ABSTRACT
Porcine reproductive and respiratory syndrome virus (PRRSV) is a positive-stranded RNA virus belonging to the family Arteriviridae. Synthesis of the viral RNA is directed by replication/transcription complexes (RTC) that are mainly composed of a network of PRRSV nonstructural proteins (nsps) and likely cellular proteins. Here, we mapped the interaction network among PRRSV nsps by using yeast two-hybrid screening in conjunction with coimmunoprecipitation (co-IP) and cotransfection assays. We identified a total of 24 novel interactions and found that the interactions were centered on open reading frame 1b (ORF1b)-encoded nsps that were mainly connected by the transmembrane proteins nsp2, nsp3, and nsp5. Interestingly, the interactions of the core enzymes nsp9 and nsp10 with transmembrane proteins did not occur in a straightforward manner, as they worked in the co-IP assay but were poorly capable of finding each other within intact mammalian cells. Further proof that they can interact within cells required the engineering of N-terminal truncations of both nsp9 and nsp10. However, despite the poor colocalization relationship in cotransfected cells, both nsp9 and nsp10 came together with membrane proteins (e.g., nsp2) at the viral replication and transcription complexes (RTC) in PRRSV-infected cells. Thus, our results indicate the existence of a complex interaction network among PRRSV nsps and raise the possibility that the recruitment of key replicase proteins to membrane-associated nsps may involve some regulatory mechanisms during infection.
IMPORTANCE Synthesis of PRRSV RNAs within host cells depends on the efficient and correct assembly of RTC that takes places on modified intracellular membranes. As an important step toward dissecting this poorly understood event, we investigated the interaction network among PRRSV nsps. Our studies established a comprehensive interaction map for PRRSV nsps and revealed important players within the network. The results also highlight the likely existence of a regulated recruitment of the PRRSV core enzymes nsp9 and nsp10 to viral membrane nsps during PRRSV RTC assembly.
INTRODUCTION
Porcine reproductive and respiratory syndrome virus (PRRSV) is the causative agent of the disease porcine reproductive and respiratory syndrome (PRRS) and represents a huge threat to the worldwide swine industry (1–3). This single-stranded positive-sense RNA virus has a genome size of approximately 15 kb and belongs to the genus Porarterivirus of the family Arteriviridae in the order Nidovirales (4, 5). Other representative members within the family are equine arteritis virus (EAV), simian hemorrhagic fever virus (SHFV), and mouse lactate dehydrogenase-elevating virus (LDV) (4, 6, 7). The 5′-end two-thirds of the viral genome is occupied by open reading frame 1a (ORF1a) and ORF1b (Fig. 1A), which specify replicase nonstructural proteins (nsps) important for viral RNA synthesis and for antagonizing host antiviral immunity (8, 9). Upon virus entry, ORF1a is translated from the incoming genome to produce replicase polyprotein pp1a, which is further matured into at least 10 nsps (e.g., nsp1α, nsp1β, nsp2 to nsp6, nsp7α, nsp7β, and nsp8) by virus-encoded proteases within nsp1α, nsp1β, nsp2, and nsp4 (7, 10, 11). In addition, there exist isoforms for nsp2, and one of them (nsp2TF) is made via a −2 frameshift mechanism (12, 13). On the other hand, translation of ORF1b, an open reading frame that specifies replicase proteins nsp9, nsp10, nsp11, and nsp12, involves a −1 ribosome frameshift (14–16).
Analysis of PRRSV nonstructural protein interactions by Y2H matrix screening. (A) Diagram of PRRSV ORF1a and ORF1b, which are posttranslationally cleaved by virus-encoded papain-like proteases (PCP1α and PCP1β), a cysteine protease (PLP2), and a 3C-like serine protease (3CL). nsp2, nsp3, and nsp5 are proteins that contain transmembrane domains. The following polymerase signature regions are indicated: RdRp, HEL, NendoU, and arterivirus-specific domain (AsD). (B) Y2H screening of interactions among PRRSV nsps. The matrix screening was performed by cotransforming into strain Y2H Gold containing prey (pGADT7) and bait (pGBKT7) vectors with the respective PRRSV nsps on high-stringency QDO plates containing X-α-Gal and aureobasidin A (QDO–X-α-Gal–AbA). All the nonstructural proteins were tested against each other, and the interaction matrix is summarized, with details in the table. (C) Summary of interactions. (D) Annotation for the interaction network of PRRSV nsps.
Synthesis of PRRSV RNA within host cells depends on the efficient and correct assembly of replication and transcription complexes (RTC) that coordinate the transcription and replication cascade. This process takes place on intracellular membranes that are remodeled to serve as a platform for protein-protein interactions and provide a favorable environment for replication (11, 17–21). Like other positive-stranded RNA viruses, this mechanism is often executed by virus-encoded hydrophobic membrane-spanning nsps (7, 17–24). In the case of PRRSV, the transmembrane proteins nsp2, nsp3, and nsp5 are likely important players. PRRSV nsp2 possesses cysteine protease activity that cleaves the nsp2-nsp3 junction site during replicase polyprotein maturation (25–27). It is also predicted to interact with nsp3, according to studies of the EAV counterparts (18, 19). The heterodimer is implicated in induction of double-membrane vesicles (DMVs), where virus replication takes place (19). nsp5 is the functionally least known membrane protein, with limited information showing that it can induce autophagic death of transfected cells and may be related to virus-induced autophagy (28, 29). Together, PRRSV nsp2, nsp3, and nsp5 are thought to serve as membrane scaffolding proteins to recruit other components of RTC to the replicase site (7, 11, 17, 21–24).
The core components of PRRSV RTC likely include all ORF1b-encoded nsps, namely, nsp9, nsp10, nsp11, and nsp12. Of these, nsp9 and nsp10 have homologues across diverse families of nidoviruses. Because of the ribosomal frameshift, PRRSV nsp9 is thought to contain nsp8 as its N terminus. Its C-terminal half specifies the function of viral RNA-dependent RNA polymerase (RdRp), whereas the N-terminal region contains a newly discovered nidovirus RdRp-associated nucleotidyltransferase domain (NiRAN) (30–32). PRRSV nsp10 encodes the function of an RNA helicase that is implicated in subgenomic-mRNA (sgmRNA) synthesis and likely participates in unwinding RNA secondary structures during replication (33–36). The enzyme is predicted to comprise four domains in succession—an N-terminal zinc binding domain (ZBD), a regulatory domain 1B, and two recA-like domains (1A and 2A, together designated HEL1)—according to structural studies of its EAV counterpart (36, 37). Importantly, both nsp9 and nsp10 have been shown to be the key virulence determinants of the Chinese highly pathogenic PRRSV (HP-PRRSV) (38). PRRSV nsp11 is an endoribonuclease (NendoU) with the catalytic sites highly conserved in arteriviruses and coronaviruses, but its function remains poorly defined in the PRRSV life cycle (39–41). In contrast, nsp12 is an arterivirus-specific protein (42). Its functions, together with those of other nsps, such as nsp6, nsp7, and nsp8, are currently not understood.
Assembly of PRRSV RTC clearly requires a network of nsps. Moreover, several PRRSV nsps (e.g., nsp2, nsp4, nsp7, and nsp8) have been found to be associated with the viral RTC during infection (43). However, the molecular details of how PRRSV nsps come together to form the viral RTC are not clear. As a critical step toward understanding this process, it is necessary to investigate the interaction network among PRRSV nsps. Information in this regard has been scarce, with only a limited number of interactions identified (e.g., nsp2-nsp1α, nsp2-nsp1β, and nsp9-nsp7) (44, 45), and the picture for a comprehensive map remained unavailable. Yeast two-hybrid (Y2H) screening, glutathione S-transferase (GST) pulldown, and coimmunoprecipitation (co-IP) are the most commonly used methods for investigating protein-protein interactions (PPI) of nidoviruses (e.g., coronaviruses) (46–49). However, all these approaches have their own limitations. The fusion expression strategy employed by Y2H screening can potentially change the conformation of a target protein, leading to either steric hindrance or exposure of protein-binding site(s). On the other hand, GST pulldown and co-IP are performed in an oxidative and detergent-containing environment in which the protein structures can potentially be perturbed. This leads to either promotion or inhibition of protein-protein interactions, which may not reflect what actually occurs within cells. In contrast, cotransfection assay (relocalization/colocalization analysis) is a highly reliable approach that tests PPI in the native form and looks for a gain of function. The protein interactions can be judged on localization changes, namely, whether a protein changes its localization in response to the presence of another protein. The cotransfection assay in general can well complement Y2H screening and co-IP and has been successfully used to investigate the interaction mechanisms of herpesvirus proteins (50–54). By combining the strengths of the three assays, we determined the protein-protein interactions among PRRSV nsps. Our studies established a comprehensive interaction map and identified key players within the network.
