A Novel Form of Rotavirus NSP2 and Phosphorylation-Dependent NSP2-NSP5 Interactions Are Associated with Viroplasm Assembly

Abstract

Rotavirus (RV) replication occurs in cytoplasmic inclusions called viroplasms whose formation requires the interactions of RV proteins NSP2 and NSP5; however, the specific role(s) of NSP2 in viroplasm assembly remains largely unknown. To study viroplasm formation in the context of infection, we characterized two new monoclonal antibodies (MAbs) specific for NSP2. These MAbs show high-affinity binding to NSP2 and differentially recognize distinct pools of NSP2 in RV-infected cells; a previously unrecognized cytoplasmically dispersed NSP2 (dNSP2) is detected by an N-terminal binding MAb, and previously known viroplasmic NSP2 (vNSP2) is detected by a C-terminal binding MAb. Kinetic experiments in RV-infected cells demonstrate that dNSP2 is associated with NSP5 in nascent viroplasms that lack vNSP2. As viroplasms mature, dNSP2 remains in viroplasms, and the amount of diffuse cytoplasmic dNSP2 increases. vNSP2 is detected in increasing amounts later in infection in the maturing viroplasm, suggesting a conversion of dNSP2 into vNSP2. Immunoprecipitation experiments and reciprocal Western blot analysis confirm that there are two different forms of NSP2 that assemble in complexes with NSP5, VP1, VP2, and tubulin. dNSP2 associates with hypophosphorylated NSP5 and acetylated tubulin, which is correlated with stabilized microtubules, while vNSP2 associates with hyperphosphorylated NSP5. Mass spectroscopy analysis of NSP2 complexes immunoprecipitated from RV-infected cell lysates show both forms of NSP2 are phosphorylated, with a greater proportion of vNSP2 being phosphorylated compared to dNSP2. Together, these data suggest that dNSP2 interacts with viral proteins, including hypophosphorylated NSP5, to initiate viroplasm formation, while viroplasm maturation includes phosphorylation of NSP5 and vNSP2.

INTRODUCTION

Globally, rotaviruses (RV) remain the leading cause of severe dehydrating diarrhea in infants and children under 5 years of age and still account for 450,000 deaths annually (1). The rotavirus virion is a nonenveloped particle composed of three concentric, icosahedral protein shells. The innermost shell contains the genome of 11 segments of double-stranded RNA (dsRNA) that encodes 6 structural proteins (VP1, VP2, VP3, VP4, VP6, and VP7) and 6 nonstructural proteins (NSP1, NSP2, NSP3, NSP4, NSP5, and NSP6). During the process of cell entry, the outermost capsid layer is removed, activating transcription from the genome within the double-layered particle (DLP). After translation of the positive-sense viral transcripts, at least 7 viral proteins (NSP2/5/6 and VP1/2/3/6) are found in discrete cytoplasmic inclusions called viroplasms. Viroplasms are the sites of virus genome replication and nascent DLP assembly.

NSP2 plays a key role in viroplasm formation. In RV-infected cells, silencing the expression of NSP2 or NSP5 using RNA interference (RNAi) technologies or intrabodies prevents viroplasm formation (2–4). A rotavirus temperature-sensitive (ts) mutant, SA11 tsE(1400) (5), contains a ts lesion in gene segment 8 (A152V) (6) that encodes NSP2 and cannot form viroplasms at the nonpermissive temperature (5). In cultured cells, NSP2 coexpressed with NSP5 forms viroplasm-like structures (VLS) in the absence of the other viral proteins (7), but neither expression of NSP2 nor NSP5 alone is sufficient to form VLS. Thus, both NSP2 and NSP5 are considered the minimum components for viroplasm formation. However, beyond these observations, the mechanism for viroplasm assembly and the specific role of NSP2 in viroplasm formation remain largely unknown.

NSP2 (∼35 kDa) is a multifunctional enzyme that performs critical functions during genome replication, such as single-stranded RNA (ssRNA) binding and ATP-independent helix unwinding, and exhibits nucleoside triphosphatase (NTPase) activity (8–10) and nucleoside diphosphate (NDP) kinase activity (11). Replication intermediates with replicase activity isolated from RV-infected cells contain NSP2 (8, 12, 13), and silencing NSP2 using the SA11 tsE(1400) mutant results in a nearly complete loss of dsRNA synthesis at the nonpermissive temperature (14). The NSP2 octamer, formed by tail-to-tail interactions of two NSP2 tetramers, is predicted to be the functional form of the protein (6, 15), and the C-terminal helical tail (CTH) has been implicated in viroplasm formation (16). In addition to its enzymatic functions, NSP2 interacts with several ligands, including NSP5, VP1, ssRNA, and tubulin (8, 17–19). The pleiotropic functionality of NSP2 may be due to Mg²⁺ and nucleotide binding-induced conformational changes in the NSP2 octamer (20).

NSP5 is an O-linked glycoprotein (21) with observed molecular masses of 26, 28, and 32 to 35 kDa that represent its many phosphorylated isoforms (22–25). NSP5 interacts with NSP2, VP1, and VP2 and is found in complexes with NSP2, NSP6, VP1, VP2, VP3, and VP6 (26–28). Knockdown of NSP5 expression using small interfering RNA (siRNA) results in reduction of (i) viral proteins, (ii) viroplasm formation, (iii) the synthesis of viral mRNAs, (iv) genomic dsRNA, and (v) the yield of virus indicating expression of NSP5 is critical during rotavirus infection (3, 29). These studies reveal that NSP5 is involved in many processes related to the dynamics and regulation of viroplasms. The majority of studies evaluating viroplasm formation have assessed the role of NSP5 but not NSP2.

To study the mechanism of viroplasm formation in RV-infected cells, we generated monoclonal antibodies (MAbs) specific for full-length NSP2 and used these antibodies to monitor the localization of NSP2 by confocal immunomicroscopy. Additionally, we used these MAbs to examine NSP2 protein interactions by immunoprecipitation. Here we report the discovery of a novel form of rotavirus NSP2 that differentially interacts in a phosphorylation-dependent manner with NSP5. This form, cytoplasmically dispersed NSP2 (dNSP2), is first detected in small puncta throughout the cytoplasm of infected cells, colocalizes with NSP5 in nascent viroplasms, and accumulates rapidly in the RV-infected cell at early times postinfection. The second previously known form, viroplasmic NSP2 (vNSP2) is detected only in viroplasms, and the amount steadily increases as viroplasms mature and increase in size. Our studies suggest a model of phosphorylation-dependent NSP2-NSP5 interaction and a possible NSP2 conformational change that may be required for viroplasm formation.

MATERIALS AND METHODS

Virus and cells.

Rotavirus strain SA11-4F was used to infect African green monkey kidney epithelial cells (MA104). MA104 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (Life Technologies) supplemented with 10% fetal bovine serum (FBS) (Sigma) at 37°C and 5% CO₂. For all infections, rotavirus SA11-4F was activated for 30 min in 10 μg/ml trypsin (Worthington) in DMEM without FBS, and MA104 cells were inoculated with trypsin-activated virus at a multiplicity of infection (MOI) of 10 for 1 h at 37°C and 5% CO₂. The inoculum was removed and replaced with DMEM without FBS and incubated for an additional 2 to 7 h (3 to 8 h postinfection [hpi]) where indicated.

Generation and characterization of monoclonal antibodies.

