Consequences of Phosphorylation in a Mononegavirales Polymerase-Cofactor System

Polymerase-cofactor interactions such as those addressed in this study are absolute requirements for mononegavirus RNA synthesis. Although cofactor phosphorylation is present in most of these interactions, its effect, if any, on this protein-protein interaction had not been addressed.

ABSTRACT

Vesicular stomatitis virus (VSV) is a member of the order Mononegavirales, which consists of viruses with genomes of nonsegmented negative-sense (NNS) RNA. Many insights into the molecular biology of NNS viruses were first made in VSV, which is often studied as a prototype for members of this order. Like those of other NNS viruses, the VSV RNA polymerase consists of a complex of the large protein (L) and the phosphoprotein (P). Recent discoveries have produced a model in which the N-terminal disordered segment of P (P_NTD) coordinates the C-terminal accessory domains to produce a “compacted” L conformation. Despite this advance, the role of the three phosphorylation sites in P_NTD has remained unknown. Using nuclear magnetic resonance spectroscopy to analyze the interactions between P_NTD and the L protein C-terminal domain (L_CTD), we demonstrated our ability to test sensitively for changes in the interface between the two proteins. This method showed that the binding site for P_NTD on L_CTD is longer than was previously appreciated. We demonstrated that phosphorylation of P_NTD modulates its interaction with L_CTD, and we used a minigenome reporter system to validate the functional significance of the P_NTD-L_CTD interaction. Using an electron microscopy approach, we showed that L bound to phosphorylated P_NTD displays increased conformational heterogeneity in solution. Taken as a whole, our studies suggest a model in which phosphorylation of P_NTD modulates its cofactor and conformational regulatory activities with L.

IMPORTANCE Polymerase-cofactor interactions such as those addressed in this study are absolute requirements for mononegavirus RNA synthesis. Although cofactor phosphorylation is present in most of these interactions, its effect, if any, on this protein-protein interaction had not been addressed. Our study is the first to address the effects of phosphorylation on P during its interactions with L in residue-by-residue detail. Since phosphorylation is the biologically relevant state of the cofactor, our demonstration of its effects on L conformation suggests that the structural picture of L during infection might be more complex than previously appreciated.

INTRODUCTION

Nonsegmented negative-sense (NNS) viruses represent numerous human pathogens, such as the Ebola and measles viruses. Vesicular stomatitis virus (VSV) is widely used as a prototype for the study of the molecular biology of NNS viruses. Studies in VSV were the first in the field to identify the unusual template of NNS viral RNA synthesis, which consists of genomic RNA, a nucleocapsid (N) protein (1, 2), and the unique polymerase, which includes both the conserved viral large (L) protein and the phosphoprotein (P) cofactor (1, 3, 4). Furthermore, VSV is relevant to human health, both as a potential oncolytic agent (5) and, more prominently, as a vaccine platform against the Ebola virus (6). Antigen presentation using VSV is being explored similarly as a strategy in vaccination against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (7–9).

The L proteins of Mononegavirales are highly conserved (10) and possess the enzymatic activities needed for mononegavirus RNA synthesis. In VSV, the L protein supplies RNA-dependent RNA polymerase (RDRP) (1, 3, 4), GDP polyribonucleotidyltransferase (PRNTase) capping (11), dispecific cap methyltransferase (MTase) (12–14), and polyadenylate synthetase (15) activities. The L protein has a modular layout consisting of a ring-shaped core (∼150 kDa) that contains the RDRP and PRNTase activities from which extend three accessory domains: the connector domain (P binding), methyltransferase domain (cap methylation), and C-terminal domain (CTD or L_CTD) (P binding) (16–18). The structures of core regions from human metapneumovirus (HMPV) (19) and respiratory syncytial virus (RSV) (20), as well as full-length structures of L from rabies virus (RABV) (21) and parainfluenza virus 5 (PIV5) (22), show that the structures of L proteins are highly similar across the members of Mononegavirales.