RESULTS
Yeast two-hybrid screening of protein-protein interactions among PRRSV nsps.To analyze the protein-protein interactions among PRRSV nsps, we employed the Y2H system. The genes coding for individual nsps (nsp1 to nsp12) (Fig. 1A) were cloned into both the prey and bait plasmids so that each interaction pair was examined from both directions (as both bait and prey). The nsp2 isoforms (e.g., nsp2N and nsp2TF), were not included, because their N-terminal regions largely overlap and also for the purpose of reducing the workload. Among 169 protein pairs tested, a total of 36 interaction combinations were identified as positive in a pairwise matrix (Fig. 1B), accounting for 21.3% (36/169) of the protein interactions. After removing the redundancy of interaction combinations, there were a total of 29 interactions. Interestingly, 19 interactions showed one directionality (Fig. 1C), indicating an influence of fusion domains on the interactions. Nevertheless, 7 pairs of interactions were detected in both directions: nsp9-nsp2, nsp9-nsp10, nsp9-nsp11, nsp9-nsp12, nsp12-nsp1β, nsp12-nsp10, and nsp12-nsp11 (Fig. 1C). In addition, self-interactions were observed for nsp1α, nsp9, and nsp12 (Fig. 1C), suggesting that these proteins could form dimeric or multimeric complexes by interacting with themselves. However, we missed the previously reported self-interactions for nsp1β and nsp11 (40, 55), which were mainly characterized by in vitro assays, suggesting that fusion domains likely have an adverse effect on the interaction. Overall, among the 29 identified interactions, 5 interaction pairs (nsp1α-nsp1α, nsp12-nsp12, nsp2-nsp1α, nsp2-nsp1β, and nsp7α-nsp9) have been reported in PRRSV (44, 45, 56, 57). However, these interactions were mainly characterized by using in vitro assays and have not been tested within mammalian cells.
The PRRSV nsp interactome-based Y2H screening suggests that nsp9 and nsp12 have the largest number of connections (Fig. 1D). The second-most-connected proteins were nsp2, nsp5, and nsp11 (Fig. 1D). The proteins with few identified binding partners were nsp1β, nsp3, nsp6, and nsp7, whereas none was identified for nsp8 (Fig. 1D). Together, the map identifies ORF1b-encoded proteins and the transmembrane proteins (nsp2 and nsp5) as the key players within the network.
Subcellular localizations of PRRSV nonstructural proteins.To further confirm the Y2H screening results, the genes coding for individual nsps (Table 1) were cloned into a eukaryotic expression vector to generate recombinant plasmids that would be used for the subsequent co-IP assay and colocalization analysis. To facilitate the detection, a hemagglutinin (HA), FLAG, or Myc epitope tag was fused to the N termini of most of the nsps. For nsp9 and its derivatives (Fig. 2A), the tag was placed at the C terminus, as we found that the subcellular localization pattern of one of the derivatives [nsp9(46–685)] was affected by the tag at the N terminus (data not shown). The expression and subcellular localization of these proteins were examined in BHK21 cells, a cell line that supports PRRSV replication in vitro and has high transfection efficiency. An immunofluorescence assay (IFA) was carried out to detect the individual proteins with antibodies to the epitope tags at 18 to 24 h posttransfection. Representative images were taken with a confocal microscope (Fig. 3), and two major localization patterns could be discerned for the PRRSV nsps. (i) The replicase proteins nsp1α, nsp1β, nsp4, nsp7, nsp8, nsp9, nsp10, and nsp11 were mostly localized in the nucleus, but with diffusion in the cytoplasm when expressed alone. (ii) The second group included nsp2, nsp3, nsp5, and nsp12. As expected, the transmembrane proteins nsp2, nsp3, and nsp5 exhibited cytoplasmic distribution. Interestingly, nsp12 was also localized to the cytoplasm, although it does not have predicted transmembrane helices. The localization patterns of these tagged proteins were also compared with those of the untagged versions by using the available antibodies specific to PRRSV nsps, and we did not observe an obvious adverse effect of tags on nsp localization (data not shown). For membrane proteins nsp3 and nsp5, specific antibodies are not available, but both proteins with tags at either the N or C terminus had the same localization pattern (data not shown).
Sequence information for nsps of PRRSV JXwn06 used in the study for plasmid construction
Domain organization and engineering of truncation constructs of PRRSV nsp9 and nsp10. (A and B) (Left) Predicted structure using the online program I-TASSER. (Right) The constructed nsp9 and nsp10 mutants. RFS, ribosomal frameshift; 1B, regulatory domain; 1A, RecA-like domain 1A; 2A, RecA-like domain 2A.
Subcellular localization of PRRSV nsps and derivatives. Expression plasmids carrying the genes for individual nsps and the corresponding derivatives were transfected into BHK-21 cells. At 18 to 24 h posttransfection, the cells were fixed, permeabilized, and stained with proper antibodies to the HA epitope, followed by staining with Alexa Fluor 488-conjugated secondary antibodies. The cell nuclei were stained with DAPI (blue) and examined by confocal microscopy. (A) Cellular localization of HA-tagged nsp1 to nsp8, nsp11, and nsp12. (B) Localization of nsp9 and its truncation mutants. (C) Intracellular localization of nsp10 and its derivatives. (D) Localization of Src peptide-tagged nsp9 and nsp10.
Analyses of interactions between PRRSV nsp9 and transmembrane proteins nsp2, nsp3, and nsp5.We analyzed the interaction network starting with nsp9 and nsp10, as they represent the key enzymes for viral RNA synthesis. The Y2H screening identified 9 binding partners for nsp9, i.e., nsp1α, nsp1β, nsp2, nsp4, nsp5, nsp7, nsp10, nsp11, and nsp12 (Fig. 1C and D). We further examined these interactions with the co-IP assay by coexpressing the interaction pairs in HEK293T human embryonic kidney cells, which support PRRSV replication, but not entry (58), and have higher transfection efficiency suitable for studying protein-protein interactions. We found that the full-length nsp9 could interact with nsp2, nsp3, nsp5, nsp10, and nsp12 (Fig. 4A, Fig. 5A, Fig. 6A and F, and Fig. 7E), but we failed to detect an interaction with nsp1α, nsp1β, nsp4, and nsp11 (data not shown). The interaction with nsp7 was not further pursued here, as it has been described previously (44).
Interaction analysis between nsp9 and nsp2. (A to E). Co-IP analysis of the interactions between nsp2 and nsp9 or its derivatives. HEK293 cells were cotransfected to express the indicated protein pairs. At 24 h after transfection, cell lysates were prepared and either analyzed by immunoblotting (WB) to measure the input expression levels or incubated with the antibodies shown at the tops of the lanes to collect interacting complexes, which were then analyzed by immunoblotting with the antibodies indicated at the left or right of each panel. (F) Relocalization and colocalization analysis. BHK-21 cells were transfected to express the indicated protein pairs. At 18 to 24 h posttransfection, the cells were fixed, stained with the appropriate antibodies to epitope tags, and examined by confocal microscopy. (G) Quantitative analysis for relocalization and colocalization relationships in cells coexpressing nsp9 or its derivatives with nsp2. The percentages of cells showing relocalization and colocalization were determined (averages are shown on top of each bar). ***, P < 0.001; ns, not significant. The error bars indicate standard deviations.
