Access to RNA Encapsidated in the Nucleocapsid of Vesicular Stomatitis Virus

Abstract

The genomic RNA of negative-strand RNA viruses, such as vesicular stomatitis virus (VSV), is completely enwrapped by the nucleocapsid protein (N) in every stage of virus infection. During viral transcription/replication, however, the genomic RNA in the nucleocapsid must be accessible by the virus-encoded RNA-dependent RNA polymerase in order to serve as the template for RNA synthesis. With the VSV nucleocapsid and a nucleocapsid-like particle (NLP) produced in Escherichia coli, we have found that the RNA in the VSV nucleocapsid can be removed by RNase A, in contrast to what was previously reported. Removal of the RNA did not disrupt the assembly of the N protein, resulting in an empty capsid. Polyribonucleotides were reencapsidated into the empty NLP, and the crystal structures were determined. The crystal structures revealed variable degrees of association of the N protein with a specific RNA sequence.

Vesicular stomatitis virus (VSV) is a member of the Rhabdoviridae, a family of viruses that also includes Rabies virus (RABV). The rhabdoviruses are a subset of the broad group of negative-strand RNA viruses (NSRV) which includes Ebola virus, influenza A virus, and mumps virus, among others. VSV has a single 11-kb RNA genome of negative polarity (15). The genome of VSV, in similarity to the genomes of other NSRVs, is perpetually encapsidated by the viral nucleocapsid (N) protein. Five proteins are encoded by the genome of VSV, including the N protein, the phosphoprotein (P), the matrix (M) protein, the glycoprotein (G), and the large polymerase subunit, L. The virion of VSV and its components have been extensively studied by structural means for decades. In recent years, through elegant studies using X-ray crystallography and nuclear magnetic resonance (NMR) techniques, structures have been published for each of the viral proteins, excluding the L protein. These studies have led to a better understanding of the role of these proteins and have also stimulated new questions about their functions. Specifically, questions still remain regarding the interaction of the viral template with the polymerase that is the complex between the L and P proteins (6).

An expression system was previously developed in Escherichia coli for the expression of the VSV N and P proteins (12). Ultimately, this system allows the large-scale purification of nucleocapsid-like particles (NLP) of the assembled N protein with encapsidated RNA. The structure of the VSV NLP (13) and its equivalent from RABV (2) were determined. The crystal structures revealed that the N protein contains an N-terminal lobe (N-lobe) and a C-terminal lobe (C-lobe). The N-lobe is preceded by an extension, the N-arm, of 22 amino acids. Also, located within the C-lobe is an extended loop, the C-loop. The N-arm and C-loop are critical for the formation of the nucleocapsid, which is created by the polymerization of the N protein concomitant with RNA encapsidation at the site of replication (25). In the nucleocapsid structure, each monomer of the N protein makes cross-molecular contacts with three neighboring monomers. These contacts include the following (each viewed with the opening of the RNA cavity facing away and the C-lobe residing above the N-lobe): (i) the interaction between the N-arm and the C-lobe on the proximal surface of the left neighboring subunit, (ii) the C-loop interaction with the C-lobe of the neighboring subunit to the right, and (iii) the N-arm interaction with the C-loop of the N protein subunit two units away on the left. These three unique interactions are repeated along the assembled nucleocapsid.

The structures of the N proteins also revealed details about how the RNA is stored, as well as the structure of the RNA in the nucleocapsid. The N- and C-lobes are angled together to form a cavity for encapsidating RNA. Each N protein was shown to encapsidate nine ribonucleotides in rhabdoviruses. The encapsidated RNA forms two quasi-helical structures, where bases one to four (counting from 5′ to 3′) are stacked to resemble a single strand of a type A RNA duplex and face the opening of the cavity, while the bases of nucleotides five, seven, and eight are stacked but face the interior of the cavity. Of the remaining bases, six is swung out to face the solvent side of the cavity and the base of nucleotide nine is located between neighboring N subunits and is transitioned between the two helical RNA structures in VSV but stacked with the first base of the adjacent subunit in the RABV structure. This base may alter its orientation when the nucleocapsid is in a more extended conformation. When the N proteins are aligned as in the nucleocapsid, they form a continuous tunnel to store the encapsidated viral RNA (9). The RNA is completely sequestered upon encapsidation and has proven to be stable against degradation by nuclease and base-catalyzed hydrolysis in previous reports (5, 12, 16, 17). This poses a dilemma for how the polymerase gains access to the RNA for the processes of transcription and replication.