Monoclonal antibodies (MAbs) were made by immunizing rotavirus antibody-negative BALB/c mice with 5 μg of full-length NSP2 protein in Freund's complete adjuvant. NSP2 was expressed in bacteria and purified as previously described (16). Mice were inoculated 4 weeks later with 2 μg of purified NSP2 protein in Freund's incomplete adjuvant. The mice were immunized three additional times over the course of 5 months. Each mouse was screened for production of high-titer antibodies to NSP2 by enzyme-linked immunosorbent assay (ELISA) (service provided by the Baculovirus/Monoclonal Antibody Advanced Technology Core Laboratory, Baylor College of Medicine, Houston, TX, USA) and Western blotting. The mouse producing the highest titer and most specific antibodies to NSP2 was sacrificed for B-cell hybridoma fusion (service provided by the Baculovirus/Monoclonal Antibody Advanced Technology Core Laboratory, Baylor College of Medicine, Houston, TX, USA). Primary hybridoma supernatants were screened for specific NSP2 antibody, and the titers of the virus were determined by ELISA. Hybridomas producing high-titer antibodies were subcloned, and each subsequent MAb was screened for the ability to detect NSP2 in RV-infected MA104 cells using confocal immunomicroscopy, and specificity for NSP2 was confirmed by reactivity using mock-infected or RV-infected cell lysates, as well as by reactivity with purified NSP2 by Western blotting and ELISA.

Monoclonal antibodies.

Two MAbs specific for NSP2 were selected for use in our studies based on the different staining patterns observed in RV-infected cells. MAb clone 1002 (12 μg/ml MAb dNSP2_CT) is referred to as “vNSP2_CT MAb,” signifying that the antibody detects viroplasmic NSP2 and binds at the C terminus (see below). MAb clone 947 (15 μg/ml MAb dNSP2_NT) is referred to as “dNSP2_NT MAb,” indicating that the antibody detects cytoplasmically dispersed NSP2 and binds at the N terminus (see below).

Additional antibodies used for confocal microscopy and immunoprecipitation experiments.

Polyclonal guinea pig anti-NSP2 (1203-2) and anti-NSP5 (1203-3) hyperimmune antisera were made using NSP2 and NSP5 purified from Escherichia coli as previously described (30) and generated by Cocalico Biologicals, Inc. Guinea pig anti-VP1 (GP539) and guinea pig anti-VP2 sera (GPE3) were made by inoculating animals with baculovirus-expressed and purified VP1 or VP2 protein using a strategy previously described (31). Rabbit anti-NSP4 (2478) has been previously described (32).

Plasmids.

Plasmid pNSP2-EGFP (EGFP stands for enhanced green fluorescent protein) was made and generously supplied by O. R. Burrone (International Centre for Genetic Engineering and Biotechnology [ICGEB], Trieste, Italy) (33). The complete NSP5 gene was PCR amplified from a pBR322 plasmid containing a full-length cDNA clone of SA11 gene 11 (22) and ligated into a TOPO vector (Invitrogen). The NSP5 gene was PCR amplified again using primers designed to insert an XhoI site upstream of NSP5 and an MluI site downstream of NSP5. The PCR product was digested with XhoI and MluI, and the fragment was gel purified and ligated into the vector pIRES (IRES stands for internal ribosomal entry site) (Clontech).

Immunofluorescence and confocal microscopy.

MA104 cells were grown to confluence on glass coverslips in 24-well plastic culture plates (Costar). The cells were either (i) infected at an MOI of 10 (see above) or (ii) transfected per the manufacturer's instructions with 2 to 3 μg total plasmid DNA (Lipofectamine 2000; Life Technologies). Cells in transfection reagent were incubated for 4 h at 37°C and 5% CO₂, after which the transfection mixture was removed, 0.5 ml DMEM with 10% FBS was added, and the cells were incubated for a total of 36 to 48 h. Both infected and transfected cells were fixed for 30 min at room temperature in 4% paraformaldehyde in phosphate-buffered saline (PBS), permeabilized in 0.5% Triton X-100 (or 0.1% saponin in 5% bovine serum albumin [BSA]), and blocked in 5% BSA before staining with antibodies. For the kinetics experiment, all cells were infected at the same time and fixed on the hour as indicated in Fig. 5. At the end of the assay, all fixed samples were processed simultaneously using the same reagents and primary and secondary antibody pools.

FIG 1.

Two pools of NSP2 are detected in rotavirus-infected MA104 cells at 8 hpi. (A) MAb 1002 (vNSP2_CT) detects punctate NSP2 in viroplasms, designated vNSP2 (secondary conjugate antibody goat anti-mouse labeled with Alexa Fluor 488 [green]). NSP5 is detected with primary polyclonal antibody guinea pig anti-NSP5 (secondary conjugate antibody goat anti-guinea pig labeled with Alexa Fluor 568 [red]). (B) A different MAb, MAb 947 (dNSP2_NT), detects a pool of cytoplasmically dispersed NSP2, designated dNSP2 (secondary conjugate antibody goat anti-mouse labeled with Alexa Fluor 488 [green]). (C) Magnification of boxed area in panel B with broken white arrow showing colocalization analysis of fluorescence intensities that are plotted in the graph (dNSP2 [green line], NSP5 [red line], and DAPI [light blue line]). (D) Simultaneous detection of dNSP2 and vNSP2 in RV-infected cells at 8 hpi. The analyzed mock- and RV-infected coverslips were replicates of cells examined in panels A and B. MAb 947 (dNSP2_NT) was detected using a secondary conjugate antibody as described above. MAb 1002 (vNSP2_CT) was directly labeled with Alexa Fluor 568 prior to incubation with the sample. The solid white arrow shows the area analyzed for fluorescence intensity that is graphed (dNSP2 [green line], vNSP2 [red line], NSP5 [blue line] [NSP5 was imaged but is not shown in the panel to the left], and DAPI [light blue line]).

(i) Primary antibodies.

Coverslips were incubated with 50% NSP2 MAb supernatant in 2.5% BSA. Additional rotavirus antibodies were added at concentrations between 1:500 and 1:1,000. Coverslips were incubated overnight at 4°C. For some experiments, MAb vNSP2_CT was directly labeled with isotype-specific mouse IgG1 Fab fragments conjugated to Alexa Fluor 568 (Zenon Alexa Fluor 568 mouse IgG1 labeling kit; Life Technologies) per the manufacturer's instructions.

(ii) Secondary antibodies.

Primary antibodies were removed, and the coverslips were washed 3 times in PBS. Coverslips were blocked again for 30 min in 5% BSA prior to adding goat anti-mouse and/or goat anti-guinea pig labeled with Alexa Fluor secondary antibodies (Life Technologies) and incubated overnight at 4°C. The following day the cells were washed twice in PBS before 4′,6′-diamidino-2-phenylindole (DAPI) (Life Technologies), 1:100 in PBS, was added for 5 min. Cells were washed a final three times in PBS before mounting.

(iii) Mounting and imaging.

Coverslips were mounted onto microscope slides using ProLong gold antifade mounting reagent (Life Technologies) and dried overnight at room temperature. All immunofluorescence images were acquired using a Nikon A1Rs confocal laser scanning microscope. Images were obtained using sequential image acquisition (to eliminate potential channel bleed of the fluorescent signals), the laser power was set from 5 to 10%, the pinhole was set at 1.0 for the 488-nm laser, and the gain for any given image ranged from 85 to 130 (maximum possible gain, 255). Mock-infected cells were imaged using the same settings as their RV-infected counterparts. Intensity profiles of selected images were analyzed, and the graphs were created with the Nikon NIS-Elements advanced research (AR) microscope imaging software.

Monoclonal antibody epitope mapping. (i) Deletion mapping.

