Interferon-Inducible Oligoadenylate Synthetase-Like Protein Acts as an Antiviral Effector against Classical Swine Fever Virus via the MDA5-Mediated Type I Interferon-Signaling Pathway

ABSTRACT Classical swine fever virus (CSFV) is the causative agent of classical swine fever (CSF), which poses a serious threat to the global pig industry. Interferons (IFNs) and IFN-stimulated genes (ISGs) play a key role in host antiviral defense. We have previously screened the porcine 2′-5′-oligoadenylate synthetase-like protein (pOASL) as a potential anti-CSFV ISG using a reporter CSFV. This study aimed to clarify the underlying antiviral mechanism of pOASL against CSFV. We confirmed that CSFV replication was significantly suppressed in lentivirus-delivered, pOASL-overexpressing PK-15 cells, whereas silencing the expression of endogenous pOASL by small interfering RNAs markedly enhanced CSFV growth. In addition, the transcriptional level of pOASL was upregulated both in vitro and in vivo upon CSFV infection. Interestingly, the anti-CSFV effects of pOASL are independent of the canonical RNase L pathway but depend on the activation of the type I IFN response. Glutathione S-transferase pulldown and coimmunoprecipitation assays revealed that pOASL interacts with MDA5, a double-stranded RNA sensor, and further enhances MDA5-mediated type I IFN signaling. Moreover, we showed that pOASL exerts anti-CSFV effects in an MDA5-dependent manner. In conclusion, pOASL suppresses CSFV replication via the MDA5-mediated type I IFN-signaling pathway. IMPORTANCE The host innate immune response plays an important role in mounting the initial resistance to viral infection. Here, we identify the porcine 2′-5′-oligoadenylate synthetase-like protein (pOASL) as an interferon (IFN)-stimulated gene (ISG) against classical swine fever virus (CSFV). We demonstrate that the anti-CSFV effects of pOASL depend on the activation of type I IFN response. In addition, we show that pOASL, as an MDA5-interacting protein, is a coactivator of MDA5-mediated IFN induction to exert anti-CSFV actions. This work will be beneficial to the development of novel anti-CSFV strategies by targeting pOASL.

Members of the OAS family exert antiviral activity via the canonical OAS/RNase L-dependent or -independent antiviral pathway. The OAS/RNase L pathway has been studied extensively (10,14). Activation of RNase L by the OAS family requires the OAS oligomerization unit and the processivity of synthesizing di-, tri-, and tetrameric 2=-5=oligoadenylates (2-5A), which in turn bind and activate RNase L. Upon binding to 2-5A, RNase L dimerizes and degrades cellular and viral RNAs, resulting in the reduction of protein synthesis (15,16). Moreover, some studies indicate that some members of the OAS family exhibit antiviral activity via several RNase L-independent pathways (12,17,18). For instance, human OASL (hOASL) exerts antiviral effects by enhancing RIG-Imediated signaling by mimicking polyubiquitin (12).
This study attempted to elucidate the role of pOASL in enhancing MDA5-mediated signaling and to clarify the mechanism by which pOASL exerts anti-CSFV activity. We have revealed a model whereby pOASL, induced upon CSFV infection, binds to MDA5 and functions as an anti-CSFV ISG via the MDA5-mediated type I IFN-signaling pathway.

RESULTS
pOASL inhibits CSFV replication. To explore the antiviral activity of pOASL against CSFV, we established PK-pOASL and PK-EGFP cells stably expressing pOASL or enhanced green fluorescent protein (EGFP), respectively. The cell growth and viability of PK-pOASL and PK-EGFP cells were indistinguishable from those of the parent PK-15 cells (Fig. 1A). We have demonstrated that overexpression of pOASL inhibits rCSFV-Fluc replication (19). The anti-CSFV activity of pOASL was examined in PK-pOASL and PK-EGFP cells upon infection with the parental CSFV strain Shimen at multiplicities of infection (MOI) of 0.1 and 1. The extracellular titers of the progeny virus (Fig. 1B), the number of viral genomic copies (Fig. 1C), and the intracellular expression level of the N pro protein (Fig. 1D) were significantly reduced in PK-pOASL cells compared to those in PK-EGFP cells at 24 or 48 h postinfection (hpi). However, the suppression of CSFV replication by OASL is different between MOI of 0.1 and 1. Considering that pOASL exerts antiviral effects against CSFV, we tried to investigate the antiviral kinetics of pOASL against CSFV replication. The intracellular viral genomic copy numbers of CSFV in infected cells were quantified at different time points postinfection. The results showed that the inhibition of CSFV at an MOI of 1 by pOASL reached a peak at 12 hpi and decreased thereafter (Fig. 1E).
