ABSTRACT
The enveloped negative-stranded RNA virus measles virus (MeV) is an important human pathogen. The nucleoprotein (N0) assembles with the viral RNA into helical ribonucleocapsids (NC) which are, in turn, coated by a helical layer of the matrix protein. The viral polymerase complex uses the NC as its template. The N0 assembly onto the NC and the activity of the polymerase are regulated by the viral phosphoprotein (P). In this study, we pulled down an N0 1-408 fragment lacking most of its C-terminal tail domain by several affinity-tagged, N-terminal P fragments to map the N0-binding region of P to the first 48 amino acids. We showed biochemically and using P mutants the importance of the hydrophobic interactions for the binding. We fused an N0 binding peptide, P1-48, to the C terminus of an N0 21-408 fragment lacking both the N-terminal peptide and the C-terminal tail of N protein to reconstitute and crystallize the N0-P complex. We solved the X-ray structure of the resulting N0-P chimeric protein at a resolution of 2.7 Å. The structure reveals the molecular details of the conserved N0-P interface and explains how P chaperones N0, preventing both self-assembly of N0 and its binding to RNA. Finally, we propose a model for a preinitiation complex for RNA polymerization.
IMPORTANCE Measles virus is an important, highly contagious human pathogen. The nucleoprotein N binds only to viral genomic RNA and forms the helical ribonucleocapsid that serves as a template for viral replication. We address how N is regulated by another protein, the phosphoprotein (P), to prevent newly synthesized N from binding to cellular RNA. We describe the atomic model of an N-P complex and compare it to helical ribonucleocapsid. We thus provide insight into how P chaperones N and helps to start viral RNA synthesis. Our results provide a new insight into mechanisms of paramyxovirus replication. New data on the mechanisms of phosphoprotein chaperone action allows better understanding of virus genome replication and nucleocapsid assembly. We describe a conserved structural interface for the N-P interaction which could be a target for drug development to treat not only measles but also potentially other paramyxovirus diseases.
INTRODUCTION
Measles virus (MeV) belongs to the Paramyxoviridae family, which includes several other human pathogens, like respiratory syncytial (RSV), mumps, and parainfluenza viruses. It has a helical ribonucleocapsid (NC) containing a nonsegmented single-strand RNA (ssRNA) genome wrapped around the outside the nucleoprotein (N) helix (1). The helical NC is active in both transcription and replication. During virus assembly, the matrix protein forms an additional helix covering the majority of the NC, potentially inhibiting transcription and promoting packaging into progeny virions (2). There are still only limited data on the detailed molecular interactions required to go from replication initiation to packaging of nascent RNA. The availability of N in a chaperoned, assembly-competent state with the phosphoprotein (P) versus the assembled helical state is thought to be critical to these processes.
N is composed of an ordered NCORE region (amino acids 1 to 391) and an intrinsically disordered NTAIL region (amino acids 392 to 525) (Fig. 1A). NCORE contains two domains (NNTD and NCTD) flanked by N- and C-terminal arms (NTarm and CTarm). A recent atomic model of the MeV NC from a cryo-electron microscopic (cryo-EM) reconstruction revealed the molecular details of N oligomerization mediated by exchange of the NTarm and CTarm between consecutive N monomers and showed the RNA-binding site on the groove between the two NCORE domains (1).
Protein constructs and N0-P complex analysis. (A) Domain structure of measles N and P proteins and protein constructs used in this study. Numbers refer to amino acid positions. (B) Coexpressed N and GST-H6-P constructs after elution from glutathione Sepharose beads. Lanes: 1, marker; 2, P1-38 and N1-408; 3, P1-48 and N1-408; 4, P1-58 and N1-408; 5, P1-68 and N1-408; 6, P1-78 and N1-408; 7, GST-H6 and N1-408. Numbers on the left are molecular size markers. (C) Electron microscopy of negatively stained N1-4080–GST-H6-P1-48 heterocomplex (left) and flowthrough from the glutathione Sepharose beads containing NC-like helical particles (right). Images are at the same scale. Scale bar, 100 nm. (D) Data from the SEC-MALLS experiment with the N1-4080–P1-48 heterocomplex after release from Ni-IMAC by enterokinase. The solid line is the absorbance profile; blue crosses show the molecular mass distribution. For comparison, the UV280 profile of the N21-4080–P1-48 chimera is shown with a dashed line.
