Physical Chemistry and Molecular Biology of mRNA and Self-Assembling Virus-Like Particles

Virus particles can be as simple as a molecule of RNA or DNA inside a spherical shell – the “capsid” – made up of multiple copies of a single protein. Further, because viruses only “become alive” when they are inside their hosts, it is possible to study them -- outside their hosts, in a test tube -- as physical objects, i.e., to do the same controlled experiments (and theory) on them that one routinely does with more familiar polymer and colloidal systems.

This approach is best illustrated with the example of one of the simplest viruses we work on – cowpea chlorotic mottle virus (CCMV) – a plant virus with a single-stranded RNA genome. Each particle of CCMV is a one-molecule-thick spherical shell consisting of (exactly!) 180 copies of a single protein, surrounding and protecting single copies of RNA genes of the virus. The 180 capsid proteins are organized into groups of 12 pentamers and 20 hexamers, involving 3 sets of inequivalent positions.

Most remarkably, this icosahedrally-symmetric (soccer-ball-like/”Bucky-ball”-like) structure can form spontaneously: one simply has to mix the purified RNA molecules and purified capsid proteins under the right conditions of pH and ionic strength and a virus forms that is infectious and indistinguishable from those found in infected plants. By changing these solution conditions, and replacing the viral RNA by non-viral RNA (and even by synthetic anionic polymers) and replacing the wildtype capsid protein by genetically- and/or chemically- modified forms of it, we are able to learn a lot about the fundamental physical chemistry of this biological process – the formation of an infectious virus, and also to prepare well-characterized and effective gene delivery systems.

RNA Compaction by Polyvalent Cations, and Polyvalent-Cation-Mediated In Vitro Packaging of "Overlong" RNAs into CCMV and BMV Spherical Capsids

The effect of polyvalent cations, e.g., spermidine (3+) and spermine (4+), on the organization of (double-stranded, ds) DNA has been well studied for many decades, both experimentally and theoretically. It is now well understood that these cations – in addition to contributing to the usual screening of self-repulsion – mediate a DNA self-attraction, arising from the breakdown of the mean-field (Poisson-Boltzmann) approximation. As a consequence, DNA undergoes a first-order/discontinuous “condensation” at a critical concentration of polyvalent cation. No systematic effort has been made to determine whether or not there is an analogue of this transition for (single-stranded, ss) RNA. Using dynamic light scattering (DLS) determinations of diffusion coefficients, and electrophoretic gel and analytical ultracentrifugation (AUC) measurements of sedimentation velocities, we are finding that the size of long (viral genome length/kb-long) RNA molecules decreases continuously with spermine concentration, with no evidence of a first-order collapse. The significant reduction in size (and effective charge) allows heterologous RNA molecules to packaged in vitro into wildtype virus-like particles, by capsid proteins that otherwise would only be able to accommodate RNA lengths comparable to their own RNA genome.

In Vitro Reconstitution of Heterologous RNA in Cylindrical versus Spherical Virus-Like Particles

The only two viruses that can be in vitro reconstituted as infectious particles from their purified components – ssRNA and capsid protein – are two plant viruses, one spherical (cowpea chlorotic mottle virus [CCMV]) and the other cylindrical (tobacco mosaic virus [TMV]). Further, each capsid protein is capable of in vitro packaging heterologous RNA, although in the TMV case it is necessary to insert a “packaging signal” – an “origin of assembly site” – into the RNA sequence, to nucleate the assembly of a helical/cylindrical shell around the RNA. The qualitative advantage of the TMV virus-like particles is the geometric fact that the curvature of the capsid is independent of its length, and therefore independent of the length of packaged RNA. Thus, unlike the case of spherical capsids, where the highly-evolved/strongly-favored protein curvature limits the amount of genetic information that can be protected by the capsid, a cylindrical geometry can accommodate any amount. Accordingly to in vitro package longer therapeutic genes – especially in the form of replicons, where they are fused to RNA replicase genes – we are beginning to synthesize and functionalize mRNA-containing TMV VLPs and compare their stability and efficacy with those of the corresponding CCMV or BMV VLPs.

