Structural dynamics of SARS-CoV-2 nucleocapsid protein induced by RNA binding

Abstract

The nucleocapsid (N) protein of the SARS-CoV-2 virus, the causal agent of COVID-19, is a multifunction phosphoprotein that plays critical roles in the virus life cycle, including transcription and packaging of the viral RNA. To play such diverse roles, the N protein has two globular RNA-binding modules, the N- (NTD) and C-terminal (CTD) domains, which are connected by an intrinsically disordered region. Despite the wealth of structural data available for the isolated NTD and CTD, how these domains are arranged in the full-length protein and how the oligomerization of N influences its RNA-binding activity remains largely unclear. Herein, using experimental data from electron microscopy and biochemical/biophysical techniques combined with molecular modeling and molecular dynamics simulations, we show that, in the absence of RNA, the N protein formed structurally dynamic dimers, with the NTD and CTD arranged in extended conformations. However, in the presence of RNA, the N protein assumed a more compact conformation where the NTD and CTD are packed together. We also provided an octameric model for the full-length N bound to RNA that is consistent with electron microscopy images of the N protein in the presence of RNA. Together, our results shed new light on the dynamics and higher-order oligomeric structure of this versatile protein.

Fig 1. The N protein forms high molecular-weight complexes with RNA but is a dimer in solution in the absence of RNA.

A- SEC showing that the untreated N protein sample (black line) elutes as a single peak in the void volume of the gel filtration column whereas the urea-treated sample (red line) elutes as a major peak of a higher elution volume. The purity of these proteins samples was evaluated by SDS-PAGE, which shows single bands of the expected size for the recombinant N protein (inset). For urea-treated samples, all experimental analysis were carried out using fractions corresponding to the center of the major peak, since the minor peak likely corresponds to the protein still carrying nucleic acid. B- Native gel electrophoresis stained with ethidium bromide (upper panel) or Coomassie blue (lower panel) showing that the N protein forms high molecular-weight complexes with bacterial RNA, as treatments with RNase A, but not DNase I, dissolve these complexes (arrowhead). The RNase A treatment also changed the oligomeric forms of N protein (arrowheads) in solution (lower panel). C- SEC-MALS analysis of urea-treated N protein in the absence or presence of polyA15 or polyA50. The chromatogram shows that the N protein treated with urea as well as in the presence of polyA15 are dimers in solution, with a mean molecular mass of 89.5 ± 6.3 kDa. In addition to the peak of dimers, urea-treated N presents a higher molecular-weight peak (338.5 ± 56.8 kDa) and a lower molecular-weight peak (18.5 ± 0.1 kDa) D- DLS measurements showing the Rh of untreated N protein carrying bacterial RNA (red boxplot), N protein treated with RNase A (brown boxplot) and N protein treated with urea in the absence (green box plot) or presence of polyA15 (blue boxplot) or polyA50 RNA (purple boxplot). Statistical comparison was performed using Dunnett-Tukey-Kramer pairwise multiple comparison test adjusted for unequal variances and unequal sample sizes and P value was estimated from the confidence interval (see Dynamic Light Scattering in Methods for details). *P<0.05 comparing urea-treated N with the other conditions. E- CD plot showing that CD curves of N protein samples treated with RNase A (blue) and urea (red) are compatible These curves are more similar to each other than the one of N protein samples carrying bacterial RNA (black).

Fig 2. Dimers of full-length N show an extended conformation in the absence of RNA.

A- Representative negative stain image of urea-treated N in the absence of RNA. Insets on the right present representative class averages from a total of 100 classes produced by the 2D classification of 3500 particles. The area corresponding to each particle is presented. B- Density plots of area distribution estimated for simulated 2D reprojections from NTD monomer structure (PDB ID: 6M3M), CTD dimer structure (PDB ID: 6WZO) or full-length dimer reprojections, or for class averages of urea-treated N. In the case of full-length dimer reprojections, when N protein structured domains (NTD or CTD) were presented as separated density regions in the reprojections, the area of these regions were computed separately. For this reason, we can see three main peaks in full-length N dimer distribution: the first peak corresponding to isolated NTD domains, the second to isolated CTD domains and the third to combinations of these domains close to each one in the 2D reprojected image (see Tools for comparing CG simulations with biophysical or electron microscopy data in Methods for details). C- Representative frames of the minimum (upper panel), mean (middle panel) or maximum (lower panel) Rg obtained from the CG simulations of the full-length N protein dimer. 2D simulated reprojections from each frame that were used to estimate area are presented at the left. N protein domains are numbered. D- Boxplot of the Rg estimated from five independent CG simulations of full-length N protein dimer in absence of RNA.

