Crystal Structure of the Rabies Virus Nucleoprotein-RNA Complex

See allHide authors and affiliations

Science  21 Jul 2006:
Vol. 313, Issue 5785, pp. 360-363
DOI: 10.1126/science.1125280


Negative-strand RNA viruses condense their genome into a helical nucleoprotein-RNA complex, the nucleocapsid, which is packed into virions and serves as a template for the RNA-dependent RNA polymerase complex. The crystal structure of a recombinant rabies virus nucleoprotein-RNA complex, organized in an undecameric ring, has been determined at 3.5 angstrom resolution. Polymerization of the nucleoprotein is achieved by domain exchange between protomers, with flexible hinges allowing nucleocapsid formation. The two core domains of the nucleoprotein clamp around the RNA at their interface and shield it from the environment. RNA sequestering by nucleoproteins is likely a common mechanism used by negative-strand RNA viruses to protect their genomes from the innate immune response directed against viral RNA in human host cells at certain stages of an infectious cycle.

Rabies virus, a member of the Rhabdoviridae family, is the causative agent of rabies, a fatal central nervous system disease (1), which constitutes a serious health problem in developing countries that lack effective vaccination programs (2). Rhabdoviridae, together with Paramyxoviridae (e.g., measles virus), Filoviridae (e.g., Ebola virus) and Bornaviridae (e.g., Borna disease virus), are RNA-containing enveloped viruses that use nonsegmented negative sense RNA as their genome. The RNA is condensed by the nucleoprotein (N) into a helical nucleocapsid (NC) (3) and this N-RNA complex constitutes the essential template for replication by the RNA-dependent RNA polymerase complex (4). The polymerase complex selects for either transcription or replication and is composed of the enzymatic active L protein and the phosphoprotein P (57). Replication produces a full-length (+) copy of the viral RNA (vRNA), which is the specific target for encapsidation by N and serves as a template for (–) RNA replication. The switch from transcription to replication is in part regulated by the abundance of free N in the cytoplasm (8). In order to understand the role of N-RNA complexes in the viral life cycle, we solved the crystal structure of a nucleoprotein oligomer from rabies virus complexed to a 99-nucleotide-long RNA segment.

Recombinant expression of most viral nucleoproteins from negative-strand RNA viruses leads to nonspecific host cell RNA encapsidation by N (9, 10), resulting in either helical or ring-like structures depending on the length of the RNA (10, 11). Rabies virus N expression in insect cells produced N-RNA rings containing 9, 10, 11, 12, or 13 copies of N as determined after purification (12). The undecameric ring produced the best diffracting crystals belonging to space group P21212. The structure was solved by single anomalous dispersion (SAD) and refined to 3.5 Å resolution with an R-factor of 27.5 (Rfree = 32.6) (table S1) (13). Two ∼550-kDa large undecameric N-RNA rings pack head to head in the crystal asymmetric unit. Each ring has an outer diameter of 160 Å, an inner diameter of 60 Å, and a height of 74 Å (Fig. 1A). The N protomer consists of two main domains, which contact nine nucleotides of single-stranded RNA, as predicted (14), that are occluded in the center of the ring (Fig. 1, B, C, and D, and Fig. 2). The N-terminal core domain (NTD; residues 32 to 233) folds into a helical arrangement composed of 6 helices connected by large loops. The C-terminal core domain (CTD; residues 236 to 356 and 396 to 450) is composed of 11 helices joined by tighter loops (Figs. 1C and fig. S1). Two regions in N (NTD: 105 to 118 and CTD: 376 to 397) are presumably flexible, because they are absent in the structure (Fig. 1C and fig. S1). In addition to the core domains, two smaller subdomains participate in domain exchange between protomers and stabilize polymerization (Figs. 1C and 3, A and B). Both the top (NTD) and bottom (CTD) domains act as “jaws” that clamp down onto the RNA strand and enclose it completely (Fig. 1D), an observation which is consistent with the fact that the RNA remains bound to N in CsCl gradients (15). The closest contact between the two jaws is between NTD residue N157 and CTD residue P435 (9 to 10 Å, depending on the protomer) (Fig. 1D). One likely reason for tight RNA sequestering is to prevent immune recognition. Tightly packaged RNA does not constitute a target for the innate immune system such as Toll-like receptors, which are present during transport of the nucleocapsid along the endosomal pathway into the cytoplasm to its replication site (16, 17) and is protected from exonuclease activities triggered by the interferon antiviral defense system (16, 18). In addition, complete genome protection could be crucial during transport of nucleocapsids to the site of virus assembly and budding.