The nucleocapsids of the negative-strand RNA viruses are unique, since they not only serve as a capsid for housing the genomic RNA but also serve along with the RNA as the active template for viral transcription and replication. Because of this, the N protein not only has a role to protect the genome, but it also has to allow the polymerase to gain access to the genome during the processes of polynucleotide synthesis. The N protein must shuffle between two conformations, one of which allows the temporary local release of the RNA to the polymerase. Thus, the nucleocapsid must be able to release and subsequently reencapsidate the genome. It has been shown through mutational studies that deletion of the N-arm diminished RNA encapsidation by the N protein but does not completely disrupt the assembly of an empty NLP (25). That study suggested that without the N-arm, perhaps the two lobes of the N protein cannot be brought together to maintain the proper cavity for encapsidating RNA. It is possible that disrupting the interaction of the N-arm with the C-lobe might lead to the conformational change that allows access to the RNA. Currently, the cause leading to this disruption is unknown but could be related to P or P/L binding to the nucleocapsid since the binding site for the P protein on the nucleocapsid is adjacent to the N-arm (11).

With the idea that N could temporarily allow access to the encapsidated RNA if the proper conditions are met, we reinvestigated the accessibility of RNA from our E. coli-expressed VSV NLPs, as well as authentic viral nucleocapsids. Here, we present the results of a series of experiments demonstrating that the RNA within each particle is susceptible to RNase A digestion. Study of the resulting particles by electron microscopy and X-ray crystallography shows that removal of the RNA did not disrupt the assembly of the N protein, resulting in an empty capsid. Empty capsids were then successfully used to reencapsidate homogeneous populations of RNA. We were able to use the empty capsid to reencapsidate polynucleotides and solve the crystal structures of the corresponding NLPs. These experiments support a model where the N protein can reversibly encapsidate and release the genomic RNA to serve as the template for transcription or replication.

MATERIALS AND METHODS

NLP expression and purification.

The expression plasmid pET N/P (12), containing the coding sequences of the N protein and the P protein, was transformed into E. coli strain BL21(DE3). The cell culture was grown at 37°C in LB broth (EM Science) in the presence of 0.1 g/liter ampicillin. When the A₆₀₀ reached 0.6 to 0.8, it was induced by 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) and was allowed to grow for 8 h at 29°C. The harvested cells were sonicated in 20 mM Tris buffer (pH 7.9) containing 500 mM NaCl. The soluble portion was purified with a Ni-affinity column according to the Novagen protocol. The N/P complex sample was eluted from the Ni column and was dialyzed in a buffer containing 100 mM citrate (pH 4.0) with 250 mM NaCl to dissociate the N and P proteins. After removal of the precipitated P by centrifugation, the sample containing only NLP was dialyzed in a 50 mM Tris buffer (pH 7.5) containing 300 mM NaCl and was further purified by size exclusion chromatography on a Superdex-200 column.

VSV infection and nucleocapsid purification.

In order to isolate the authentic viral nucleocapsid, 50 plates (100 mm) of BHK-21 cells were infected by VSV stock (S1) at a multiplicity of infection (MOI) of 4 and were allowed to absorb for 45 min. Cells were then incubated at 33°C for 16 h. Infected cells in medium were harvested by centrifugation at 1,500 rpm for 3 min. Cells were resuspended in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.4) to a concentration of 2.5 × 10⁷ cell/ml. Cells were disrupted by three cycles of freezing and thawing and then centrifuged at 6,500 rpm for 10 min to remove cellular debris. The supernatant was collected and purified by CsCl gradient centrifugation. A 20% to 40% (wt/vol) CsCl gradient was made in TNE (20 mM Tris-HCl, 100 mM Nacl, 1 mM EDTA, pH 7.4) using Beckman SW 41 tubes. Three milliliters of 5% sucrose (in TNE) was added on top of the CsCl gradient. Three milliliters of cell lysate was laid upon each tube and centrifuged at 32,000 rpm for 16 h at 12°C. Three major bands were visible. The bottom band, which contained the RNP, was collected for RNase A digestion. Purity of the nucleocapsid was checked on a low-Bis (0.13% Bis) 10% sodium dodecyl sulfate-polyacrylamide gel. The resulting gel was stained with Coomassie brilliant blue.

NLP and nucleocapsid RNase A digestions.

Five mg/ml purified NLP was digested by RNase A (final concentration, 25 μg/ml) for 1 h at room temperature or 30°C. Samples were also digested by RNase A (final concentration, 1.0 mg/ml) for 1 h at room temperature, 30°C, 37°C, 42°C, or 60°C. One mg/ml purified VSV nucleocapsid was digested by RNase A (final concentration, 25 μg/ml) at 30°C for 1 h or RNase A (final concentration, 1.0 mg/ml) at 37°C for 1 h. RNA extraction was performed following the Trizol reagent (Invitrogen) protocol. A 0.3-mg protein sample was used for each extraction. Briefly, after phase separation by chloroform, the RNA in the aqueous phase was precipitated by addition of isopropyl alcohol. The precipitated RNA was washed, briefly dried, and visualized by running a 10% polyacrylamide-8 M urea gel with ethidium bromide staining.