P is a modular protein comprised of an ordered tetramerization domain, MD (amino acids 304 to 377), forming a parallel four-helix coiled-coil (3), and an extreme C-terminal domain (CTD), XD, alternating with disordered regions (Fig. 1A). For transcription and replication, the RNA polymerase (L) in complex with P attaches to the NC via an interaction between the XD domain in P and the molecular recognition element (MoRE) (Fig. 1A) in N (4–7). The three-helix bundle in XD binds a helix from N′s MoRE element to facilitate this interaction (7). P has a second role: it binds NCORE through its N-terminal soyuz1 motif (8) and performs a chaperone function required to keep newly synthesized N from binding to cellular RNA (9). This RNA-free N0 is then transferred from the N0-P complex to the nascent NC by a currently unknown mechanism.
Whereas the XD and MD domains of P have been well characterized, the interaction between the P N terminus and the N in the N0-P complex is less well described. The dual function for P has been established for many viruses of the Mononegavirales order, and the crystal structures of the vesicular stomatitis virus (VSV), Ebola virus, and Nipah virus (NiV) N0-P complexes have been solved (10–12). In VSV N0-P, the P N-terminal amino acids 17 to 31 formed an amphipathic α-helix and occupied a hinge region in N adjacent to the RNA-binding site (13). In NiV N0-P, P amino acids 1 to 35 formed two α-helices separated by a kink (11). Interestingly, the NiV P binding site does not overlap the predicted RNA binding groove; therefore, the chaperoning mechanism of P appears to be remarkably different from that in VSV. In the present study, we addressed MeV N0-P complex formation and structure. We expressed and purified MeV N0-P complexes from Escherichia coli in a monodisperse form and mapped the location of the N-binding region on P to the first 48 N-terminal amino acids. Then we designed a chimeric N-P protein that was crystallized to reconstitute the N0-P complex and solved the structure at a resolution of 2.7 Å. We also characterized the mode of interaction between the P N terminus and N0 and showed the importance of hydrophobic interactions. Based on the structural data, we describe conformational changes upon RNA binding and propose a model for the preinitiation complex for RNA replication and transcription.
MATERIALS AND METHODS
Cloning and expression.All constructs were derived from reverse-transcribed N and P genes of an MeV wild-type isolate (a gift from I. Davidkin, Helsinki, Finland) (2). The N coding sequence was identical to the GenBank sequence for the Halonen strain (accession number U01996). The P coding sequence differed from GenBank sequence AF266288 for the Edmonston strain by three nucleotides: one was synonymous, and the other two resulted in the amino acids G225 and D492. Truncated P constructs were generated by PCR and cloned into NcoI and XhoI sites in pET41(a). The P constructs had an N-terminal glutathione S-transferase followed by a hexahistidine sequence (GST-H6 tag) for purification. N1-408 and N21-408 were constructed similarly but were cloned into pET22(b) with a stop codon added to the 3′ end and did not contain any tags. For the N21-408–P1-48 chimera, the N21-408 and P1-48 coding sequences were amplified by PCR to generate megaprimers with overlapping sequences. Then the megaprimers were annealed and extended. The product was amplified with primers coding for an N-terminal H6-TEV tag, MGSSHHHHHHENLYFQ|S, where the tobacco etch virus (TEV) protease recognition sequence is underlined and the cleavage site is shown by a vertical line (14). A stop codon was introduced at the 3′ end. The product was cloned into NcoI and XhoI sites in pET22(b). Mutations in the P constructs were introduced by site-directed mutagenesis.
Proteins were expressed in E. coli Rosetta (DE3) (Merck Millipore). Expression was induced at an optical density at 600 nm (OD600) of 0.5 with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) and was allowed to proceed for 16 to 20 h at 22°C. Cells were collected by low-speed centrifugation and frozen at −80°C as pellets until use.
Protein purification.N1-4080 and GST-H6-P1-48 were coexpressed in E. coli Rosetta (DE3). The cell pellet was resuspended in buffer A (20 mM Tris-HCl, 500 mM NaCl, 20 mM imidazole, 2 mM CaCl2 [pH 8.0]) supplemented with 200 μg/ml of lysozyme and one EDTA-free protease inhibitor tablet/25 ml (Thermo Scientific). Cells were lysed with a French press at 22,000 lb/in2, cell debris was spun down by low-speed centrifugation (11,000 × g for 15 min at 4°C), and the resulting supernatant was incubated with Ni-loaded IMAC beads (GE Healthcare) for 45 min at room temperature. After a washing with buffer A, the beads were exchanged into buffer B (20 mM Tris-HCl, 150 mM NaCl, 10 mM MgCl2, 2 mM ATP [pH 8.0]) and incubated at 37°C for 10 min. Next, the beads were exchanged into buffer C (20 mM Tris-HCl, 150 mM NaCl, 2 mM CaCl2 [pH 8.0]). The N1-4080–P1-48 heterocomplex was then released from the beads by an overnight digestion with enterokinase light chain (New England BioLabs). For 2 ml of Ni-IMAC beads with protein from 1 liter of cell culture, 0.16 μg of enzyme was used. The released protein was then concentrated with Millipore Amicon Ultra-4 30-kDa-cutoff spin concentrators and polished with size exclusion chromatography (SEC) using a Superdex 200 column (GE Healthcare). Peak fractions were collected and concentrated to the desired concentration with the same concentrator.