In contrast to the L proteins, P proteins from each family within the Mononegavirales have both structured and unstructured components. From N terminus to C terminus, the VSV P protein is composed of the following regions: a structured nucleocapsid (N) protein chaperone (23, 24), a disordered region that binds L and supplies polymerase cofactor activity (P N-terminal disordered [P_NTD] region) (18, 25, 26), a structured homodimerization interface (27–29), a short, flexible linker (26), and a C-terminal structured domain that binds the assembled N-RNA complex (30, 31) (Fig. 1A). P_NTD contains a subregion, referred to as the L stimulatory domain (P_LSD). This stretch of residues 81 to 106 appears to mediate initiation competence (25, 32) and essentially corresponds to the region of P_NTD bound to the RDRP and connector domains of L (17, 18). P proteins and functional equivalents thereof (such as VP35 in Ebola virus) are present throughout the order. Although domain organization and oligomerization state differ between viral genera, P proteins can be broadly described as flexible, charged, and multimeric cofactors of the L protein polymerases.

FIG 1.

Overview of VSV P and the polymerase-cofactor complex. (A) Structural layout of VSV P, labeled and color-coded by domain. A linear sequence cartoon is shown below. The phosphorylation targets in the P N-terminal disordered (P_NTD) cofactor region (residues S60, T62, and S64) are circled. (B) Structure of the VSV L-P_NTD complex, labeled and color-coded by domain, with P_NTD shown in blue as in panel A. The white dashed line shows the putative approximate position of VSV P residues 57 to 81 near the L C-terminal domain (L_CTD), which includes the phosphorylation targets. A linear sequence cartoon of L is shown below. The PDB accession codes for the structures visualized are as follows: 3PMK (P chaperone), 2FQM (P homodimerization), 3HHW (P template-binding), and 6U1X (L-P_NTD complex).

P has been shown to be phosphorylated in cellula, and a subset of phosphorylation sites have been localized to P_NTD (residues S60, T62, and S64) (33, 34). These modifications are mediated by casein kinase II (35). P protein derived from the virus particle is phosphorylated (36), although the sequence in which phosphorylation takes place on the P_NTD sites relative to the phosphorylation targets in the N-terminal (37) and C-terminal (38) regions of the protein is unclear. Phosphorylation has a protranscriptional effect (33, 35, 39), but what exactly the modification does has not been definitively established. In the existing structural studies of the VSV L-P_NTD complex (16–18, 25), only nonphosphorylated P was used. Furthermore, the phosphorylation targets are missing in the L-P_NTD cryoelectron microscopy (cryo-EM) structure (18). Adding to the complexity of this picture, it was found that P expressed in Sf21 insect cells stimulated initiation by L more effectively than did P expressed in Escherichia coli (32), an effect possibly explainable by phosphorylation. In an in vitro transcription system, Sf21 cell-expressed P with mutagenized phosphorylation sites was less active than the wild type (40), though, interestingly, more active than E. coli-expressed P in previous observations, potentially indicating that phosphorylation events both inside and outside the P_NTD region could contribute to cofactor activity. Phosphorylation of P proteins and of functional analogs thereof has been observed in other members/families of the Mononegavirales (41, 42), although its role in polymerase-cofactor interactions has, similarly, not been characterized. Furthermore, mutants of VSV P with modified phosphorylation sites are known to decrease viral RNA synthesis (39, 43) and abrogate the ability of P to prevent cellular RNA from binding the N protein (40). Taking these observations as a whole, the lack of information on how the phosphorylation of the cofactor P affects its interactions with the polymerase is a notable gap in the knowledge of the field. A graphical depiction of the L-P_NTD complex with emphasis on the potential position of the phosphorylated residues is provided in Fig. 1B.

In order to gain a more complete picture of the VSV L-P_NTD complex, we sought to determine a role for phosphorylation in the L-P_NTD interaction. Using a biophysical approach, we demonstrated that nuclear magnetic resonance (NMR) spectroscopy can be used to probe for phosphorylation-induced changes to L-P interactions with resolution at the amino acid level. We report that phosphorylation alters the L_CTD-P_NTD interaction and increases the conformational heterogeneity of the L-P_NTD complex. We supplemented these studies with a minigenome reporter system, which allowed us to quantify the effects of mutations in the P protein on viral RNA synthesis and revealed the importance of a conserved motif in the L_CTD-P_NTD interaction. Taken as a whole, our data suggest that P_NTD phosphorylation modulates the balance of L between its compact and noncompact conformational states, potentially representative of initiation and elongation, respectively.