To further examine the antiviral effects of pOASL on CSFV, three specific small interfering RNAs (siRNAs) were used to silence the expression of endogenous pOASL in PK-15 cells, resulting in efficient reduction of pOASL expression ( Fig. 2A and B). Subsequently, the replication efficiency of rCSFV-Fluc or Shimen was assessed upon silencing pOASL. Compared with the nontargeting control siRNA (siNC)-transfected control cells, the intracellular firefly luciferase (Fluc) activities and the expression of the N pro protein were increased in PK-15 cells transfected with a mixture of three siRNAs targeting pOASL (sipOASLs) (Fig. 2C and F). Similarly, the extracellular viral genomic copy numbers and viral titers were increased significantly in these cells ( Fig. 2E and D). The results demonstrate that knockdown of pOASL promotes CSFV replication.
pOASL is activated upon CSFV infection. Since pOASL is able to inhibit CSFV replication, we attempted to examine the effects of CSFV infection on the expression of pOASL. PK-15 cells were infected with Shimen, and the transcription of pOASL was determined by real-time quantitative reverse transcription-PCR (qRT-PCR). As a control, PK-15 cells were treated with different doses of beta interferon (IFN-␤). In accord with predictions, IFN-␤ activated the expression of pOASL in a dose-dependent manner (Fig.  3A). Similarly, the transcription level of pOASL was increased in the PK-15 cells infected with Shimen at an MOI of 0.1 at 3, 6, 12, 24, 36, and 48 hpi (Fig. 3B), as well as at MOI of 0.1 and 1 at 6 hpi (Fig. 3C). However, the expression of pOASL was lower at 48 hpi than at 36 hpi at an MOI of 0.1 and lower at an MOI of 1 than at an MOI of 0.1 at 6 hpi. We also showed that pOASL mRNA was expressed in all tested CSFV-infected organs, including heart, liver, spleen, lung, kidney, tonsils, and lymph nodes. High-level expression of pOASL was detected in liver, spleen, tonsils, and lymph nodes (Fig. 3D).
The anti-CSFV effects of pOASL are independent of RNase L. OAS are able to synthesize 2-5A, which leads to RNA degradation by activating RNase L (22). Considering that pOASL has an N-terminal OAS-like domain, we speculated that its antiviral activity likely depends on RNase L. We knocked down the RNase L expression in PK-15 cells using a mixture of three siRNAs targeting RNaseL (siRNaseLs) ( Fig. 4A and B). In both PK-EGFP and PK-pOASL cells, knockdown of RNase L resulted in increased luciferase activities (Fig. 4C), extracellular viral titers (Fig. 4D), and viral genomic copy numbers (Fig. 4E) compared to those in siNC-transfected cells after infection with rCSFV-Fluc or Shimen. When RNase L was silenced, however, pOASL still exerted anti-CSFV actions, as demonstrated by lower viral genome RNA levels and virus yields in PK-pOASL than in PK-EGFP cells. These data suggested that the anti-CSFV activity of pOASL is independent of the RNase L pathway.
pOASL enhances the type I IFN response. The results presented above reveal that the anti-CSFV activity of pOASL is independent of the classical RNase L pathway. Hence, whether pOASL exerts antiviral actions by an alternative mechanism needs to be investigated. It has been documented that several ISGs exert antiviral effects via various signaling pathways (12,23,24). To define the signaling pathway affected by pOASL, we conducted a dual-luciferase reporter assay. The results indicated that pOASL increased the promoter activities of IFN-␤ ( pOASL interacts with the RNA sensor MDA5. It has been reported that hOASL interacts with RIG-I and enhances RIG-I-mediated type I IFN signaling (12). CSFV has been verified to trigger the RIG-I-and MDA5-dependent signaling pathway (25). Indeed, our data also demonstrated that pOASL activates the type I IFN response (Fig. 5). Based on the above-mentioned results, we hypothesized that pOASL plays a role in RIG-I-or MDA5-mediated signaling. The interaction between pOASL and porcine RIG-I (pRIG-I) or MDA5 (pMDA5) was investigated using a coimmunoprecipitation (co-IP) assay. The results showed that the Flag-tagged pOASL interacted with the Myc-tagged pMDA5, but not with the Myc-tagged pRIG-I ( Fig. 6A and B). Furthermore, the Myc-tagged pMDA5 was shown to coimmunoprecipitate with the Flag-tagged pOASL, but not the Flag-tagged pOAS1 (Fig. 6C). Glutathione S-transferase (GST) pulldown assay further was examined by Western blotting (WB) using a rabbit anti-N pro PAb (1:500). GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was included as an internal reference control. N pro expression in the cell lysates was evaluated using Odyssey application software, version 3.0 (YQ09-0052; Li-Cor, USA). (E) Anti-CSFV activity of pOASL at early times after viral infection. PK-pOASL or PK-EGFP cells were infected with CSFV at a multiplicity of infection of 1 and harvested at 0, 3, 6, 9, 12, 15, and 18 h postinfection. The intracellular genomic copy numbers of CSFV were quantified by real-time RT-PCR. Each sample was run in triplicate. The error bars represent standard deviations. *, P Ͻ 0.05; **, P Ͻ 0.01. confirmed that GST-pOASL, but not GST, interacted with pMDA5 (Fig. 6D). Furthermore, the pOASL-pMDA5 interaction was confirmed by confocal assay (Fig. 6E). To exclude the involvement of RNA in the pOASL-pMDA5 interaction, the cell lysates were treated with RNases A and T1, followed by co-IP. The results indicated that the interaction of pOASL and pMDA5 was independent of RNA ( Fig. 6F).