For N21-4080–P1-48 chimera purification, a cell pellet containing the tagged chimera was resuspended in buffer D (20 mM Tris-HCl, 150 mM NaCl, 2 mM MgCl2, 20 mM imidazole [pH 8.0]) supplemented with 10 μg/ml of lysozyme, 1 μg/ml of DNase I (Sigma-Aldrich), and 0.5 mM Pefabloc (Roche). Cells were lysed, and the lysate was cleared and incubated with Ni-IMAC as described above. After a washing with buffer D, protein was eluted with buffer D supplemented with 0.2 M imidazole. Eluted protein was incubated with TEV protease (purified in-house) overnight at 4°C. Cleaved protein was purified on a Superdex 200 column in buffer E (20 mM Tris-HCl, 150 mM NaCl [pH 8.0]), the monomer peak was collected, and uncleaved protein was removed by passing through an Ni-IMAC column. Purified protein was concentrated as described above.
N0-P heterocomplex interaction experiments.To find the minimal length of P that stably interacted with N1-408, 5 different P constructs were coexpressed with N1-408 as described above. The cells were lysed by sonication in phosphate-buffered saline (PBS) supplemented with 0.17 mg/ml of lysozyme. The cleared lysates were incubated with glutathione beads in PBS for 30 min at room temperature. After 3 washes with PBS, the proteins were eluted with 10 mM reduced glutathione in 50 mM Tris-HCl (pH 8.0).
To test the stability of the interactions, the purified N1-4080–GST-H6–P1-48 heterocomplex was bound in 50 mM sodium phosphate (pH 7.4), 10 mM imidazole, and 300 mM NaCl to Ni-IMAC beads and eluted with 20 mM Tris (pH 8.0) supplemented with one of the following: NaCl at 0, 0.5, 1, or 2 M; KCl at 0.5, 1, or 2 M; urea at 2, 4, or 8 M; or Triton X-100 at 0.1% or 1%.
To probe N21-4080 heterocomplex formation with mutated GST-H6-P constructs, cell lysates of individually expressed proteins were mixed and incubated overnight at 4°C. Lysate samples were incubated with Ni-IMAC beads and washed with lysis buffer E. The samples were eluted with buffer E supplemented with 0.2 M imidazole. Eluates were analyzed by SDS-PAGE.
SEC-MALLS experiment.To analyze the exact stoichiometry of the P1-48–N1-408 heterodimeric complex, 47 μl of the complex (1 mg/ml) released by enterokinase digestion was run on a Superdex 200 10/300 GL column coupled into UV, refractive index, and multiangle laser light scattering (MALLS) detectors (Wyatt Technology). The molecular weight of the complex was then calculated based on the refractive index and MALLS signals using ASTRA 6 software (Wyatt Technology).
Electron microscopy of negatively stained samples.Samples were pipetted on glow-discharged carbon coated copper grids and stained with 1% (wt/vol) sodium phosphotungstate (pH 7.0). Grids were imaged with an FEI F20 transmission electron microscope, and images were collected with a Gatan Ultrascan 4000 charge-coupled-device (CCD) camera.
Structure determination.Crystals of the N21-4080–P1-48 chimeric protein were grown by sitting-drop vapor diffusion (22°C) by mixing 200 nl of protein (8 mg/ml) with 200 nl of reservoir (0.1 M sodium citrate [pH 5.2], 3% polyethylene glycol 8000). Crystals were cryoprotected in mother liquor containing 20% glycerol and flash-frozen in liquid nitrogen. Diffraction data were collected at the Diamond Light Source beamline I03. The data set was processed and scaled using the xia2 package (with XDS and AIMLESS) (15, 16). A summary of the data collection is given in Table 1. The structure was solved by molecular replacement using PHASER (17). The N-terminal domain (NTD) and C-terminal domain (CTD), corresponding to amino acids 31 to 261 and 262 to 71, respectively, of NiV N (PDB code 4CO6 [11]), were used as search models separately. The model was rebuilt using several cycles of autobuilding and refinement with PHENIX (18) and manual rebuilding with COOT (19). No density was observed for the N regions from 21 to 30, 119 to 120, and 133 to 139 and the P region from 39 to 48, and therefore, they were left out of the model. The last refinement cycles were done using TLS parameters (two TLS groups). The final refinement statistics are summarized in Table 1. The final R factors (Rwork/Rfree) of the refined structure are 21.1%/26.6% (Table 1). In the Ramachandran plot, 90% of the residues in the structure are in the most favored regions.
