Structural and Functional Characterization of the Mumps Virus Phosphoprotein

Abstract

The phosphoprotein (P) is virally encoded by the Rhabdoviridae and Paramyxoviridae in the order Mononegavirales. P is a self-associated oligomer and forms complexes with the large viral polymerase protein (L), the nucleocapsid protein (N), and the assembled nucleocapsid. P from different viruses has shown structural diversities even though their essential functions are the same. We systematically mapped the domains in mumps virus (MuV) P and investigated their interactions with nucleocapsid-like particles (NLPs). Similar to other P proteins, MuV P contains N-terminal, central, and C-terminal domains with flexible linkers between neighboring domains. By pulldown assays, we discovered that in addition to the previously proposed nucleocapsid binding domain (residues 343 to 391), the N-terminal region of MuV P (residues 1 to 194) could also bind NLPs. Further analysis of binding kinetics was conducted using surface plasmon resonance. This is the first observation that both the N- and C-terminal regions of a negative-strand RNA virus P are involved in binding the nucleocapsid. In addition, we defined the oligomerization domain (P_OD) of MuV P as residues 213 to 277 and determined its crystal structure. The tetrameric MuV P_OD is formed by one pair of long parallel α-helices with another pair in opposite orientation. Unlike the parallel orientation of each α-helix in the tetramer of Sendai virus P_OD, this represents a novel orientation of a P_OD where both the N- and the C-terminal domains are at either end of the tetramer. This is consistent with the observation that both the N- and the C-terminal domains are involved in binding the nucleocapsid.

INTRODUCTION

The phosphoprotein (P) is a multifunctional protein encoded in the genomes of negative-strand RNA viruses (NSVs) of the Rhabdoviridae and Paramyxoviridae in the order Mononegavirales. P performs several essential functions in virus replication. It is the cofactor in the viral RNA-dependent RNA polymerase (RdRp), a complex that also includes the virally encoded large (L) protein. L harbors the catalytic functions for RNA synthesis and mRNA capping. P also directly binds the nucleocapsid, the active template for viral RNA synthesis. Through this interaction, P delivers the RdRp to the nucleocapsid. In the absence of P, the RdRp cannot recognize the nucleocapsid or gain access to the viral genome. The domain responsible for P to bind the nucleocapsid, the phosphoprotein nucleocapsid binding domain (P_NBD), has been mapped to the C-terminal region for a number of NSVs, including vesicular stomatitis virus (VSV), rabies virus (RABV), measles virus, Sendai virus (SeV), Mokola virus, and mumps virus (MuV) (1–5). Three-dimensional structures of this domain revealed a certain degree of structural homology among different viruses (5–9). The crystal structure of VSV P_NBD in complex with a nucleocapsid-like particle (NLP) clearly showed that the P binding site in the nucleocapsid is formed by two neighboring nucleocapsid protein (N) subunits (10). The large extended loop in the C-lobe and a single α-helix (α13) of one N subunit and the same extended loop in an adjacent N subunit make up the U-shaped P binding site, with α13 at the bottom of the cleft. This high-affinity P binding site could only be created by N subunits that are assembled together in the nucleocapsid. By this mechanism, the cofactor P can dock the RdRp specifically to the nucleocapsid in order to use it as the template for viral RNA synthesis. In addition, P forms a stable complex with N that is free of RNA, termed N⁰-P. This encapsidation-competent complex of N and P exists prior to incorporation of N into the nucleocapsid (11). During virus replication, the N subunit in the N⁰-P complex is used to concomitantly encapsidate the newly synthesized viral RNA (12). The domain essential for keeping N free of RNA in the N⁰-P complex has been mapped to the N-terminal region (phosphoprotein N⁰ binding domain [P_N°D]) of P for several NSVs (11, 13–15). A crystal structure of VSV P_N°D in complex with a VSV N mutant showed that P_N°D sits in the cavity where the viral RNA would be accommodated in the nucleocapsid (16). Occupation of the cavity may be instrumental in allowing P to keep N free of RNA in the N⁰-P complex.

Another important feature of P is that its functional form is a self-associated oligomer. Self-association is required for supporting viral RNA synthesis (17) but is not necessary for the association of P with other viral proteins such as L or N. The domain responsible for P oligomerization (P_OD) is located in the central region of P. P_N°D and P_NBD are linked to P_OD through flexible loops, which appears to be a common modular structure for all NSV P proteins. The length of P differs greatly among NSVs, and there is no homology for P_OD. Furthermore, the structure and the mode of self-association also vary considerably from one P to another. In the case of SeV P (568 amino acids), P_OD corresponds to residues 320 to 433. Each SeV P_OD subunit donates a single long helix to form a tetrameric coiled coil, with all helices being parallel to one another. Short helices at the N-terminal end of SeV P_OD contribute to additional tetrameric interactions. For VSV P (265 amino acids), P_OD comprises residues 107 to 177 (18, 19). VSV P_OD dimerizes by domain swapping of a β-hairpin from each subunit that participates in collating a β-sheet of four antiparallel strands on each side. Two parallel α-helices held together by a number of hydrophobic interactions form the core of the VSV P dimer. By comparison, the oligomerization mode of RABV P_OD is quite different from that of VSV P_OD, despite the two viruses being closely related members of the rhabdovirus family. RABV P_OD spans residues 92 to 131 in P (297 amino acids) and forms a dimer (20). However, each polypeptide contains two antiparallel helices linked by a loop. The dimer of RABV P_OD involves a four-helix bundle between two parallel subunits. This particular orientation places P_N°D and P_NBD at the same end of RABV P, in contrast to other P proteins in which the two domains are at distant opposite ends of the oligomer (19–21). In VSV, it has been shown that a chimeric P dimer with one P mutant lacking P_N°D and another mutant lacking P_NBD can fully support VSV RNA synthesis (17). In the case of the RABV P dimer, it is not clear what the functional implications are when all P_N°D and P_NBD domains are at the same end.

