Research ArticleVIROLOGY

An ultraweak interaction in the intrinsically disordered replication machinery is essential for measles virus function

See allHide authors and affiliations

Science Advances  22 Aug 2018:
Vol. 4, no. 8, eaat7778
DOI: 10.1126/sciadv.aat7778

Abstract

Measles virus genome encapsidation is essential for viral replication and is controlled by the intrinsically disordered phosphoprotein (P) maintaining the nucleoprotein in a monomeric form (N) before nucleocapsid assembly. All paramyxoviruses harbor highly disordered amino-terminal domains (PNTD) that are hundreds of amino acids in length and whose function remains unknown. Using nuclear magnetic resonance (NMR) spectroscopy, we describe the structure and dynamics of the 90-kDa N0PNTD complex, comprising 450 disordered amino acids, at atomic resolution. NMR relaxation dispersion reveals the existence of an ultraweak N-interaction motif, hidden within the highly disordered PNTD, that allows PNTD to rapidly associate and dissociate from a specific site on N while tightly bound at the amino terminus, thereby hindering access to the surface of N. Mutation of this linear motif quenches the long-range dynamic coupling between the two interaction sites and completely abolishes viral transcription/replication in cell-based minigenome assays comprising integral viral replication machinery. This description transforms our understanding of intrinsic conformational disorder in paramyxoviral replication. The essential mechanism appears to be conserved across Paramyxoviridae, opening unique new perspectives for drug development against this family of pathogens.

INTRODUCTION

Measles virus (MeV) is a nonsegmented negative-sense RNA virus belonging to the family of Paramyxoviridae that includes a number of emerging human pathogens with dangerously high mortality rates for which there is currently no treatment. In addition to the viral polymerase (L), paramyxoviral replication machinery is composed of the nucleoprotein (N), which encapsidates the viral genome, and the tetrameric phosphoprotein (P) (13), a cofactor of L that interacts with N at different stages of the viral replication process.

Paramyxoviral P proteins are essential for replication and transcription, interacting with N at different stages of the viral cycle (49) and exhibiting very long intrinsically disordered N-terminal domains (PNTDs), whose function is not understood (10, 11). The first 40 amino acids of P bind tightly to N that is maintained in a monomeric state (N0P) (12, 13) before encapsidation of the viral RNA to form the nucleocapsid (NC). X-ray structures of heterodimeric constructs of N0P1–40 have been determined for MeV (14), Nipah virus (NiV) (15), parainfluenza virus 5 (16), and metapneumoviruses (17). In all cases, the flexible domains of N that are required for assembly into NCs (18, 19), including the intrinsically disordered C-terminal domain of N (NTAIL) and most of the disordered domain of PNTD (more than 400 amino acids in NiV and 300 amino acids in MeV), were removed to facilitate crystallization. The functional N0P assembly, however, comprises extensive disorder (more than 450 amino acids per heterodimer), whose role has never been studied at a molecular level. The presence of this level of disorder, in viruses whose genetic information is normally so parsimoniously exploited, remains unexplained.

To investigate the role of the disordered PNTD, we designed an MeV N0P construct comprising integral N, which is fused to the entire 300–amino acid PNTD (20). This approach allowed us to use solution nuclear magnetic resonance (NMR), small-angle x-ray scattering (SAXS), ensemble simulation, and NMR exchange spectroscopy to investigate the conformational behavior of the 90-kDa N0PNTD complex, including the intrinsically disordered regions (IDRs) PNTD, NARM, CARM, and NTAIL (in total of more than 450 amino acids), and to identify the role of these flexible domains in the molecular processes of replication and transcription (see fig. S1 for domain description). The atomic resolution description of such a highly dynamic complex presents significant challenges for structural biology, requiring approaches that explicitly account for their time- and ensemble-dependent conformational heterogeneity (2123) and the dynamic exchange between free and bound forms of the proteins (2427). Here, we apply these approaches to describe the behavior of PNTD alone and upon formation of the N0P complex.

Rather than containing apparently nonfunctional sequences, PNTD harbors a previously hidden linear motif that interacts ultraweakly but specifically with the surface of N. The two N0PNTD binding sites are separated by 150 highly dynamic residues and differ in affinity by orders of magnitude, allowing PNTD to transiently wrap around the surface of N, exchanging rapidly between compact and more extended forms. This long-range allosteric coupling possibly enhances the chaperone function by frustrating the interaction of NCORE (the folded domain of N) with RNA or host factors. Cell-based MeV minigenome studies combined with mutagenesis demonstrate that despite the inherently weak affinity of the newly discovered interaction, this mechanism is essential for viral transcription and replication and therefore also for successful host infection. The interaction motif appears to be conserved within paramyxoviruses, providing a new target for treating these dangerous human diseases.