Data collection and refinement statisticsa
Structure analysis.All the structure illustrations were prepared using UCSF Chimera software (20). Interface surface was estimated using the PDBePISA server (21). Calculation of the relative angle between the N domains in N0-P versus NC structure (PDB code 4UFT) was done using Modeller software (22) as described earlier (23). Structure alignments and root mean square deviation (RMSD) value calculations were made using UCSF Chimera. Dali multiple structural alignment (24) was used to generate the corresponding primary sequence alignment, followed by phylogenetic tree generation by PHYLIP in Unipro UGENE software (25).
Protein structure accession number.Final refined coordinates and structure factors have been deposited in the Protein Data Bank (PDB) under accession code 5E4V.
RESULTS
Mapping the interaction of the MeV NCORE0 with the N-terminal region of P and crystallization of the complex.In order to obtain a well-structured NCORE0-P complex, we analyzed the protease sensitivity of N by limited trypsin proteolysis (data not shown). Based on mass spectrometric analysis of the fragments observed, we cloned a C-terminally truncated construct containing the first 408 amino acids, N1-408, thus excluding the disordered C-terminal NTAIL region (7).
To screen for NCORE interaction in the N-terminal region of P, we used coexpression of N1-408 together with GST-hexahistidine (GST-H6) fusions with P1-38, P1-48, P1-58, P1-68, or P1-78 (Fig. 1A) and analyzed the interactions by GST affinity chromatography. All of the P constructs readily interacted with N1-408 and could be clearly seen in SDS-PAGE (Fig. 1B). Thus, the P N-terminal interaction site with NCORE resides within the first 38 amino acids. We analyzed negatively stained N1-4080–GST-H6–P1-48 eluate and the flowthrough with electron microscopy. In the eluate we observed a monodisperse solution of a small complex (Fig. 1C, left), whereas in the flowthrough, NC-like helical particles were readily visible (Fig. 1C, right). Probably, the NC-like particles contained N assembled on nonspecific cellular RNA (26). After GST-H6 tag cleavage, the purified complex was eluted from gel filtration as a single peak corresponding to a 1:1 heterodimer and was verified by SEC-MALLS to be 52 ± 2 kDa in size (Fig. 1D). This complex appeared not to contain nucleic acid, as the A260/280 was 0.55, whereas the expected ratio for pure protein is ∼0.6. Despite extensive efforts, the heterodimeric complex failed to crystallize. Hence, we designed a chimeric construct, H6–TEV–N21-4080–P1-48, in which the N-terminal region of P was directly fused to the C terminus of the NCORE domain lacking its NTarm region (Fig. 1A). The chimera was readily expressed as a soluble protein and purified. The gel filtration mobility (Fig. 1D) and the A260/280 ratio of the chimera were similar to those of the heterodimeric complex, with an additional dimer peak. The solution state of the N21-4080–P1-48 chimera suggests that the P1-48 sequence is bound to the P-binding site reconstituting the N0-P complex, preventing the formation of helical complexes.
Crystal structure of the MeV NCORE0-P complex.The MeV N21-4080–P1-48 chimera was crystallized in the space group P3121 as a dimer with the P1-48 sequence swapped between chimera monomers. We determined the structure at a resolution of 2.7 Å by molecular replacement using the NiV N0-P complex structure with PDB code 4CO6 (11) as a starting model (Fig. 2 and Table 1). The amino acid sequence of the N21-4080–P1-48 chimera could be traced starting from N residue 31 to P residue 38 with the exception of N residues 119 and 120 and 133 to 139. The buried surface interface in the crystallized dimer was 6,520 Å2, indicating a stable interaction interface for the dimer as seen in gel filtration.
Crystal structure of MeV N0-P complex. Shown is a cartoon representation of the chimeric N21-4080–P1-48 structure. (A) The crystallized dimer. Monomer 1 is sky blue and orange; monomer 2 is blue-green and orange-red. (B) The interaction of monomer 1 N (sky blue) and monomer 2 P (orange-red) fragments composing one N0-P heterocomplex. Secondary structure elements are labeled.