For this report, we carried out systematic studies of the domains of MuV P and their possible interactions with the nucleocapsid. Previously, MuV P expressed as a recombinant glutathione S-transferase (GST) fusion protein was shown to bind a truncated N (N₃₉₈) that was described as capable of assembling into NLP (4). In addition, deletion of the last 49 residues in MuV P abolished NLP binding, and the C-terminal domain (C343-391) alone was shown to be capable of binding NLP (4). In contrast, nucleocapsid binding by the P protein from other paramyxoviruses reportedly requires interactions with the C-terminal tail region of N (22–25). The previous report appears to suggest that MuV P contains a P_NBD that is similar to other paramyxoviruses, but its binding site on the nucleocapsid may be located in a different region of N. The crystal structure of the C-terminal domain of MuV P showed that it contains three packed α-helices that may be induced to fold when P binds the nucleocapsid (7). We expanded the study of nucleocapsid binding to include all domains in MuV P. Our structural and functional characterization of MuV P is discussed here with comparison to other NSV Ps.

MATERIALS AND METHODS

Materials.

Restriction enzymes and T4 DNA ligase were purchased from New England BioLabs. pET28b and BL21(DE3) were purchased from Novagen. Primers for PCR were purchased from Invitrogen and Integrated DNA Technologies.

Plasmid construction.

The P genes were amplified from a cDNA clone of MuV strain 88-1961 (GenBank accession no. AF467767.2) by PCR. Primers for amplification on the 5′ end of the gene fragments contained NheI restriction sites, while the 3′ primers contained XhoI sites. Sequences of the primers can be sent upon request. The P gene or gene fragments and the pET-28b plasmid were each digested with restriction enzymes NheI and XhoI and gel purified. The purified digested P genes were individually ligated in frame with the N-terminal polyhistidine tag, resulting in the vector MuV P-pET28b or MuV (P clone ID)-pET28b.

Protein expression and purification.

Expression vectors were transformed into Escherichia coli strain BL21(DE3). Bacteria were cultured in Luria-Bertani broth at 37°C until the optical density at 600 nm reached 0.6. Protein expression was induced with 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for 18 h at 18°C. The cells were harvested by centrifugation and resuspended in binding buffer A containing 20 mM Tris (pH 7.9), 500 mM NaCl, and 5 mM imidazole. The cells were disrupted by sonication and then centrifuged for 1 h at 18,000 rpm. Soluble fractions were collected and passed through an Ni-affinity column (Chelating Sepharose Fast Flow; GE Healthcare). The loaded Ni-affinity column was washed with 5 column volumes (CV) of binding buffer A. The loaded column was washed with 5 CV of wash buffer A containing 20 mM Tris (pH 7.9), 500 mM NaCl, and 50 mM imidazole. Samples were eluted with elution buffer A containing 20 mM Tris (pH 7.9), 500 mM imidazole, and 500 mM NaCl. Proteins were further purified by size-exclusion chromatography (Sephacryl S-75 or S-200; GE Healthcare) in size exclusion buffer containing 20 mM Tris (pH 7.5) and 500 mM NaCl.

Purification of trypsinized MuV NLP.

The MuV N protein was expressed, and NLP was purified as previously reported (26). NLP has a ring structure composed of 13 N subunits and a single strand of random RNA. The purified NLP was allowed to incubate overnight at room temperature with trypsin at a 1-mg/100-mg ratio (trypsin:N). Trypsin was then inactivated by the addition of phenylmethylsulfonyl fluoride (1 mM). The digested NLP (NLP₃₇₉) was concentrated to 8 mg/ml and purified by size-exclusion chromatography using a S-400 column (HiPrep Sephacryl S-400; GE Healthcare). The identity of the digested N was confirmed by N-terminal amino acid sequencing and mass spectrometry analysis. The size of the NLP₃₇₉ was consistent with that of a ring structure similar to the full-length NLP.

Identification of protease-resistant fragments of the MuV P protein.

The full-length P protein was digested by using three different proteases: trypsin (TPCK [tolylsulfonyl phenylalanyl chloromethyl ketone] treated; Worthington), α-chymotrypsin (Sigma), and proteinase K (Fisher). The amounts of enzymes used per mg of the P protein were 0.2 U, 0.1 U, and 0.2 μg for trypsin, chymotrypsin, and proteinase K, respectively. The P protein was digested with each protease for a maximum of 2 h at 37°C. During this time course, aliquots of each digestion mixture were taken at 5-min intervals and denatured for analysis in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels. SDS-PAGE gels of digestion reactions were electroblotted onto polyvinylidene difluoride membranes and stained with Coomassie blue. Bands corresponding to stable P fragments were cut out and sent to the Protein Analysis Core at the University of Texas Medical Branch (UTMB; Galveston, TX) for N-terminal amino acid sequencing. Products at optimal time points from the limited digestions of P were purified using the Gelfree 8100 fractionation system (Protein Discovery) in accordance with the manufacturer's protocol. Fractions containing pure digestion products were subjected to mass spectrometry analysis at the Protein Analysis Core at UTMB.