RESULTS

The P_NTD-L_CTD interface includes the phosphorylated region of P_NTD.

Recent biophysical and cryo-EM studies have shown that the P_NTD-L interactions in VSV take place in the RDRP, connector, and C-terminal domains of the L protein (17, 18). However, the residues that are sites of phosphorylation in P_NTD (S60, T62, S64) (Fig. 1) were not resolved in those structural studies. Because of its residue-level resolution and extreme sensitivity, we employed NMR to study the interaction between P_NTD and L_CTD.

To determine the effect of phosphorylation on the interaction of P_NTD with L_CTD, we used an NMR technique known as chemical shift perturbation (CSP) analysis. This technique was used previously by our lab to detail the interaction between the L and P proteins from VSV with resolution at the level of single amino acids (17). The first step in this analysis is to obtain chemical shift assignments for the protein of interest. To this end, P_NTD was isotopically labeled with ¹⁵N and ¹³C to enable us to obtain the high-quality triple-resonance spectra required for these assignments. We successfully assigned peaks for 62 of 66 nonproline residues in the unphosphorylated and phosphorylated P_NTD spectra, using the same methods as in our previous study (17). Once the backbone resonances were assigned, we obtained two heteronuclear spin quantum coherence (HSQC) spectra for the CSP analysis. The first HSQC spectrum that we recorded consisted of the ¹⁵N-labeled P_NTD by itself, while the second HSQC spectrum was ¹⁵N-labeled P_NTD in the presence of a nonisotopically labeled construct consisting of the VSV L methyltransferase and C-terminal domains, termed MT/CT. Two equivalent experiments were then performed with phosphorylated P_NTD. To perform the CSP analysis, we quantified both changes to position and changes to the signal strength of amide chemical shifts upon the addition of the MT/CT construct. Using this method, we found significant changes to a stretch of amino acids in P (residues 49 to 68), specifically a signal depletion of 80% or more, upon MT/CT addition, of all residues in the stretch except for residues 64, 65, and 67 (residue 66 was a proline and therefore was not visible in this experiment) (Fig. 2). We noted that our assignment of residues 49 to 68 was longer than what had been observed in the L-P_NTD complex structure of residues 49 to 56. In that structure, however, residues 57 to 68 (in addition to others) were not resolved (18). Our findings concur with those of the previous study (18) and encompass the residues observed in that structure but suggest a larger binding region, observed potentially due to the more sensitive nature of the NMR experiments. Because the C terminus of the known binding site would direct any extension away from the methyltransferase, we assert that the interaction detected likely is not affected by the methyltransferase portion of the MT/CT construct.

FIG 2.

NMR of the MT/CT-P_NTD interaction. (A) HSQC spectra of P_NTD (residues 35 to 106) either in an apo state (black) or with L_MT/CT (residues 1596 to 2109) added (green). (B) The experimental setup used for panel A was repeated, but P_NTD was phosphorylated. Apo spectra (red) and L_MT/CT-added spectra (blue) are shown. In panels A and B, boxes highlight examples of residues that undergo signal depletion upon L_MT/CT addition. (C) Signal intensities of L_MT/CT-added spectra normalized to those of apo state spectra for unphosphorylated (black) and phosphorylated (green) P_NTD. The red horizontal line indicates the signal loss cutoff for determining an interaction in the unphosphorylated spectra (80% or more).

The presence of phosphorylation in P_NTD is known to be a positive regulator of viral RNA synthesis (33), although it has been demonstrated in an in vitro context that an overabundance of P protein expressed in Sf21 cells (which express casein kinase II) can be an RNA synthesis initiation suppressant (32). Having demonstrated that the region containing the phosphorylation targets S60, T62, and S64 interacts with L_CTD, we used a sample of P_NTD that had been triphosphorylated in vitro and examined using mass spectrometry (Fig. 3) to repeat the experiments described above (Fig. 2B and C). We noted that signal loss upon the addition of MT/CT was mitigated about 20 to 40% by phosphorylation in the region of interest. In P residues 49 to 68, only residue D61 showed a greater amount of signal loss in the phosphorylated-P_NTD experiment than in the nonphosphorylated-P_NTD experiment, and this difference was only about 1%. Taken as a whole, our NMR experiments show that residues 49 to 68 of P are involved with interactions with the L_CTD and that phosphorylation of P_NTD modulates this interaction. It should be noted that the parameters of the NMR experiments that were used to assess the effect of phosphorylation on the interaction between P_NTD and MT/CT were equivalent. This equivalence in experimental design allows a robust and direct comparison of the spectra for the phosphorylated and nonphosphorylated forms of P_NTD. Furthermore, the spectra of phosphorylated and unphosphorylated P_NTD are essentially superimposable with the exception of the phosphorylated residues; in other words, changes to binding upon phosphorylation likely did not come from a major structural change to P_NTD upon phosphorylation prior to L binding.