NCORE0 is primarily an α-helical protein with two domains, the NTD (amino acids 31 to 265) and the CTD (amino acids 266 to 372) (Fig. 2B) separated by a hinge. The NTD is formed by α-helices αN1 to αN9, one 3/10 helix ηN1, and parallel β-sheet βN1-βN2 with the adjacent short β-strand βN3 (Fig. 2). The CTD is formed by helices αN10 to αN15 and four 3/10 helices, ηN2 to ηN5. The CTarm (amino acids 373 to 408) continues as helices αN15 to αN17, with αN15 and αN16 adopting a helix-turn-helix conformation. P1-48 forms two helices (Fig. 2 and 3). The first helix, αP1, is a continuation of the αN17 helix and binds the partner molecule in the crystallized dimer to the groove formed by helices αN10 and αN11. The second helix, αP2, contacts helix ηN2.
Hydrophobic interactions in N0-P binding. (A) View of N21-408–P1-48 binding interface in cartoon representation. The P (orange-red) residues interacting with N (sky blue) are shown in stick representation and labeled. Colors represent residues conserved throughout the Paramyxovirinae as follows: violet, acidic; green, polar; blue, hydrophobic; and orange, glycine (8). (B) Alignment of P N termini of MeV, NiV, and PIV5. Asterisks indicate residues making contacts with N0. Conserved residues have a colored background. (C) SDS-PAGE of protein released from the N1-408–GST-H6–P1-48 heterodimer complex bound to Ni-IMAC beads when subjected to different conditions. Lanes: 1, marker; 2, 0 M NaCl; 3, 0.5 M NaCl; 4, 1 M NaCl; 5, 2 M NaCl; 6, 0.5 M KCl; 7, 1 M KCl; 8, 2 M KCl; 9, 2 M urea; 10, 4 M urea; 11, 8 M urea; 12, 0.1% Triton X-100; 13, 1% Triton X-100. (D) SDS-PAGE of pulldown of N21-408 by GST-H6–P1-48 and its mutants. Lanes: 1, markers; 2, control N21-408 only; 3, control GST-H6–P1-48wt only; 4, N21-408 plus GST-H6–P1-48wt; 5, N21-408 plus GST-H6–P1-48 L13D; 6, N21-408 plus GST-H6–P1-48 L13A; 7, N21-408 plus GST-H6–P1-48 I16D; 8, N21-408 plus GST-H6–P1-48 L19E; 9, N21-408 plus GST-H6–P8-48. Note that some of the P mutations affected the electrophoretic mobility of the tagged constructs due to impaired SDS binding to the protein molecules with changed net charge (compare lanes 5 and 7 with lane 3). The deletion construct P8-48 migrates faster, reflecting its shorter amino acid sequence. Numbers on the left in panels C and D are molecular size markers.
Interaction of NCORE0 with P is mainly hydrophobic.Conservation of the NCORE0 binding interface for P and of the P N-terminal region in some paramyxoviruses has been described previously (8, 11). In the MeV NCORE0-P complex, the binding interface is mostly composed of conserved hydrophobic residues (Fig. 3A and B, in blue). To biochemically probe the binding of the P N-terminal region to NCORE0, we screened for dissociation of N1-408 from GST-H6–P1-48 bound on Ni-IMAC beads under different conditions and looked for release of N1-408. The screen was designed to include conditions which would hinder either ionic or hydrophobic interactions between the proteins. NaCl or KCl concentrations ranging between 0 and 2 M did not cause significant release of N1-408, whereas 0.1% and 1% Triton X-100 caused release of N1-408 from the complex (Fig. 3C) to levels similar to those obtained with 4 and 8 M urea, respectively.
To evaluate the role of P's hydrophobic amino acid residues in N0 binding, we expressed N21-408 lacking the NTarm. We found that N21-408 was insoluble and formed NC-like particles upon expression (data not shown), similar to N1-408, but could bind to P1-48 in vitro. We pulled down N21-408 with GST-H6-tagged wild-type P1-48 and its mutants (Fig. 3D). While wild-type P1-48 can efficiently bind N21-408, replacements of hydrophobic by negatively charged amino acids (L13D, I16D, and L19E) severely affected the interaction. Replacement of L13 by the small amino acid Ala also strongly affected the interaction, possibly due to the increased solvent accessibility of the binding interface. In line with this observation, we still observed residual binding of the shorter P8-48 peptide where most of the interacting hydrophobic amino acids were retained (Fig. 3D, lane 9). Thus, hydrophobic interactions make a major contribution to P N-terminal region and NCORE0 binding.