Bioinformatics analysis.

The secondary structure of MuV P protein was predicted using Jpred3 (27). PIV5 V protein (accession no. AAA47882) was used in the Jpred3 homologue search.

Crystallization and data collection of MuV P_OD.

P161-277 was expressed and purified as described above and concentrated to 5 to 7 mg/ml. Crystallization conditions at 20°C were screened using Crystal Screen 1&2 (Hampton). An initial crystal hit was found in the handing drop setup with a reservoir solution containing 0.1 M sodium acetate (pH 4.6) and 2 M NaCl. Typical crystals appeared in hanging drops after 10 to 14 days at 20°C. Crystals were cryoprotected in a cryosolution containing the reservoir solution supplemented with 25% glycerol. For the preparation of uranyl acetate derivatives, crystals were soaked stepwise in the reservoir solution containing 0.5, 1.0, 1.5, and 2.5 mM uranyl acetate at ∼10-min intervals. Uranyl acetate soaked crystals were cryoprotected by the same protocol as that used for the native crystals, except for the addition of 2.5 mM uranyl acetate to the cryosolution. Native crystals diffracted X-rays to a resolution of 2.2 Å when exposed to South East Regional Collaborative Access Team (SER-CAT) synchrotron beamline BL-22 BM X-rays at the Advanced Photon Source (APS).

Structure determination and refinement.

The phases of X-ray diffraction data were determined by SIRAS using the program CNS (28). The uranium atom in the uranyl acetate derivative was used as an anomalous scatterer. An initial model was manually built into solvent flattened maps. The model was then subjected to several rounds of the Autobuild routine in PHENIX (29). The resulting model was subjected to several cycles of manual rebuilding using COOT (30) and refinement with PHENIX (29). TLS refinement was performed using the PHENIX program (31, 32). The structure has been deposited in the Protein Data Bank under the code 4EIJ. Solvent accessible surfaces were calculated using CNS (28). Structure figures were created using PyMOL (33). The data collection and refinement statistics are shown in Table 1.

Table 1.

Crystal X-ray data, phasing, and refinement statistics

Parametera	BL-22 BMb
	Native crystal	Uranyl derivative
Data collection statistics
X-ray source	SER-CAT, APS, IL	SER-CAT, APS, IL
Wavelength (Å)	1.0	1.0
Space group	R32	R32
Cell dimensions
a, b, c (Å)	80.348, 80.348, 165.17	78.919, 78.919, 165.447
α, β, γ (°)	90, 90, 120	90, 90, 120
Resolution (Å)	50–2.2 (2.24–2.2)	50–2.9 (2.96–2.9)
Rsym	0.062 (0.38)	0.081 (0.47)
Avg I/σ(I)	26.695 (4.145)	23.11 (3.0)
Completeness (%)	99.9 (100)	99.6 (97.6)
Redundancy	4.6 (4.7)	3.5 (3.7)
No. of reflections	49,262	16,324
No. of unique reflections	10,717 (528)	4,601 (233)
Phasing statistics
Heavy atom sites		3×U(O)2
Resolution range (Å)		50–2.56 (2.76–2.56)
Figure of merit, acentric/centric		0.6566/0.3542 (0.6584/0.3531)
Phasing power, iso/ano		1.3471/1.3104 (1.2658/1.4415)
Rcullis, iso/ano		0.599/0.7606 (0.6059/0.9150)
Refinement statistics
Molecules (ASU)	2
Resolution (Å)	26.6–2.2 (2.31–2.2)
Avg B-factor (Å2), protein/water	43.05/46.52
Coordinate error (maximum likelihood based)	0.26
No. of reflections used	10,308
Rwork/Rfree	0.1814/0.2302
Completeness (%)	96.15
No. of atoms
Protein	942
Water	58
Glycerol	1
RMSD
Bond length (Å)	0.006
Bond angle (°)	0.887
Ramachandran plot (%), favored regions	98.3

^aR_sym = Σ|(I − <I>)|/Σ <I>, where <I> is the observed intensity; R_work = Σ(||F_obs| − k|F_model||)/Σ(|F_obs|). R_free was obtained for a test set of reflections (9.98%).

^bValues for the highest-resolution shell are indicated in parentheses.

Contact area between the helices.

In order to estimate how tight the interactions are between the different helices, calculations of the buried surface areas were performed in CNS (28) (Table 2). The total buried surface area is similar for each of the interchain interactions. Among the three possible pairs of contacts, chain A-B proved to have the highest percentage of buried surface. However, the negligible differences among the three pairs strongly suggest that the oligomeric state of MuV P_OD is a tetramer.

Table 2.

Surface area calculations

Chain	Surface (Å2)		% Buried
	Buried	Total
Chain A		6,382
Chain B		5,824
A-B	2,771	9,436	29.3
A-A′	2,570	10,193	25.2
B-B′	2,601	9,050	28.7
Tetramer	5,065	13,809	36.7

Glutaraldehyde cross-linking.

Protein cross-linking studies were carried out with the addition of 0.1% glutaraldehyde stock solution diluted in phosphate-buffered saline. Increasing concentrations of glutaraldehyde were added to 50 μl of P161-277 (3 mg/ml) and allowed to incubate for 4 h or overnight on ice. The reactions were quenched by addition of 18 μl of SDS-PAGE sample buffer. The samples were denatured and analyzed by SDS-PAGE.