Fig. 1.

Overall structure of the N-RNA complex. (A) Ribbon diagram of the 11-nucleotide oligomer N-RNA ring structure as viewed from the bottom. Each N protomer is colored differently. The RNA is shown as a black coil. (B) View of the inside of the ring structure; only six protomers are depicted in different colors with the NTD on the top and the CTD on the bottom. The path of the RNA is shown as a coil in a clockwise 5′ to 3′ orientation. The top arrow indicates that the NTDs do not interact with each other; the bottom arrow points toward the extensive interface between CTDs. (C) Ribbon diagram of the N protomer; the NTD is shown in dark blue and the CTD in light blue. The helical secondary structure elements are numbered consecutively. (D) Space-filling model of the N protomer reveals that the RNA is completely clamped at the interface of the NTD and the CTD and thus is not accessible as a template for the polymerase.

Fig. 2.

Nucleoprotein-RNA interaction. (A) Stereo view of a close up of the interaction between N and the RNA. The RNA is shown as stick model and the protein moiety as a coil; the NTD is depicted in dark blue and the CTD in light blue. The nucleotides with bases facing the solvent are drawn in yellow and the bases facing the protein in green. The central nucleotide at the tip of the left-hand helical RNA segment is shown in red. Residues contacting the phosphates and ribose moieties are shown as sticks. (B) Schematic drawing highlighting the protein-RNA interactions with the phosphates and the ribose moieties. Color coding and orientations of the bases are the same as in (A). Most interactions are with the ribose-phosphate backbone.

Fig. 3.

Two hinge regions stabilize polymerization of N. (A) Space-filling model of the N-RNA ring as viewed from the side. Each protomer is shown in a different color. This indicates that the NTD hinge from protomer M reaches over to M+1 and its CTD hinge reaches over to M–1. This arrangement leads to the interaction of both M–1 NTD and M+1 CTD with each other and the surface of protomer M. (B) A close up of this interaction. Protomer M is shown in a light gray for clarity; both hinge regions are drawn as coils; and key residues implicated in interaction are shown as sticks.

The experimental electron density map showed continuous density representing the sugar-phosphate backbone and averaged densities for the bases indicating a 5′-3′ direction in clockwise orientation in the ring-structure (Figs. 1B and 2A). Although each N-RNA ring bound short random RNA from the expression host, differences between purines and pyrimidines could be discerned in the electron density map based on the angles and the sizes of the nucleotides, and they were modeled as either adenine or cytosine (Fig. 2A). The RNA strand is twisted clockwise in an irregular left-handed helix along the inner perimeter of the ring within the continuous cleft made up by the NTD (top) and CTD (bottom) interface in the ring structure (Fig. 1B and fig. S2). For each protomer, the RNA strand is roughly split into two halves, which wrap around the NTD jaw; this arrangement provides most of the interactions and forces the RNA to bulge out at the tip of the NTD jaw (Figs. 1D and 2A). Each nucleotide makes polar contacts either via its phosphate group (seven out of nine phosphates are recognized), as predicted (19), or its ribose moiety (two out of nine are recognized), which is modeled in the C3′ endo sugar pucker conformation (Fig. 2, A and B). The first three bases point toward the solvent and stack onto each other and onto the last two bases of the preceding N protomer (M–1). Nucleotide 5 is involved in a kink that allows the base of nucleotide 4 to stack onto the bases from nucleotides 6 and 7 pointing toward the protein moiety, while its own base points away from the protein. The last two bases then point again away from the protein (Fig. 2, A and B). Most of the basic residues involved in RNA coordination are conserved between rabies virus and vesicular stomatitis virus (genus Vesiculovirus), indicating a conserved RNA coordination network (fig. S1).

There are no significant interactions between the NTDs within the ring structure (Fig. 1B). Their position is mainly determined by the connection to the CTD, the bound RNA, and extensive crystal contacts between NTDs of the two rings in the crystal asymmetric unit. In contrast, the CTDs share a large interaction surface (2700 Å2 total buried surface) (Fig. 1B) that is determined by hydrophobic and van der Waals contacts as well as multiple polar interactions (total of 15) involving many main chain contacts (total of nine contacts).

Two small subdomains emerge from NTD and CTD (Fig. 1C) and reach over to neighboring protomers, contacting them either clockwise (NTD; M+1) or counter clockwise (CTD; M–1) and so establish domain swapping (Fig. 3A). The extreme N terminus folds into a short β hairpin that nestles between the 3/10 helix η1 and helix 8 (fig. S1) of the CTD from M+1. The main contacts are made between the N11 and E266, as well as E20 and R254 in some monomers, in addition to hydrophobic interaction (V10, I22) (22). The β hairpin is followed by a stretch of residues (23 to 29) that are completely solvent exposed and constitute a potential hinge region between the subdomain and the core NTD, while Y28 and Y30 are used as anchor residues (Fig. 3B).