Electron microscopy.

Protein samples were diluted to a concentration of 0.05 mg/ml. These samples were then negatively stained with 2% aqueous uranyl acetate for 20 s. Electron photomicrographs were taken on a conventional transmission electron microscope.

NLP reconstitution.

Five milligrams purified NLP (5 mg/ml) was treated by RNase A (Qiagen) (final concentration, 1.0 mg/ml) overnight at 42°C. The digested NLP was repurified by a Superdex S200 column and concentrated to 5 mg/ml. RNase inhibitor (Qiagen) was added to the purified empty NLP at a concentration of 1 U/μl, and the mixture was incubated at room temperature for 30 min. In the presence of RNase inhibitor (Qiagen), the empty NLP was incubated with poly(rA) (Midland) at a molar ratio of 1:5 for 15 min at 42°C. The mixture was then cooled to room temperature for 15 min. The oscillation of temperature was repeated for a total of four times. The newly encapsidated NLPs were applied to the S200 column. This step was to eliminate the majority of exogenous RNA contamination. NLPs were then purified by ion-exchange chromatography on a Mono Q 5/50 GL column (GE). NLPs containing RNA (as monitored by measurements of the ratio of A₂₆₀ to A₂₈₀ [A_260/280]) were pooled, and the last purification on the S200 column was performed. The procedure to encapsidate polyribonucleotide was also performed with poly(rU), poly(rC), or poly(rG) using the same procedure as described above.

Crystallization and structure determination.

Crystallization of VSV NLP has been previously reported (10). Purified N protein devoid of RNA and N protein with encapsidated poly-RNA (A, C, G, or U) were crystallized by the hanging drop vapor diffusion method (20) in 4 to 6% polyethylene glycol (PEG) 3,350 (wt/vol) containing 250 mM sodium chloride buffered with 100 mM sodium acetate (pH 4.5) at 20°C. Crystals were harvested and stepwise soaked in mother liquor containing uranyl acetate to a final concentration of 2.5 mM. Crystals were cryoprotected stepwise in 10% PEG 3,350 (wt/vol) containing 250 mM sodium chloride and 2.5 mM uranyl acetate buffered with 100 mM sodium acetate (pH 4.5) supplemented with glycerol at concentrations of 10, 12.5, 15, and 18% (vol/vol). Prior to flash freezing, the crystals were washed in a final solution 20% PEG 3,350 (wt/vol) containing 250 mM sodium chloride, 2.5 mM uranyl acetate, and 20% glycerol and buffered with 100 mM sodium acetate (pH 4.5). Diffraction data were collected at the Stanford Synchrotron Radiation Laboratory on either beamline 9-2 or 11-1 on a MAR325 charge-coupled device (CCD) detector with an oscillation of 0.3° and a crystal-to-detector distance of 450 mm. Raw data frames were processed with the HKL2000 software package (22), and structure factors were calculated with TRUNCATE (8). Phasing for each structure was determined by molecular replacement with the MolRep software program (24) using the previously determined VSV N structure (Protein Data Bank accession number 2GIC). For these studies, RNA was removed from the model coordinates. Model rebuilding was performed with the COOT software program (7). Refinement was performed with PHENIX.REFINE (1). No outliers were found in the Ramachandran plot. For refinement statistics, see Table 2.

Protein structure accession number.

The atomic coordinates have been deposited in the Protein Data Bank (PDB), www.rcsb.org. PDB identifier (ID) codes are as follows: 3PTO, 3PTX, 3PUO, 3PU1, and 3PU4, corresponding to empty NLP, NLP-poly(rA), NLP-poly(rC), NLP-poly(rG), and NLP-poly(rU), respectively).

RESULTS

Removal of RNA by RNase A.