Sedimentation velocity analysis.

Sedimentation velocity experiments were performed in an XL-A analytical ultracentrifuge (Beckman Coulter) in two-channel Epon centerpieces. The data were collected at 280 nm and a speed of 40,000 rpm in an An60Ti four-hole rotor at 20°C. The concentration of protein loaded was ∼1 mg/ml. The data (Table 3) were analyzed with the program SEDFIT using both the c(s) and c(s,ff0) models (http://www.analyticalultracentrifugation.com/default.htm). The c(s,ff0) model was used for molecular weight estimation of the sedimenting species. The buffer density, buffer viscosity, and protein partial specific volume used in these calculations were estimated with the program SEDNTERP (http://bitcwiki.sr.unh.edu/index.php/Main_Page).

Table 3.

Sedimentation velocity statistics

Sample	S value	RMSD	f/f0	Mol mass (kDa)
P161-277	2.6	0.009049	1.8525	57.8
P full	3.8	0.01062	2.6522	165.5

Pulldown assay.

In order to map N-P interactions, a pulldown assay was performed. A small column containing 50 μl of charged Chelating Sepharose Fast Flow beads was first saturated with the P protein or a P fragment. After allowing the sample to flow through the column, the beads were washed with 10 CV of binding Buffer B, containing 20 mM Tris (pH 7.9), 50 mM NaCl, and 5 mM Imidazole. NLP, or NLP₃₇₉, in binding Buffer B was added to the loaded column at a 1:1 molar ratio (N:P) and incubated at room temperature for 15 min. The NLP or NLP₃₇₉ solution was allowed to flow through. Next, the loaded column was washed with 10 CV of binding Buffer B, followed by wash with 5 CV of wash Buffer B containing 20 mM Tris (pH 7.9), 50 mM NaCl, and 50 mM Imidazole. The protein was eluted with 5 CV of elution Buffer B containing 20 mM Tris (pH 7.9), 50 mM NaCl, and 500 mM Imidazole. The eluate was denatured in SDS-PAGE sample buffer and electrophoresed in an SDS–12% PAGE gel. Gels were stained with 1% Coomassie brilliant blue dye. The full-length MuV P protein was used as a positive control for binding NLP. The purified NLP or NLP₃₇₉ in the absence of any P fragments was used as a negative control.

BIAcore 2000 analysis.

Real-time binding analysis of the MuV P protein to the MuV NLP and its truncated form NLP₃₇₉ were analyzed using surface plasmon resonance on a BIAcore 2000 instrument (GE Healthcare). Purified NLP or N₃₇₉ ligands were immobilized on the carboxylated surface of a CM5 sensor chip using an amine-coupling kit. Analyte full-length MuV P and P fragments P1-194, P161-277, and P286-391 were flowed over the immobilized NLP or NLP₃₇₉ CM5 chip at a flow rate of 20 μl/min at 25°C. Regeneration of the chip surface after each cycle was performed using (0.015 M HEPES in 0.5 M NaCl). The change in the surface plasmon resonance angle is displayed as response units, where 1,000 response units (RU) is equal to 1 ng of analyte bound per nm² on the sensor surface. The experiments were carried out in triplicates and kinetic association (K_A) and dissociation (K_D) rate constants were deduced using the 1:1 Langmuir kinetic model with BIA evaluation software.

RESULTS

Digestion studies of MuV P.

In order to study the structure and function of the MuV P protein, a single vector was constructed to express the MuV P protein in E. coli. After expression, His-tagged recombinant P was purified from the soluble fraction of the lysate. The P protein was further purified by size-exclusion chromatography and analyzed by SDS-PAGE. Purified P migrates to a position consistent with that observed in previous publications, ∼5 kDa higher than the calculated molecular mass of 41.5 kDa (26). Slow migration has also been observed for VSV P (34), suggesting that NSV P proteins appear to migrate more slowly in SDS-PAGE.

The full-length P protein was shown to be susceptible to digestion by residual impurities over a brief time period following purification. In order to define domains within MuV P, limited proteolytic digestion studies were performed. Purified full-length P protein was digested by using three different proteases: trypsin, chymotrypsin, and proteinase K. Digestion products were examined by SDS-PAGE, which showed that each protease had a different digestion pattern (Fig. 1A to C). In order to identify the N and C termini of the stable fragments, individual bands corresponding to digested products were isolated and analyzed by N-terminal amino acid sequencing and mass spectrometry. The results for the limited digestion studies are summarized in Fig. 1D. Trypsin digestion resulted in a major stable fragment corresponding to residues 135 to 391 with an N-terminal sequence of MINRF and a molecular mass of 27.7 kDa. Chymotrypsin digestion resulted in a major stable fragment corresponding to residues 194 to 391 with an N-terminal sequence of AHPSP and a molecular mass of 21.7 kDa. Although bands corresponding to other partially digested products were seen on SDS-PAGE gels for both trypsin and chymotrypsin digestions, the molecular masses of these fragments could not be determined by mass spectrometry. Proteinase K digestion resulted in two stable fragments. Identification of mass by mass spectrometry was not successful. The residue numbers for the two K fragments (K1 and K2), therefore, were designated based on their N-terminal sequence, and the molecular masses were approximated from their position on SDS-PAGE. The N-terminal sequence for K2 was SVISA and likely corresponds to residues 214 to 391. The N-terminal sequence for K1 was somewhat ambiguous, but the sequence in cycles 3 through 5 was PSP. This sequence with the size observed on SDS-PAGE of ∼22 kDa is consistent with a fragment encompassing residues 194 to 391. The chymotrypsin and K1 fragments are thus the same and consistent with the cleavage pattern of each enzyme, where each cleaves adjacent to the carboxyl group of bulky or aromatic amino acids. The sequence at this cleavage site of P is Y/AHPSP.