A second small subdomain emanates from the CTD as a coil region followed by helix 13 that reaches over to a neighboring protomer (M–1) in counterclockwise fashion (Fig. 3, A and B). The connection from helix 13 back to the core of the main CTD is disordered in all protomers (Fig. 1C). The CTD subdomain is much shorter and is firmly attached to the CTD by a hydrophobic core (F350, F349, F355), which is followed by a shorter potential hinge region (residues 351 to 356). The following loop region leading to helix 13 interacts with M–1 (R361 to E403; R357 to E403) and is in close contact with β strand 1 from M+1 (Fig. 3B).

The structural overlay of the Calpha atoms of the 11 protomers reveals that the two swapped domains and an NTD region present the highest level of flexibility (fig. S3); this supports the notion that the two subdomains could act as potential hinges and lead to lateral opening of the ring. This flexible domain linkage thus permits not only the formation of differently sized ring-like structures but also the assembly of the nucleocapsid that adopts a diameter of ∼75 nm, containing ∼53 protomers per helical turn (12).

The tight sequestering of the RNA suggests that the rabies virus genomic RNA has to dissociate in order to become a template for the polymerase. However, simultaneously, the RNA must remain close to N, because the polymerase stays attached to N during its activity (4, 5). Although each protomer spans a distance of 22 Å between nucleotide ends (fig. S2), the nine nucleotides could be stretched out to a length of ∼49 Å. Thus, a local dissociation of the RNA from one or several N protomers could provide sufficient space for binding of the polymerase complex (5) (fig. S4).

Phosphoprotein P links the polymerase complex to N and binding requires that S389 be phosphorylated within a highly mobile region on the CTD of N (Fig. 1C and fig. S1) (21). This flexible region is disordered in the structure (Fig. 1C), but may become ordered upon P binding (22), as has been observed for the measles virus N-P interaction (23). P binding in the region of S389 (Fig. 3, A and B) may thus affect both subdomains including the hinge regions and could transfer a signal to the NTD, which would lead to the vertical opening of the NTD-CTD clamp, facilitating RNA exposure. The absence of any protein-protein contacts between NTDs (Fig. 1B) would facilitate a large NTD motion required for this process. In contrast, the CTD conformation is locked in by the large CTD-CTD interaction surface and by the tethering of the hinge loops. The latter interactions will thus maintain the contacts between the N protomers during the passage of the polymerase and the displacement of the RNA. Maintaining contact between N and RNA is required for rebinding of RNA and thus prevents the formation of double-stranded RNA during transcription and replication, which would otherwise require a helicase activity that is not encoded in the viral genome. Furthermore, an extensive production of double-stranded RNA would induce toll-like receptor 3 (TLR3) recognition and trigger immune reactions (16).

Even though N spontaneously associates with RNA in vivo in the absence of other viral proteins, a number of studies have shown that the 5′ region of the viral genomic sequence regulates RNA packaging into a functional nucleocapsid (11, 24). Because the structure does not reveal any sequence specificity in RNA recognition, an alternative RNA-binding mechanism must exist. In the first protomer of the assembled nucleocapsid, the side extending the NTD hinge loop will be exposed and will not participate in polymerization (Fig. 1C); consequently, this region may provide a surface for sequence-specific RNA recognition. Conversely, at the 3′ end of the packaged genome, the CTD hinge loop (Fig. 1C) may adopt a different conformation that may facilitate access of the polymerase to the 3′ end, which is the start site for replication and transcription (25).

Electron micrographs of nucleocapsids from measles virus (Paramyxoviridae) and Marburg virus (Filoviridae), as well as an RNA-free crystal structure from the Borna virus (Bornaviridae) nucleoprotein, suggest that these nucleoproteins also adopt a two-domain structure (26, 27). This suggests that these enveloped viruses use an RNA-sequestering mechanism similar to that observed for the rabies virus N-RNA complex. The N-RNA polymer has thus evolved as the ideal template for the polymerase activity, which exposes the genomic RNA only temporarily to the host cell defense systems during replication. The tight sequestering of RNA observed in the crystal structure suggests further that the closed N-RNA conformation might be stabilized or frozen by small molecules, which could thus act as antiviral agents preventing rabies virus replication.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article