RESULTS

NMR spectroscopy describes the conformational behavior of PNTD

We characterized the conformational behavior of PNTD at the molecular level using NMR spectroscopy (fig. S2). Protein backbone resonances were assigned, and representative multiconformational ensembles were generated on the basis of 1H, 15N, and 13C (CO, Cα, and Cβ) backbone chemical shifts, using a combination of flexible-meccano and the genetic algorithm ASTEROIDS (Fig. 1 and fig. S2) (28, 29). The resulting ensemble faithfully predicted experimental ensemble-averaged residual dipolar couplings (RDCs) that are not used in the description (Fig. 1). This site-specific description of the conformational sampling of free PNTD reveals that α helices contained in the first 37 residues (α1 and α2), which are involved in formation of the N0P complex, appear as transient helical structures within the unbound protein, as previously observed for vesicular stomatitis virus P (30). Two additional regions have similar or higher propensities to form helices in the unbound PNTD: residues 87 to 93 (α3) and 189 to 198 (α4; Fig. 1A and fig. S2). Spin relaxation reveals increased rigidity in α2, α4, and the hydrophobic region around position 110 that is implicated in the binding of STAT (signal transducers and activators of transcription) (31), while long, highly dynamic segments link α1/2 to α3 and α3 to α4 (Fig. 1).

Fig. 1 P1–304 structural dynamics.

(A) Population of αR conformation of P1–304 as calculated from an ASTEROIDS ensemble based on chemical shifts and SAXS (gray bars) compared to those expected from a statistical coil ensemble (flexible-meccano, black lines). (B) NHN RDCs (1DNH-N) of P1–304 obtained from alignment in a liquid crystal made from polyethylene glycol (PEG) and 1-hexanol (gray bars). Red line shows NHN RDCs back-calculated from the ASTEROIDS ensemble selected against chemical shifts only. (C) Heteronuclear {1H}-15N Overhauser effects (nOes) of P1–304, measured at a 1H frequency of 600 MHz and 25°C. (D) Representation of the differential flexibility along the sequence of PNTD. The image shows the color and width of the ribbon as a function of the heteronuclear nOe in (C) (thick, red ribbon corresponds to more flexible regions; thin, blue ribbon to less flexible regions). Annotations indicate the presence of helical elements. The protein is highly dynamic, and the actual conformation is not representative of the ensemble of conformers.

The highly disordered N0P complex contains a hidden dynamic interaction site with N

We used NMR to investigate the N0PNTD complex by titrating 15N-labeled P1–304 with unlabeled P1–50N1–525 (Fig. 2). P1–304 is able to displace P1–50 from the P1–50N1–525 complex, supporting the observation that this complex can be purified from a mixture of P1–304 and P1–50N1–525 (figs. S3 and S4A). In addition to decreased peak intensities in α1/2, known to engage in the N0P complex (1315, 32), a 20-residue sequence is found to interact with N, comprising the α4 helical motif. Small local increases in relaxation rates were also observed, centered on aromatic residues F65 and Y110 to Y113, suggesting the existence of additional transient hydrophobic interactions with N within the complex.

Fig. 2 P1–304 interaction with N.

(A) R2 of P1–304 (gray bars) and P1–304N1–525 (red lines) as purified from a Superdex 200 column (fig. S3) measured at a 1H frequency of 950 MHz. (B) Interaction profile of P1–304 with P1–50N1–525. Titration admixtures included 25 (gray), 50 (red), 100 (green), and 150 μM (blue) final concentration of P1–50N1–525 and P1–304 at a final concentration of 50 μM. Shown are normalized peak intensities (I/I0). (C) Interaction profile of P1–304,HELL→AAAA with P1–50N1–525. Concentrations and colors are the same as in (B). (D) 1H-15N heteronuclear single-quantum coherence (HSQC) spectrum of P1–304 in absence (blue) and presence (red) of P1–50N1–525. ppm, parts per million. (E) 1H-15N HSQC spectra of P1–304 (red), P1–50N1–525 (green), and P1–304N1–525 (blue).