Persistent RNA Replicons and Defective Interfering RNAs: Competition to be Replicated

For many reasons, RNA replication is cytotoxic, whether or not the RNA being replicated is the infectious genome or a non-infectious form of it like the replicons we work with that involve the open reading frame for the RNA-replication-related proteins but not for any of the structural proteins. In the case of both the Nodamura and the Sindbis replicons, for example, the expression of reporter genes is much stronger than for reporter-gene mRNAs alone, and it peaks and dies out after a day or two instead of just a few hours. In an effort to learn about RNA replication, and to provide less cytotoxic/longer-lived amplification of mRNAs for protein expression, we are working with a mutant of the Sindbis replicon developed in the lab of Charles Rice. A single point mutation in the RNA-replicase polyprotein results in a 50-fold lower level of replication and – most dramatically – in a persistent level of replication. Further, the expression of the replicon survives the growth of the cells into a confluent monolayer and their passaging through many generations. We will be using this system to screen the relative efficacies of different antiviral candidates in the form of defective interfering molecules, i.e., RNAs that are templates for and replicated by the RNA-dependent RNA polymerase (RdRp) – RNA replicase – of the replicon and which therefore compete for RdRp and nucleotides, inhibiting the replication of the replicon (here, a stand-in for the infectious genome).

3D Cryo-Electron Microscopy (CryoEM) Reconstructions of RNA Inside Viral Capsids

We use high-resolution cryo-electron microscopy – in particular, asymmetric reconstruction methods, in which no symmetry is assumed – to investigate the structure of single-stranded RNA genes inside a multipartite virus, Brome Mosaic Virus (BMV). Multipartite here refers to the fact that the RNA genome consists of two or more (three in this case) molecules, each of which is packaged in a different, otherwise-identical, capsid. We study separately each of the three kinds of virions involved and find that the RNA is highly disordered, suggesting a different method of assembly for multipartite viruses compared to the monopartite viruses MS2 and Qβ, which have been shown to have highly ordered genomes within their capsids. This can be understood in the context of their differing viral life-cycles, as BMV must package separately each of several different RNA molecules and has been shown to replicate and package them in replication-factories, while MS2 and Qβ depend on “packaging signals” throughout the viral RNA to selectively package their monopartite genomes.

Beren, C., Cui, Y., Chakravarty, A., Yang, X., Knobler, C. M, Zhou, Z. H., Gelbart, W. M. (2020) Genome organization and interaction with capsid protein in a multipartite RNA virus. PNAS, 117 (20), 10673-80.

In collaboration with Prof. Hong Zhou's group (MIMG, UCLA) and Prof. A. L. N. Rao's group (Plant Pathology, UC Riverside)

Distribution of RNA Ends in Multiplet-Capsid Structures

It is a remarkable and convenient fact that purified capsid proteins (CPs) of some viruses like Cowpea Chlorotic Mottle Virus (CCMV) and Brome Mosaic Virus (BMV) can spontaneously self-assemble into virus-like particles (VLPs) upon being mixed with purified single-stranded RNA (ssRNA) of different sequences and lengths. Earlier work in our group showed, however, that if the length of the RNA exceeds that of the longest viral gene (~3200 nt) by more than a thousand nucleotides (nts), single molecules of RNA are shared between two or more T=3 or T=4 capsids (multiplets) [1]. Also, theoretical work suggests that in (thousands-of-nts-) long ssRNAs the separation between the 5' and 3' ends of the molecules is small (2-3nm) and constant, independent of their particular sequence and length [2]. We are performing experiments to investigate both the nature of multiplet packaging (e.g., are the ends of over-long RNA in the same or in different capsids?), and the separation between the ends of un-packaged ssRNAs, by conjugating their ends with small (<3nm) gold nanoparticles, thereby allowing them to be observed directly by cryo-electron microscopy imaging.

[1] Garmann, R. F., Comas-Garcia, M., Knobler, C. M., & Gelbart, W. M. (2015). Physical principles in the self-assembly of a simple spherical virus. Accounts of chemical research, 49, 48-55.