Comparison of the MeV RNA-bound helical form with the chaperoned form.Direct comparison of the MeV NCORE0-P and helical NC structures reveals several factors that could contribute to P's chaperone activity; these include conformational changes (Fig. 4) as well as the position of P and RNA binding (Fig. 5 and 6). The largest difference between the two structures is that there is a relative domain movement in the MeV N in the NCORE0-P complex, compared to the helical RNA-bound NC form (Fig. 4A and B) (1). By aligning either the CTD or NTD only, we measured an ∼40° relative rotation of the two domains, with the hinge occurring between αN9 and αN10 (Fig. 4). The RMSD for the individual domains in the two different MeV N conformations were calculated (Table 2). This comparison indicated that the 4.3-Å resolution cryo-EM structure agrees well with our crystal structure, and the changes seen between the structures could be interpreted reliably (Fig. 4A and B). Besides bending of the hinge between helices αN9 and αN10, helix αN6 forming the lower lobe of the RNA binding cleft differs between the two different states. The helix undergoes both a shift and a rotation around its axis by half a turn (Fig. 4D). The helix movement increases the proximity of the two sides of the interdomain cleft in NCORE0-P and therefore collapses the NC RNA-binding site. In addition, the surface electrostatic charge distribution changes quite dramatically depending on the αN6 conformation. In NCORE0-P, a new negatively charged groove is evident on the NTD surface that could potentially bind RNA (Fig. 5). It has a contribution from the conserved R194 that interacts with the RNA backbone in NC (1). In NC, Y199 stacks with Y260, a key residue that regulates RNA binding pocket size. In NCORE0-P, Y199 faces the exterior and W196 occupies the space instead, thus potentially participating in the local stacking configuration.
MeV N0-P versus NC structure. Tube-and-plank representation of overlays of N0-P (N, sky blue; P, orange-red) and NC model (light gray) with their CTDs (A) or NTDs (B) aligned. (C) Cartoon representation close-up view of hinge region boxed in panel A. (D) Cartoon representation close-up view of αN6 helix boxed in panel B. The arrow on the right side shows hypothetical turn direction of αN6 helix upon RNA binding.
Potential alternative RNA-binding site. (A) Three consecutive N protomers of MeV NC, with the second protomer in a surface representation. The orientation is such that the outer surface of the nucleocapsid is facing the viewer. (B) N0 surface representation in the same orientation as the second N protomer shown in panel A. Missing side chains in N0 were added manually. Surface models are colored according to the electrostatic surface charge (positive, blue; negative, red). The positively charged patch in panel B is highlighted by a black box. The scale bar shows the electrostatic charge values.
P interferes with NC assembly. The P1-48 fragment overlaps both the NTarm and CTarm of NC. Shown is a cartoon representation of superposed N0-P (N0, sky blue; P, orange-red) and NC (gray; PDB code 4UFT), NTarm of the Ni−1 protomer (yellow), and CTarm of the Ni+1 protomer (pink). Molecules were aligned using the NCTD domains.
RMSD values between N structures of MeV N0-P and MeV NC, NiV N0-P, or PIV5 N-RNA complexesa
Alignment of the NCTD domains of NCORE0-P and NC (1) models (Fig. 4A and C) shows that the P N terminus would clash with helix αN9 of N in the NC conformation; thus, the alternative conformation is favored. When we consider the superposition of NCTDs in the context of the assembled NC, we can discern a direct effect of P binding (Fig. 6). The αP1 helix overlaps the NTarm of the Ni−1 protomer and the αP2 helix overlaps the CTarm of the Ni+1 protomer. Hence, P could specifically inhibit the association of adjacent protomers to the growing NC helix by steric hindrance, rather than by competing with the RNA-binding site.
Comparison to other virus nucleoproteins.Comparison of the N structures from MeV and other viruses of the Mononegavirales order reveals their structural similarity (Fig. 7). Each protein is composed of two domains with a single interdomain connection. Phylogenetic analysis based on Dali multiple structural alignment (24) shows that the two structurally closest to MeV N are NiV N and parainfluenza virus 5 (PIV5) N; MeV N shows 32% amino acid sequence identity to NiV N and 24% identity to PIV5 N.
Phylogenetic analysis of N structures. Shown is a phylogenetic comparison of the MeV N0 structure with the other published Mononegavirales nucleoprotein structures as a nonrooted tree. The scale bar corresponds to the number of expected substitutions per amino acid site between nodes. Atomic models are aligned by their CTDs and shown in ribbon representation. PDB codes are given above the virus abbreviations.