Exchange NMR reveals the detail of the additional N interaction site

To further investigate this additional interaction site, intermediate stoichiometric complexes were formed where 15N-labeled P1–304 is in exchange between bound and unbound forms (Fig. 2). While the bound α1/2 region is in slow exchange, manifest through duplication of the peaks on the edges of the helical region (fig. S4, B to D), the region around α4 exhibits chemical exchange kinetics in the millisecond regime between free and bound PNTD, as evidenced by 15N Carr-Purcell-Meiboom-Gill (CPMG) relaxation dispersion (33) measured at substoichiometric ratios (50 μM P1–304 and 25 μM P1–50N1–525; fig. S5).

Using a shorter construct of PNTD (P140–304) that does not contain the N-terminal interaction site, the α4 site is shown to bind independently from α1/2 (Fig. 3 and fig. S5, C and D). This analysis reveals much more detail concerning the exact nature of the interaction motif, particularly in identifying an additional linear motif (181DVETA185, termed δ) immediately preceding helix α4, which also participates in the interaction (Fig. 3B). This motif exhibits no detectable intrinsic propensity or secondary structure in the free form of PNTD. 15N relaxation dispersion experiments measured at two magnetic field strengths (600 and 950 MHz) allowed us to estimate the intrinsic dissociation constant (Kd) for the combined δα4 interaction with N0P as 614 ± 218 μM (Fig. 3D). Both α4 and δ motifs have the same effective Kd and exchange kinetics, indicating that they participate in a concerted interaction with N0P. NMR titration studies using the N-terminal region of PNTD, without the δα4 motif (P1–160), show that the transient hydrophobic interaction sites centered on residues 65 and 110 are still present even in the absence of the second binding motif (fig. S6).

Fig. 3 δα4 interaction with N.

(A) Intensity ratios of 15N P140–304 peaks extracted from 1H-15N HSQC spectra in the presence of 5% (gray bars), 10% (red), or 20% (blue) P1–50N1–525 with respect to the unbound P140–304. Concentration of P140–304 remained constant at 200 μM throughout the titration. (B) R of P140–304 alone (gray bars) and with 20% P1–50N1–525 (blue) recorded at 16.5 T. (C) CPMG relaxation dispersion of 15N P140–304 (200 μM) in the presence of 20% P1–50N1–525: ΔR2,eff was determined at 14 T as the difference in effective R2 at CPMG frequencies of 31 and 1000 Hz. (D) CPMG relaxation dispersion of 15N P140–304 (200 μM) in the presence of 20% P50N525 at 22.3 T and 14 T. Data from eight sites throughout δ and α4 were fitted simultaneously, assuming a two-site exchange process giving kex = 624 ± 88 s and population, 4.7 ± 0.5%. Example sites from residues S180 (δ) and L193 (α4) are shown.

Identification of a conserved interaction motif within PNTD helix α4

To delineate the core interaction site involved in δα4 binding to N, we compared the sequences and secondary structure predictions of PNTD from different paramyxoviruses, revealing that a helix resembling α4 is predicted throughout the family (Fig. 4A). A comparison of 259 nonredundant, curated PNTD MeV sequences also shows clear conservation of sequence identity in α1, α3, and the δα4 motif (Fig. 4B), including two consecutive bulky hydrophobic residues (isoleucine, leucine, or valine). On the basis of these comparisons, the central interaction motif, 191HELL194 was mutated (HELL→AAAA). This mutant no longer interacts with N through δα4 (Fig. 2C and fig. S7), although the helical propensity remains intact (fig. S7C). Comparison of MeV P sequences also underlines the strongly conserved acidic nature of the long dynamic loop connecting residues 125 and 170 (comprising 10 conserved acidic side chains and no conserved basic side chains).

Fig. 4 Conservation and functional impact of the δα4 motif.

(A) Alignment of phosphoproteins from different related viruses (respective UniProt identifiers are shown on the left, and morbilliviruses are shown above the dashed line). Full PNTD sequences aligned with MUSCLE (35). Positions of the helical elements predicted using PSIPRED (57) are shown as boxes (red, negatively charged; blue, positively charged). (B) Conservation of MeV sequences of PNTD. Comparison of 259 nonredundant sequences curated in Genbank (60). The position of α helices, additional interaction sites δ and α, STAT interaction site, and the RNA editing site defining the start position of the unique domain of V are indicated. Image obtained using WebLogo software (http://weblogo.berkeley.edu/logo.cgi). (C and D) Functional impact of HELL→AAAA mutation in cellulo. (C) Ability of MeV P constructs to associate with MeV N protein in cultured human cells, as determined by Gaussia luciferase–based protein complementation assay. Gray zone indicates the threshold NLR (normalized luminescence ratio) value. (D) Inability of HELL→AAAA P mutant to support the expression of two reporter genes from either a (+) or (−) strand MeV minigenome when coexpressed with MeV N and L proteins using a reverse genetic assay [relative light units (RLU): luciferase reporter activity]. NanoLuc, nanoluciferase.