Interaction analysis between nsp9 and nsp3. (A to D) Co-IP analysis of the interactions between nsp3 and nsp9 or its derivatives. HEK293 cells were cotransfected to express the indicated protein pairs. At 24 h after transfection, cell lysates were prepared and either analyzed by immunoblotting to measure the input expression levels or subjected to co-IP analysis. (E) Relocalization and colocalization analysis. BHK-21 cells were transfected to express the indicated protein pairs. At 18 to 24 h posttransfection, the cells were fixed, stained with the appropriate antibodies to epitope tags, and examined by confocal microscopy. (F) Quantitative analysis for relocalization and colocalization relationships in cells coexpressing nsp9 or its derivatives with nsp3. The percentages of cells showing relocalization and colocalization were determined. ***, P < 0.001. The error bars indicate standard deviations.
Interaction analysis between nsp9 and nsp5 or nsp12. (A to G) Co-IP analysis of the interactions between nsp9 and nsp5 or nsp12 or its derivatives. HEK293 cells were cotransfected to express the indicated protein pairs. At 24 h after transfection, cell lysates were prepared and either analyzed by immunoblotting to measure the input expression levels or subjected to co-IP analysis. The antibody used for immunoblotting is indicated at the right of each panel. (H and I) Relocalization and colocalization analysis. BHK-21 cells were transfected to express the indicated protein pairs. At 18 to 24 h posttransfection, the cells were fixed, stained with the appropriate antibodies to epitope tags, and examined by confocal microscopy. (J) Quantitative analysis for relocalization and colocalization relationships in cells coexpressing nsp9 or its derivatives with nsp5. The percentages of cells showing relocalization and colocalization were determined. ***, P < 0.001. The error bars indicate standard deviations.
Interaction analyses between nsp10 and nsp2, nsp3, nsp5, nsp9, and nsp12. (A to G) Co-IP analysis of interactions between nsp10 or its derivatives and nsp2, nsp3, nsp5, nsp9, and nsp12. HEK293 cells were cotransfected to express the indicated protein pairs. At 24 h after transfection, cell lysates were prepared and either analyzed by immunoblotting to measure the input expression levels or subjected to co-IP analysis. The antibody used for immunoblotting is indicated at the right of each panel. (H to M) Relocalization and colocalization analysis. BHK-21 cells were transfected to express the indicated protein pairs. At 18 to 24 h posttransfection, the cells were fixed, stained with the appropriate antibodies to epitope tags, and examined by confocal microscopy. (N to P) Quantitative analysis for relocalization and colocalization relationships in cells coexpressing nsp10 or its derivatives with nsp2, nsp3, and nsp5. The percentages of cells showing relocalization and colocalization were determined. ***, P < 0.001. The error bars indicate standard deviations.
The interactions of nsp9 with nsp2, nsp3, nsp5, and nsp12 were further examined by relocalization assay. We found that nsp12, a cytoplasmic protein, could relocate well nsp9 from the nucleus to the cytoplasm within intact mammalian cells (Fig. 6I). In contrast, the three transmembrane proteins (nsp2, nsp3, and nsp5) were poorly capable of finding the full-length nsp9 (Fig. 4F, Fig. 5E, and Fig. 6H, top row). Thus, there appears to be a discrepancy between the results of the co-IP experiment and the cotransfection assay. To gain further evidence that nsp9 can interact with these transmembrane proteins within cells, we performed mutagenesis studies. To guide the engineering of nsp9 truncation constructs, we modeled the PRRSV nsp9 structure (Fig. 2A, left) with I-TASSER (59). Bioinformatics analysis suggests that the RdRp domain is localized to the C-terminal region (nsp9 amino acids [aa] 230 to 685), while the NiRAN domain is contained within the N terminus (aa 66 to 182) (31). The two domains are joined by a small linker region (aa 183 to 229). Based on this information, we created 5 mutants (Fig. 2A, right). Specifically, the construct nsp9(230–685)-HA includes the complete putative RdRp domain, whereas nsp9(431–685)-HA represents a further truncated form of RdRp. On the other hand, nsp9(1–229)-HA represents the NiRAN-containing N-terminal region of nsp9. In addition, nsp9(1–45)-HA represents the nsp8 portion and nsp9(46–685)-HA represents the ORF1b-coded portion of nsp9. When expressed alone, nsp9(1–45)-HA, nsp9(230–685)-HA, and nsp9(431–685)-HA were mainly localized to nuclei, but with a diffuse pattern in the cytoplasm, whereas the fragments nsp9(1–229)-HA and nsp9(46–685)-HA exhibited cytoplasmic distribution as puncta or speckles (Fig. 3B). It should be mentioned that in a small percentage of cells (25%), nsp9(1–229)-HA was found mainly in the nucleus (Fig. 3B).
When coexpressed with either nsp2, nsp3, or nsp5, the RdRp-containing fragment [nsp9(230–685)-HA] underwent massive redistribution (Fig. 4F, Fig. 5E, and Fig. 6H); it was absent from the nucleus and relocalized to the cytoplasm (Fig. 4F, Fig. 5E, and Fig. 6H). In contrast, nsp9(1–229)-HA did not respond well to either nsp2, nsp3, or nsp5, showing a limited colocalization relationship with nsp2, nsp3, or nsp5 (Fig. 4F, Fig. 5E, and Fig. 6H). In addition, the truncation fragments nsp9(1–45) and nsp9(46–685) were not responsive to any of the three transmembrane proteins. The quantitative analyses suggest that the transmembrane proteins mainly interact with the RdRp domain (Fig. 4G, Fig. 5F, and Fig. 6J). Further truncation of the RdRp domain [nsp9(431–685)-HA] increased the relocalization efficiency (Fig. 4G, Fig. 5F, and Fig. 6J), suggesting that the binding region is localized to the C-terminal half. Moreover, these interactions were verified by the co-IP assay (Fig. 4B, C, and D; Fig. 5B and C; Fig. 6B, C, and D). In contrast, the nsp9 N-terminal region [nsp9(1–229)-HA] did not appear to be involved (Fig. 4E, Fig. 5D, and Fig. 6E). Thus, the RdRp domain is the main region to mediate the interactions with membrane proteins.
Interactions between nsp10 and transmembrane proteins nsp2, nsp3, and nsp5.The interactions of nsp10 with nsp1α, nsp2, nsp9, and nsp12 revealed by the Y2H assay (Fig. 1C) were further confirmed by the co-IP assay (Fig. 7A, B, E, and G). In the cotransfected cells, nsp10 was found to colocalize with nsp1α in the cytoplasm (Fig. 7H, top row), but not with nsp2, nsp3, nsp5, or nsp12 (Fig. 7I to K and M, top row). Thus, like nsp9, a similar discrepancy also occurs with nsp10, raising the possibility that it takes a truncation strategy similar to that used for nsp9 to turn on the interaction of nsp10 with these membrane proteins within cotransfected mammalian cells.
Structural modeling revealed three distinct domains for nsp10 (Fig. 2B, left) (36, 37), and we accordingly constructed four truncation mutants (Fig. 2B, right). The fragment HA-nsp10(1–127) contains the complete ZBD and 1B domain, whereas HA-nsp10(128–441) contains only the HEL domain. The other two constructs contain the disrupted 1B domain [HA-nsp10(1–100) and HA-nsp10(101–441)]. When expressed alone, HA-nsp10(1–100) and HA-nsp10(101–441) were mainly localized to the nucleus, but with diffusion in the cytoplasm (Fig. 3C). For the other two mutants [HA-nsp10(1–127) and HA-nsp10(128–441)] that contain an undisrupted 1B domain, two localization patterns could be discerned. In some cells (65% to 75%), they behaved just like wild-type (WT) HA-nsp10, whereas in another population of cells (25% to 35%), they exhibited nuclear localization, but with cytoplasmic puncta (Fig. 3C). In cotransfected cells, HA-nsp10(128–441) responded to nsp2, as evidenced by its complete translocation from nucleus to cytoplasm and perfect colocalization (Fig. 7I). In about 32% of cells coexpressing nsp2 and HA-nsp10(128–441), the nsp10 mutant showed relocalization and colocalization (Fig. 7N). This interaction was also verified by the co-IP experiment (Fig. 7C). In contrast, nsp2 did not relocalize HA-nsp10(1–127) (Fig. 7I and N). Further tests with the truncation fragments HA-nsp10(1–100) and HA-nsp10(101–441) revealed that addition of a partial 1B domain to the helicase domain (aa 128 to 441) could dramatically increase relocalization efficiency (about 60%) (Fig. 7I and N), suggesting that the 1B domain can somehow exert an effect on the HEL domain. Similar results were obtained for the interactions of nsp10 with nsp3 and nsp5 (Fig. 7J, K, O, and P).