To our knowledge, MeV and VSV are the only Mononegavirales members with both the N0-P and N-RNA complexes available. Compared to the 40° rotation transition between the N0-P and the RNA-bound states in MeV N, in VSV N, the RSMD reported between the two states was less than 1 Å (10), reflecting the fact that both VSV states were crystallized in a ring form that probably constrained the conformation. In contrast to MeV, VSV P blocks the RNA-binding site rather than binds on the opposite side of the molecule. For the other N0-P complexes reported, NiV N0-P is the closest, with a binding site similar to that for MeV N0-P (Fig. 3 and 6), and relative domain positioning (referred to as an “open conformation” in NiV [11]). None of the other RNA-bound states are from a helical NC structure; rather, they are all ring structures, but in PIV5, RSV, and rabies virus, the RNA-bound states also indicate domain positioning similar to that of MeV NC (“closed state”) (11, 27). These comparisons emphasize the importance of the flexibility in the interdomain region in regulating N′s interactions with other viral components.
DISCUSSION
Here, for the first time in paramyxovirus research, our data allow direct comparison of the structures of the nucleoprotein from the same virus in two functional states: a P-bound naive state and an RNA-bound helical assembly. Our X-ray crystallographic atomic model at higher resolution confirms, complements, and improves upon the recently published cryo-EM reconstruction (1). The overall fold of MeV N is most similar to those of NiV and PIV5 N proteins (Fig. 7) (11, 27). In addition, the structure of MeV P1-48 and its binding site are very similar to those of NiV N0-P complex (11). Noting the extensive, conserved hydrophobic interactions of the P protein helix αP1 and N (8, 11), we showed by mutation of the hydrophobic residues in P and by biochemical analyses that we could indeed impair the binding interaction.
Our findings suggest that both N domains mostly preserve their fold upon transition from the N0-P to the NC state. Notably, N in both N0-P and NC has a flexible region between residues 118 and 140 composed of a well-defined αN4 helix (residues 124 to 130) flanked by unresolved regions. In NiV N, the αN4 helix is longer and only one unresolved region was left, whereas in PIV5, there are no gaps here. This region is on the outer surface of the NC. Hence, this flexible region could interact with the flexible C terminus of N or with the polymerase complex.
How does P act as a chaperone?The roles of P are at least 2-fold: first, to act as a chaperone to keep N0 in its RNA-free, soluble, monomeric form, and second, to position the polymerase complex for polymerization. In the role as a chaperone, it has been proposed that in the NiV N0-P complex, binding of P to N locks the open conformation by rigidifying the NCTD structure (11). Our model, however, suggests a significant impact from steric interference between the P N terminus and NNTD (Fig. 6). Alignment of the NCTD domains of the MeV N0-P and NC (1) models (Fig. 4A and C) shows that the P N terminus will clash with helix αN9 of N in the RNA-bound conformation, thus favoring the RNA-free conformation in N0-P. The flexibility of the N molecule is therefore an inherently important part of our model, compared to the published NiV model (11). We have additional evidence that in the NC, N can assume different conformations. The pitch of the protease-treated NC used for high-resolution structure determination is 5.0 nm (1) and imposes a rigidity on the helix that was important for image processing. However, the recombinant full-length protein forms flexible helices with pitches ranging from 5.0 to 6.6 nm (28, 29), and those imaged inside virions have a pitch of 6.4 nm (2). In the latter, the rigidity of the NC helix is enforced by interaction with an outer layer of matrix protein. Where the matrix is lacking, the NC is flexible. From the current work, at least two flexible regions could affect the twist and pitch, the twisting of the two domains (28), induced by the interdomain hinge region described above, and the conformation of the αN6 helix. Confident assignment of amino acids Trp196 and Tyr199 in this helix in both the cryo-EM and X-ray electron densities showed rotation and elongation of the αN6 helix (Fig. 4D), reflecting the intrinsic flexibility in this part of the molecule. Noticeably, in both the NiV N0-P and PIV5 N-RNA structures (11, 27), a loop preceding the corresponding helix is unresolved; this loop flexibility further supports the inherent mobility of the αN6 helix.
What determines whether N binds P or RNA?The RNA binding in the NC state is favored by specific arrangement of the amino acid residues from both the NNTD and NCTD (1). In the N0-P state, the rotation of the N domains forces the overlap of these two binding surfaces; hence, we hypothesize that the RNA binding affinity is reduced. This is supported by two observations. First, regarding the solubility of the chimera in an E. coli cell lysate, we found a predominance of RNA-free monomers and dimers rather than helical assemblies, even in the presence of E. coli RNA, compared to what was found with N expressed alone. Hence, the N0-P interaction hampered NC assembly and binding to RNA in our study. Second, the surface charge distribution of the chimera is altered, changing and shrinking the position of the positively charged surface in the N0-P compared to the NC. This suggests that N′s affinity for P in our constructs was higher than for RNA. There is probably a balance in the cell, during infection, dictated by the local concentrations of the relevant components and the avidity of N for RNA and its neighboring N subunits that together orchestrate the assembly of the NC. The flexibility of N facilitates its exchange between its binding partners, P and RNA. P can further regulate helix assembly through sterically impeding both side-to-side and vertical growth of the helix through occupying the same sites as both the NTarm of the Ni−1 protomer and CTarm of the Ni+1 protomer (Fig. 6).