The δα4 motif induces a more compact conformation of N0P

We investigated the influence of the HELL→AAAA mutation on the overall dimensions of the P1–304N1–525 complex using SAXS. Ensemble analysis based on SAXS and NMR data from wild-type (WT) and mutated forms reveals that the P1–304N1–525 ensemble requires a bimodal distribution of radii of gyration (fig. S8, A to D), likely due to interactions between δα4 and N. Mutation of the HELL motif leads to significant quenching of the population of more compact conformations, in agreement with the abrogation of binding around δα4. The presence of the 125–amino acid disordered NTAIL domain potentially masks the impact of compaction of N0P on SAXS data. We therefore carried out a similar analysis on N1–405P1–304 (in the absence of NTAIL) further emphasizing the effect of δα4, exhibiting average radii of gyration of WT and HELL→AAAA forms of N1–405P1–304 of 45 ± 6 and 58 ± 8 Å, respectively (fig. S8, E and F).

Functional relevance of the HELL interaction site in cellula

The functional importance of the δα4 N0P interaction site for viral function was determined by measuring polymerase replication in vivo using cell-based MeV genome replication assays exploiting reverse genetics with a dual-luciferase reporter (9). In these assays, the viral polymerase function relies on simultaneously providing a host cell with the antigenomic or genomic RNA, N, P, and the polymerase (L), and polymerase-mediated transcription is determined by the quantification of the luciferase activities. Irrespective of whether antigenomic or genomic RNA was provided, luciferase activity observed with PWT was completely suppressed when using PHELL→AAAA (Fig. 4D). All P variants were expressed (fig. S9A), and N:P binding and P oligomerization was unaffected by mutation (Fig. 4C and fig. S9B), indicating that the abrogation of transcription/replication results unambiguously from the absence of the δα4 motif.

δα4 binds N on its N-terminal lobe

Although NCORE backbone resonances are not observed in 1H-15N correlation spectra of N0P, a truncated construct comprising the N-terminal lobe NNTD (34) was used to identify the interaction site (fig. S10, A and B) that is located in a contiguous region spanning residues 96 to 127, situated on the opposite side of N to the known N0P binding site. The HELL→AAAA mutant of P140–304 did not interact with P1–50N1–525 (fig. S7), and the interaction profile of P140–304 closely resembles that obtained from P1–50N1–525, confirming that NNTD contains the full interaction site for δα4 (Fig. 5).

Fig. 5 N-δα4 binding site and N0P model.

(A) Ratio of peak intensities from 1H-15N HSQC spectra of free 15N P140–304 and 15N P140–304 bound to unlabeled N1-261 (gray bars) superimposed with an intensity ratio of 15N P140–304 unbound and bound to P1–50N1–525. (B) Transverse relaxation of 15N13C2DN1–261. (C) Secondary structure propensities (SSPs) for N1–261. Red bars indicate interaction regions with P140–304. (D) 1H-15N HSQC spectra of 15N, 2DN1–261 alone (red) and in the presence of 50 μM (blue) and 170 μM (cyan) unlabeled P140–304. (E) Cartoon of interaction of P1–304 (yellow and blue) with NCORE (gray). Red surface denotes the interaction region on N. Dashed red line indicates the dynamic region of NCORE. IDRs are marked as dotted lines. Yellow indicates transient helices. (F) Representation of conformational space available to P1–304 (blue) when both α1/2 (yellow spheres) and δα4 (yellow spheres) are bound to NCORE (gray surface; red shading shows the α4 interaction region on NCORE). All conformations shown have both interaction sites in proximity to the respective N binding sites. Gray indicates NTAIL.

DISCUSSION

Paramyxoviral phosphoproteins are essential for viral replication and transcription, acting as essential cofactors for the polymerase complex. The architecture of paramyxoviral P proteins is highly conserved over Paramyxoviridae, comprising a tetrameric coiled-coil domain and a long intrinsically disordered PNTD, whose length varies between 215 and 470 amino acids. Although the first 40 N-terminal amino acids are known to bind N0, the role of the remainder of PNTD has remained unknown, and no rationale exists for the conserved requirement of these long unfolded chains in all paramyxoviral P proteins.