Unlike nsp2, nsp3, and nsp5, PRRSV nsp12 can relocate not only the C-terminal domain of nsp10, but also the N-terminal domain, both of which occurred with high efficiency (Fig. 7M). The fact that nsp12 can relocate the N-terminal region of nsp10 [nsp10(1–127)] also strengthens the interaction specificity of nsp10 with nsp2, nsp3, and nsp5. Together, the results suggest that the nsp10 helicase domain is the common region for binding these four proteins, while the N-terminal domain is only for nsp12.
Analysis of the interaction between nsp9 and nsp10.Both Y2H and co-IP studies identified an interaction between nsp9 and nsp10 (Fig. 1B and C; Fig. 7E and F). However, it is hard to judge the interaction within cells due to their similar distribution patterns (Fig. 3). Thus, we employed a relocalization assay. Both proteins were tagged in the N terminus with a plasma membrane anchor peptide (10 amino acids) from the Rous sarcoma virus (RSV) oncoprotein Src (60), and the resultant mutants were named Src.nsp9-HA and Src.nsp10-HA, respectively (Fig. 3). When expressed alone, Src.nsp9-HA and Src.nsp10-HA were localized to both the plasma membrane and the cytoplasm (Fig. 3D). In the cotransfected cells, however, nsp9 and nsp10 did not respond to the respective Src.nsp10-HA and Src.nsp9-HA (Fig. 7L), indicating that nsp9 and nsp10 do not interact with each other within cells. A potential caveat, however, is that the Src tag may have an adverse effect on protein interaction, which needs to be further pursued in the future. In the second assay, we used an nsp10 truncation mutant [HA-nsp10(128–441)] to test the interaction. As this mutant showed puncta in the cytoplasm in a small percentage of cells (Fig. 3), nsp9 or its derivatives would follow this pattern if they interact within cells. It was found that nsp9-FLAG responded to HA-nsp10(128–441), moving out of the nucleus to become colocalized in the cytoplasm (Fig. 7L). Thus, it appears that a conditional interaction between them exists within cells.
nsp12 is an unexpected interaction hub.nsp12 is highly conserved among PRRSV strains, but its function in the virus life cycle has remained enigmatic. Unexpectedly, the Y2H screening revealed nsp12 as a key interaction hub (Fig. 1C and D) for interaction with a variety of nsps, including nsp1α, nsp1β, nsp2, nsp3, nsp5, nsp6, nsp9, nsp10, and nsp11 (Fig. 1B, C, and D). The co-IP assay could confirm its interactions with nsp1β (Fig. 8A), nsp2 (Fig. 8B), nsp9 (Fig. 6F and G), nsp10 (Fig. 7G), and nsp11 (Fig. 8C) but failed to detect an interaction with nsp1α, nsp3, and nsp5 (data not shown). The interaction with nsp6 was not tested by co-IP due to its very small size (only 16 aa). As its interactions with nsp9 and nsp10 were described above (Fig. 6F and I; Fig. 7G and M), we focus here on other binding partners. In the cotransfected cells, nsp12 exhibited robust colocalization with nsp1β (Fig. 8E) and nsp11 (Fig. 8J) and to a lesser extent with nsp1α (Fig. 8D). Moreover, nsp12 could induce massive relocalization of these three proteins to the cytoplasm (Fig. 8D, E, and J). Consistent with the Y2H result, nsp12 colocalized well with the membrane proteins nsp2 (Fig. 8F), nsp3 (Fig. 8G), and nsp5 (Fig. 8I). However, it did not interact with either nsp3 or nsp5 in co-IP assays, in which it might not be in a favorable conformation for the interactions. Overall, our data suggest that nsp12 can interact with a variety of nsps (Fig. 8K), highlighting its role as an interaction hub.
PRRSV nsp12 is an unexpected interaction hub. (A to C) Co-IP analysis of nsp12 with its binding partners. HEK293 cells were cotransfected to express the indicated protein pairs. At 24 h after transfection, cell lysates were prepared and analyzed by immunoblotting to measure the input levels or subjected to further co-IP analysis. (D to J) Relocalization and colocalization analysis. BHK-21 cells were transfected to express the indicated protein pairs. At 18 to 24 h posttransfection, the cells were fixed, stained with the appropriate antibodies to epitope tags, and examined by confocal microscopy. (K) Summary of the PRRSV nsp12 interaction network.
The membrane proteins nnsp2, nsp3, and nsp5 form a subnetwork and mainly interact with ORF1b-encoded replicase proteins.The membrane proteins nsp2, nsp3, and nsp5 are thought to serve as a platform to recruit core nsps to viral RTC (18, 19). Consistent with this idea, the Y2H screening demonstrated nsp2 to mainly interact with nsp9, nsp10, nsp11, and nsp12, in addition to nsp3 (Fig. 1B and C). These interactions were also confirmed by the co-IP experiment (Fig. 4A, Fig. 7B, and Fig. 8B), except for that with nsp11 (Fig. 9E). In the transfected cells, nsp2 interacts with nsp9 and nsp10, but in an apparently conditional manner, as described above (Fig. 4 and 7), whereas it colocalized well with nsp12 (Fig. 8F) and nsp3 (Fig. 9H). However, we did not observe a colocalization relationship between nsp2 and nsp11 (Fig. 9K), an observation that was similar to the co-IP result (Fig. 9E). A recent study has identified nsp1α and nsp1β as the binding partners for nsp2 (45), but they were missing in our Y2H screen. We confirmed these interactions by co-IP assay (Fig. 9A and B). However, they did not colocalize within transfected cells (Fig. 9F and G).
Analyses of the binding partners for transmembrane proteins nsp2, nsp3, and nsp5. (A to E) HEK293 cells were cotransfected to express the indicated protein pairs. At 24 h after transfection, cell lysates were prepared and analyzed by immunoblotting to measure the input expression levels or subjected to co-IP analysis. (F to L) BHK-21 cells were transfected to express the indicated protein pairs. At 18 to 24 h posttransfection, the cells were fixed, stained with the appropriate antibodies to epitope tags, and examined by confocal microscopy. (M) Summary of the confirmed interaction network of nsp2, nsp3, and nsp5.
PRRSV nsp5 was found to interact with nsp1α, nsp7, nsp9, nsp11, and nsp12 in the Y2H system (Fig. 1B and C), but the co-IP and cotransfection assays failed to confirm these interactions, except for the above-described interaction with nsp9 (Fig. 6A). We also investigated whether nsp5 interacts with the membrane proteins nsp2 and nsp3, which were not detected in the Y2H assay. The co-IP and cotransfection experiments identified an interaction between nsp5 and nsp2 (Fig. 9D and I). Despite the fact that an interaction between nsp5 and nsp3 was not detected by the co-IP assay (data not shown), a perfect colocalization relationship could be observed in cotransfected cells (Fig. 9J), suggesting that they likely interact within intact cells. Finally, we tested whether nsp2, nsp3, and nsp5 can form a complex in cotransfected cells. The coexpression of nsp2 and nsp3 was judged by the characteristic dispersed puncta formation (Fig. 9H), and nsp5 was found to be localized to these puncta (Fig. 9L), indicating that they likely form a triple complex. The confirmed interaction relationships are summarized in Fig. 9L, and the network map suggests that the membrane proteins nsp2, nsp3, and nsp5 likely form a subnetwork and mainly interact with ORF1b-encoded core replicase proteins.