FIG 3.

Phosphorylation of P_NTD. An intact phosphorylated sample of P_NTD was analyzed using mass spectrometry. The red chevron indicates a peak cluster at a charge state of +5 and an m/z centroid of roughly 1,676 (corresponding to a mass of 8,375 Da, against a predicted molecular weight of 8,374 Da for the triphosphorylated species). The white chevron indicates a cluster with an m/z centroid of 1,660 (corresponding to a mass of 8,295 Da, against a predicted molecular weight of 8,294 Da for the diphosphorylated species). Black chevrons indicate where centroids for the monophosphorylated and unphosphorylated species would be expected (m/z 1,644 and 1,628, respectively). Higher-mass species, likely corresponding to adducts, are highlighted with a red bracket.

A conserved and functionally important region of P_NTD is responsible for the L_CTD–P_NTD interaction.

In the VSV L-P_NTD structure, residues 49 to 56 of P are resolved in contact with L_CTD; however, the continued length of P_NTD is presumably proximal to the CTD (18). This additional segment of P putatively bridges the gap between residues 56 and 82, terminal residues of resolved portions of P_NTD. Our NMR analysis indicated that the P_NTD-L_CTD interaction may extend as far as residue 68, so we decided to examine the conservation and potential functional consequences of residues 49 to 68.

We compared the sequences of phosphoproteins from other members of the genus Vesiculovirus and from members of the closely related Sprivivirus genus. We identified a conserved PSYΩ (where Ω is an aromatic residue) motif corresponding to VSV residues 51 to 54 followed by the relatively acid-rich stretch flanking the phosphorylation sites of S60, T62, and S64 (Fig. 4).

FIG 4.

Conservation of P_NTD among vesiculoviruses and spriviviruses. Phosphoprotein sequences from each of the viruses listed were aligned against that of VSV, and the portions aligned against VSV P residues 49 to 68 are shown here. A conserved motif and the three phosphorylation sites of VSV P are marked. Abbreviations: VSIV, vesicular stomatitis Indiana virus; VSNV, vesicular stomatitis New Jersey virus; VSAV, vesicular stomatitis Alagoas virus; CARV, Carajas virus; CHPV, Chandipura virus; COCV, Cocal virus; ISFV, Isfahan virus; MARAV, Maraba virus; PIRYV, Piry virus; JURV, Jurona virus; PERIV, Perinet virus; YUBOV, Yug Bogdanovac virus; PFRV, pike fry virus; SVCV, spring viremia of carp virus.

We next examined the functional consequences of alanine substitutions in the PSYΩ motif and other amino acids in the region implicated by our NMR studies. We used a minigenome reporter system, in which cells are transfected with plasmids encoding the VSV N, P, and L proteins, along with a plasmid encoding a VSV-based minireplicon containing a single reporter luciferase gene. This system had previously allowed us to examine the consequences of disrupting connector-P_NTD interactions (17). We quantified reporter activity normalized to a transfection control and noted a number of functionally important substitutions. Substitutions P51A, S52A, and Y53A showed near-total functional loss. Interestingly, F54A also showed reporter activity suppression of ∼40%. Reporter activity loss for the three phosphorylation sites increased toward the C terminus; in other words, S64A had more suppression than T62A, which, in turn, had more than S60A. The loss of two acidic residues, D61A and E65A, resulted in activity losses of roughly 25% and 50%, respectively. Because casein kinase II targets residues with acidic content toward the carboxy terminus, these two substitutions might have had an effect on phosphorylation. Surprisingly, P66A, an alanine substitution at a nonconserved position, showed near-total reporter activity loss. Two modest increases to reporter activity, resulting from the substitutions D58A and D59A, were also observed. We also replaced the three phosphorylation sites with either alanines (P3A) or phosphomimetic glutamic acids (P3E). As expected (40, 43), P3A demonstrated a loss of activity, while P3E maintained roughly 20% of activity (Fig. 5A). Western blotting was used to validate proper expression of P mutants (Fig. 5B). Taking our findings as a whole, we can confirm that both conserved and VSV-unique residues in the N terminus of P_NTD have functional consequences in terms of viral RNA synthesis.

