B virus (HBV) ..............................159, 167–172 HFn. See Human heavy chain ferritin (HFn)

Brendan P. Orner (ed.), Protein Cages: Methods and Protocols, Methods in Molecular Biology, vol. 1252, DOI 10.1007/978-1-4939-2131-7, © Springer Science+Business Media New York 2015

189

PROTEIN CAGES 190 Index HIV. See human immunodeficiency virus (HIV) HIV Δ16–99 Gag ...................................................5, 9–10 Human heavy chain ferritin (HFn) ............ 18, 21, 22, 24 Human immunodeficiency virus (HIV) ...................... 2–4

I Icosahedral virus.....................2, 3, 61, 62, 122–124, 155 Inductively coupled plasma mass spectrometer (ICP-MS) ................................................ 42, 43, 48 Interfacial concavities................................................52, 55 Iron nanoparticle................................................ 41, 43, 44 Isosurface.............................................................. 144–146

K Kinematic assumption.......................................... 151, 185

M Maghemite. See Magnetite Magnetic nanoparticle .................................................... 40 Magnetic relaxivity ....................................................40–49 Magnetic resonance imaging (MRI)............... 39–49, 144 Magnetite ........................................................... 18–22, 24 MALDI. See Matrix-assisted laser desorption ionization (MALDI) Mapman smoothing filter .................................... 145, 174 Matrix-assisted laser desorption ionization (MALDI) ............................................................. 36 Mechanical properties ......................................... 115, 116, 120–123, 126, 127, 129, 131–133 MED4 carboxysome ....................................................... 73 Metabolic engineering .................................................... 69 Metal binding ..........................................................91–100 Microtubules ................................................159, 177–180 MRI. See Magnetic resonance imaging (MRI)

4-(2-Pyridylazo)resorcinol (PAR) .............. 91, 93, 95, 98

R RAVE software suite ..................................................... 145 Retrovirus .......................................................................... 2 Ross River virus (RRV)........................................ 3, 5, 8, 9

S SEC. See Size exclusion chromatography (SEC) Seeded growth.................................................... 29–30, 35 Self-assembly .......................2, 27, 28, 51, 72–73, 79–81, 103, 131 Sindbis virus ...................................................................... 2 Size exclusion chromatography (SEC) .................. 18, 22, 28, 32, 34, 43, 65, 66, 80, 83, 86, 105, 106, 110 Surface plasmon ..........................................................1, 35 Surface topographic analysis.....................................53, 54 Symmetry.......................... 3, 28, 52, 55–57, 85, 91, 122, 123, 127, 133, 162–172, 180, 181, 183 Synthetic biology ............................................................ 69 Synthetic operon ............................................................. 71 SYPRO Orange .....................................92, 93, 95, 96, 99

T TEM. See Transmission electron microscopy (TEM) Templated self-assembly .............................................4, 12 tF. See Thermophilic ferritin (tF) Thermal stability .................................57, 93, 95–96, 103 Thermodynamics........................................................... 110 Thermophilic ferritin (tF)............................28–32, 34–36 Thermotoga maritima ................................. 62, 63, 65, 66 Transmission electron microscopy (TEM) ...............9–11, 24, 28, 30, 33–36, 42, 44, 48, 74–76, 80 Tryptophan (Trp) fluorescence ...................................... 36

N

U

Nanoparticle-protein interaction ................................... 27 Nonuniform topography ....................140, 141, 161, 185

Ultracentrifugation ..................22, 23, 62, 66, 72, 74–76

V O Oligomer detection......................................................... 86

P P22. See Bateriophage P22 PAR. See 4-(2-Pyridylazo)resorcinol (PAR) Pdu BMC. See Propanediol utilization bacterial microcompartment (Pdu BMC) Plant virus ..................................................................2, 160 Propanediol utilization bacterial microcompartment (Pdu BMC)....................................................73, 76 Protein engineering ........................................................ 69 Protein-protein interaction......2, 51–54, 57, 80, 85, 103

Vibrating sample magnetometer (VSM) .................42, 44 Virtual saturation mutagenesis ....................................... 55 Virus immobilization .................................................... 122 Virus-like particle .................2–4, 9–11, 19, 62, 121, 122 Visualization toolkit software ..................... 145, 146, 174 VLP. See Virus-like particle (VLP) Von Mises stress .........................171, 172, 175, 176, 179 VSM. See Vibrating sample magnetometer (VSM) VTK. See Visualization toolkit (VTK)

Y Young’s modulus ...................... 125–126, 131, 152, 172, 176, 178