Fig 1.

Limited digestion of MuV P by different proteases. In panels A to C, lanes 1, 2, and 3 correspond to the molecular mass ladder, undigested P protein, and digested P protein, respectively. (A) Proteinase K digestion resulted in two stable fragments (K1, residues 194 to 391; K2, residues 214 to 391) (lane 3). (B) Trypsin digestion resulted in a major stable fragment (T1, residues 135 to 391) (lane 3). Lanes 4 to 7 correspond to the different fractions separated using the Gelfree 8100 fractionation system. (C) Chymotrypsin digestion resulted in a major stable fragment (C1, residues 194 to 391) (lane 4). The identity of the major stable fragments was revealed by N-terminal amino acid sequencing and molecular weight determination, and is summarized in panel D. The molecular masses for fragments T1 and C1 were determined through mass spectrometry. The molecular masses of fragments K1 and K2 were approximated from their position on SDS-PAGE gels. Protease cut sites (*) represent the sequences identified using N-terminal sequencing. (D) Summary of bioinformatics analysis using Jpred3. Predicted helical regions are shown in the schematic to the right of “Helix”. Predicted beta-strand regions are shown in the schematic to the right of “β-strand”. Homology to PIV5 V protein is shown in the schematic to the right of “PIV5 V”.

Bioinformatics analysis.

Secondary structure prediction by Jpred3 identified nine different regions as probable helical regions (Fig. 1E). The prediction also identified four different regions that were likely to have β-strand conformation. In addition, the Jpred3 search identified PIV5 V as a homologue of MuV P. PIV5 V has a 37% identity to MuV P, located within the first 155 residues and 164 residues of MuV P and PIV5 V, respectively. These data were used, along with the results of protease digestion, to design a clone library of MuV P fragments. These protein fragments were expressed and purified by the same methods used for the full-length P. The yield of purified protein per liter varied for each fragment. A list of cloned P fragments and a summary of protein expression results are shown in Fig. 2. P fragments that were consistent with the results of the protease digestion have a better expression level in the soluble fraction.

Fig 2.

Summary of MuV P fragments. A schematic representation of MuV P modules is shown at the top. For expression levels, “+” indicates <15 mg/liter, “++” indicates 15 to 25 mg/liter, and “+++” indicates >25 mg/liter of purified recombinant protein. Fragments were derived from a combination of protease digestion and bioinformatics analysis.

Structure of the P oligomerization domain.

Several fragments were subjected to crystal screens. Crystals were grown with fragment P161-277, and the crystal structure was determined to 2.2-Å resolution. Only residues 213 to 277 could be fit correctly into the electron density maps. The traced polypeptide represented an appropriate protein volume for the unit cell with a solvent content of 60.04%, assuming a partial specific volume of 0.74 ml/g. Composite omit maps were created to confirm that modeling was carried out correctly. The lack of residues 161 to 212 is likely due to digestion in situ during crystallization since crystal formation only occurred after 2 weeks. The asymmetric unit contains two chains (A and B) that cover residues 213 to 273 and residues 215 to 277, respectively (Fig. 3A). Both chains are composed of a single long α-helix. However, chain A, unlike chain B, contains a kink at Gly 246. Chain A also contains a stretch of amino acids with extended conformation at the C-terminal end (residues 272 to 277) of its long helix.

Fig 3.

Structure of MuV P_OD. (A) Crystal structure of two parallel extended α-helices (chain A in green and chain B in cyan). Chain A contains residues 215 to 277, and a kink at Gly246 is noted. Chain B contains residues 213 to 273. (B) P_OD tetramer as found in the crystal. The crystal structures of P_ODs from SeV (C), VSV (D), and RABV (E) are shown for comparison. The range of residues for each P_OD is labeled. In panels C to E, each independent polypeptide chain is a different color.

The crystallized MuV P fragment forms a tightly packed tetramer through noncrystallographic and crystallographic symmetry contacts (Fig. 3B). We therefore define this tetramer as MuV P_OD. This tetrameric coiled-coil structure is reminiscent of SeV P_OD (Fig. 3C). However, the tetramer formed by SeV P_OD is composed of four parallel helices, whereas the tetramer formed by MuV P_OD is composed of two sets of parallel helices that are in opposite orientation (Fig. 3B). In the MuV P_OD tetramer that is primarily formed with hydrophobic interactions (Table 3), there are two zippers of charged side chain interactions at each end in addition to hydrophobic surface contacts among the helices. Between the parallel pair of helices, the zipper is formed by Asp229, Glu236, and Asp240 (chain A) and Arg231 and Lys238 (chain B) as mentioned above (Fig. 4A, inset A). Between the antiparallel pair of helices, the zipper is formed by Lys253 and Lys260 (chain B) and Asp229, Glu236, and Asp240 (chain B′) (Fig. 4B, inset B). The helices wrap around each other and seem to clamp one another through these charge zippers at both ends of the tetrad.

Fig 4.