We used NMR spectroscopy to investigate the conformational behavior of MeV PNTD in its free state and to describe the structural, dynamic, and kinetic behavior of this highly disordered 90-kDa complex. Although unfolded, PNTD exhibits transient helical propensities in the N-terminal N binding site (α1/2), around residues 87 to 93 (α3) and 189 to 198 (α4), linked by long, highly dynamic segments (Fig. 1).

In addition to the known N-terminal binding concerning the first 40 amino acids of PNTD, a 20-residue sequence, comprising the α4 191HELL194 motif, preceded by the additional unstructured motif δ, also interacts with N. The HELL interaction motif, as well as the helix itself, appears to be conserved within morbilliviruses and representative paramyxoviruses, in particular the two bulky hydrophobic residues at the C terminus of the motif (Fig. 4A). The distance between the α1/2 and putative δα4 binding sites is, however, highly variable among different viruses—for example, NiV PNTD is considerably longer (470 amino acids) than in MeV and is predicted to exhibit helical propensity at positions 340 to 348, containing the sequence RELL. Despite near-negligible identity among PNTD from different paramyxoviruses, the MUSCLE sequence alignment algorithm (35) successfully aligns these conserved motifs, even when comparing entire PNTD sequences of very different lengths. A number of physical characteristics are also conserved; for example, the interaction motif is preceded by a negatively charged region (4 acidic residues out of 12 in MeV PNTD) and followed by a positively charged region (4 basic residues out of 14; Fig. 4B). Similarly, analysis of nonredundant MeV sequences reveals that δα4 is one of only four strongly conserved linear sequences in the 300–amino acid PNTD (Fig. 4B).

The δα4:NCORE interaction occurs independently of the presence of α1/2, as demonstrated by a truncation mutant of PNTD. Using NMR relaxation dispersion, we have investigated the kinetics of the interaction, revealing that the intrinsic interaction strength is ultraweak, with a Kd around 600 μM. The effective affinity within the PNTD:NCORE complex is expected to be significantly higher because of the elevated effective population of δα4:NCORE that is maintained by the much higher-affinity α1/2:NCORE interaction. The α1/2:NCORE interaction is in slow exchange on the NMR time scale and has sufficiently slow dissociation rates that the complex can be copurified on a size-exclusion column.

The binding site of δα4 on N (residues 96 to 127) is located at the extremity of its N-terminal lobe, which is the most evolutionarily variable part of NCORE within paramyxoviruses and the Morbillivirus genus (15). A single-chain antibody that binds to this region of NCORE from vesicular stomatitis virus, a related member of the Rhabdoviridae family, is a potent inhibitor of virus transcription and replication (36), while a homologous region in respiratory syncitial virus binds the C terminus of the P protein and is the target for inhibitory small-molecule development (37). In MeV, the affected binding surface on N, situated at the opposite end of NCORE to the α1/2 binding site, comprises an exposed β-strand (Fig. 5), adjacent to a highly dynamic surface loop that is devoid of any secondary structure in solution. This subdomain exhibits conformational heterogeneity between known structures, adopting a helix in crystalline N0P (14) [and in the homologous NiV structure (15)] and a surface loop in the cryo–electron microscopy NC structure (18). A recent study identified this domain as insertion-tolerant, allowing the molecular recognition element of NTAIL to be grafted into the domain while maintaining viral replication (38).

On the basis of our observations, we conclude that, contrary to current understanding, whereby the N0P complex comprises a single N:P binding site that is sufficient for chaperone activity, MeV PNTD interacts with monomeric N via two distinct linear regions, separated by 150 amino acids, and with affinities that differ by many orders of magnitude. The presence of two N binding sites on PNTD helps to rationalize the presence of long intrinsically disordered PNTD domains (11) throughout the paramyxoviral family. The length of the long dynamic linker connecting the two interaction sites provides sufficient degrees of freedom to sample very different conformations and, in particular, to allow for significant conformational disorder within the complex (Fig. 5), even when both sites are occupied. The relative weakness of the δα4 interaction, therefore, leads to rapid exchange between more voluminous free and more compact bound forms of P40–300, while α1/2 remains bound. When simultaneously bound to N, the α1/2 and δα4 binding sites are positioned at opposite extremities of the NCORE structure, so that the conformational fluctuations of the acidic loop between the binding sites on PNTD are able to maximally frustrate access to the surface of N, potentially inhibiting interaction with RNA and consequent self-assembly with other N monomers as well as interactions with host proteins that may play a role in the immune response (39).