pOASL enhances the MDA5-mediated type I IFN response. Because pOASL interacts with pMDA5 and activates the IFN-␤ signaling pathway, we endeavored to investigate whether the action of pOASL is mediated via the pMDA5 pathway. We tried to determine whether pOASL coexpressed with pMDA5 affects the transcription of IFN-␤ and ISGs in PK-15 cells. The results showed that the coexpression of pOASL with pMDA5, but not with pRIG-I, markedly upregulated the mRNA transcriptional levels of IFN-␤, myxovirus resistance protein 1 (Mx1), and guanylate-binding protein 1 (GBP1) (Fig. 7A to C). Collectively, these studies support the idea that pOASL positively regulates pMDA5-mediated type I IFN signaling.
Antiviral activity of pOASL is abolished in MDA5 knockdown cells. MDA5 is one of the primary sensors of viral RNA for CSFV infection (25). In light of the fact that pOASL displayed antiviral activity against CSFV depending on pMDA5-mediated type I IFN signaling (Fig. 7), we silenced pMDA5 expression using siRNAs to further confirm the results. Marked reduction of pMDA5 expression was observed in the pMDA5-silenced cells (Fig. 8A). As expected, significant reduction in CSFV replication was observed in the siNC-transfected PK-pOASL cells. Upon silencing of pMDA5, however, pOASL almost lost its ability to inhibit CSFV replication in PK-pOASL cells, as assessed by the viral genomic copy numbers and the viral titers ( Fig. 8B and C). In contrast, in the cells with knockdown of pRIG-I (Fig. 8D), the anti-CSFV activity of pOASL was not affected (  and F). These data indicate that the anti-CSFV activity of pOASL is dependent on pMDA5.

DISCUSSION
In this study, we clarified the antiviral mechanism by which pOASL binds to pMDA5 and enhances pMDA5-mediated antiviral signaling. We showed that pOASL efficiently suppressed CSFV growth in PK-15 cells ( Fig. 1 and 2). Interestingly, we found that pOASL expression was activated in vitro at an earlier phase after viral infection and then declined (Fig. 3) and that the anti-CSFV effects of pOASL depend on enhancement of the type I IFN response, but not on the canonical RNase L pathway ( Fig. 4 and 5). Notably, we demonstrated that pOASL interacts with pMDA5 and enhances pMDA5mediated type I IFN antiviral signaling to antagonize the replication of CSFV (Fig. 6 to  8). Taken together, these findings suggest that pOASL is an anti-CSFV ISG via the pMDA5-mediated type I IFN-signaling pathway.