Tetrameric interactions. (A) Chain A-B interactions. In addition to extensive surface contacts, the main interactions between chains A and B also include a zipper of charged residues: Asp229, Glu236, and Asp240 (chain A) and Arg231 and Lys238 (chain B). In panel A, chain A is on the left, and chain B is on the right. The other two chains in the tetramer are monocolored. (B) Chain B-B′ interactions. Residues Lys253 and Lys260 from chain B (left) also form a zipper of charged residues—Asp229, Glu236, and Asp 240—from the antiparallel chain B′ (right). This clamping by the zippers occurs on both ends of the tetramer. The tetramer in panel B is rotated left-handed by 90° compared to the tetramer in panel A.

Glutaraldehyde cross-linking.

In order to verify the tetrameric state of the MuV P protein, cross-linking reactions were performed using glutaraldehyde. Initially, full-length P was used in cross-linking experiments. However, because MuV P is highly susceptible to nonspecific digestion (35), the resulting cross-linked products were composed of a mixture of different P fragments, and no clear data could be discerned (data not shown). Since the majority of the known protease sensitive sites are contained within the terminal regions, P161-277 was used in place of the full-length P protein, and the resulting cross-linked products were much easier to identify. After analysis in SDS-PAGE gels, bands representing monomers, dimers, trimers, and tetramers could be clearly discerned (Fig. 5A).

Fig 5.

Oligomerization state of MuV P. (A) Cross-linking with glutaraldehyde. MuV P161-277 was cross-linked using glutaraldehyde in order to assess the oligomerization state. Cross-linking products were analyzed in SDS-PAGE gels. Lanes 1 to 5 correspond to 4 h of incubation with glutaraldehyde at the following concentrations: 0, 0.05, 0.06, 0.1, and 0.2%. Lanes 6 and 7 correspond to overnight incubation with 0.2 and 0.3% glutaraldehyde, respectively. Bands representing P161-277 monomers (P*1), dimers (P*2), trimers (P*3), and tetramers (P*4) can be seen at ∼15, ∼30, ∼45, and ∼60 kDa, respectively, and are noted to the right of the gel. (B) Sedimentation velocity analysis. Sedimentation coefficients were determined, and molecular masses were derived for P161-277 and full-length P (Table 3). The distribution plots showed a single species for each sample. The derived molecular mass for P161-277 (57.8 kDa) corresponded to the theoretical mass of a P161-277 tetramer (57.6 kDa). Furthermore, the molecular mass derived for full-length P (165.5 kDa) was very close to the theoretical molecular mass for a MuV P tetramer (168 kDa).

Sedimentation velocity analysis.

To further confirm the oligomeric state determined by the P_OD crystal structure, we carried out ultracentrifugation analyses to measure the sedimentation coefficient of the purified P protein and fragment P161-277. For both P161-277 and full-length P, the data revealed a single dominant species (Fig. 5). The molecular mass derived from the sedimentation coefficient is 57.8 kDa for P161-277, which corresponds to the theoretical mass of a P161-277 tetramer (57.6 kDa). Furthermore, the molecular mass derived from sedimentation coefficient is 165.5 kDa for full-length P, which is very close to the theoretical molecular mass for a MuV P tetramer (168 kDa). We therefore concluded that the purified MuV P protein or P161-277 is a tetramer, in accordance with the results from the crystal structure.

Nucleocapsid binding.

A number of recombinant P fragments were expressed and purified (Fig. 1 and 2). In order to determine which parts of the MuV P protein interact with the nucleocapsid, a pulldown assay was developed. Purified P fragments were first added to a small Ni-affinity column in saturating amounts. After allowing the supernatant to flow through, the beads were washed once to remove any unbound protein. NLP of MuV prepared as previously reported (26) was added to the loaded column at a 1:1 molar ratio (N:P) and allowed to incubate at room temperature for 15 min. The NLP solution was then allowed to flow through. The loaded column was washed to remove any unbound protein. The proteins were then eluted from the column, and the products were denatured and electrophoresed in SDS-PAGE gels. A summary of the results from the pulldown assays is shown in Fig. 6. These assays confirmed that the extreme C-terminal domain, P342-391, is a nucleocapsid binding domain as previously reported (4). Other C-terminal fragments including residues 251 to 391, 286 to 391, and 342 to 391 each bound NLP (Fig. 6). Unexpectedly, the results showed that the N-terminal region of MuV P (P_1-194) was also capable of binding NLP. This has not been reported previously for a NSV P protein and suggests that the functional domains of MuV P may be different from those of other NSVs.

Fig 6.

NLP pulldown assays. In order to determine which P fragments could interact with NLP, pulldown assays were performed. The headers of the gels in panels A, B, D, and E indicate whether the P protein (or P fragments) saturated column was incubated with NLP (+) or NLP₃₇₉ or not (−). Amino acid numbers corresponding to individual P fragments are labeled on the footer of the gels. The positions for N protein (labeled “N” in panels A and B) and trypsinized N (labeled “NLP₃₇₉” in panels D and E) are noted (A and B), and trypsinized NLP (N₃₇₉) are noted by arrows on the right sides of each gel (D and E). (C) Negative controls. Lanes 1 and 2 show the results for negative controls for NLP and trypsinized NLP₃₇₉, respectively, with no P fragments added to the column.