Coexpression of nsp2 and nsp3 does not enable interaction with full-length nsp9 or nsp10 in transfected cells.It is possible that efficient recruitment of nsp2 and nsp3 takes place only on replication-associated membrane structures. It has been reported that EAV nsp2 and nsp3, the PRRSV counterparts, suffice to induce double-membrane vesicles (18, 19, 24), which resemble those in infection. Now that PRRSV nsp2 and nsp3 had been shown to interact with each other (Fig. 9C and H), we tested the effect of their coexpression on the recruitment of nsp9 or nsp10. To reduce the number of plasmids in transfection, we initially used a plasmid encoding nsp2-nsp3 polyprotein for cotransfection. It was found that nsp2 was not completely processed from the precursor and displayed a rather diffuse localization pattern in the cytoplasm in a small percentage of cells (Fig. 10A). Therefore, separate plasmids were used to express the individual nsp2 and nsp3. The coexpression induced typical dispersed puncta in the cytoplasm (Fig. 9H; Fig. 10B, top row) that resembled the structures seen in PRRSV-infected cells. Moreover, this structure was different from that induced by either nsp2 or nsp3 alone (Fig. 3). In the triple-transfection assay, we judged the coexpression of nsp2 and nsp3 by the appearance of characteristic dispersed puncta structures and then looked for the relocalization of nsp9 or nsp10 or their derivatives. Unfortunately, the coexpression did not significantly increase the relocation efficiency of either the full-length nsp9 (Fig. 10B) or nsp10 (Fig. 10C) or that of the RdRp domain or the helicase domain (Fig. 10D and E). Interestingly, we observed dramatically increased colocalization with nsp9(1–229)-HA (Fig. 10B and D), although neither nsp2 nor nsp3 colocalized well with this mutant in doubly transfected cells (Fig. 4F and Fig. 5E). We also tried quadruple transfections (nsp2 plus nsp3 plus nsp9 plus nsp10), but the result was not informative due to the lack of antibodies for the triple staining (data not shown). Thus, the recruitment of the core enzymes to membrane proteins is much more complicated than we thought, and this deserves rigorous investigation in the future.
Coexpression of nsp2 and nsp3 does not increase interaction efficiency with full-length nsp9 and nsp10. BHK-21 cells were transfected to express the indicated protein pairs. At 18 to 24 h posttransfection, the cells were fixed, stained with the appropriate antibodies to epitope tags, and examined by confocal microscopy. (A) (Left) Localization pattern of nsp2 in cells expressing nsp2-nsp3 polyprotein. (Right) Processing of nsp2 and nsp3 by Western blotting with antibodies to nsp2. (B) Triple expression of nsp9 and its mutants with nsp2 and nsp3 in BHK-21 cells. (C) Triple expression of nsp10 and its derivatives with nsp2 and nsp3 in BHK-21 cells. (D and E) Quantitative analysis for relocalization and colocalization relationships in cells coexpressing nsp9 or nsp10 or its derivatives with nsp2 and nsp3. The percentages of cells showing relocalization and colocalization were determined. R, rabbit polyclonal antibodies; M, mouse monoclonal antibodies. ***, P < 0.001; ns, not significant. The error bars indicate standard deviations.
Identification of self-interacting nonstructural proteins.Oligomerization may be important for certain individual nsps to perform their functions in PRRSV infection, and such self-interactions have been identified for nsp1α, nsp1β, and nsp11 by gel filtration assay (40, 55, 56). Interestingly, our Y2H system missed nsp1β and nsp11, a failure that could be due to an adverse effect of partner fusion on the self-interaction. Instead, we identified nsp9 and nsp12 as two novel self-interacting nsps, in addition to the already reported nsp1α. These interactions were confirmed by the co-IP assay, which showed that antibodies to one tag could pull down the same protein, with different tags for all three proteins (Fig. 11A, B, and C). Thus, the number of PRRSV self-interacting nsps was expanded to at least 5: nsp1α, nsp1β, nsp9, nsp11, and nsp12.
Self-interaction analysis of PRRSV nsp1α, nsp9, and nsp12 by co-IP. (A to C) HEK293 cells were cotransfected to express the indicated protein pairs. At 24 h after transfection, cell lysates were prepared and either analyzed by immunoblotting to measure the input expression levels or subjected to co-IP analysis.
Summary of the interactome of PRRSV nsps.Based on the screening by three methods, we constructed a comprehensive interactome for PRRSV nsps (Fig. 12). The resulting map reveals a focus of ORF1b-centered interactions that are mainly connected by transmembrane proteins nsp2, nsp3, and nsp5. In addition, the map suggests that the three transmembrane proteins may serve as scaffolding proteins to recruit ORF1b-encoded core replication proteins and other components for assembly of PRRSV RTC.
Summary of the interactome of PRRSV nsps. The interactions among PRRSV nsps were summarized according to Y2H, co-IP, and colocalization/relocalization analyses. The viral proteins were grouped into different colors, and the lines represent the results from single assays or combinations of different examination assays (left), with details shown (right).
The majority of nsps are recruited to viral RTC during PRRSV infection.Having identified the nsp interactions in vitro, we examined the localization patterns of PRRSV nsps in infection. A previous study reported the association of nsp2, nsp4, nsp7, and nsp8 with viral RTC (43), and here, we tried to identify the components associated with PRRSV RTC in a more comprehensive way. To do that, MARC-145 cells were infected with the HP-PRRSV strain JXwn06 at a multiplicity of infection (MOI) of 0.1. At different times postinfection, the cells were fixed, permeabilized, and stained with a panel of antibodies to PRRSV nsps. Meanwhile, we used anti-double-stranded RNA (dsRNA) antibodies to indicate the virus replication sites. Representative images were captured with a confocal microscope. As shown in Fig. 13, the result at 24 h postinfection revealed that both nsp2 and nsp9 were colocalized well with dsRNA (Fig. 13A), indicating that both are recruited to the viral RTC. Moreover, nsp4, nsp7, nsp8, nsp10, nsp11, and nsp12 were all found at viral RTC (Fig. 13A). In contrast, nsp1α and nsp1β were mainly localized in the nucleus and rarely colocalized with nsp2 in the cytoplasm (Fig. 13A). We do not have antibodies to nsp3, nsp5, and nsp6. However, since nsp3 and nsp5 interact with nsp2, they are expected to be associated with viral RTC. Thus, the majority of nsps are recruited to viral RTC during PRRSV infection.
A majority of PRRSV nsps are recruited to viral RTC during infection. (A) MARC-145 cells grown on coverslips were infected with HP-PRRSV JXwn06 at an MOI of 0.1. At 24 h after infection, the cells were fixed and doubly stained with appropriate antibodies to PRRSV nsps or dsRNAs and examined by confocal microscopy. Representative images were taken and processed with image J. (B to D) Coimmunoprecipitation analysis of nsp2 and its binding partners from PRRSV-infected MARC-145 cells. R, rabbit polyclonal antibodies; M, mouse monoclonal antibodies.
Figure 13A also shows that PRRSV nsp2 was colocalized well with nsp9 and nsp10 in the infected cells, indicating that they come together during infection. Their interactions were also confirmed by the co-IP assay using virus-infected cell lysates (Fig. 13B, C, and D). Thus, these results present a stark contrast to the poor colocalization relationship in cotransfected cells (Fig. 3 to 7) and raise the possibility that other viral proteins may play a regulatory role in promoting the recruitment of nsp9 and nsp10 to membrane proteins during PRRSV infection.
DISCUSSION
Revelation of the nonstructural protein interaction network is a key step in understanding the assembly of PRRSV RTC. In this regard, the experiments described in this article revealed several salient findings, as follows: (i) there exists a complex network among PRRSV nsps, with interactions centered on ORF1b-encoded nsp9, nsp10, and nsp12, which are mainly connected by transmembrane proteins nsp2, nsp3, and nsp5; (ii) nsp12 is a novel interaction hub; (iii) the majority of PRRSV nsps are associated with viral RTC during infection; and (iv) the recruitment of the core enzymes nsp9 and nsp10 to membrane-associated nsps may involve a regulatory mechanism by other viral proteins during infection. The relevant insights and significance are discussed below.