When the N protein of VSV is coexpressed with the P protein in E. coli, NLP is produced in which a single strand of RNA of E. coli origin is encapsidated (12). The crystal structure of the NLP, containing 10 subunits of the N protein and a single 90-mer of RNA with random sequence, shows that the RNA is buried in the continuous cavity formed by the assembly of N protein (13). Given that the RNA encapsidation cavity has a closed conformation in the crystal structure, it appears that the RNA may be inaccessible when encapsidated. However, a protein-RNA complex like the NLP is a dynamic structure that allows movements of a protein domain to expose the RNA inside. In the case of the rhabdoviruses, the N-terminal domain of the N protein is held to its neighboring subunits by an N-terminal arm that interacts with the C-terminal domain of the following subunit (contact I) and the extended loop of the C-terminal domain in the subunit further down (contact III). Our recent results showed that deletion of the N-terminal arm diminished RNA encapsidation by the NLP but did not completely disrupt the assembly of an empty NLP (25). It is therefore conceivable that the N-terminal domain may open and expose the RNA if the N-terminal arm is temporarily released. By incubation of the purified NLP with RNase A at a final concentration of 1 mg/ml for 1 h, the RNA was shown to be digested. The level of degradation increased as the temperature was increased from room temperature to 60°C (Fig. 1a). At 60°C, the NLP contained no RNA, as shown in Fig. 1a, lane 6. The empty NLP retained the ring-like structure, as shown by electron microscopy (EM) (Fig. 1c). Similarly, the viral nucleocapsid isolated from VSV-infected cells was also treated with RNase A at 37°C for 1 h, and the viral RNA was completely removed from the nucleocapsid (Fig. 1d). The assembly of the VSV nucleocapsid remained unchanged with or without RNA, as shown by EM (Fig. 1f). These observations are consistent with those of NLP and nucleocapsid of RABV (5, 12, 16, 17). The difference is that in the case of the VSV NLP, no short pieces of RNA were left as in the RABV NLP based on the A_260/280 ratio of the RNase A-treated NLP (Table 1). The RABV N protein has a longer α-helix near the RNA cavity, which may be the reason for retaining the short pieces of RNA (18). The images of the nucleocapsid are also consistent with those of electron micrographs from previously published studies (23).

FIG. 1.

RNA analysis and electron microscopy of VSV nucleocapsid-like particles (NLP) and viral nucleocapsids digested with RNase A. (a) RNA electrophoresis. Purified VSV NLP treated with RNase A (1 mg/ml, final concentration) at room temperature (lane 3), 37°C (lane 4), 42°C (lane 5), or 60°C (lane 6) for 1 h. The RNA from each sample was phenol extracted and analyzed by a 10% polyacrylamide-8 M urea gel. A control of the RNA from the untreated NLP is shown in lane 2. NLP (lanes 3 to 6) in panel b were treated for 1 h with an RNase A final concentration of 25 μg/ml (lanes 3 and 5) or 1 mg/ml (lanes 4 and 6). Digestion experiments were carried out at room temperature (lane 3 and 4) or 30°C (lane 5 and 6). A control of the RNA from the untreated NLP is shown in lane 2. (c) Electron micrographs of the empty NLP (lower panel) following treatment with RNase A at 60°C for 1 h. Empty particles retained the ring-like structure characteristic of RNA-containing NLP (upper panel). (d) The purified VSV nucleocapsid was treated with RNase A (1 mg/ml) at 37°C for 1 h (lane 3). RNA was completely removed compared to RNA isolated from the untreated nucleocapsid (lane 2). (e) The purified VSV nucleocapsid was treated with RNase A (25 μg/ml) at 30°C for 30 min (lane 3). The untreated sample is shown in lane 2. (f) Shown are electron micrographs of purified VSV nucleocapsid (upper panel) and RNase A-treated VSV nucleocapsid (lower panel). RNA molecular mass markers are shown in lanes a1, b1, d1, and e1, with sizes listed to the left.

TABLE 1.

Ratios of UV absorbance at 260 nm to that at 280 nm for N protein and N protein with specific RNA species

Protein	A260/280
Empty N protein	0.71
N protein with poly(rA)	1.25
N protein with poly(rC)	0.95
N protein with poly(rG)	0.96
N protein with poly(rU)	0.85
N protein with random E coli RNA	0.97

The nuclease susceptibility of RNA encapsidated by VSV NLPs and nucleocapsids is in contrast to previous experiments by our group and others that showed the VSV nucleocapsid is resistant to nuclease digestion (4, 5, 12, 16, 17, 19). The current experiments were performed with a much higher concentration (1 mg/ml) of RNase A, which is 40-fold more than the amounts used in the studies described in references 4 and 19. In addition, digestions were performed at temperatures exceeding 37°C. In order to demonstrate that these factors facilitate RNase susceptibility and also to demonstrate that our current nucleocapsid samples share the hallmarks of decades of documented experiments, we performed digestion experiments on NLP and purified VSV nucleocapsids at a concentration of 25 μg/ml RNase A. Results of these experiments are shown in Fig. 1B and E, respectively. As with previously published experiments, this subset of experiments showed that the N protein in our purified NLP and nucleocapsid samples could protect RNA in the same manner as previously reported. At room temperature and 30°C, RNA within the NLP is protected at the lower RNase A concentration while NLPs with 40-fold more RNase A showed some degradation (Fig. 1B). Purified VSV nucleocapsids incubated with lowered RNase concentrations at 30°C were also largely protected against RNase digestion.

Reconstitution of NLP with sequence-specific RNA.