Full-length NSP2 cDNA was amplified by PCR from pNSP2-EGFP (described above) and used as a template for generating the truncation mutants. A series of deletion mutants were constructed by systematically deleting 25 or 50 amino acids (aa) from the N or C terminus of NSP2. The deletion constructs were cloned into the NcoI and XhoI restriction enzyme sites of the pET46 vector (Novagen) for expression in bacteria (services provided by EPOCH Life Science, Missouri City, TX). Each plasmid was expressed in bacteria, and the resulting protein was electrophoresed on SDS-polyacrylamide 4 to 20% gradient gels. Proteins were detected using guinea pig polyclonal antibody (PAb) serum to NSP2 and mouse MAbs dNSP2_NT or vNSP2_CT. Proteins were visualized using goat anti-guinea pig or goat anti-mouse secondary antibodies conjugated to alkaline phosphatase. Blots were developed using a 5-bromo-4-chloro-3-indolylphosphate (BCIP)/Nitro Blue Tetrazolium (NBT) color development substrate method.

(ii) Peptide array.

NSP2 12-mer peptides, with 4-amino-acid overlaps, were synthesized as previously described (34). Replicate blots were incubated with either one of the NSP2 MAbs and washed extensively in PBS with 0.05% Tween 20, and the MAb was detected using goat anti-mouse secondary antibody conjugated to alkaline phosphatase. Blots were developed using a BCIP/NBT color development substrate method.

Western blots and immunoprecipitations.

RV-infected or mock-infected MA104 cell monolayers in T150 cm² or T175 cm² flasks were harvested 8 h postinfection by freezing the flasks at −20°C. The flasks were thawed on ice, and the cells were solubilized in radioimmunoprecipitation assay (RIPA) buffer (Thermo Scientific Pierce) containing protease inhibitors (complete, mini, EDTA-free protease inhibitor cocktail tablets) with or without phosphatase inhibitors (phosphatase inhibitor cocktail 2 [aqueous solution] [Sigma-Aldrich]; phosphatase inhibitor cocktail 3 [dimethyl sulfoxide {DMSO} solution] [Sigma-Aldrich]) and with or without phosphatase (calf intestinal alkaline phosphatase; Life Technologies), added per the manufacturer's instructions. Cells were scraped from the flasks, and the cell debris was pelleted by centrifugation at 10,000 rpm (13,800 × g) in a Beckman Coulter J2-HS centrifuge. The supernatants were aliquoted into cryotubes and frozen at −20°C until used.

(i) Western blots.

RV-infected and mock-infected cell lysates were transferred to microcentrifuge tubes containing sample buffer with or without β-mercaptoethanol (βME). Samples were heated for 3 min at 100°C and loaded onto 4 to 20% polyacrylamide gradient gels (Bio-Rad) for electrophoretic separation in Tris-glycine-SDS buffer (Bio-Rad). All samples within a given experiment and probed with the same primary antibody were run on the same polyacrylamide gel; however, empty lanes containing 1× sample buffer were placed between each sample to prevent lane spillover. Following electrophoresis, the proteins were transferred to nitrocellulose membranes using the iBlot 7-min blotting system (Life Technologies). Blots were incubated in primary antibodies overnight at 4°C, and secondary antibodies (goat anti-mouse IgG-alkaline phosphatase [Sigma-Aldrich], goat anti-guinea pig IgG-alkaline phosphatase [Sigma-Aldrich], goat anti-rabbit IgG-alkaline phosphatase [Sigma-Aldrich], and Clean-Blot IP reagent alkaline phosphatase [Thermo Scientific Pierce]) for 2 h. Blots were developed using a BCIP/NBT color development substrate method.

(ii) Immunoprecipitations.

Aliquots (0.5 to 1 ml) of RV-infected and mock-infected cell lysates were transferred to microcentrifuge tubes and precleared by incubating the lysates in 50 to 100 μl of protein G agarose bead resin (Roche Molecular Systems) for 1 h while rotating at 4°C. Beads were pelleted by centrifugation (12,000 × g for 30 s). Supernatants were transferred to new tubes, and 2 to 10 μg of precipitating antibody was added. For larger experiments, the amounts of reagents were increased proportionately. Lysate-antibody mixtures were incubated for 4 h, or overnight, at 4°C with rotation. Protein G agarose bead resin (100 to 150 μl) was added to the antigen-antibody complexes and incubated overnight at 4°C with rotation. Antigen-antibody-protein G complexes were pelleted the following day by centrifugation (12,000 × g for 30 s). The supernatant was discarded, and the complexes were washed in a no-salt wash buffer (0.01 M Tris-HCl [pH 7.2], 0.1% [vol/vol] NP-40). Samples were transferred to microcentrifuge tubes and prepared for Western blot analysis as described above.

Purification and biotinylation of octameric NSP2 for binding kinetics study.

NSP2 of rotavirus strain SA11 was expressed and purified as previously described (16). The purified octameric NSP2 was concentrated to 5.7 mg/ml in PBS and biotinylated with a 5:1 molar ratio of EZ-link NHS-LC-LC-biotin (Thermo Scientific). The NSP2/biotin mixture was incubated at room temperature for 30 min. After the biotinylation reaction, the excess nonreacted biotin was removed using a Centricon centrifugal filter device (Millipore).

Binding kinetics analysis using Octet.

The apparent K_d (apparent dissociation constant) of each NSP2 monoclonal antibody was determined by biolayer interferometry using an Octet RED96 system (ForteBio, Inc.) (35). This system is conceptually similar to surface plasmon resonance (SPR). Binding experiments were carried out using HBS-P buffer (10 mM HEPES [pH 7.4], 150 mM NaCl, 5 mM MgCl₂, 0.05% surfactant P20). The biotinylated octameric NSP2 was immobilized on streptavidin (SA)-coated biosensors and incubated with 4 concentrations of MAbs (20 to 160 nM and 3.7 to 29.4 nM for dNSP2_NT MAb and vNSP2_CT MAb, respectively) at 30°C. Data acquisition included 5 steps: (i) baseline (120 s); (ii) loading of NSP2 (180 s); (iii) second baseline (120 s); (vi) association of MAb dNSP2_NT or MAb vNSP2_CT for the measurement of association rate (k_on) (300 s); and (v) dissociation of MAb dNSP2_NT or MAb vNSP2_CT for the measurement of dissociation rate (k_off) (300 s). Binding curves were fit to a 1:1 binding model by global fitting to calculate the K_d.

Mass spectrometry (MS).

A large volume (8 ml) of RV-infected cell lysate was prepared in RIPA buffer containing protease and phosphatase inhibitors, divided equally, and incubated with MAb dNSP2_NT or MAb vNSP2_CT covalently bound to protein G agarose beads (Thermo Pierce Scientific) overnight at 4°C. Immunoprecipitations and elutions were performed per the manufacturer's instructions. Proteins were eluted into four 250-μl fractions, and 50 μl of elution 2 from each immunoprecipitation (plus uninfected control) was sent for mass spectrometry analysis (mass spectrometry services provided by Pathway Discovery Mass Spectrometry/Identification of Protein Complexes, Advanced Technology Core Laboratory, Baylor College of Medicine, Houston, TX, USA).

RESULTS

Two pools of NSP2 are detected in RV-infected cells.