Model for the formation of a preinitiation complex.According to the current paramyxovirus models, both transcription and replication are initiated at the 3′ end of the genomic RNA (30). The linear unidirectional organization of the “herringbone” NC means that the 3′ and 5′ ends of the NC do not present the same molecular surface due to this polarity. In addition, the transition between the bulk of the helix to the tip means that there is an extra potential site for P binding on the last molecule of the NC at the 3′ end (Fig. 8). The specific architecture of the pointed 3′ end of the NC could thus facilitate the recognition of the initiation site and assembly of the preinitiation complex through the interactions of P, L, N, and RNA. We propose a simple model for formation of a preinitiation complex, as shown in Fig. 8. In this model, the first interactions occur between the RNA polymerase complex, L-P, and NC through P's XD domains (Fig. 1A) in a low-affinity interaction with the flexible extended NTAILs (5, 31). This transient interaction allows one-dimensional diffusion of the polymerase complex along the NC. The accurate positioning on the tip occurs when the P N-terminal region binds to a vacant NTarm binding site at the NC's 3′ end. Binding of P may initiate NC uncoiling, as has been observed with mumps virus (32), to facilitate the release of the genomic RNA 3′ end from the RNA-binding groove. RNA release from the NC 3′ end by P is indirectly supported by the ability of the P N-terminal region to dissociate the N21-408 assembly where effectively all NTarm sites are vacant (Fig. 3D). Upon RNA 3′-end release, it may transiently bind to the exposed positively charged patch on N by its sugar-phosphate backbone (Fig. 5). The polymerase complex is then positioned for the entry of the first 6 nucleotides of the RNA 3′ end in to the active site of L (33–35). Bipartite promoter recognition by the polymerase complex is required for genomic RNA replication (36). In analogy to mumps virus, this may require further uncoiling of the NC, promoted by P (32). Elongation will also require NC uncoiling to expose the RNA. The processivity of the polymerase complex will promote this, and additional P may be injected into the helical assembly, resulting in local NC uncoiling and template RNA exposure according to the cartwheel model (37). N could be recycled onto the NC once the polymerase complex has passed due to the transient association with the P CTD. The presence of assembled matrix on the NC during early stages of infection will necessitate additional disassembly which is as yet not understood.
Schematic model of RNA synthesis preinitiation complex formation. (A) Model of nonspecific binding of the polymerase complex to NC. The NC is composed of genomic RNA (black ribbon) and the N protomers, in which the NCORE domains are depicted as sky blue peanut shapes, NTarm in yellow, and NTAIL in pink. A complex of polymerase subunit L (gray) with the P tetramer (orange-red) binds the NC through the interaction of P's XD domains with MoRE motifs in NTAILs. There is a vacant P/NTarm site on the 3′ protomer of the NC (dark blue crescent). (B) Model of the preinitiation complex formation. One of the P N termini occupies the P/NTarm site on the 3′ protomer of the NC. Binding results in release of the RNA 3′ end, which is then positioned in the polymerase catalytic site to form a preinitiation complex.
In conclusion, our MeV N0-P structure and its comparison to the previously reported NC state provide insight into MeV NC assembly and polymerase activity.
ACKNOWLEDGMENTS
We thank the following people for their help: Eevakaisa Vesanen for electron microscopy, Juho Kellosalo for SEC-MALLS experiments, Tuula Nyman for mass spectrometry, and the Finnish Instruct National Associate Center, the National Biocenter Finland Cryo-Electron Microscopy Unit, the Crystallization Facility, and the Proteomics and Metabolomics Unit for the use of their facilities.
FOOTNOTES
- Received 11 November 2015.
- Accepted 20 December 2015.
- Accepted manuscript posted online 30 December 2015.
- Address correspondence to Sarah J. Butcher, sarah.butcher{at}helsinki.fi.
↵* Present address: Lassi Liljeroos, GlaxoSmithKline Vaccines, Siena, Italy.
Citation Guryanov SG, Liljeroos L, Kasaragod P, Kajander T, Butcher SJ. 2016. Crystal structure of the measles virus nucleoprotein core in complex with an N-terminal region of phosphoprotein. J Virol 90:2849–2857. doi:10.1128/JVI.02865-15.
REFERENCES
- Copyright © 2016, American Society for Microbiology. All Rights Reserved.