The δα4 interaction is abrogated by mutation of four central residues (191HELL194) of the α4 motif, although its helical propensity is maintained under alanine mutation, demonstrating that its helical nature alone is not sufficient to ensure interaction. Mutation of α4 also removes the more compact, enwrapped form of N0P from the equilibrium as shown from small-angle scattering. We demonstrated the importance of the δα4 N0P interaction site for viral function by measuring polymerase replication in vivo using recombinant minigenome assays. Only the N-genomic RNA NC can be used as a template, so that the observed lack of reporter gene expression from the genomic (−) stranded RNA demonstrates that the HELL→AAAA mutation prevents assembly of N and RNA into a functional NC and subsequent transcription. The possibility that HELL→AAAA specifically targets replication and not transcription is supported by the previous demonstration that a PNTD-deleted P protein remains active in transcription in the closely related Sendai paramyxovirus (40).

Despite its weak affinity, the δα4 interaction is therefore essential for viral function, reinforcing recent observations of the importance of ultraweak interactions for intrinsically disordered protein function, as illustrated for example in the case of linear motifs present in the nuclear pore complex (41). In the case of N0P, the combination of distinct affinities on PNTD, regulated by its highly disordered nature, coordinates allosteric coupling between the two interaction sites linking opposite ends of NCORE. The bipartite N:PNTD interactions may also facilitate NC assembly or provide the molecular basis of formation of organelles that stimulate viral replication (42). We note that the observations made here in the context of PNTD are potentially equally valid for the genome-edited viral protein V (4346) that comprises PNTD until the editing site (residue 229), followed by a folded zinc finger domain, which is unique to V.

In summary, an atomic resolution description of the structure and dynamics of the highly disordered 90-kDa N0P complex of MeV, comprising 450 intrinsically disordered amino acids, reveals the presence of a hitherto unidentified N interaction site, 150 amino acids distant from α1/2 that is essential for viral function. The newly identified motif exhibits an intrinsic affinity that is remarkably low but nevertheless specific, constraining the unfolded chain to rapidly interchange between different states while bound at the N-terminal site, thereby affording protection against N-RNA, N-N, or N–host factor interactions. We discover that this linear motif is essential for MeV replication and transcription, opening up new perspectives for drug development against these increasingly important human pathogens. Paramyxoviruses share similar replication machinery with filoviruses, including Ebola, suggesting that the discovery of this essential allosteric mechanism may have yet further-reaching consequences for human health.

MATERIALS AND METHODS

Cloning, protein expression, and purification

N0P constructs comprising full-length PNTD were generated by purifying a heterodimeric N1–525P1–304 complex with a TEV (tobacco etch virus) cleavage site between the two proteins, similar to the production of the shorter P1–50N1–525 described previously (20), or by mixing TEV-cleaved P1–50N1–525 with P1–304 at equal concentrations and subsequent purification of the P1–304N1–525 complex by size exclusion chromatography (SEC; fig. S3). TEV cleavage was performed before a final SEC (Superdex 200, GE Healthcare) in NMR buffer [50 mM Na-phosphate (pH 6), 150 mM NaCl, and 2 mM dithiothreitol (DTT)]. P1–304 was cloned into pET41c(+) between the Nde I and Xho I cleavage sites, where the Xho I site was ligated with a cleaved Sal I site of the insert, yielding a construct with a C-terminal 8His-tag. P1–304 was expressed in Escherichia coli RosettaTM (λDE3)/pRARE (Novagen) overnight at 20°C after induction at an optical density of 0.6 with 1 mM isopropyl-β-d-thiogalactopyranoside. Cells were lysed by sonication and subjected to standard Ni purification in 20 mM tris (pH 8) and 150 mM NaCl. The protein was eluted from the beads in 20 mM tris (pH 8), 150 mM NaCl, and 400 mM imidazole and was then concentrated and subjected to SEC (Superdex 200) in NMR buffer. The HELL mutation was inserted into P by site-directed mutagenesis, and P1–304,HELL→AAAA was expressed and purified the same way. Shorter P constructs (P1–100, P1–160, P140–304, and N1–261) were cloned into pet41c(+) between Nde I and Xho I sites and were subjected to the same expression and purification procedure; the final SEC step was conducted on a Superdex 75 (GE Healthcare) column. N1–525P1–304 was cloned in two steps, first inserting N1–525-TEVsite between the Nde I and Not I cleavage sites of pET41c(+), followed by insertion of P1–304 between the Not I and Xho I sites into the resulting vector. The Xho I site was ligated with a Sal I site of the P1–304 insert. Expression of unlabeled protein was performed in LB medium. Protein labeled for NMR (15N and 13C) was expressed under the same conditions in M9 minimal medium. For deuterated protein (N1–261), the M9 medium was made in D2O.