FIG 5.

Minigenome reporter analysis of mutagenized VSV P. (A) A minigenome dual-luciferase reporter assay was carried out with the listed amino acid substitutions in P. Results for wild-type P, used as a positive control (+), and a negative control without P (–P) are provided as well. The horizontal axis break separates two batches of experiments. Error bars represent 95% confidence intervals for three replicates. (B) Each of the samples used in the minigenome reporter assay was examined for VSV P content by Western blotting, with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a control.

Phosphorylation of P_NTD alters its conformational regulatory effect on L.

Given our NMR studies, which showed that phosphorylation alters the P_NTD-L_CTD interaction, and the current model of full-length P_NTD as a one-copy steric restraint on the RDRP, connector, and CTD domains of L, we examined the effects of phosphorylated and unphosphorylated P_NTD on L protein conformation. We approached this by using negative-stain electron microscopy. Imaging was carried out for apo L, L with unphosphorylated P_NTD, and L with phosphorylated P_NTD. From each data set, ∼3,000 to 7,000 particles were picked and used to generate 2-dimensional (2D) class averages (Fig. 6). The 2D class averages were categorized as “compact” or “noncompact,” where compact classes showed a condensed, bean-shaped morphology while noncompact classes displayed a ring-shaped core with either poorly resolved or punctiform density in proximity. In previous studies (16, 25), the ring-shaped core was found to correspond to the RDRP and PRNTase core of the L protein, and the puncta to the accessory domains.

FIG 6.

Negative-stain electron microscopy of the L-P_NTD complex. (A) Twelve class averages each from samples of apo L, L with phosphorylated P_NTD, and L with unphosphorylated P_NTD. Each class average is bordered in red or blue, corresponding to its assignment to the compact or noncompact state, respectively. The number given for each class average corresponds to the fraction of particles in the final data set that it represents. (B) Examples of class averages from each sample are shown for the compact and noncompact states. (C) Table summarizing the percentages of particles in the compact and noncompact states in each sample.

We found that the unphosphorylated L-P_NTD complex overwhelmingly preferred the compact state (no noncompact classes were present). In contrast, the apo L sample showed only low levels of the compact state (∼7%), with most classes showing punctiform accessory domains or a general lack of definition in the accessory domains. The phosphorylated sample showed a general increase in heterogeneity and was found to contain a mixture of compact and noncompact states (roughly 82% and 18%, respectively). These results indicated that the sample with the phosphorylated cofactor had a greater amount of conformational heterogeneity than the unphosphorylated equivalent, specifically in the accessory domains of the L protein.

DISCUSSION

Many of the fundamental insights into negative-strand virus molecular biology have been generated from studying VSV. As polymerase studies specifically approach increasing levels of structural detail, it is necessary to characterize the role of known posttranslational modifications of the polymerase components L and P in order to obtain a better mechanistic understanding. In this study, we have demonstrated that the phosphorylation of the P cofactor has an effect on L-P interactions.

The modular layout of the L protein has proven advantageous in isolating different components of the protein for study. In both this report and a previous report detailing connector-P interactions (17), we have demonstrated that solution NMR allows for the sensitive study of L-P interactions on a residue-by-residue basis, without requiring the purification of full-length L. This study, in particular, also shows that NMR studies can detect interacting residues that might not be visible to high- or intermediate-resolution cryo-EM due to mobility, partial occupancy, or other issues. Although this study used fragments of L and P rather than full-length proteins, the regions of each protein implicated by our NMR results were consistent with available structural information (18). In the cryo-EM reconstruction of the L-P_NTD complex, the L_CTD-binding region of P_NTD positions the C-terminal residues of the cofactor away from the methyltransferase domain. Our study detects further interaction by P_NTD with the C-terminal residues, so we assert that our inclusion of the methyltransferase in our construct did not result in alterations to the interaction. It should, however, be noted that the use of fragments can yield information on local protein-protein interactions but may undersample the interactions of the more-intact protein-protein system.