We characterized a number of MAbs generated against full-length NSP2 in order to visualize NSP2 and NSP5 in viroplasms in RV-infected cells using confocal microscopy. We sought an NSP2 MAb that clearly detected viroplasms based on punctate staining in the cytoplasm and colocalization with antibody against NSP5. To select the MAbs for our studies, RV-infected MA104 cells were fixed and permeabilized at 8 h postinfection. NSP2 was detected using each of 56 different hybridoma supernatants chosen for high ELISA optical density (OD) values against bacterially expressed and purified NSP2 (optical density at 450 nm [OD₄₅₀] of 1.9 to 4.0). The majority (39/56) of our antibody panel detected viroplasms as expected, while a smaller subset (17/56) detected what appeared to be diffuse NSP2 distributed throughout the cytoplasm of the infected cells. To confirm this, two prototype MAbs representing each phenotype were characterized further. RV-infected cells were probed for NSP2 using either one of the selected MAbs against NSP2, as described above, and for NSP5 using a polyclonal antibody (PAb) serum and imaged by confocal microscopy. Images representative of the infected cells at 8 h postinfection are shown in Fig. 1. MAb clone 1002, with an ELISA reading of OD₄₅₀ of 2.6, detected punctate viroplasmic NSP2 (vNSP2) that colocalized with NSP5 (Fig. 1A), and this antibody was named MAb vNSP2_CT. MAb clone 947, with an ELISA reading of OD₄₅₀ of 3.5, detected diffusely distributed dispersed NSP2 (dNSP2) that did not predominantly appear to colocalize with NSP5 (Fig. 1B), and this antibody was named MAb dNSP2_NT.

Magnification of the images of dNSP2 with NSP5-containing viroplasms suggested that dNSP2 was closely associated with NSP5 but only marginally colocalized (Fig. 1C, magnification of orange box in Fig. 1B). However, further image analysis examined the fluorescent intensity of dNSP2 using a defined path through the viroplasms in the image (white arrow), which is presented as a fluorescence intensity profile in the graph (Fig. 1C, right panel). This analysis revealed that, although not readily visible to the naked eye, dNSP2 is a component of viroplasms, and greater amounts of dNSP2 are in the cytoplasm. We concluded that at 8 hpi, viroplasms contain dNSP2, vNSP2, and NSP5, but it was not clear what role the two forms of NSP2 played in viroplasm formation.

The extent of colocalization between the two forms of NSP2 was also examined by probing RV-infected cells simultaneously for NSP2 with both MAbs (Fig. 1D). The images clearly show partial colocalization of dNSP2 and vNSP2 at 8 hpi based on the orange color in the merged image and supported by the fluorescence intensity analysis. To ensure that the phenotype we observed was not an artifact of our fixation or permeabilization method, MA104 cells were fixed each time in freshly prepared 4% paraformaldehyde for 30 min and permeabilized in two different detergents (ionic and nonionic) with the same results. In addition, the distribution of NSP2 detected with the MAbs was not an artifact of MA104 cells based on probing for dNSP2 and vNSP2 in RV-infected Caco-2 cells at 6 and 10 hpi where similar results were observed (data not shown).

New MAbs bind the N and C termini of NSP2.

To begin to characterize the MAbs that, apparently, detect two forms of NSP2 by confocal microscopy, we mapped the MAb binding sites by probing full-length NSP2 and truncated NSP2 proteins (Fig. 2A) on Western blots (Fig. 2B). As expected, anti-NSP2 PAb detected full-length NSP2 and each of the truncated NSP2 proteins. MAb dNSP2_NT failed to detect any NSP2 peptide lacking the first 50 amino acids (aa) of the protein, suggesting that the epitope resides within amino acids 1 to 50 in the N terminus. Peptide array analysis could not further define the dNSP2 epitope, suggesting that the N-terminal epitope is nonlinear (data not shown). We therefore defined the MAb dNSP2_NT epitope as being between amino acid 1 and 142 based on the smallest truncated protein recognized by deletion mapping. In contrast, MAb vNSP2_CT did not detect NSP2 peptides lacking the C terminus, suggesting the epitope was located within amino acids 293 to 317. Peptide array analysis confirmed and refined the map to amino acids 293 to 300 (data not shown). Although we could not more finely map the epitope for MAb dNSP2_NT, it is clear the epitope is N terminal and distinctly separate from the C-terminal epitope for MAb vNSP2_CT. Thus, these two MAbs were named MAb dNSP2_NT and MAb vNSP2_CT, respectively.

FIG 2.

Antigenic mapping of NSP2 MAbs using deletion constructs and modeling the domains on the NSP2 crystal structure. (A) Schematic of the NSP2 deletion plasmids. FL, full length. (B) Western blot of expressed NSP2 truncated proteins from panel A. Each plasmid was expressed in bacteria, and the resulting protein was electrophoresed on an SDS-polyacrylamide 4 to 20% gradient gel and transferred to nitrocellulose. Proteins were detected using guinea pig PAb serum to NSP2 and mouse MAbs to dNSP2_NT or vNSP2_CT. Proteins were visualized using goat anti-guinea pig or goat anti-mouse secondary antibodies conjugated to alkaline phosphatase. (C) Binding analysis and predicted locations of MAb binding on the NSP2 octamer. Binding kinetics were determined by biolayer interferometry (BLI). Sensograms for NSP2-MAb dNSP2_NT binding (left) and NSP2-MAb dNSP2_CT (right) at the indicated concentrations. (D) Surface representation of the NSP2 octamer with one of the subunits colored to show the predicted MAb binding sites. (Left) The predicted binding location of MAb dNSP2_NT is shown in blue (aa 1 to 50 [dark blue]; aa 51 to 142 [light blue]), and the predicted location of the MAb vNSP2_CT binding shown in red (aa 293 to 300) can be seen on the side. Amino acids 301 to 317 defining the remainder of the C terminus is colored orange. The rest of the subunit is colored yellow. (Right) The NSP2 octamer is rotated about the vertical axis to show the MAb vNSP2_CT (red) binding site head on.

Each NSP2 MAb binds accessible, surface-exposed regions of NSP2.

Octameric NSP2 has enzymatic activity, forms viroplasms, and is the functional form of the protein (6, 15). Therefore, we next evaluated the binding of each MAb to bacterially expressed and purified octameric NSP2 using Octet. Our analysis demonstrated that both MAbs bind the NSP2 octamer tightly and with high affinity (K_d of 20 nM for MAb dNSP2_NT and K_d of 4 nM for MAb vNSP2_CT [Fig. 2C]).

Based on the binding results with octameric NSP2, the predicted MAb binding sites were modeled on the known structure of the NSP2 octamer to evaluate the location and accessibility of the N and C termini. Figure 2D illustrates the predicted MAb dNSP2_NT binding domain (aa 1 to 50 [dark blue]; aa 51 to 142 [light blue]) and MAb vNSP2_CT epitope (aa 293 to 300 [red]; the remainder of the C terminus is in orange for reference) on one NSP2 monomer (yellow) within the octamer. Modeling indicates that both MAbs can bind to surface-exposed epitopes on the NSP2 octamer.

New monoclonal antibodies are NSP2 specific and recognize different epitopes.

At every stage of the selection process, each of the new NSP2 MAbs showed the same reactivity with bacterially expressed and purified recombinant NSP2 by immunoblotting and ELISA (Fig. 2B and data not shown). However, detection of NSP2 outside the viroplasm in RV-infected cells (Fig. 1B) was unexpected, and we were not aware of any report of cytoplasmically diffuse NSP2. Therefore, we evaluated whether a nonviral cytoplasmic protein in infected cells might cross-react with the MAbs by performing Western blotting with mock-infected and RV-infected cell lysates. Cell lysate samples were prepared and analyzed by SDS-PAGE separation using Laemmli sample buffer with β-mercaptoethanol (βME) to reduce intra- and intermolecular disulfide bonds and probed for NSP2 using each MAb (and a guinea pig anti-NSP2 PAb that consistently detects NSP2). In RV-infected cell lysates analyzed under reducing conditions, a band corresponding to the same molecular mass as recombinant NSP2 (∼34 to 35 kDa) was detected by Western blotting using MAb dNSP2_NT (Fig. 3A, left panel), but no NSP2 was detected by MAb vNSP2_CT in a duplicate blot (Fig. 3A, right panel). When the same lysate was prepared for PAGE separation under nonreducing conditions in sample buffer without βME and without urea in the running buffer, two bands of NSP2 were detected by both MAbs (Fig. 3B). NSP2 was consistently detected in lysates with the anti-NSP2 PAb (data not shown), and neither MAb detected proteins in mock-infected cell lysates or other bands in RV-infected cell lysates (Fig. 3A and B), indicating that the MAbs are specific for NSP2 and do not cross-react with a cellular protein in mock- or RV-infected cells. All subsequent immunoblots were performed under nonreducing conditions to ensure detection of NSP2 by both MAbs.