Fig 3. Representative negative stain micrographs of untreated N protein samples containing bacterial RNA.

A- Detail of large helical-like structures with 20–30 nm width. B- Representative micrograph showing a wide range of particles and aggregates. Smaller particles, with less than 20 nm in diameter are shown (colored boxes). In insets, class averages (1 to 7) of picked particles with sizes up to 20 nm in width from one hundred electron micrographs: class-1 (toroid-like particles), class-2 (square-like particles), classes 3 to 5 (elliptical and round-like particles), and classes 6 and 7 (rod-like particles). The percentage of each class average in relation to all picked particles are indicated.

Fig 4. CG dynamics simulations of full-length N dimers in the presence of RNA of varying lengths.

A- Representative frames, obtained from CG simulation, of mean Rg for N protein dimer in complex with a single strand RNA molecule with 10, 30 or 60 nucleotides. N protein monomers are colored in red and green whereas the RNA is colored in yellow. B- Density plot of the Rg distribution (in nm) calculated from five independent CG simulations of the N protein dimers in the presence of single-strand RNA of 10 to 70 nucleotides (nt) long. The median of distribution is indicated by a vertical line. C–Density plot of the frequency of contacts between N protein and any RNA nucleotide during the CG simulation. For instance, a frequency of 100% for N protein residue 260 means that this residue, from any of the N protein monomers, makes contact with at least one nucleotide of the RNA segment in all the analyzed frames of the CG simulations. The distance cut-off used to consider a protein-RNA contact in CG model was 10 Å between the alpha carbon of the protein residues and any atom of the RNA.

Fig 5. The full-length N dimer undergoes domain compaction in the presence of RNA.

A- 2D classification of 24137 toroid-like particles picked with xmipp3. Thirty class averages used for the 3D model are shown in the main panel, whereas raw projections of particles representing the classes are presented in the insets. B- 3D density map reconstruction of the toroid-like particles with the flexible fitting of the N dimer atomic model derived from the CG simulations performed with the 60 nt-long RNA. Upper inset shows SR linker (residues 183 to 195 in sticks) contact with RNA whereas the lower inset shows the interaction between NTD N-terminal tail (residues 26 to 42 in sticks) and RNA. One N protein monomer is colored in orange and the other one is colored in cyan. RNA is colored in yellow. Flexible regions of the NTD and CTD tails outside the density map was not presented for clarity. The complete fitted structure is presented in S4B Fig. C- The same N protein orientations as presented in B, but only showing structured regions (NTD and CTD) of N dimer without its non-structured regions (NTD tail, SR-linker and CTD tail). The two NTD units are represented above the CTD dimer. Upper inset shows, as spheres, positively charged residues of the CTD region (residues 248 to 280) that was previously demonstrated to be implicated in RNA binding [28,36]. Lower inset shows the superposition of the NTD from the structural model with the NMR NTD structure complexed with a 10-mer RNA in magenta (PDB code 7ACT [23]).

Fig 6. 2D Classification and 3D density map reconstruction of the square-like particles.

A- 2D classification of square-like particles picked with xmipp3. Class averages are sorted by the number of class members (number increase up and left). Raw projections of particles that compose the classes are shown on insets. B- Last frame of N protein octamer CG simulation in the presence of 60-nt RNA for each dimeric unit. Reprojections from the 3D structure are shown on the right. C- 3D-density map of the square-like particles in comparison to the map of toroid-like particles, both rendered at 3-sigma contour level.

Fig 7. All-atom molecular dynamics simulation of the N protein C-terminal tail.

A- and B- Contact maps for each individual monomer (a or b) and C- for C-terminal tail dimer. D-, E- and F- Representative structures of contact maps. The monomer contact maps (a or b) is an average of the whole trajectory of five MDs, considering only the last 50 ns of each trajectory for making the dimer contact map. The scale is defined as the contact persistence from 0 (none) to 1 (along all the simulation). D- and E- Structure of each monomer colored from N-terminal in red to C-terminal in blue, some Cα atoms in green VDW representation as reference of the residue numbers. F- A representative structure of the dimer. Each monomer is in new cartoon representation, one in yellow and the another in green. In purple, the hydrogen atoms from the residues 395 to 402 suggested to participate in the dimer interface (grey transparent regions in C, see [27]). Spheres in blue and in red indicate the N- and C-terminal of the C-terminal tail used in the simulation.