In addition to using NLP formed by the full-length NLP (540 amino acids), binding assays were also conducted using a trypsin digested form of the MuV NLP (NLP₃₇₉). The digested NLP₃₇₉ was used to determine whether P binding could be abolished in NLP lacking the C-terminal region of N. Similar to the full-length NLP, NLP₃₇₉ was also bound by both the N-terminal (1 to 194 and 51 to 194) and C-terminal (251 to 391, 286 to 391, and 342 to 391) regions of MuV P (Fig. 6D and E).

Affinity of P fragments binding NLP.

Based upon pulldown experiments, we concluded that MuV P contains two separate nucleocapsid binding domains: the N-terminal domain (P1-194) and the C-terminal domain (P286-391). In order to further characterize the interactions between the two separate nucleocapsid binding domains and NLP, surface plasmon resonance analysis was conducted. The recombinant full-length P, P1-194, P161-277, and P286-391 were investigated for their ability to interact with immobilized NLPs or NLP₃₇₉. All analytes, except for P161-277 as a negative control, exhibited measurable binding (Fig. 7 and Table 4). Kinetics experiments were conducted at multiple concentrations (Fig. 7). Modeling the association and dissociation of the P analytes with NLP using the 1:1 Langmuir binding model led to estimated K_A values of 2.02 × 10⁵ M⁻¹ for N1-194, 2.35 × 10⁶ M⁻¹ for P286-391, and 2.23 × 10⁶ M⁻¹ for the full-length P (Table 4). Using K_D calculations, P286-391 also exhibited a relatively higher affinity for NLP with an estimated dissociation constant of 4.26 × 10⁻⁷ M compared to N1-194 (4.95 × 10⁻⁶ M). Next, we tested the ability of P to bind immobilized NLP₃₇₉ (Fig. 7E to H). Modeling the association and dissociation of the P analytes with NLP₃₇₉ using the 1:1 Langmuir binding model led to estimated K_A values of 2.17 × 10⁶ M⁻¹ for P1-194, 1.01 × 10⁵ M⁻¹ for P286-391, and 1.61 × 10⁶ M⁻¹ for the full-length P (Table 4). Of the estimated K_A and K_D values, P1-194 exhibited a relatively higher affinity for NLP₃₇₉ with an estimated dissociation constant (K_D) of 4.6 × 10⁻⁷ M compared to P286-391, which had a K_D value of 9.88 × 10⁻⁶ M. Control surface plasmon resonance experiments with P161-277, which contains the oligomerization domain (OD), did not show measurable interactions with full-length NLP or NLP₃₇₉. The surface plasmon resonance data indicate a 1.3-fold-higher affinity in P binding the full-length NLP versus NLP₃₇₉ but shows a slower dissociation rate (Fig. 7A and E). The curves of Fig. 7F and G show that there are very fast on and off rates for both P1-194 and P286-391 to NLP₃₇₉ compared to the full-length NLP.

Fig 7.

Evaluation of N-P interactions using surface plasmon resonance. (A to D) Full-length NLP was immobilized, and P protein analytes (full-length P, P1-194, P286-391, and P161-277, respectively) were injected over the CM5 chip surface. All analytes displayed a measurable binding response (1 RU = 1 pg/mm²). (E to H) N₃₇₉ was immobilized, and P protein analytes (full-length P, P1-194, P286-391, and P161-277, respectively) were injected over the CM5 chip surface. All analytes displayed a measurable binding response. All experiments were carried out in triplicate. K_A and K_D values are listed in Table 4.

Table 4.

Binding kinetics

Sample	KA (M−1)	KD (M)
P1-194 + N full	2.02 × 105	4.95 × 10−6
P286-391 + N full	2.35 × 106	4.26 × 10−7
P full + N full	2.23 × 106	4.48 × 10−7
P1-194 + N379	2.17 × 106	4.60 × 10−7
P286-391 + N379	1.01 × 105	9.88 × 10−6
P full + N379	1.61 × 106	8.65 × 10−7

DISCUSSION

It has been anticipated that MuV P has a modular structure composed of separated functional domains. Here, we present an experimental approach to more accurately map the functional domains of MuV P. In our proteolytic studies, the N-terminal region of P preceding residue 194 appeared to be less stable than the region that follows since more cleavage sites were found here. This is consistent with observations of other NSV Ps in which the N-terminal region is more disordered compared to the middle and C-terminal regions of P (11, 23, 36, 37). A previous study determined the structure of an extreme C-terminal segment (residues 343 to 391) of MuV P (7). Our proteolytic digestion study did not identify this smaller region of P as a stable fragment. A possible explanation for not being able to map P343-391 in the present study is that this C-terminal region is part of a larger C-terminal domain. The P343-391 region was previously proposed to be MuV P_NBD by analogy to other P proteins. P343-391 may be a molten globule since its crystallization required the addition of stabilizing agents (7). On the other hand, the protein expression data did point to a potential linker between P_OD and the C-terminal domain to be from residues 277 to 286. This could suggest that the C-terminal domain of MuV P is larger than previously thought. P proteins of other NSVs were reported to contain flexible regions between P_N°D, P_OD, and P_NBD (1, 21, 36, 38–43).