Our NMR studies implicated P_NTD residues 49 to 68, which included a highly conserved PSYΩ motif, in interactions with L_CTD. Mutagenesis in these residues was examined in a minigenome reporter system. Substitution of P51, S52, or Y53 greatly suppressed reporter activity, while the F54A mutation, though still significant, had a more modest effect. The other highly suppressive mutant described in this study is the P66A mutant. Because prolines lack an amide hydrogen, they are not visible to HSQC NMR, and as such it cannot be said if the signal suppression indicative of an interaction that we observed for VSV P_49–68 could be observed for P66 specifically. The existing VSV L-P_NTD structure could not place this residue, presumably due to structural disorder, but it likely lies between the resolved CTD and RDRP binding sites for P_NTD. We hypothesize that residue P66 either enables an important structural element needed for the L-P_NTD interaction or directly interacts with L itself. It is important that minigenome reporter activity is a function of gene expression, which itself is a function of multiple activities. These include transcription, transcript maturation, N-RNA complex formation, and replication of the minigenome. A minigenome reporter does not identify the step of gene expression being suppressed, but because our NMR experiments indicated that our region of interest bound L_CTD, we hypothesize that our mutants impaired either the formation or the conformational regulation of the L-P complex. On the latter note, previous work (16, 18) has suggested that the unphosphorylated L-P_NTD conformation is associated with initiation and has presumed that movement of the MTase domain would be required for transcript maturation.

Although residues in and around the phosphorylation site had been suggested to form a metal-binding pocket, since EDTA was necessary to visualize this signal using HSQC (24), this was not a result that our studies recapitulated. Amide peaks in both phosphorylated and unphosphorylated samples remained clustered along the ¹H axis, indicating that both were in a disordered state, meaning that other structural rearrangements were also unlikely upon phosphorylation.

Using class averages generated from negative-stain EM, we found that the conformation of the L-P_NTD complex in solution appears to change upon phosphorylation of P_NTD. Specifically, a loss of definition for the accessory domains or the separation of the accessory domains from the core in the phosphorylated sample indicated that additional heterogeneity was present relative to that in the sample prepared with unphosphorylated P_NTD. This resulted in conformations more similar to the noncompact forms predominant in the apo L sample. Given that our NMR observations were consistent with a loss of interaction strength upon phosphorylation, and given that conformational compaction is a well-characterized effect of the unphosphorylated and more strongly binding P_NTD, it seems intuitive to suggest that the phosphorylated cofactor dissociates from the L_CTD binding site more frequently. A visual model for this change to L-P_NTD interactions is presented in Fig. 7. It should be noted that a full dissociation of P_NTD could also explain the emergence of the noncompact apo-like states observed in our EM studies. Regardless, interactions between the L and P components can still occur in the phosphorylated state.

FIG 7.

Model for L-P_NTD conformation and phosphorylation. In the apo state, the accessory domains of L are not coordinated with the core, resulting in high conformational heterogeneity. The P_NTD binding events in the L_CTD, RDRP domain, and connector domain restrain the movement of the accessories relative to the core. When phosphorylation interferes with the P_NTD-L_CTD interaction, part of the L-P_NTD complex samples a state in which the movement of the accessory domains is not comparably restrained, resulting in greater conformational heterogeneity.

The results of this study are particularly interesting when contextualized with the emerging structural picture of polymerase initiation in the Mononegavirales. In VSV and RABV, a conserved priming-capping loop is thought to be required in a forward conformation for the initiation of RNA synthesis, but this conformation is presumptively prohibitive of RNA product ejection (18, 21, 32). In RSV and HMPV, however, this loop can be observed in a retracted state, while the accessory domains remain highly flexible (19, 20). Furthermore, studies in PIV5 (22) show a compacted state similar to that of VSV or RABV, but one in which the position of the methyltransferase and C-terminal domains has changed. In this structure the priming-capping loop is retracted, but a different “intrusion” loop is forward. While the existence of comparable states in the VSV L-P system has not yet been demonstrated, it is tempting to speculate that the accessory domains of VSV L sample multiple functionally relevant conformations. The increased conformational heterogeneity observed with cofactor phosphorylation may play a role in regulating the switch to these conformations. In the future, additional structures of the polymerase, polymerase-template, and polymerase-product complexes will shed light on this matter.