FIG 3.

NSP2 detected in mock- and rotavirus-infected cell lysates by Western blotting and modeling the disulfide bond in NSP2. (A and B) Sample lysates prepared with β-mercaptoethanol (βME) and urea (+) (A) or prepared without βME or urea (−) (B) were boiled for 3 min and analyzed by gradient SDS-PAGE. Proteins in the resulting gels were transferred to nitrocellulose and probed for NSP2 with either MAb dNSP2_NT or MAb vNSP2_CT hybridoma supernatant and detected with goat anti-mouse secondary antibodies labeled with alkaline phosphatase. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) detection was used as a loading control. The molecular masses (in kilodaltons) are shown on the right of each panel. (C) Licorice representation of NSP2 monomer is shown with the N- and C-terminal domains colored in green and blue, respectively. Four cysteine residues in NSP2 are shown as spheres and indicated by black arrows. (D) The side chains of a pair of Cys residues (Cys8 and Cys85) in NSP2 structure (PDB accession no. 1L9V) are close enough to engage in disulfide bond formation with small changes in the side chain orientations. The side chains of Cys are represented with balls and sticks, with their carbon and sulfur atoms colored red and yellow, respectively. (E) Modeling of a possible disulfide bond between Cys8 and Cys85 by rotating their side chains. The side chains are colored by the same scheme as described above for panel D.

MAb vNSP2_CT detected NSP2 in RV-infected cells only under nonreducing conditions (without βME), which suggested that the vNSP2 epitope might require a disulfide bond for detection (Fig. 3B). Full-length bacterially expressed NSP2 was readily detected with MAb vNSP2_CT under reducing conditions (Fig. 2B), indicating that the vNSP2 epitope is linear and not conformational, and this was confirmed by peptide array analysis. This indicated that NSP2 expressed during rotavirus infection differs from recombinant, bacterially expressed NSP2 and suggested that NSP2 expressed in mammalian cells might be posttranslationally modified or conformationally distinct. Additionally, we observed two bands of NSP2 under nonreducing conditions (Fig. 3B) that may represent both a reduced and nonreduced form of NSP2. Modeling predicts a disulfide bond in NSP2 (Fig. 3C to E). On the basis of these results, we concluded that these MAbs are specific for NSP2, they recognize distinct epitopes on NSP2, and MAb dNSP2_NT detects a previously unreported form of NSP2.

Dispersed and viroplasmic NSP2 represent two distinct forms of the protein.

We performed additional experiments to evaluate whether MAb dNSP2_NT and MAb vNSP2_CT detect the same form of NSP2. Both MAbs precipitated NSP2 in similar quantities based on Western blot analysis of the precipitates detected by an anti-NSP2 PAb serum (data not shown). Therefore, reciprocal immunoprecipitations (IPs) from mock- and RV-infected cell lysates were performed using either MAb dNSP2_NT, MAb vNSP2_CT, or an isotype control, to capture NSP2, and Western blots were then probed for NSP2 with the nonprecipitating MAb or an anti-NSP2 PAb (Fig. 4). NSP2 was precipitated by MAb vNSP2_CT and readily detected with MAb dNSP2_NT (Fig. 4A). In contrast, dNSP2 precipitated by MAb dNSP2_NT was not detected by MAb vNSP2_CT (Fig. 4B, top panel). A companion blot was probed for NSP2 using guinea pig anti-NSP2 PAb serum (Fig. 4B, bottom panel). This experiment showed that MAb dNSP2_NT did precipitate NSP2 from the infected cell lysate. No NSP2 was detected by either MAb in mock-infected lysate, and a MAb isotype control also did not precipitate NSP2 from infected or noninfected lysates (Fig. 4C). These results confirmed the following. (i) There are two different forms of NSP2 in the infected cell. (ii) dNSP2 differs from vNSP2 at the C terminus.

FIG 4.

Monoclonal antibodies precipitate two forms of NSP2. (A to C) Immunoprecipitations were performed with mock- and rotavirus-infected cell lysates using either MAb vNSP2_CT that detects viroplasmic NSP2 (A), MAb dNSP2_NT that detects dispersed NSP2 (B), or an isotype control MAb matched to vNSP2_CT isotype (C). The precipitated samples were analyzed by SDS-PAGE under nonreducing conditions, and Western blots were probed for NSP2 using the reciprocal MAb (dNSP2_NT or vNSP2_CT) or a guinea pig anti-NSP2 PAb. The NSP2 primary antibodies were detected with either goat anti-mouse or goat anti-guinea pig alkaline phosphatase secondary conjugates.

Kinetics of viroplasm formation.

Both forms of NSP2 were initially visualized by confocal microscopy and characterized by Western blotting in cells infected for 8 h (Fig. 1, 3, and 4). This single snapshot at 8 hpi suggested that dNSP2 might be a precursor to vNSP2. To test this hypothesis and an alternative possibility that dNSP2 might result from a breakdown of viroplasms, we determined the localization of both forms of NSP2 and NSP5 during the replication cycle by performing a kinetic analysis of mock- and RV-infected cells by confocal microscopy from 3 to 7 h postinfection (Fig. 5). No signal was detected in mock-infected cells as shown previously (data not shown). Representative RV-infected cells from each time point were imaged, and the fluorescence intensity for selected paths (white arrows) was analyzed (Fig. 5, graphs). At the earliest time point (3 hpi), dNSP2 was visualized colocalizing with NSP5 in distinctive puncta in the cytoplasm suggestive of nascent viroplasms. The fluorescence intensity profile of the 3 hpi image (Fig. 5, 3 hpi, right panel) confirmed these observations. At 4 hpi, dNSP2 began to accumulate in the cytoplasm, all NSP5-containing puncta colocalized with dNSP2, and a small amount of vNSP2 also appeared in viroplasms. By 5 and 6 hpi, a notable increase in vNSP2 that colocalized with NSP5 was observed. By 7 hpi, vNSP2 was the major form of NSP2 in the viroplasm. The confocal images and the intensity profiles suggest that dNSP2 is synthesized early and associates with NSP5 to form nascent viroplasms. In contrast, very little vNSP2 is detected in viroplasms prior to 4 hpi, but vNSP2 rapidly accumulates after 4 hpi becoming the major component of the mature viroplasm at 7 hpi. The ratio of fluorescence intensity of dNSP2 to NSP5 and vNSP2 to NSP5 (Fig. 5, bottom graph) were determined for each time point. Comparison of the ratios within a time point reveals that early in infection the ratio of dNSP2 to NSP5 is greater than that of vNSP2 to NSP5. At 5 and 6 hpi, the ratios of dNSP2 to NSP5 and vNSP2 to NSP5 are similar. However, at 7 hpi, the ratio of vNSP2 to NSP5 is greater than the ratio of dNSP2 to NSP5, indicating an inverse ratio of dNSP2 to NSP5 and vNSP2 to NSP5 during the course of infection. These results do not support the idea of dNSP2 arising from disintegrating vNSP2 viroplasms but rather that dNSP2 may initiate viroplasm formation and excess dNSP2 accumulates with time in the cytoplasm outside viroplasms if it cannot enter viroplasms.