Crystallization studies clearly defined that MuV P_OD comprises residues 213 to 277 when P161-277 was used in crystal growth. The structure of MuV P_OD was determined and shows a stable tetramer containing one pair of long parallel α-helices (64 residues) and a second pair of parallel helices in opposite orientation. The presence of P tetramers in solution was confirmed by analytical ultracentrifugation and cross-linking with glutaraldehyde (Fig. 5). The unique organization of this tetrameric coiled-coil where parallel helices are joined with antiparallel helices has not been reported for a native protein before. In this distinctive assembly, interactions between the parallel helices are different from those between the antiparallel helices. In contrast, interactions between any neighboring pair of helices are identical in an all parallel or all antiparallel coiled-coil. The helices in the MuV P_OD tetramer are held together by a large hydrophobic contact area between each monomer (Table 2) and two zippers of charged side chain interactions (Fig. 4). This interesting orientation of antiparallel helices places the N-terminal and the C-terminal regions at both ends of MuV P_OD. This particular structure may well suit the functional requirements of MuV P as discussed below.

We have shown that P binds the full-length MuV NLP and truncated NLP₃₇₉. These observations confirm that the P binding site in the MuV nucleocapsid does not fully require the N-tail, the same as that previously reported (4). In lieu of the apparent dispensability of this N-tail, nucleocapsid binding by MuV P may involve several regions of both N and P. Given this and the novel structure of MuV Pod, P may well be different from other NSV Ps in binding the nucleocapsid. With this open-minded approach, His-tagged fragments covering the full range of MuV P were tested for binding a well-prepared NLP composed of 13 N subunits and a piece of random RNA as previously described (26). In the His-tag pulldown assays, full-length MuV P and the previously identified C-terminal domain, P343-391, were shown to bind NLP as expected. What was unexpected is that the N-terminal region of MuV P from residues 1 to 194 could also bind NLP. This indicates that specific nucleocapsid binding may require both the N-terminal and the C-terminal regions of MuV P, a binding mode that is different from all other known NSV Ps. The NLP binding by the N-terminal and C-terminal regions was confirmed by surface plasmon resonance experiments. This analysis showed that the C-terminal region of P has a higher affinity for NLP. When the N-tail was removed by trypsin digestion (NLP₃₇₉), the C-terminal region of P still binds NLP₃₇₉ with a lower affinity. The affinity of binding NLP₃₇₉ by the C-terminal region of P is similar as that of the N-terminal region of P binding to the full-length NLP. These observations suggest that the C-terminal region of P interacts with the N-tail and additional regions of N, and the N-terminal region of P interacts with regions of N not including the N-tail. However, there is a 10.7-fold increase in binding NLP₃₇₉ by the N-terminal region of P versus the full-length NLP. This increase could point to the accessibility of the N⁰ binding site for P upon removal of the N-tail. Evidence shows that this site is located in the RNA cavity for other N proteins (11, 36, 37). The N⁰ binding motif of P could potentially bind in this site if encapsidated RNA was displaced from NLP₃₇₉. The overall affinity of the full-length P for NLP is lower than that of the individual domains. The affinity values for the full-length tetrameric P may be underestimated because the NLP is representative of a single turn of the helical nucleocapsid and thus does not have the multiple binding sites available in the helical nucleocapsid (described below). In addition, the full-length P was not very stable in solution.

Comparisons of known structures of P uncovered three modes of P oligomerization (Fig. 3B to E). In the SeV P tetramer, oligomerization results in a coiled-coil structure of four long parallel α-helices (21). Similarly, the subunits in the VSV P dimer are also parallel to each other even though the β-sheet on each side is formed by domain swapping (19). The parallel orientation of P subunits in the oligomers places the N-terminal regions on one end and the C-terminal regions on the other. Since specific nucleocapsid binding in these viruses requires only the C-terminal region of P, the N-terminal region may be placed far away from the C-terminal region. In the RABV P dimer, however, the N-terminal and C-terminal regions are each positioned at the same end due to the U-shaped structure that links two antiparallel helices in each P_OD subunit (20). There is no indication that the N-terminal region of RABV P is involved in binding the nucleocapsid. It is not clear what functional implications this particular orientation of RABV P has. MuV P represents a third mode of oligomerization in which parallel and antiparallel P subunits are associated together to form a tetramer. In the MuV P tetramer, the dimensions of the helix bundle are similar to those of the coiled-coil structure of the SeV P tetramer, but the orientations of the helices is different. The antiparallel orientation of the MuV P subunits places two N-terminal regions and two C-terminal regions on each end of the tetramer. This organization may be functionally required because both the N-terminal and the C-terminal regions of P are involved in binding the nucleocapsid, as shown in this report. The two terminal regions should be in close proximity to each other in order to effectively bind the nucleocapsid. The full P binding sites for each terminal regions in the nucleocapsid remain to be identified for MuV. The results shown in Fig. 6 and 7 suggest that there are P binding sites within residues 1 to 379 in addition to the C-terminal tail of N. This mode of binding requires that P bind at least two or more N subunits either in the same turn or in successive turns in the helical nucleocapsid. Since the conserved sequences for initiation of replication (PrE I and II) are located in two successive turns of the nucleocapsid (44), it is possible that P plays a role in recognition of the two sites by the viral RdRp. Our results again demonstrated that NSV P proteins all have a modular structure with separated functional domains and are present as oligomers (Fig. 8). Although NSV P proteins harbor analogous functions, each may have a very different domain organization and can interact with other viral proteins, especially the nucleocapsid, in a very different manner.

Fig 8.

Diagram of the domain organization for four different NSV Ps. P_OD is typically located in the central region of P. P_N°D and P_NBD are linked to P_OD through flexible loops. This appears to be a common modular structure for all NSV P proteins.