MATERIALS AND METHODS

Molecular cloning.

The sequences of VSV L_1596–2109 and VSV P_35–106 were amplified from whole-gene plasmids using primers with the sequences 5′-ACGGTCTCAAGGTATGAGCTATCCCCCTTGGG and 5′-CCGCTCGAGTTAATCTCTCCAAGAGTTTTCCTCATGTAG (VSV L) and 5′-ACGGTCTCAAGGTTCCAATTATGAGTTGTTCCAAGAG and 5′-CACCGCTCGAGTTACGAAGTGAATACAACATCCACGTC (VSV P). Amplification products were cloned into a SUMOpro Gene Fusion Technology Kanamycin (SpGFTk) vector (LifeSensors, Malvern, PA) using the BsaI and XhoI restriction enzymes and T4 DNA ligase. These plasmids were labeled Sp-L1596-2109 and Sp-P35-106, respectively.

Protein expression and purification.

Plasmids Sp-L1596-2109 and Sp-P35-106 were transformed into Escherichia coli SoluBL21 cells (Genlantis, San Diego, CA) using heat shock. Cells were grown under kanamycin selection. Overnight starter cultures were prepared using a single transformant colony in either LB broth for conventional expression (Fisher, Fair Lawn, NJ) or M9 minimal medium for isotopic labeling using [¹⁵N]ammonium chloride and [¹³C]glucose (Cambridge Isotope Laboratories, Andover, MA). Starter cultures were grown overnight at 37°C, pelleted, resuspended in fresh medium with antibiotic, and diluted 1:100 into expression cultures. LB and M9 cultures were grown to optical density at 600 nm (OD₆₀₀) values of 0.4 to 0.6 and 0.7, respectively. Cultures were then cooled to 18°C, and protein expression was induced by adding isopropyl-β-d-1-thiogalactopyranoside (IPTG) to a final concentration of 0.5 mM. Induced cultures were incubated overnight at 18°C.

Induced cultures were pelleted at 6,000 × g for 15 min at room temperature. Pellets were resuspended in binding buffer (500 mM sodium chloride, 10% [vol/vol] glycerol, 5 mM imidazole, 20 mM Tris [pH 7.9]) at a ratio of 30 ml binding buffer for each 1 liter of culture pelleted. Initial purification was performed using nickel affinity chromatography. Cell pellets were sonicated and pelleted at 27,000 × g for 30 min at 4°C. Supernatants were passed over 5 ml Chelating Sepharose Fast Flow beads (GE Healthcare, Pittsburgh, PA) per 2 liters of expression culture. Beads were washed using 10 column volumes (CV) binding buffer and 5 CV wash buffer (binding buffer with 60 mM imidazole). P_35–106 was eluted using eluting buffer (binding buffer with 1 M imidazole). L_1596–2109 was eluted using stripping buffer (500 mM sodium chloride, 10% [vol/vol] glycerol, 20 mM Tris, 100 mM EDTA [pH 7.9]).

L_1596–2109 was exchanged to HA buffer [500 mM sodium chloride, 10% (vol/vol) glycerol, 20 mM HEPES, 1 mM Tris(2-carboxyethyl)phosphine (TCEP), 0.02% (wt/vol) sodium azide (pH 7.0)] and loaded onto a 5-ml HiTrap heparin column (GE Healthcare, Pittsburgh, PA). The protein was eluted with a gradient of HA buffer and HB buffer (equivalent to HA but with 1 M sodium chloride). The heparin eluate was treated with 2 μg SUMO protease to 1 mg of the protein of interest and was incubated overnight at 4°C. Completion of digestion was validated by SDS-PAGE. The digested protein was purified on a HiLoad 16/600 Superdex 200 pg column (GE Healthcare, Pittsburgh, PA) using HB buffer.