Conserved interactions among arteriviruses and coronaviruses.Although employing a similar expression strategy for replication, arteriviruses are quite divergent from coronaviruses, sharing only a limited number of homologues. The proteins identified in this regard mainly include the transmembrane proteins (arterivirus nsp2, nsp3, and nsp5 versus coronavirus nsp3, nsp4, and nsp6), the main protease (arterivirus nsp4 versus coronavirus nsp5), and the core replicase proteins (arterivirus nsp9 [RdRp], nsp10 [helicase], and nsp11 [endoribonuclease] versus coronavirus nsp12 [RdRp], nsp13 [helicase], and nsp15 [endoribonuclease]). By comparative analyses of the nsp interactions between the two families according to the currently published data (20, 30, 46–48, 61, 62), the conserved interactions can be classified in the following categories: (i) complex formation among the three transmembrane proteins (arterivirus nsp2, nsp3, and nsp5 versus coronavirus nsp3, nsp4, and nsp6), (ii) interaction between viral RdRp and the helicase protein (arterivirus nsp9-nsp10 versus coronavirus nsp12-nsp13), (iii) a connection of RdRp to the transmembrane proteins (e.g., arterivirus nsp9-nsp2 versus coronavirus nsp12-nsp3), (iv) interaction of RdRp with the main protease (arterivirus nsp9-nsp4 [Y2H assay] versus coronavirus nsp12-nsp5), and (v) self-interaction of RdRp. These conserved interactions highlight the evolutionary convergence on the replication strategies at the protein-protein interaction level and point to its potentially critical role in regulating viral replication.
Insight into the interactions of nsp9 and nsp10 with membrane proteins.Our studies identified a total of 24 new interactions, and several of them were solidly confirmed by different methods (Fig. 12 and 13). However, it should be mentioned that the negative results may be due to the limitations of the assays themselves and do not mean that the interaction cannot occur during infection. As a majority of the nsps were localized to viral RTC, a more sensitive method to detect the in situ interactions in infection may be required in the future. Among the identified interactions, perhaps the most interesting are those taking place between the viral core enzymes (nsp9 and nsp10) and the membrane proteins (nsp2, nsp3, and nsp5). The full-length nsp9 interacted well with nsp2, nsp3, and nsp5 in co-IP assays, but they were poorly capable of finding each other within mammalian cells. This is not due to the potential effect of the epitope tag at the C terminus, as similar results could be observed for the untagged version of nsp9 in both BHK-21 and MARC-145 cells (data not shown). The discrepancy is likely attributable to the altered conformation of nsp9 arising from an artificial condition, the oxidative and detergent-containing environment in vitro, leading to exposure of the interaction sites to binding partners, whereas within intact mammalian cells nsp9 exists in its native form. Rigorous proof of this hypothesis requires a forced genetics approach by which the N-terminal region [the truncation fragments nsp9(230–685) and nsp9(431–685)] was removed. When coexpressed, the nsp9 RdRp-containing fragment readily translocated from nucleus to cytoplasm and became colocalized with nsp2, nsp3, or nsp5. These interactions were also confirmed by the co-IP assay (Fig. 4 to 6). In contrast, the N-terminal fragment of nsp9 [nsp9(1–229)] did not respond well to these proteins (Fig. 4 to 6). A similar scenario was also found between nsp10 and its membrane-associated binding partners nsp2, nsp3, nsp5, and nsp12 (Fig. 7). In this case, activation of the interactions appears to require deletion of the N-terminal ZBD and the 1B domain (Fig. 7).
Although it is possible that the N-terminal deletion can result in the creation of a totally new and irrelevant binding activity within the remaining C-terminal sequence of nsp9 or nsp10, we think this is a very unlikely scenario because of the high efficiency with which the truncated protein could find its binding partners within the vast expanses of the cells, even leading to its exclusion from the nucleus. For the interaction between nsp10 and nsp2, we actually observed partial colocalization in transfected cells (Fig. 7I, top row), while the truncation of nsp10 could further increase its interaction efficiency with nsp2 (Fig. 7N). Moreover, the interactions were verified by the co-IP assay (Fig. 7C and D). Thus, the simplest interpretation of these results is that the N termini of nsp9 and nsp10 serve as negative regulators for the interactions. Deletion of the N-terminal domain of nsp9 or nsp10 might have caused the structural perturbation at the level of possible refolding or destabilization or posttranslational modifications of the RdRp and HEL domains, thereby exposing the binding sites.
The N-terminal region likely has an effect on the structure of the C-terminal domain through intramolecular interactions. At least for EAV nsp10, structural studies have shown that the ZBD interacts with the helicase domain (37). Moreover, the regulatory domain 1B interacts with the subunits of the helicase domain to form an RNA-binding channel (37). PRRSV nsp10 likely possesses a similar property, according to structure modeling. Thus, it is possible that removal of the PRRSV ZBD affects the structure of the HEL domain. A strong piece of evidence stems from the phenotype of nsp10(128–441) and nsp10(101–441). As opposed to nsp10(128–441), addition of a partial 1B to the HEL domain [the truncation construct nsp10(101–441)] significantly increased interaction efficiency with membrane proteins nsp2, nsp3, and nsp5 (Fig. 7N, O, and P). This result indicates that 1B can exert an effect on the HEL domain to promote the interaction, which is also consistent with the finding that EAV nsp10 1B interacts with the HEL domain (37). For nsp9, it is expected that there is cross talk between the NiRAN domain and RdRp (63). We actually have some evidence to support this possibility (unpublished data). The resolution of the crystal structure of PRRSV nsp9 may provide direct evidence for the possibility.
The inability of the full-length nsp9 and nsp10 to interact with membrane proteins within mammalian cells may also be attributed to the “incorrect” membrane environment that prevents interactions from taking place. It is known that arteriviruses induce specific membrane structures in infection (24), and coexpression of nsp2 and nsp3 is sufficient to induce similar membrane structures (18, 24). We tested this possibility by triple expression, but unfortunately, an increase in the relocalization of nsp9 or nsp10 to nsp2/nsp3-induced structures was not apparent. However, it should be mentioned that the membrane protein nsp5 is also part of the network, and induction of the authentic viral membrane structures may need the involvement of this molecule. The unavailability of antibodies to do triple or quadruple staining prevents further exploration.
This mechanism of interaction between nsp9 or nsp10 and PRRSV membrane proteins likely resembles that found in herpes simplex virus (HSV). We have previously reported the coordinated assembly of tegument proteins UL11, UL16, and UL21 on the cytoplasmic tail of glycoprotein E (gE) of HSV-1 (54). It was found that the gE tail interacts well with UL16 in the in vitro binding assay, but in cotransfected cells, the full-length gE does not colocalize with UL16 unless the C-terminal half of UL16 is removed (52). A regulated interaction also occurs between UL16 and the membrane-associated UL11 (53). We further found that the gE-UL16 interaction is normally enabled by UL11 whereas the UL16-UL11 interaction is activated by UL21 (54). In the case of PRRSV, despite the poor colocalization relationship in doubly transfected cells, PRRSV nsp2 exhibited a perfect colocalization relationship with nsp9 and nsp10 (Fig. 13A) in virus-infected cells. Moreover, these interactions could be confirmed by the co-IP assay using virus-infected cell lysates (Fig. 13B, C, and D). Thus, it is conceivable that other single viral nsps or combinations or nsp cleavage intermediates (e.g., nsp3-4, nsp5-7, nsp5-8, and nsp3-8) (43) may normally serve as the switch for the intermolecular interactions. Currently, such efforts are under way and will be fundamental to understanding the assembly of PRRSV RTC.