NMR experiments

Unless otherwise noted, all experiments were acquired in NMR buffer [50 mM Na-phosphate (pH 6), 150 mM NaCl, and 2 mM DTT] at 25°C. The spectral assignments of the different 13C- and 15N-labeled P constructs were obtained using sets of triple resonance experiments correlating Cα, Cβ, and CO resonances at a 1H frequency of 600 MHz. N1–261 was assigned as a 13C-, 15N-, and 2D-labeled sample using band-selective excitation short transient–transverse relaxation-optimized spectroscopy (BEST-TROSY) triple-resonance experiments recorded at a 1H frequency of 700 MHz (47). The spectra were processed with NMRPipe (48), and automatic assignment was performed with the program MARS (49) and manually verified. Secondary chemical shifts were calculated using the random coil values from refDB (50). For the measurement of RDCs, 13C, 15N-labeled P1–304 was aligned in polyethylene glycol (PEG) and 1-hexanol, yielding a D2O splitting of 25 Hz (51). RDCs were measured using BEST-type HNCO experiments that allow for spin-coupling measurements in the 13C dimension at a 1H frequency of 800 MHz (52).

Interactions between different proteins were performed as described in the main text and figures. Peak intensities (I) and 1H as well as 15N chemical shifts were extracted from 1H-15N HSQC or 1H-15N TROSY spectra (for N1–261). Combined chemical shift differences (ΔCS) were calculated asEmbedded Image(1)15N R1, R, {1H}-15N heteronuclear Overhauser effects (nOes) were acquired at different fields, as indicated in the figure legends. 15N R1 relaxation rates were obtained by sampling the decay of magnetization at 5 to 10 delays between 0 and 1.71 s. R relaxation rates were measured using 5 to 10 delay times between 0.001 and 0.25 s. The spin-lock field was 1500 Hz, and R2 was calculated from R1 and R, considering the resonance offset (53).

15N relaxation dispersion was measured at a P1–304 concentration of 50 μM in the presence of 25 μM unlabeled P1–50N1–525 at 600 MHz and using 14 points at CPMG frequencies between 31 and 1000 Hz using a constant-time relaxation of 32 ms (33). 15N relaxation dispersion of P140–304 was measured at a concentration of 200 μM in the presence of 40 μM P1–50N1–525 at 1H frequencies of 600 and 950 MHz under otherwise the same conditions as for P1–304. Data recorded at both fields were fit together using ChemEx (https://github.com/gbouvignies/chemex) and a two-state exchange model. The Kd value was calculated from the concentrations of P140–304 and P1–50N1–525 as well as the fraction of bound P140–304 (pb), as obtained from the fit and assuming a 1:1 binding stoichiometry. The error of the Kd value was estimated by error propagation from the fitting error of pb and assuming a 5% error in concentration determination of both proteins. Nonoverlapped residues 180, 181, 182, 184, 185, 193, 196, and 199 covering both δ and α4 were included in the fit.

Small-angle x-ray scattering

SAXS experiments were performed on P1–304N1–525 complexes purified, as described above. Samples were adjusted to three different concentrations between 0.25 and 3.5 mg/ml and were measured at 20°C on the BM29 beamline at the European Synchrotron Radiation Facility (ESRF), Grenoble, France. Scattering was recorded at a wavelength of 0.992 Å, and samples were exposed 10 times for 2 s. All frames were analyzed for radiation damage and excluded if necessary. All other frames of sample and buffer were averaged respectively, and buffer scattering curves were subtracted from the scattering curves of the samples.

Ensemble description of P1–304 and P1–304N1–525

A statistical coil ensemble of P1–304 comprising 10,000 structures was generated using flexible-meccano. Two hundred conformations that best described the experimentally obtained N, HN, Cα, Cβ, and CO chemical shifts were selected from the ensemble using the genetic algorithm ASTEROIDS. A new ensemble of 8500 conformers was generated on the basis of the Φ and Ψ angles of the selected conformers, supplemented with 1500 conformers from the initial statistical coil ensemble (28, 29). This new ensemble was subjected to another round of ASTEROIDS selection, and the iteration step was repeated eight times until the ensemble converged with respect to the chemical shifts. Ensemble averaged chemical shifts were calculated using SPARTA (54), and SAXS curves were obtained using CRYSOL (55). Another round of ASTEROIDS was used to include, in addition to chemical shifts, the SAXS curves into the selection.