[2] Yoffe, A. M., Prinsen, P., Gelbart, W. M., & Ben-Shaul, A. (2010). The ends of a large RNA molecule are necessarily close. Nucleic acids research, 39, 292-299.

In Vitro Synthesis of Enveloped Virus-Like Particles (EVLPs)

Most mammalian viruses are enveloped by a lipid membrane, which enhances their stability and facilitates their entry into host cells. On the other hand, the plant viruses like CCMV and BMV that we have been studying and utilizing for gene delivery are unenveloped – their RNA genomes are enclosed and protected by a protein shell only. In order to enhance the biocompatibility and targeting of a new delivery system of VLPs, we are wrapping these particles with a lipid bilayer membrane to create enveloped virus like particles (EVLPs). In addition to driving the wrapping by introducing oppositely-charged lipid, we use NHS-ester and maleimide lipids to “seed” the nucleation of bilayer envelope through interaction with lysines and cysteines on the VLP exteriors.

RNA Homopolymer Packaging by CCMV - the Role of RNA Secondary Structure on Self-Assembly

We investigated the effect of RNA secondary structure on CCMV spontaneous self-assembly by looking at how CCMV capsid protein (CP) self-assembles around polyadenylic acid (polyA), a homopolymer that has no self-complementarity but exhibits strong base-stacking and helical structure. Consistent with earlier work in our group with polyU [1], we find that – instead of T=3/180-subunit capsids (26 nm) arising from viral RNA or from heterologous RNA with “normal composition” (i.e., comparable numbers of A, U, G, and C bases) – the virus-like particles that form have T=2/120-subunit size (22 nm) [2]. These results point up the extent to which the spontaneous/preferred curvature of capsid protein is not an intrinsic property of the protein but rather one that depends on the kind of RNA with which it interacts (e.g., with or without normal composition and hence secondary structure formation). Further, unlike polyU, the stiffness of poly A resulting from its base-stacking and helical ordering makes it much less efficiently packaged by capsid protein.

[1] Beren, C., Dreesens, L. L., Liu, K. N., Knobler, C. M., & Gelbart, W. M. (2017). The effect of RNA secondary structure on the self-assembly of viral capsids. Biophysical journal, 113, 339-347.

[2] Thurm, A. R., Beren, C., Duran-Meza, A. L., Knobler, C. M., Gelbart, W. M. (2019) RNA homopolymers form higher-curvature virus-like particles than do normal-composition RNAs. Biophysical Journal, 117 (7), 1331-41.

Catching the Ends of Fluctuating RNA Outside Intact Capsids

We hypothesize that a positive-sense RNA virus does not necessarily have to disassemble (“uncoat”) in its host cell cytoplasm in order for translation of its genome to begin. Instead, we consider the possibility that translation is initiated while the protective capsid is still intact, so that the RNA genome is handed over directly from the protection of its capsid to protection of the ribosomal machinery. This is in fact the scenario suggested by the “co-translational delivery” mechanism suggested many years ago by T. A. M. Wilson, in the context of plant viruses like TCV, and pursued later by A. Helenius in the context of mammalian viruses like Sindbis. More explicitly, we are trying to demonstrate that the 5’ end of the genome fluctuates outside the capsid – through one of the many 1.5-to-2nm-sized holes between capsomers – and is bound by translation initiation factors like eIF4E which then engage it with the ribosomal subunits. To this end we are carrying out several experiments in parallel. One involves incubation of BMV virions with eIF4E proteins that are either fluorescently labeled or bound to beads coated with anti-eiF4E antibodies. Another involves biotinylating the 5’ end of fluorescently-labeld viral RNA, in vitro packaging it into virions, and them incubating them over a streptavidin coated surface and measuring the linear-with-time increase in bound virions (see figure). In another, related, experiment, we prepare “cherry bomb” constructs [1] that consist of BMV VLPs with one end of their RNA stuck outside of the capsid because of it being hybridized to a long enough complementary DNA oligo before in vitro packaging; we then catch this “stem” with a magnetic bead and measure the force-versus-distance profile as we pull on it in a magnetic tweezer apparatus.