FIG 5.

Time course of RV infection in MA104 cells. Each panel shows a representative RV-infected cell fixed with 4% paraformaldehyde (PFA) at the indicated times postinfection. All samples were permeabilized and probed with MAb dNSP2_NT, MAb vNSP2_CT, and PAb NSP5 simultaneously using the appropriate secondary antibody. MAb dNSP2_NT was detected with goat anti-mouse labeled with Alexa Fluor 488 (green), PAb anti-NSP5 was detected with goat anti-guinea pig labeled with Alexa Fluor 633 (blue), and MAb vNSP2_CT was directly labeled with Alexa Fluor 568 (red); DAPI was used for Merge panels (light blue). (Right graphs) Fluorescence intensity graphs for colocalization analyses corresponding to the white arrows in the merged image for each time point with colors that correspond to those in each panel. (Bottom graph) Fluorescence intensity ratios of dNSP2 to NSP5 (white bars) and vNSP2 to NSP5 (black bars) in viroplasms (6 to 11 viroplasms per time point) were calculated at each time point.

MAbs dNSP2_NT and vNSP2_CT detect viroplasm-like structures.

Expression of NSP2 and NSP5 alone in cells results in the formation of viroplasm-like structures (VLS). VLS have been used as surrogates for viroplasm assembly and permit the study of NSP2 and NSP5 interactions in the absence of other viral proteins (7, 28, 33, 36–40). We next evaluated the ability of vNSP2_CT and MAb dNSP2_NT to detect NSP2 in cells transfected to express NSP2 and NSP5 or NSP2 alone (Fig. 6). Both MAbs detected punctate NSP2 in VLS in cells expressing NSP2-NSP5 (Fig. 6A and B, left panels). In contrast, both MAbs detected NSP2 dispersed throughout the cytoplasm in cells expressing NSP2 alone. Together, these data show that both MAbs detect NSP2 in VLS, as seen in the RV-infected cells (Fig. 5) (7).

FIG 6.

Detection of viroplasm-like structures (VLS) with MAbs dNSP2_NT and MAb vNSP2_CT. MA104 cells were cotransfected with plasmids expressing NSP2 and NSP5 (left panels) or a plasmid expressing NSP2 alone (right panels). NSP2 was detected using either MAb vNSP2_CT (A) or MAb dNSP2_NT (B) and visualized using a secondary conjugate antibody goat anti-mouse labeled with Alexa Fluor 488. NSP5 was detected using guinea pig PAb anti-NSP5 serum and visualized with a secondary goat anti-guinea pig antibody conjugated to Alexa Fluor 633 and artificially colored red for ease of visualization.

dNSP2 and vNSP2 interactions with NSP5 are phosphorylation dependent.

Having MAbs that distinguish two different forms of NSP2 could aid in elucidating different functions of these NSP2s. To examine this possibility, we performed immunoprecipitations with mock- and RV-infected cell lysates to determine whether the same viral and cellular proteins interact with dNSP2 and vNSP2. Because NSP5 phosphorylation is a known aspect of rotavirus replication, we examined the interactions of both forms of NSP2 with hypophosphorylated and phosphorylated isoforms of NSP5. For this experiment, aliquots of the same mock- and RV-infected cell lysates were treated as follows. (i) One portion of the mock- and RV-infected cell lysates was left untreated. (ii) Another portion was treated with phosphatase inhibitors to preserve the many phosphorylated isoforms of NSP5. (iii) The remaining portion was incubated with calf intestinal phosphatase (CIP) to remove as many phosphate modifications as possible. NSP2 was immunoprecipitated from each of the three treated or untreated infected cell lysates with MAb dNSP2_NT and MAb vNSP2_CT. The precipitated complexes were probed for NSP5 (Fig. 7). In the untreated RV-infected cell lysate, NSP5 retained some phosphorylation based on detection of the 26- and 28-kDa bands (glycosylated and hypophosphorylated forms) and ∼32-kDa band (hyperphosphorylated forms) (Fig. 7A, Input lane). In contrast, NSP5 detected in the RV-infected lysate treated with phosphatase inhibitors was more highly phosphorylated (Fig. 7B, Input lane), while NSP5 in CIP-treated RV-infected lysates contained only the hypophosphorylated forms (Fig. 7C, Input lane). When NSP2 was immunoprecipitated with MAb dNSP2_NT from each of the three lysates, NSP5 was detected in the precipitates from the untreated and CIP-treated RV-infected lysates (Fig. 7A and C), but not from the phosphatase inhibitor-treated lysates (Fig. 7B), indicating that dNSP2 does not interact with hyperphosphorylated NSP5. In contrast, when NSP2 was immunoprecipitated with MAb vNSP2_CT from each of the three lysates, NSP5 was not detected in the precipitates from the untreated and CIP-treated RV-infected lysates (Fig. 7A and C). Significantly, NSP5 was detected in precipitates from the phosphatase inhibitor-treated lysates (Fig. 7B), indicating that viroplasmic NSP2 interacts only with the more highly phosphorylated isoforms of NSP5. These results clearly show that there are biochemical and functional differences between the two forms of NSP2 recognized by the dNSP2_NT and vNSP2_CT MAbs and the forms of NSP5 that dNSP2 and vNSP2 interact with, suggesting that (i) the conformation of NSP2 may be affected by the interactions with the different isoforms of NSP5 and (ii) viroplasm formation could be regulated by a phosphorylation-dependent mechanism.

FIG 7.

Monoclonal antibodies to NSP2 coimmunoprecipitate NSP5. Mock- and rotavirus-infected cell lysates were prepared in RIPA buffer containing protease inhibitors. (A to C) Aliquots of the lysates were either left untreated (A) or incubated with phosphatase inhibitors (B) or calf intestinal phosphatase (C). Immunoprecipitations (IPs) were performed with MAb dNSP2_NT or MAb vNSP2_CT. Input lysates and IP samples were analyzed by SDS-PAGE, and Western blots were probed for NSP5 using a PAb guinea pig anti-NSP5 serum. Duplicate gels were probed for NSP2 using PAb anti-NSP2, shown in the bottom blots. Anti-NSP5 and anti-NSP2 polyclonal antibodies were detected with a secondary goat anti-guinea pig IgG antibody conjugated to alkaline phosphatase. The molecular masses (MM) 26-28 kDa indicate the glycosylated, hypophosphorylated forms of NSP5 and MW 32-35 kDa indicates the migration range of hyperphosphorylated NSP5.

Both forms of NSP2 interact with VP1 and VP2.

We next evaluated the interaction of MAb dNSP2_NT and MAb vNSP2_CT with VP1 and VP2, major components of viroplasms. Both VP1 and VP2 interacted with dNSP2 and vNSP2 by immunoprecipitation (Fig. 8A). To ensure that the NSP2 MAbs were specifically precipitating RV proteins, the MAb precipitates were probed for NSP4 that is not found in viroplasms. Although abundant NSP4 was detected in the input lysate, NSP4 was not detected in the precipitates from either MAb (Fig. 8B).

FIG 8.

Monoclonal antibodies to NSP2 coimmunoprecipitate VP1 and VP2. Rotavirus- and mock-infected cell lysates were prepared in RIPA buffer with protease inhibitors, and immunoprecipitations were performed with either MAb dNSP2_NT or MAb vNSP2_CT. (A and B) Proteins in the precipitated samples were analyzed by SDS-PAGE under nonreducing conditions, and Western blots were probed with PAb guinea pig anti-VP1, PAb guinea pig anti-VP2, and PAb guinea pig anti-NSP2 sera (A) or with PAb rabbit anti-NSP4 (B). The primary antibodies were detected with either goat anti-mouse or goat anti-guinea pig alkaline phosphatase secondary conjugates.