To build a conformational model of P1–304N1–525 and N1–405P1–304, the structure of NCORE and the first 50 residues of P in the complex were adapted from the x-ray structure of MeV (14) and NiV (15) N0P. The N-terminal flexible arm of N, NTAIL, and PNTD were subsequently built onto the structure using flexible-meccano with the helical sampling of the NTAIL molecular recognition element, as previously described (56), as well as the Φ and Ψ angles of the last selection of P1–304 to describe PNTD within the complex. The order by which the different chains were built onto the folded structure did not make a difference for the final ensemble of 100,000 conformers. ASTEROIDS selections with sizes of 5 to 50 conformers were then selected on the basis of the SAXS scattering curve. The same procedure was applied to describe the scattering curve of P1–304,HELL→AAAAN1–525 and P1–304,HELL→AAAAN1–405. To illustrate the interaction between the HELL motif within P1–304 and N, 200 conformers were filtered that fulfill a distance constraint of 5 Å between the P1–304 HELL motif and its binding site on N (Fig. 5F).

Identification of the HELL interaction site by sequence analysis

Various paramyxovirus phosphoproteins were aligned in a multiple sequence alignment using MUSCLE (35). Although sequence conservation in IDRs is weak, the algorithm identified a HELL-like motif within all sequences that coincided with the position of predicted helices of all viruses using the program PSIPRED (57).

In cellula analyses

The expression of P protein constructs was verified by Western blot using a polyclonal rabbit MeV P protein as detailed (9). The ability of P and domains of P to associate with N protein in cellula was tested by the Gaussia-based protein complementation assay, as detailed elsewhere (8, 9, 20). Briefly, plasmids encoding P protein fused to either the C terminus or N terminus of the N-terminal glu1 domain, and N protein fused to the C terminus of glu2 domain by polymerase chain reaction and In-Fusion (Clontech) recombination. Glu1 and glu2 fusion constructs were transfected in human embryonic kidney–293T cells (from CelluloNet BioBank BB-0033-00072, SFR BioSciences), and Gaussia luciferase activity was measured after 2 days of culture. The results were expressed in normalized luminescence ratio as described (58). Each test was carried out three times with each condition applied in triplicate.

Minigenome assay was performed in BSR-T7 cells, which constitutively express the T7 polymerase (59), as detailed previously (9). The minigenome included the leader and trailer sequences separated by a first gene encoding the firefly luciferase and a second gene encoding the Gaussia luciferase, the expression of which relies on the edition of the gene transcript. The two genes are separated by an intergenic junction elongated by 324 nucleotides (9). Briefly, plasmids encoding for MeV minigenome RNA of (+) or (−) polarity, N, L, and P protein under the control of T7 promoter were transfected together in BSR-T7 cells. Two days later, the expression of the two-luciferase reporter gene was recorded by luminescence measurement in the presence of their specific substrate. Each test was performed three times with each condition performed in triplicate.

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/4/8/eaat7778/DC1

Fig. S1. Scheme showing the different domains of N and P proteins referred to in the main text.

Fig. S2. NMR of P1–304.

Fig. S3. Purification of P1–304N1–525 complexes.

Fig. S4. Interaction of N with α1/2 of P.

Fig. S5. Analysis of P interaction dynamics with N1–525.

Fig. S6. Interaction of P1–160 with N.

Fig. S7. P140–304 interaction with P1–50N1–525.

Fig. S8. Defining the conformational ensemble of P1–304N1–525 from SAXS curves.

Fig. S9A. Expression of MeV P protein constructs used for functional studies.

Fig. S9B. Functional assays using MeV minireplicon.

Fig. S10. NMR spectroscopy of N-terminal domain of NCORE in interaction with P1–304.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.

REFERENCES AND NOTES

Acknowledgments: Funding: We thank GRAL (ANR-10-LABX-49-01) and Finovi, the Grenoble Instruct–ERIC Center (ISBG; UMS 3518 CNRS-CEA-UGA-EMBL) with support from FRISBI (ANR-10-INSB-05-02) within the Grenoble Partnership for Structural Biology and the ESRF. S.M. acknowledges the European Molecular Biology Organization for a long-term fellowship (ALTF 468-2014) and the European Commission (EMBOCOFUND2012, GA-2012-600394) via Marie Curie Action. Author contributions: S.M., M.R.J., D.G., R.W.H.R., and M.B. conceived and designed the project and wrote the manuscript. S.M., G.C., S.I., S.G., D.M., M.R.J., C.L., and D.G. carried out experiments. All authors analyzed data, discussed results, and commented on the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from authors.
View Abstract

Navigate This Article