Insight into nsp12-centered interactions.PRRSV nsp12 is a mysterious protein that has a predicted size of about 150 amino acids; it is arterivirus specific but poorly conserved in the primary sequences among arteriviruses (42). A recent report proposed that nsp12 may be equivalent to coronavirus nsp16, which specifies the function of 2′-O-methyltransferase; however, the results of in vitro studies failed to support such a hypothesis (42). Our studies revealed that nsp12 is an interaction hub connecting other PRRSV nsps (Fig. 8), including the transmembrane proteins nsp2, nsp3, and nsp5 and the core enzymes nsp9 to nsp11, as well as nsp6, nsp1α, and nsp1β. This discovery is intriguing; connection to so many nsps suggests that the molecule likely plays a critical role in PRRSV replication.
nsp12 is likely critical for recruiting nsp11 to RTC during RNA synthesis. Our assays showed that it had a robust interaction with nsp11 in transfected cells, in which it induced massive relocalization of nsp11 from the nucleus to the cytoplasm. In addition, nsp12 interacts with nsp9 without regulation by viral factors. On the other hand, nsp11 appears to have a limited number of binding partners. The Y2H screening revealed 5 binding partners (nsp2, nsp5, nsp6, nsp9, and nsp12) for nsp11, but only the interaction with nsp12 was confirmed by co-IP and colocalization analyses. However, as an endoribonuclease, the biological function of nsp11 in viral RNA synthesis has remained unclear. Studies of EAV nsp11 imply that it is critical for viral replication, as mutation of certain conserved sites either is lethal to the virus or reduces virus yields up to 5 log units (64). For PRRSV nsp11, we showed that nsp11 is associated with viral replication sites (Fig. 13). Based on the nsp interactome, we propose that nsp11 is likely recruited to RTC via nsp12 to perform a yet-to-be-determined function.
Potential scaffolding role for membrane proteins nsp2, nsp3, and nsp5.Replication of positive-stranded RNAs involves membrane modulation that is often executed by virus-encoded transmembrane proteins. In coronaviruses, the membrane proteins nsp3, nsp4, and nsp6 are important for virus-induced membrane modifications (20), whereas in the family Arteriviridae, EAV nsp2 has been shown to interact with nsp3 to induce extensive membrane webs when coexpressed alone in mammalian cells (18). Also, it has been proposed that these membrane proteins also serve as membranous scaffolding proteins for the core enzymes (e.g., RdRp and helicase) for RTC assembly (21, 24). For PRRSV, we know little in this regard. Our analyses of the nsp interaction network (Fig. 13) revealed several important pieces of information: (i) there is a subnetwork among transmembrane proteins nsp2, nsp3, and nsp5; (ii) the key binding partners for nsp2, nsp3, and nsp5 are oriented toward the ORF1b-encoded replicase proteins; and (iii) the recruitment of the key replicase proteins nsp9 and nsp10 to membrane proteins does not occur in a straightforward manner. Thus, these data support a model in which nsp2, nsp3, and nsp5 form a complex to modulate intracellular membranes and then serve as membrane anchors to further recruit core replicase proteins to the replication site. Moreover, this recruitment likely operates in a regulated manner to ensure the orchestrated loading of RTC components.
PRRSV nsp4, nsp6, nsp7, and nsp8.PRRSV nsp4 performs as a 3C-like main protease that has multiple functions in the virus life cycle (65–68). It plays important roles in antagonizing host antiviral immunity and in the maturation of replicase polyproteins pp1a and pp1ab (65–70). According to EAV replicase processing, PRRSV nsp4 was supposed to mediate 9 cleavages in the nsp3-to-nsp12 region (69, 70). Moreover, it is associated with the viral RTC (Fig. 13), suggesting a role in viral RNA synthesis (43). Unfortunately, we did not screen out the interaction partners for nsp4 with high confidence. The Y2H screening results indicated that nsp4 could interact with nsp5, nsp9, and nsp12, but it could not be confirmed by the assays, such as colocalization analysis. It is possible that their interactions involve complicated regulation or that nsp4 is recruited to viral RTC as cleavage intermediates (e.g., nsp3-nsp4). PRRSV nsps6-nsp8 has the fewest binding partners. Moreover, nsp6 and nsp8 are predicted to be very small peptides. We do not have antibodies to nsp6. Immunostaining of nsp8 showed that it was colocalized with nsp2, suggesting that it is recruited to viral RTC during infection. For nsp7, we expressed nsp7α- and nsp7β-coding regions as a single gene, and this may have had an adverse effect on detection of interactions with other potential binding partners in the Y2H screening. Nevertheless, by Y2H assay, we could detect an interaction of nsp7 with nsp9, consistent with a previous report that nsp9 is a binding partner of nsp7α (44). In any case, the interaction network of these three proteins and their biological functions are worth further exploration in the future.
MATERIALS AND METHODS
Cells, viruses, and commercial antibodies.HEK293 cells, BHK-21 cells, and MARC-145 cells were grown in Gibco Dulbecco’s modified Eagle medium (DMEM) (Invitrogen, CA, USA) containing 10% fetal bovine serum (FBS) (Invitrogen, CA, USA) at 37°C with 5% CO2. The type II HP-PRRSV strain JXwn06 was used in the study (38, 71). Infected MARC-145 cells were grown in DMEM supplemented with 2% FBS.
All the restriction enzymes were purchased from New England Biolabs Inc. (Beverly, MA, USA). The plasmids pGADT7, pCMV-HA-C, and pCMV-Myc were obtained from Clontech (Mountain View, CA, USA). Rabbit anti-HA monoclonal antibody (MAb) (3724S) was purchased from Cell Signaling Technology (Danvers, MA, USA). Rabbit anti-FLAG polyclonal antibody (PAb) (F7425), mouse anti-FLAG MAb (F1804), mouse anti-HA MAb (H3663), and rabbit anti-Myc PAb (C3956) were all purchased from Sigma-Aldrich (St. Louis, MO, USA). Horseradish peroxidase (HRP)-conjugated goat anti-mouse PAb and HRP-conjugated goat anti-rabbit PAb were purchased from ZSGB-Bio (Beijing, China). Alexa Fluor 488-conjugated goat anti-rabbit IgG(H+L) F(ab′)2 fragment, Alexa Fluor 568-conjugated goat anti-rabbit IgG(H+L) F(ab′)2 fragment, Alexa Fluor 488-conjugated goat anti-mouse IgG(H+L) F(ab′)2 fragment, and Alexa Fluor 568-conjugated goat anti-mouse IgG(H+L) F(ab′)2 fragment were all purchased from Molecular Probes (Invitrogen, CA, USA).
PRRSV nsp-specific antibodies.The mouse MAbs specific for PRRSV nsp1β, nsp2 (E3G11), nsp4, nsp9, nsp10 (4D9), and nsp11 (3F9) were prepared in our laboratory and have been reported previously (72–76). Mouse anti-nsp12 MAb (1E5) was kindly provided by Changjiang Weng (Harbin Veterinary Research Institute) (57). The rabbit polyclonal antibodies against PRRSV nsp1α, nsp2, nsp7, and nsp8 were prepared in our laboratory. The nsp1α antibodies were raised against nsp1α-His6 that was purified from Escherichia coli BL21 cells. The nsp2 polyclonal antibodies were obtained by immunizing rabbits with immunogen containing the nsp2 protease domain (aa 12 to 323) with a His6 tag that was expressed and purified from E. coli BL21 cells. nsp7 and nsp8 were expressed as a fusion protein with GST in E. coli BL21 cells, and the purified recombinant GST-nsp7 and GST-nsp8 were used as immunogens. The nsp9 antibodies were made by using the ORF1b-coded portion of nsp9 (aa 46 to 685) tagged with StrepII [nsp9(46–685)-StrepII] as an immunogen that was expressed and purified from E. coli BL21 cells. The reactivities and specificities of the respective antibodies were tested under conditions of transfection and infection, while mock-transfected or mock-infected cells served as a negative control. Normal rabbit serum was also used as a negative control. All the antibodies showed good reactivity and specificity.
Yeast two-hybrid screen.Yeast two-hybrid screening was performed as previously described (77). Briefly, the Matchmaker Gold yeast two-hybrid system (Clontech, Mountain View, CA) was used according to the manufacturer’s instructions. The genes coding for the individual nsps (Table 1) were cloned into vectors pGBKT7 and pGADT7 to generate bait and prey plasmids. The autoactivation and toxicity of bait fusion vectors were tested in the Y2H Gold yeast strain (Clontech, Mountain View, CA) according to the manufacturer’s instructions. To screen protein-protein interactions, the respective bait (pGBKT7) and prey (pGADT7) plasmids were cotransformed into the Y2H Gold yeast strain, and cotransformants were selected on high-stringency quadruple-dropout (QDO) plates. The positive colonies were reselected on high-stringency QDO plates containing 0.04 mg/ml 5-bromo-4-chloro-3-indolyl-α-d-galactopyranoside (X-α-Gal) (Clontech, Mountain View, CA) and 0.07 g/ml aureobasidin A (AbA) (Clontech, Mountain View, CA). Y2H Gold yeast cells transformed with pGBKT7-53 (Clontech, Mountain View, CA) and pGADT7-T (Clontech, Mountain View, CA) served as positive controls, and cotransformation with pGBKT7-Lam (which encodes the Gal4 DNA binding domain [BD] fused with lamin) and pGADT7-T served as the negative control. All the plates were incubated at 30°C for 2 to 5 days to observe the colony colors. Clones that appeared as blue colonies on QDO–X-α-Gal–AbA agar plates were considered positive results. All QDO–X-α-Gal–AbA-positive interactions were assessed to identify duplicates and to verify that the interactions were genuine.