An empty NLP was generated by removal of the RNA by incubation with RNase A overnight at 42°C. The question was then asked if it was possible for the empty particles to encapsidate RNA. So, a method was developed to reconstitute an NLP with RNA that has a specific RNA sequence using the empty NLP. By reconstituting an NLP with sequence-specific RNA, we could be able to study the interactions between the N protein and RNA with a specific sequence. Purified empty NLP was incubated with poly(rA) (250 nucleotides [nt] or longer) in a 1:5 molar ratio for 15 min at 42°C in the presence of an RNase inhibitor. The mixture was cooled at room temperature for 15 min. The two-step procedure was repeated three additional times. The reconstituted NLP was purified by size exclusion and ion-exchange chromatography. The final purification by size exclusion chromatography showed that the reconstituted NLP eluted at the same time as the original RNA-encapsidating NLP. SDS-PAGE gel analysis also showed that the N protein was intact following RNase digestion and reconstitution (Fig. 2a). Likewise, A_260/280 measurements of purified particles authenticated that RNA was indeed encapsidated into the formerly empty particles (Table 1). The same procedure was applied to reconstitute an NLP with poly(rC), poly(rG), and poly(rU). The reconstituted NLPs were crystallized under the same conditions as for the E. coli-expressed NLPs. An example of crystals grown from NLPs reconstituted with poly(rG) RNA is shown in Fig. 2b. The length of the poly-RNA species used for the reconstitution experiments was longer than the 90-mer required to fill the empty NLP. Analysis of the RNA isolated from reconstituted NLP-poly(rA) showed that the RNA was approximately 90 bases in length (data not shown). This suggested that RNA outside the NLP was still susceptible to nonspecific degradation. The relatively higher A_260/280 ratio may reflect the more stable base stacking of poly(rA) (see below).

FIG. 2.

(a) SDS-PAGE of NLP. Lane 1, protein molecular mass marker; lane 2, untreated VSV NLP; lane 3, empty NLP after RNase A treatment; lane 4, reconstituted VSV NLP containing poly(rA) RNA. (b) Image of crystals of NLP with encapsidated poly(rG) RNA.

N-RNA structures.

In order to observe any changes in the NLP upon encapsidation of specific RNA, reconstituted NLPs were the subject of crystallographic studies. Empty NLPs were used to encapsidate oligomers of either poly(rA), poly(rC), poly(rG), or poly(rU) RNA. Each sample, in addition to empty NLPs, was crystallized, and X-ray diffraction data were collected. Subsequently, crystal structures of each of these samples were determined. The structure of N encapsidated with each specific RNA is shown in Fig. 3. Data and refinement statistics are listed in Table 2 . The overall structure of each of these NLPs is similar to those previously published (13, 25). The stoichiometric state of the NLP remained that of a 10-mer with each monomer of the N protein encapsidating the typical nine bases of RNA or a 90-mer per nucleocapsid-like particle. Each of the three types of cross-molecular interactions between N subunits in the nucleocapsid is still observed; thus, there is no change in the nucleocapsid structure upon encapsidating the new species of RNA. To confirm that RNA was indeed not present in the empty NLPs, particles prior to reencapsidation of polyribonucleotides were crystallized. The refined structures indeed confirm that RNA was not present in the empty NLPs. This concurred with A_260/280 measurement, which also suggested a lack of RNA (Table 1).

FIG. 3.

Crystal structures of the VSV N protein encapsidated with specific RNA species are shown in panels A through D, while panel E shows the structure of the empty capsid. A single representative copy of the N protein with poly(rA) RNA is shown in panel A. The N-terminal lobe and C-terminal lobe are colored green and yellow, respectively. RNA is shown in stick model form with bases numbered 1 to 9 (5′ to 3′). A single ribonucleotide, encapsidated by the preceding N subunit within the nucleocapsid-like particle, is labeled 9*. Electron density (2f_o - f_c) is shown as blue mesh at a contour of 1.5 σ, carved at a radius of 2.7 Å around the RNA. (B) N with poly(rC) RNA is shown with density contoured at 1.2 σ. (C) N with poly(rG) RNA is shown with density contoured at 1.0 σ. (D) N with poly(rU) RNA is shown with density contoured at 1.0 σ. (E) Density is shown at a contour of 1.2 σ. Spherical densities potentially correspond to water molecules.

TABLE 2.