Both forms of NSP2 interact with tubulin, but only dNSP2 interacts with acetylated tubulin, and both interactions are abolished by phosphorylation.

Previous reports showed that NSP2 interacts with tubulin (19, 41), so we determined whether this function was specific to the diffuse or viroplasmic form of NSP2. Analysis of immunoprecipitates from the phosphatase inhibitor-treated and CIP-treated lysates from Fig. 7 showed that alpha and beta tubulin coprecipitated with both dNSP2 and vNSP2 MAbs when RV-infected lysates were treated with CIP (Fig. 9), but neither form of tubulin was precipitated in the presence of phosphatase inhibitors (data not shown). When the same lysates were probed for acetylated alpha tubulin, a modified form of tubulin generally associated with stabilized microtubules, acetylated alpha tubulin coprecipitated only with the dNSP2_NT MAb, and only from CIP-treated lysates (Fig. 9 and data not shown). Acetylated alpha tubulin did not coprecipitate with the vNSP2_CT MAb under any treatment condition, suggesting no direct or indirect interaction with the vNSP2_CT MAb. These results provided further evidence differentiating the two forms of NSP2 by showing that only the dNSP2_NT MAb pulls down NSP2 that interacts with acetylated alpha tubulin. In addition, the tubulin interaction was seen only in dephosphorylated lysates, again implicating phosphorylation-dependent regulation of viroplasm assembly.

FIG 9.

Monoclonal antibodies to NSP2 coimmunoprecipitate alpha and beta tubulin. NSP2 was precipitated from CIP-treated rotavirus-infected lysates prepared in RIPA buffer with protease inhibitors, using either MAb dNSP2_NT or MAb vNSP2_CT. Proteins coprecipitating with NSP2 were analyzed by SDS-PAGE, and Western blots were probed with rabbit anti-alpha tubulin, rabbit anti-beta tubulin, or rabbit anti-acetylated alpha tubulin. Primary antibodies were detected using secondary goat anti-rabbit IgG conjugated to alkaline phosphatase. Western blot analysis of NSP2 from the same CIP-treated lysates is shown in Fig. 7C.

The C-terminal helix of NSP2 is required for VLS assembly.

Crystallographic analyses demonstrated that the C-terminal helix (CTH) of NSP2 can adopt an open or closed conformation (16). In the closed conformation, monomers of NSP2 within the octamer have the CTH oriented in a position close to the rest of the molecule (Fig. 10A). In the open conformation, two of the eight NSP2 monomers have the CTH oriented away from the main body of the octamer (only one monomer is shown in color [yellow]; CTH shown in orange). NSP2 monomers in the open conformation can interact with the open CTH of adjacent octamers resulting in domain swapping and NSP2 octamer chain formation. Hu et al. (16) predicted that the CTH and chain formation might be critical for viroplasm formation. Using MAb dNSP2_NT, which detects punctate VLS in cells transfected with full-length NSP2 and NSP5 (Fig. 6), VLS formation was assessed using confocal microscopy. MA104 cells were cotransfected with a plasmid expressing a C-terminal truncated form of NSP2 (NSP2 with amino acids 1 to 292 [NSP2_1-292]) that lacks the entire CTH, and a second plasmid expressing NSP5 (Fig. 10B). Both NSP2_1-292 and NSP5 were diffusely distributed throughout the cytoplasm, and no VLS was observed by MAb dNSP2_NT, supporting the requirement for the CTH in viroplasm formation.

FIG 10.

NSP2 lacking the C-terminal helix (CTH) is cytoplasmically dispersed. (A) An NSP2 octamer highlighting a single subunit with the CTH (red) in the closed and open conformations. The coloring scheme is the same as described in the legend to Fig. 2. (B) Immunofluorescent confocal image of MA104 cells transfected with plasmids expressing truncated NSP2_1-292 lacking the CTH and full-length NSP5. NSP2_1-292 was detected using MAb dNSP2_NT that detects dispersed NSP2 in virus-infected cells and VLS in cells coexpressing NSP2-NSP5. The NSP2 MAb was detected with a secondary goat anti-mouse antibody conjugated to Alexa Fluor 488. NSP5 was detected using guinea pig PAb and visualized with goat anti-guinea pig antibody conjugated to Alexa Fluor 568.

NSP2 can be phosphorylated at S313 in the C terminus.

Our data up to this point indicated that two forms of NSP2 exist in RV-infected cells and that the two forms appear to differ at the C terminus. Additionally, the results of several of our experiments suggested a C-terminal posttranslational modification or conformational change in vNSP2. To further explore the differences between NSP2 detected by MAb dNSP2_NT and MAb vNSP2_CT, mass spectrometry (MS) analysis was performed on NSP2 immunocomplexes prepared from RV-infected cell lysates in the presence of phosphatase inhibitors. These analyses revealed that NSP2 could be phosphorylated at serine 213 in the CTH. Both phosphorylated dNSP2 and phosphorylated vNSP2 were detected; however, phosphorylated vNSP2 was found in greater abundance than dNSP2 (ratio 40:1), further suggesting that viroplasm formation is regulated by phosphorylation.

MS analyses of NSP2 immunoprecipitated complexes also confirmed the results of our Western blot analysis that indicated that MAb dNSP2_NT and MAb vNSP2_CT coprecipitated VP2 with NSP2 (Fig. 8). The MS analysis also showed that the VP2 found in complex with vNSP2, but not dNSP2, is phosphorylated at S389. This is the first report of which we are aware of phosphorylated VP2, and intriguingly, it is in a complex with phosphorylated NSP2 and hyperphosphorylated NSP5. Unexpectedly, MS did not detect VP1 in the NSP2 immunoprecipitates, although we reproducibly demonstrated an NSP2-VP1 interaction by detecting immunoprecipitates by Western blotting. This incongruity might be explained by inadequate trypsin digestion of VP1 and the subsequent failure of large VP1 peptides to be detected by MS. However, the analysis did confirm that both dNSP2 and vNSP2 interact with alpha and beta tubulin. Additionally, VP3 was detected in dNSP2 and vNSP2 immunoprecipitates, supporting previous reports that NSP2 is associated with VP3 in replication intermediates (13).

DISCUSSION

Rotavirus genome replication and double-layer particle assembly occur in discrete punctate structures associated with lipid droplets (42, 43), called viroplasms, which contain structural proteins VP1, VP2, VP3, and VP6 and nonstructural proteins NSP2, NSP5, and NSP6. Although viroplasm formation requires NSP2 in association with NSP5, the exact mechanism(s) that initiates viroplasm assembly remains unknown. Here we report the following. (i) Two conformationally and biochemically distinct forms of NSP2 exist in RV-infected cells. (ii) One form of NSP2, dispersed NSP2 (dNSP2), is associated with nascent viroplasms, while the other form, viroplasmic NSP2 (vNSP2), accumulates in viroplasms concurrent with their increase in size throughout the replication cycle and is the predominant form in mature viroplasms. (iii) The C-terminal helix (CTH) on NSP2 is required for VLS formation. (iv) Phosphorylation-dependent cellular and viral protein modifications are important for interactions with NSP2 in RV-infected cells, suggesting a potential phosphorylation-dependent mechanism for viroplasm formation.