[1] Garmann, R. F., Sportsman, R., Beren, C., Manoharan, V. N., Knobler, C. M., & Gelbart, W. M. (2015). A simple RNA-DNA scaffold templates the assembly of monofunctional virus-like particles. Journal of the American Chemical Society, 137, 7584-7587.

In collaboration with Prof. Jim Bowie's group (Chemistry & Biochemistry, UCLA) and Prof. David Bensimon's group (Chemistry & Biochemistry, UCLA and Physics, ENS Paris)

Characterization of Assembly Mechanism through Cryo-Electron Tomography (CryoET) of Individual RNA-Capsid Protein Assembly Intermediates

For many unenveloped viruses whose genomes are single-stranded RNA, infectious particles – nucleic acid inside a protein shell (capsid) – form spontaneously from their purified components. There are two prevailing scenarios for this assembly process: (i) nucleation and growth; and (ii) adsorption and rearrangement. According to the nucleation and growth mechanism, viral capsid proteins (CPs) first form an ordered critical nucleus on the genome – a partial capsid – and then grow by sequential addition of CPs from solution until the spherical/icosahedrally-symmetric capsid shell is completed. In the adsorption and rearrangement mechanism, on the other hand, CPs first adsorb onto the genome in a saturating and disordered fashion and then collectively rearrange to form the ordered capsid without further addition of CPs from solution. Molecular dynamics simulations predict that the assembly mechanism can be tuned by solution conditions that alter the relative strengths of CP-CP and CP-genome interactions [1]. We aim to test this idea experimentally for the case of a particularly well-characterized virus, Brome Mosaic Virus (BMV). Specifically, we want to demonstrate a cross-over in assembly mechanism by changing solution conditions – namely, ionic strength and pH – and directly imaging protein-RNA intermediates using Cryo-Electron Tomography (CryoET). We will collect tomography data on our samples in order to time-resolve the structures of single protein-RNA complexes, thereby generating 3D reconstructions of individual assembly intermediates. In the nucleation and growth pathway, we should see partial capsids appearing in CP-RNA complexes after a lag time, quickly evolving into complete capsids. By contrast, when the adsorption and rearrangement mechanism becomes operative we should see saturated but disordered adsorption of CPs on RNA followed by ordering into the icosahedral capsid. This direct visualization study would demonstrate how relative strengths of CP-CP and CP-RNA interactions can be tuned by solution conditions to determine the mechanism of assembling viruses.

[1] Perlmutter, J. D., Perkett, M. R., & Hagan, M. F. (2014). Pathways for virus assembly around nucleic acids. Journal of Molecular Biology, 426 (18), 3148-65.

In collaboration with Prof. Jose Rodriguez (Chemistry & Biochemistry, UCLA) and Prof. Michael Hagan (Physics, Brandeis)

Atomistic Molecular Dynamics Studies of Protein-RNA Interactions

Our goal is to determine the role of the positively charged N-terminal tail for Cowpea Chlorotic Mottle Virus (CCMV) in the binding of capsid protein to RNA. We use atomistic molecular dynamics simulations and various free energy calculation methods to determine binding energies for the N-terminal tail to different RNAs and RNA motifs. In particular, we want to explain the drastically different protein-RNA interactions and packaging scenarios observed experimentally for "normal-composition" RNAs with extensive duplex formation and single-stranded loops, versus for single-letter RNAs like polyU, for which no intramolecular base-pairing is possible [1, 2]. Ultimately, we would like to develop a theory that can predict how RNA sequence and consequent secondary structure affects the encapsidation process.

[1] Beren, C., Dreesens, L. L., Liu, K. N., Knobler, C. M., & Gelbart, W. M. (2017). The effect of RNA secondary structure on the self-assembly of viral capsids. Biophysical journal, 113, 339-347.

[2] Thurm, A. R., Beren, C., Duran-Meza, A. L., Knobler, C. M., Gelbart, W. M. (2019) RNA homopolymers form higher-curvature virus-like particles than do normal-composition RNAs. Biophysical Journal, 117 (7), 1331-41.

In collaboration with Prof. Robijn Bruinsma's group (Physics, UCLA)