Plasmid construction.The genes coding for the nonstructural proteins and the corresponding mutants were cloned from HP-PRRSV strain JXwn06 and engineered into the eukaryotic vectors pCMV-Myc and pGADT7. All the nonstructural proteins with HA-tagged, FLAG-tagged, and pGBKT7 bait plasmids were prepared in our laboratory. The proteins were tagged with epitope tags (e.g., HA and FLAG) at the N or C terminus. The plasmid pNsp9-HA was a derivative of the plasmid pCMV-HA-C, which encodes the full-length nsp9 with an HA tag at its C terminus. pSrc-nsp9-HA is a derivative of pNsp9-HA that has the Src membrane binding peptide (MGSSKSKPKDAL) at the N terminus of nsp9. The plasmid pSrc-nsp10-HA was made in a similar way. All PRRSV genes in the created plasmids are under the control of the cytomegalovirus (CMV) promoter, and a Kozak core sequence is also included to allow optimal expression. All the plasmids were constructed by standard molecular biological techniques and confirmed by sequencing. The sequence information for individual nsps in this study is provided in Table 1.
Confocal microscopy.BHK-21 cells grown on coverslips in 6-well plates were cotransfected with Myc-tagged, HA-tagged, or FLAG-tagged nonstructural proteins using Lipofectamine 2000 DNA transfection reagent (Invitrogen, CA, USA). MARC-145 cells grown on coverslips in 6-well plates were infected with PRRSV JXwn06 at a multiplicity of infection (MOI) of 0.1. At 18 to 24 h posttransfection or postinfection, the cells were fixed with 3.7% paraformaldehyde for 10 min at room temperature (RT), washed with phosphate-buffered saline (PBS) three times (5 min for each wash), permeabilized with 0.1% Triton X-100–2% bovine serum albumin (BSA) for 10 min, and blocked with 2% BSA-PBS for 30 min (RT). The cells were then incubated with the proper primary antibodies for 1 h in a humid chamber (RT) and then washed with PBS three times (5 min for each wash). After that, the cells were incubated with the appropriate secondary antibodies, including Alexa Fluor 488-conjugated goat anti-rabbit IgG(H+L) F(ab′)2 fragment, Alexa Fluor 568-conjugated goat anti-rabbit IgG(H+L) F(ab′)2 fragment, Alexa Fluor 488-conjugated goat anti-mouse IgG(H+L) F(ab′)2 fragment, and Alexa Fluor 568-conjugated goat anti-mouse IgG(H+L) F(ab′)2 fragment, for another 1 h (RT) and then washed with PBS three times (5 min for each wash). Nuclear DNA was stained with 4′,6-diamidino-2-phenylindole (DAPI) (Molecular Probes). The images were captured under a Nikon A1 confocal microscope and processed using Image J. The primary antibodies used included mouse anti-HA (1:1,000; Sigma), rabbit anti-Myc (1:1,000; Sigma), mouse anti-FLAG (1:1,000; Sigma), rabbit anti-FLAG (1:1,000; Sigma), rabbit anti-HA (1:2,000; CST), nsp1α PAb (1:1,000), nsp1β MAb (1:1,000), nsp2 MAb (1:1,000), nsp2 PAb (1:3,000), nsp4 MAb (1:1,000), nsp7 PAb (1:2,000), nsp8 PAb (1:2,000), nsp9 MAb (1:2,000), nsp9 PAb (1:2,000), nsp10 MAb (1:2,000), nsp11 MAb (1:1,000), nsp12 MAb (1:1,000), and mouse anti-dsRNA (1:200; SCICONS). All the secondary antibodies were used at a dilution of 1:1,000.
Coimmunoprecipitation.HEK293 cells were grown on 6-well plates at a confluence of 70% and cultured for 24 h before cotransfection with Myc-tagged, HA-tagged, or FLAG-tagged plasmids using Lipofectamine 2000 DNA transfection reagent (Invitrogen, CA, USA), and MARC-145 cells were infected with PRRSV JXwn06 at an MOI of 0.1. At 24 h posttransfection or postinfection, the cells were lysed using NP-40 buffer (50 mM Tris, 150 mM NaCl, 0.5% NP-40, 0.5 mM EDTA) containing 1 mM phenylmethylsulfonyl fluoride (PMSF) and 1 mg/ml protease inhibitor cocktail (p8340; Sigma) for 30 min at 4°C with gentle rotation. The total cell lysates were clarified by centrifugation at 12,000 rpm for 20 min. The cell supernatants were then precleared with protein A/G agarose beads (Santa Cruz, CA, USA) and immunoprecipitated using rabbit anti-Myc PAb (C3956; Sigma), rabbit anti-FLAG PAb (F7425; Sigma), or mouse anti-HA MAb (H3663; Sigma) or mouse MAbs to PRRSV nsp2 or nsp9 in conjunction with protein A/G beads (SC-2003; Santa Cruz) overnight at 4°C with rotation. The immunoprecipitation pellets were collected by centrifugation and gently washed three times with NP-40 buffer, boiled in 5× loading buffer for 5 min, and subjected to SDS-PAGE for Western blot analysis.
Western blotting.Proteins were fractionated by SDS-PAGE and then blotted onto polyvinylidene difluoride (PVDF) membranes (Millipore, USA) before being blocked overnight at 4°C with 5% nonfat milk in PBS. The membranes were probed with specific primary antibodies for 8 h (RT). After being washed with washing buffer (0.1% Tween 20 in PBS) six times (5 min for each wash), the membranes were probed with anti-mouse or anti-rabbit secondary antibodies conjugated with horseradish peroxidase (ZSGB-Bio, Beijing, China), which were diluted in PBS containing 5% nonfat milk. After being washed six times (5 min for each wash), membrane-bound antibodies were detected using enhanced chemiluminescence (ECL) detection reagents (Vigorous, Beijing, China). The primary antibodies used were as follows: mouse anti-HA (1:5,000; Sigma), rabbit anti-Myc (1:5,000; Sigma), mouse anti-FLAG (1:5,000; Sigma), nsp2 PAb (1:1,000), nsp9 MAb (1:1,000), nsp10 MAb (1:1,000), and nsp12 MAb (1:1,000). Horseradish peroxidase-conjugated goat anti-mouse secondary antibodies and goat anti-rabbit secondary antibodies were used at a dilution of 1:10,000.
Bioinformatics prediction.The I-TASSER online service tool was used to model the structures of PRRSV strain JXwn06 nsp9 and nsp10 (59). The nsp9 and nsp10 structures were referred to poliovirus precursor protein 3CD (Protein Data Bank [PDB] ID 2IJD) and EAV nsp10 (PDB ID 4N0N), respectively, and then processed using SPDBV software (SPDBV version 4.0.1).
Statistical analysis.Statistical significance was evaluated by using one-way analysis of variance (ANOVA). Statistical analyses were performed using GraphPad Prism software (version 5.0).
ACKNOWLEDGMENTS
This work was supported by the National Natural Science Foundation of China (31472189), a National Key Basic Research Plan Grant from the Chinese Ministry of Science and Technology (2014CB542700), the China National One-Thousand Youth Talents Program (1051-21986001), and the earmarked fund for the China Agriculture Research System (CARS-35) from the Chinese Ministry of Agriculture.
FOOTNOTES
- Received 26 June 2018.
- Accepted 25 September 2018.
- Accepted manuscript posted online 3 October 2018.
- Copyright © 2018 American Society for Microbiology.