To address the role of NSP2 in viroplasm assembly, we generated and characterized monoclonal antibodies (MAbs) to full-length NSP2. Using unique MAbs that map to either the N or C termini of NSP2, we identified two forms of NSP2. dNSP2 is a previously unreported form of NSP2 that localizes to nascent viroplasms, remains associated with maturing viroplasms, accumulates in a dispersed form throughout the cytoplasm over time, and forms a protein complex with VP1, VP2, VP3, alpha/beta tubulin, acetylated alpha tubulin, and hypophosphorylated NSP5. vNSP2 is observed exclusively in viroplasms and also forms a protein complex that includes VP1, VP2, VP3, and alpha/beta tubulin. However, vNSP2 interacts only with the hyperphosphorylated isoforms of NSP5 and does not interact with acetylated tubulin.

Our biochemical and kinetics analyses suggest a mechanism of viroplasm initiation and assembly coordinated by phosphorylation, possibly involving a conformational change from one form of NSP2 into another. We postulate a model in which dNSP2 associates with hypophosphorylated NSP5, initiating viroplasm formation. Cytoplasmically dispersed dNSP2 associates with unphosphorylated NSP5 and possibly other viroplasm-bound proteins and initiates nascent viroplasm assembly via a dNSP2-microtubule interaction. By an as yet undetermined mechanism, perhaps involving cellular casein kinase-like enzymes (44–46), NSP2, NSP5, and VP2 are phosphorylated. Phosphorylation of NSP2 or VP2 and/or increased phosphorylation of NSP5 may induce a conformational change in dNSP2, resulting in the transition of dNSP2 to vNSP2 triggering viroplasm maturation via open CTH interactions.

In support of our model, both dNSP2 and vNSP2 coimmunoprecipitated NSP5 from RV-infected cell lysates; however, these interactions were dependent on the phosphorylation state of NSP5, and possibly the phosphorylation state of NSP2. dNSP2 interacted exclusively with hypophosphorylated NSP5, while vNSP2 interacted only with hyperphosphorylated NSP5 isoforms. We noted that the faster migrating 26- and 28-kDa bands, indicative of hypophosphorylation, also precipitated with vNSP2 in the presence of phosphatase inhibitors, but NSP5 readily forms dimers and may form phospho-heterodimers in which only one monomer is hyperphosphorylated, conferring the ability to interact with viroplasmic NSP2. NSP5 is hyperphosphorylated in transfected cells only when coexpressed with NSP2 (17), and hyperphosphorylation requires casein kinase 1 alpha (CK1α)-mediated phosphorylation of serine 67 on NSP5 (44, 45). Together, these results support a phosphorylation-dependent mechanism for viroplasm formation in virus-infected cells.

In further support of a model of phosphorylation-dependent viroplasm assembly, mass spectrometry analysis of rotavirus proteins coimmunoprecipitated with each form of NSP2 from virus-infected cell lysates revealed new data that both NSP2 and VP2 are phosphorylated. NSP2 was phosphorylated at S313 in the CTH domain. Both dNSP2 and vNSP2 were phosphorylated; however, more phospho-NSP2 was associated with viroplasms than with dispersed, cytoplasmic NSP2 (ratio of 40:1). NSP2 was previously shown to exhibit a transient phosphorylation of H225 that occurs during NTP hydrolysis (47), and this phosphate is transferred quickly to receptor NDPs. We are unaware of any other report of phosphorylated NSP2. VP2 phosphorylated at S389 was detected only in vNSP2 immunoprecipitates. To our knowledge, phosphorylated VP2 has not been reported either. The precise role(s) of phosphorylated NSP2 and VP2 in initiating or regulating viroplasm formation will need to be investigated in future studies. Interestingly, temperature-sensitive (ts) mutants of VP2 are phenotypically similar to ts mutants of NSP2. Thus, VP2 and NSP2 ts mutants fail to form viroplasms, exhibit decreased viral protein expression, and fail to synthesize dsRNA and virus progeny (5), suggesting a role for both NSP2 and VP2 in viroplasm formation. When VP2 is coexpressed with NSP5 in transfected cells, VP2, like NSP2, hyperphosphorylates NSP5 and forms VP2-NSP5 VLS (27, 28). VP2-mediated hyperphosphorylation of NSP5 and VLS formation support our model that viroplasm formation is phosphorylation dependent.

VP1 is another protein that may regulate viroplasm formation. Interaction of VP1 and NSP2 has been reported previously (18, 38), and our results show that both dNSP2 and vNSP2 coprecipitate VP1. Our MS results did not confirm an NSP2-VP1 interaction, although we and others reproducibly demonstrated this interaction by immunoprecipitation and Western blotting (18). In cells coexpressing VP1 and NSP2-NSP5, or VP2-NSP5, VP1 is recruited to VLS, and the degree of NSP5 hyperphosphorylation is reduced (28). It may be that the role of VP1 is to modulate the phosphorylation of NSP5, suggesting that hyperphosphorylation of NSP5 is regulated.

Crystallographic studies of NSP2 proposed a possible role of the CTH in viroplasm formation (16). The results of our biochemical assays indicated that dNSP2 and vNSP2 differ in presentation of the C-terminal MAb epitope on the CTH and that the CTH is required to form VLS in cells cotransfected with NSP2 and NSP5. On the basis of our biochemical assays and three-dimensional (3D) modeling, we predict that in RV-infected cells, MAb vNSP2_CT binds an exposed epitope on the CTH of vNSP2 in the open conformation, and this epitope is inaccessible on the CTH of dNSP2 that is in the closed conformation. It was previously demonstrated that the C-terminal 7 amino acids of NSP2 are dispensable for dsRNA synthesis in a complementation assay (48). This deletion would remove phosphoserine 313 on vNSP2. Because the deletion did not negatively affect RNA synthesis, it is unlikely that NSP2 phosphorylation is important for replication. The specific role of phosphorylated NSP2 in viroplasm assembly remains to be determined.

NSP2 has been shown to interact with tubulin (19), and we found that both dNSP2 and vNSP2 interact with alpha and beta tubulin but only in untreated or CIP-treated RV-infected cell lysates. Neither dNSP2 nor vNSP2 coprecipitated alpha or beta tubulin from RV-infected cell lysates treated with phosphatase inhibitors. This indicates the following. (i) The dNSP2-tubulin interaction requires that NSP2 or another interacting protein in the complex be unphosphorylated. (ii) Tubulin is no longer complexed with vNSP2 despite abundant tubulin surrounding viroplasms in the form of microtubules (19, 41, 49). We also detected acetylated alpha tubulin associated with dNSP2 from untreated and phosphatase-treated RV-infected cell lysates but not with vNSP2. Acetylated alpha tubulin is generally associated with stabilized microtubules and is observed surrounding viroplasms (49). Nocodazole treatment of RV-infected cells destabilizes microtubules, and the viroplasms observed in these cells are small and do not fuse into larger viroplasms (41). This suggests that dNSP2, with its associated interacting proteins, binds to microtubules and may traffic associated viroplasmic components to sites of viroplasm assembly. Once in the viroplasm, the microtubule-NSP2 interaction is no longer required.

The two forms of NSP2 characterized in this study are named dNSP2 and vNSP2 due to their differential distribution at 8 hpi based on staining with the new MAbs. However, our results indicate a role for dNSP2 in viroplasm initiation and a role for vNSP2 in viroplasm maturation, suggesting a dynamic interaction between the two forms. Detailed follow-up studies will more precisely define the role each form plays in the initiation and growth of viroplasms. When these studies are completed, the names for the two forms of NSP2 may change.

Our results show for the first time that two forms of NSP2 exist and are differentially associated with viroplasm assembly and maturation. Additionally, our results suggest that viral and cellular proteins destined ultimately for localization into viroplasms may be assembled in a coordinated association with NSP2. We propose that phosphorylation and a conformational change in the C terminus of NSP2 regulate these associations and the switch from nascent viroplasm assembly to mature viroplasm formation.