Data collection and refinement statistics

Parametera	Value for crystal structure
	N	N/poly(rA)	N/poly(rC)	N/poly(rG)	N/poly(rU)
Space group	P21212	P21212	P21212	P21212	P21212
Unit cell dimensions (Å)
a	165.6	168.0	164.7	163.2	165.5
b	234.1	237.1	235.6	233.2	235.1
c	76.0	76.3	75.9	75.6	75.5
Resolution	50-3.00	50-3.00	50-3.10	45.1-3.14	35-3.00
High-resolution bin	3.11-3.0	3.11-3.0	3.21-3.1	3.26-3.14	3.11-3.0
No. of reflections (total/unique)	154,721/51,615	259,923/48,171	158,332/46,101	190,442/42,182	191,739/51,903
Completeness (%)	86.7 (39.4)	82.5 (33.5)	84.1 (68.0)	82.6 (64.7)	86.5 (45.2)
I/σI	24.67 (2.40)	25.25 (2.99)	20.51 (2.35)	17.115 (2.74)	28.59 (3.16)
Rmerge	0.071 (0.320)	0.096 (0.420	0.083 (0.488)	0.094 (0.55)	0.066 (0.311)
Rcryst	0.2579	0.2117b	0.2504	0.2592	0.2506
Rfree	0.2907	0.2498b	0.2900	0.3051	0.2883
Mean B value (Å2)	100.49	117.89	104.81	97.18	88.22
Model
No. of atoms	16,467	17,527	17,437	17,572	17,437
No. of residues	2,076	2,086	2,086	2,086	2,086
No. of ribonucleotides	-	45	45	45	45
RMS deviations
Bonds (Å)	0.011	0.010	0.011	0.010	0.011
Angles (°)	1.375	1.448	1.443	1.354	1.392

^aI/σI, ratio of the mean of the intensities of all reflections to the mean of the standard uncertainties of the intensities of all reflections; R_merge, R_merge = ΣhΣn|I − <I>|/ΣhΣn<I>, where I is the observed intensity and <I> is the averaged integrated intensity taken over n measurements for reflection h; R_cryst, R_work = Σh||Fo| − |Fc||/Σh|Fo|, where Fo is the observed amplitude and Fc is the calculated amplitude from the model; R_free, an R factor for a selected subset of the reflections that was not included in prior refinement calculations; RMS, root mean square.

^bIncludes TLS refinement.

The previously published NLP structure contained a random series of RNA encapsidated during the expression of the N protein in E. coli (13). In that structure, the encapsidated RNA has the appearance of two quasi-helical structures with two additional bases unstacked. In contrast to the VSV structure, the nucleotide number nine was shown to be stacked with the first base in the RABV NLP structure (2). The RNA in the NLP-polyribonucleotide structures presented here has the two characteristic quasi-helical structures. However, the ninth base of each subunit in the structures containing poly(rA) and poly(rC) has an orientation like that observed in the RABV NLP structure for all of its subunits in the NLP. The ninth base in the poly(G)and poly(U) structure also adopts this orientation in some subunits but is more similar to the previously determined VSV NLP structure in other subunits. Since the density for this base is reduced in these two structures [poly(G) and poly(U)], it appears as though there could be a mixture of the two orientations and in some instances it could be modeled either way.

Each of the polynucleotide strands has the common ribose-phosphate backbone structure. The positively charged residues that coordinate with the backbone phosphate groups largely remain the same in the polyribonucleotide NLPs. Therefore, the structural distinction between RNA strands in NLP is the different bases themselves. In the structure of NLP-poly(rA), the electron density for the RNA is much better defined than that for the mixed ribonucleotide sequence in the previous structure. The structure of NLP-poly(A) is therefore considered to be the best-defined structure of the N-RNA complex. The bases of poly(rA) in NLP are closely stacked in the two quasi-helical motifs. The strong base stacking may keep the poly(rA) structure more rigid in the nucleocapsid. There are three direct interactions observed between 3 of the 45 bases of poly(rA), which is encapsidated by five N subunits and resides in the encapsidation cavity. Each of the bases is in the ninth position and interacts with the atom OD1 of Asn187 of its corresponding subunit. Potential interactions between bases of the RNA and the protein are listed in Table 3. In addition, a uranyl ion was found to coordinate with phosphate groups of nucleotides five and six, leaving the side chain of Arg146 free. The uranyl ion is observed in each of the structures, and as a result the orientation of Arg146 is common among all of the poly-RNA structures but is different from the previously published structure. The structure of poly(rC) in NLP is very similar to that of poly(rA). Both of these structures appear to have a more tightly packed set of RNA half-helices. There are two potential interactions between a single base and two protein side chains. This interaction is observed in only one of the five subunits. This sole cytidine has potential bonds between atoms O2 and N4 with atoms NE and NZ of Arg317 and Lys410, respectively. No potential interactions were observed between the bases of poly(rU) and the N protein. However, one difference is observed for the structure of NLP-poly(rU), as the side chain of Arg317 adopts a rotamer different from that of all previous structures. As a result, this residue does not make contact with the phosphate of ribonucleoside U5. The bases of the poly(rG) structure showed the largest number of interactions with the protein, 11, involving 9 individual bases. There is not a readily observable repeated pattern to the interactions of the bases with the N protein. On the whole, there is a slight variation in the conformation of the N subunits among the five independent subunits in the crystallographic asymmetric unit. This could lead to the lack of common interactions between a single poly-RNA species and each independent N subunit (Table 3). Thus, crystal packing may be a factor responsible for the differences in interactions. Generally speaking, interactions between the N protein and RNA are fluid. For poly(rG) and poly(rU), the RNA structure is much less defined, as the bases of both poly(rG) and poly(rU) are more irregularly stacked although they seem to be somewhat aligned. Ultimately, the degree of base stacking observed in the structure of NLP-poly-RNA is consistent with the calculated stacking energy between two bases in a single-strand RNA (3). The reduced stacking of poly(G) in NLP relative to the theoretical calculation may result from the hydrophilic property of the guanine base.