P_35–106 was treated with SUMO protease as described above. Following a second round of nickel affinity chromatography to remove the SUMO tag, P was exchanged to RQA buffer (150 mM sodium chloride, 20 mM Tris, 0.02% [wt/vol] sodium azide [pH 7.5]) using a HiPrep 26/10 desalting column (GE Healthcare, Pittsburgh, PA) and was purified using anion exchange on a 1-ml Resource Q column (GE Healthcare, Pittsburgh, PA) with a gradient of RQA buffer and RQB buffer (equivalent to RQA buffer but with 1 M sodium chloride). The ion-exchange eluate was purified on a HiLoad 16/600 Superdex 75 pg column (GE Healthcare, Pittsburgh, PA) using HB buffer.

Full-length L was expressed by infecting Sf21 cells with a baculovirus carrying a polyhistidine-tagged L construct (44) and was purified using nickel affinity chromatography as described previously (45).

In vitro phosphorylation of VSV P.

Purified P_35-106 was exchanged to phosphorylation buffer (150 mM sodium chloride, 10 mM magnesium chloride, 1 mM β-mercaptoethanol, 25 mM Tris [pH 7.5]) using a HiPrep 26/10 desalting column. To the sample was added ATP to a final concentration of 1 mM and casein kinase subunit IIα to a final concentration of 5 ng/μl enzyme to 30 μM P_35–106. Phosphorylation was validated using mass spectrometry (Fig. 3), and the phosphorylated protein was repurified using size exclusion chromatography as described above.

Nuclear magnetic resonance spectroscopy.

NMR experiments were carried out at the Central Alabama High-Field NMR facility on a Bruker Avance III spectrometer with a TCI CryoProbe. Experiments were carried out at 10°C on a 600-MHz instrument. Data were analyzed using NMRPipe (46) and NMR View (47). Before analysis, proteins were exchanged into NMR buffer (200 mM sodium chloride, 10% [vol/vol] glycerol, 20 mM HEPES, 1 mM TCEP [pH 6.8]) with a HiPrep 26/10 desalting column. Resonance assignments were carried out using standard experiments (48).

Amino acid sequence alignments.

Sequence alignments were performed using ClustalX2 (49), and phosphoprotein amino acid sequences of the following viruses were retrieved from the UniProt database: vesicular stomatitis Indiana virus (P04880), vesicular stomatitis New Jersey virus (I7CGJ1), vesicular stomatitis Alagoas virus (B3FRL2), Carajas virus (A0A0D3R1C6), Chandipura virus (E3T3B5), Cocal virus (B3FRK7), Isfahan virus (Q5K2K6), Maraba virus (F8SPF1), Piry virus (Q01769), Jurona virus (I1SV83), Perinet virus (I1SV88), Yug Bogdanovac virus (K4FFQ0), pike fry virus (C3VM12), and spring viremia of carp virus (A8VJA2).

Minigenome reporter system.

A minigenome assay was carried out as described previously (17). The reporter assay was performed using three technical replicates. The expression levels of mutagenized VSV P were examined by Western blotting as described previously (11). Mutagenesis on the VSV P plasmid was carried out using a Q5 site-directed mutagenesis kit (New England BioLabs, Ipswich, MA).

Electron microscopy.

VSV L was loaded onto 3-nm-thick continuous carbon grids over lacey carbon grids (Electron Microscopy Sciences, Hatfield, PA) at a concentration of 0.15 μM, with and without 1.5 μM phosphorylated or unphosphorylated VSV P_35–106. Dilutions to final concentrations were carried out in phosphorylation buffer. Grids were washed with water prior to staining with 0.75% (wt/vol) uranyl formate. Grids were imaged on an FEI Tecnai F20 transmission electron microscope using a Gatan K3 direct electron detector and an operating voltage of 200 keV. The images were collected at a nominal magnification of ×40,200, corresponding to 1.24 Å/pixel. Images were prepared in EMAN, version 2.2 (50), using e2proc2d.py, and data processing was carried out using RELION, version 3.0.8 (51). Final 2D class averages were made using 3,708 particles and 13 classes for apo L, 7,392 particles and 21 class averages for the unphosphorylated sample, and 3,312 particles and 12 class averages for the phosphorylated sample.