TABLE 3.

Potential interactions between nucleotide bases and the VSV N protein

Base no.	Atom	Distance (Å)	Residue (chain)	Atom
Poly(rA)
27	N6	3.12	187Asn(A)	OD1
36	N6	3.17	187Asn(E)	OD1
45	N6	3.12	187Asn(D)	OD1
Poly(rC)
40	O2	2.79	317Arg(E)	NE
40	N4	2.68	410Lys(E)	NZ
Poly(rU)
None
Poly(rG)
6	N7	2.69	408Arg(D)	NH1
9	N2	2.81	212Ser(D)	N
9	N2	2.92	211Ala(D)	O
11	N2	2.70	179Arg(C)	NH2
14	O6	2.75	312Arg(C)	NE
19	N2	2.94	187Asn(B)	OD1
19	N2	3.05	187Asn(B)	ND2
23	O6	2.90	312Arg(B)	NH2
24	N7	2.93	408Arg(B)	NH1
35	N7	3.07	143Arg(A)	NH2
45	N2	2.88	162Asn(E)	OD1

DISCUSSION

The active template for transcription and replication for the negative-strand RNA viruses is the genomic RNA, along with the associated nucleocapsid. Naked RNA is an insufficient template for these processes of polynucleotide synthesis. The viral RNA is encapsidated at the site of replication by a form of the N protein that is in complex with the P protein, known as N°P. As the N protein is encapsidating the RNA, the individual subunits of N associate with each other, forming three unique cross-molecular contacts in the nucleocapsid. The contacts include an interaction between the extended N-terminal arm (N-arm) of one N subunit and the C-lobe on the proximal surface of the left neighboring subunit (when viewing the N surface opposite the RNA encapsidation cavity and the C-lobe over the N-lobe). The two additional contacts are an interaction of the C-loop with the C-lobe of the neighboring subunit to the right and an interaction of the N-arm with the C-loop of the N protein subunit two units away on the left. These contacts provide stability to the nucleocapsid, but they are also important for the formation of the RNA encapsidation cavity. The linear association of the N subunits in the nucleocapsid allows the formation of a continuous tunnel that houses the viral RNA. The RNA, as observed in previous crystal structures, is inaccessible to the polymerase if the conformation of the N protein does not change.

In the current study, RNA encapsidated in the authentic nucleocapsid isolated from VSV-infected cells and 10 subunit nucleocapsid-like particles (NLP) from our E. coli expression system were reinvestigated for susceptibility to RNase A in a series of digestion studies. It had previously been shown that encapsidated RNA was stable against degradation by nuclease and base-catalyzed hydrolysis in reports by our group and others (4, 5, 12, 16, 17, 19). In this report, we have demonstrated that RNA can be digested from VSV nucleocapsids with a substantially increased concentration of RNase A concurrent with elevated temperature (Fig. 1a and c). With the current conditions, it was possible to digest RNA from both nucleocapsid structures. Following the removal of RNA, the nucleocapsid and the NLPs were examined by electron microscopy and each nucleocapsid structure appeared to remain intact and was indistinguishable from its RNA-containing predecessor (Fig. 1c and f). This is particularly interesting, since it shows that the nucleocapsid can in some way allow exposure of the RNA without disassembling the global nucleocapsid structure. The structure of the empty NLP was subsequently determined by crystallographic techniques and was shown to be devoid of RNA while maintaining all of the intermolecular interactions of the encapsidated NLP. The empty NLP has a structure similar to that of an RNA encapsidation-defective mutant of N harboring a Ser290 → Trp mutation which was previously determined (25). Both the empty NLP and the mutant structures, however, suggest that information necessary to form the capsid structure is inherent primarily in the N protein alone.

Based on the wild-type structures of the N proteins of VSV and RABV, there has been speculation about how the N protein might undergo a conformational change in order to expose the encapsidated RNA. The RNA encapsidation cavity is located between the N- and C-lobes of the N protein. The contacts between the N-arm and both the C-lobe and C-loop are important for holding the two lobes in the proper orientation to form the cavity. This is inferred from the static crystal structure but is also alluded to by mutagenesis studies in which the N-arm has been removed (25). N missing the first 22 amino acids lost the ability to encapsidate RNA but retained the ability to assemble larger structures. It has been speculated that if the interactions between the N-terminal arm and the neighboring N molecules were disrupted, the N-lobe would cease to maintain the orientation required to form the RNA encapsidation cavity with the C-lobe. Ultimately, this would lead to exposure of the RNA to the polymerase or in the current studies to RNase. The question arises, then, of how the RNA is exposed in our current experiments. The answer could likely be related to the method used to digest the RNA, in which samples were not only treated with larger amounts of RNase A but were also subjected to higher temperatures. The increase in temperature may serve to destabilize the nucleocapsids enough to open up, but not enough to destroy, their global structure. An interesting possibility to release the N-lobe would be through destabilization of the interaction of the N-arm with the C-lobes of adjacent N proteins.

The empty capsids resulting from these studies are interesting because they allow the possibility of introducing specific RNA sequences to the nucleocapsids. During the expression of N and P in the bacterial milieu, N encapsidates random sequences from the diverse pool of RNA in the cell. Specific sequences were introduced to this expression system but weren't encapsidated by the N protein, presumably due to the excess of nonspecific RNA in the cell (unpublished data). In an effort to see if the empty NLP were encapsidation competent, we introduced poly species of RNA [i.e., poly(rA), etc.] to empty NLPs. Following a series of purification steps to eliminate excess RNA from the samples, NLPs were tested for the presence of RNA by A_260/280 measurements. These experiments confirmed that empty NLPs could encapsidate RNA. Ultimately, the resulting NLPs were crystallized and their structures were determined. The resulting structures confirmed the presence of RNA and were essentially indistinguishable from the previously reported N-RNA structures. The RNA adopted the two quasi-helical structure motifs observed previously (Fig. 3).

The current model for how the polymerase gains access to the genomic RNA is unknown. The RNA is inaccessible to the polymerase when the N protein is in the closed conformation. The RNA must be locally removed in order to be read, and then it must be put back inside the nucleocapsid to be used again for the following rounds of polynucleotide synthesis. The experiments presented here show that RNA can be removed and subsequently can be encapsidated by the N protein, thus giving more credibility to such a model. The ability to experimentally dictate the sequences that the N protein encapsidates in vitro will allow us to look at how N interacts with VSV-specific sequences in the future. The sequences could include the genomic termini as well as intergenic junctions.

In the structure of NLPs containing a specific poly-RNA sequence, we had the first glimpse on how an N subunit could interact with a specific RNA sequence. Along the length of the viral genome, each N subunit in the nucleocapsid encounters a unique sequence of 9 nt. For the most part, the N-protein subunits associate with each other to form the capsid to encapsidate any RNA sequence. However, there are potential specific interactions between the N protein and specific regions in the viral RNA that may influence viral transcription or replication. For instance, mutation of Phe348 to Ala348 diminished transcription without compromising replication (14). Similarly, mutation of Tyr289 also only diminished transcription (21). It is unclear how the N mutant could encapsidate viral RNA and allow the initiation of replication but would not support the initiation of transcription from the same template. It could be possible that the accessibility of the replication origin or the promoter of transcription is regulated by interactions between the N protein and the RNA sequence. In the structure of NLP-poly(rA), we observed a rigid RNA structure with two motifs of closely stacked bases. The cavity in the N protein provides the environment for the base stacking in the RNA encapsidated in the nucleocapsid. The sequence-specific stacking energy may add additional stabilization to rigidify the RNA motifs, as well as the nucleocapsid overall. The signal for initiation of transcription following the conserved GA spacer (GA3′UUGUCnnUAG5′) may represent a unique structure of base stacking that allows the polymerase complex to form a stable assembly to initiate mRNA synthesis. This sequence occupies two N subunits in the nucleocapsid. It will be interesting to see if GA and UAG, or any pairs of conserved sequences, are located in the first motif (nucleotides one to four) of stacked bases, since they face the opening of the RNA encapsidation cavity, which might present a unique spot for the polymerase to recognize. The mutations in the N protein that altered transcription may change the stability of the base stacking, which may lead to the failure of forming a transcription initiation assembly by the polymerase complex. Initiation of replication may require a different assembly that is regulated by a different mechanism in which the role of the N protein is also different. Last, the structure of the encapsidated RNA could also affect the mechanism of transcriptional termination. Since the stacking of poly(rU) is much less tight (Fig. 3d), this may form a soft region where the polymerase complex may fall off the template, thus terminating transcription. Additional structures with VSV-specific sequences could shed light on these possibilities.