The IFN-inducible OASs are important components of the antiviral activity of IFNs and are involved in other cellular processes, including cell growth, apoptosis, gene regulation, cell differentiation, RNA splicing, and DNA replication (26,27). In humans, there are three functional OAS genes (OAS1 to OAS3), including 8 to 10 OAS isoforms due to alternative mRNA splicing (28). In mice, in addition to OAS2 and OAS3, there are eight OAS1 genes (29,30). The pig and cattle genomes contain three OAS genes, OAS1, OAS2, and OASL. The human OASL gene has one OAS unit and two ubiquitin-like repeats on the C terminus (31). While two tandemly duplicated OASL genes have been identified in the dog genome, only a single OASL ortholog has been found in the cattle or pig genome (26,32). The porcine and bovine OASL genes contain a premature stop codon, resulting in truncated proteins that lack the typical C-terminal ubiquitin domains. Indeed, we demonstrated that the porcine OASL gene contains only a single OAS domain (33). Viral infection usually upregulates the expression of ISGs, which inhibit viral replication. It has been demonstrated that both OAS1 and OASL mRNA levels peak around 10 h after influenza A virus infection and then decline, and the decline observed in virus-infected cells was probably due to the synthesis of viral inhibitors of the innate immune system, such as the influenza A virus NS1 protein (34,35). Similarly, our data also showed that the mRNA level of pOASL peaked around 12 h after infection and then declined subsequently. Specifically, the mRNA level of pOASL was lower at 48 hpi than at 24 hpi at an MOI of 0.1 (Fig. 3B) and lower at an MOI of 1 than at an MOI of 0.1 (Fig.  3C) at 6 hpi. A possible explanation is that CSFV has the ability to escape the antiviral action of pOASL. It has been shown that OASL is rapidly induced upon viral infection through interferon regulatory factor 3 (IRF3), as well as IFN signaling (11,36), and that CSFV N pro interacts with IRF3 and induces its proteasomal degradation to prevent type I IFN induction (34,37). Thus, high doses of CSFV may increase the chance of IRF3 degradation, leading to decreased expression of pOASL.
With hundreds of ISGs induced by IFN, we assume that any step of the viral life cycle (entry, uncoating, transcription, translation, assembly, or egress) could be targeted for inhibition (38). The OAS family are well-known molecules that regulate the early phase of viral infection (39), and the mouse Oas1b acts mainly on the replication stage of West Nile virus (WNV) (17). In this study, we investigated the antiviral kinetics of pOASL on CSFV replication. The data demonstrated that the viral genomic RNA copy numbers were decreased in PK-pOASL cells compared to those in PK-EGFP cells from 6 to 12 hpi (Fig. 1E). However, our experiments do not define the exact stage(s) of the CSFV life cycle that pOASL targets. Thus, we need to further investigate the exact stage of CSFV replication that pOASL targets using different methods in the future.
hOASL interacts with RIG-I and enhances RIG-I-mediated antiviral signaling (12). Similar to hOASL, mouse Oasl2 exhibits antiviral activity, and its loss results in the reduction of IFN signaling and enhances RNA virus replication (12). However, it has been demonstrated that mouse Oasl1 binds to the 5= untranslated region (UTR) of IRF7 to inhibit IRF7 translation, thus reducing IFN induction and increasing viral replication (40). In this study, we showed that the interaction of pOASL with pMDA5 enhances the pMDA5-mediated type I IFN-signaling pathway, which increases IFN induction and efficiently inhibits CSFV replication. The OAS proteins bind to virus-derived double-stranded RNA (dsRNA) and result in allosteric activation of OAS, giving rise to the synthesis of 2-5A from ATP, which activates RNase L. Upon 2-5A binding, RNase L dimerizes and degrades cellular, as well as viral, RNA in the cell, causing the inhibition of protein synthesis (14,15). The OAS/RNase L system is a component of IFN-regulated innate antiviral immunity. Our study revealed that pOASL inhibits viral replication at the early stages of the CSFV life cycle. We also demonstrated that pOASL contains only a single N-terminal OAS-like domain and lacks C-terminal double ubiquitin domains. To investigate the sensitivity of CSFV to the pOASL antiviral pathway at the molecular level, we supposed that pOASL exerting an antiviral effect is associated with the OAS/RNase L system, based on sequence alignment. The results showed that RNase L knockdown by specific siRNAs did not result in an obvious change of CSFV replication. Thus, we conclude that pOASL exerts antiviral effects independently of the canonical RNase L pathway but depending on the noncanonical pathway.
It has been indicated that hOASL plays a critical role in resistance to HCV infection in vivo (41). The hOASL protein displays antiviral activity and activates RIG-I by mimicking polyubiquitin. HCV and CSFV are both single-stranded, positive-sense RNA viruses belonging to the family Flaviviridae. Hence, we suppose that pOASL exerts antiviral actions also depending on activation of the type I IFN-signaling pathway. Indeed, in this study, we revealed that pOASL could enhance the type I IFN-signaling pathway.
hOASL interacts and colocalizes with RIG-I and enhances RIG-I signaling via its C-terminal ubiquitin-like domain (12). Our study indicated that pOASL interacts with pMDA5, but not pRIG-I, and enhances pMDA5-mediated IFN induction. Our study reveals a mechanism by which pOASL exerts antiviral action via the MDA5-mediated IFN-signaling pathway. In conclusion, we demonstrate that pOASL is an anti-CSFV ISG that depends on the MDA5-mediated type I IFN-signaling pathway.
Plasmid construction and cell transfection. The pOASL gene (GenBank accession no. NM_001031790) was cloned into the vector pCMV-HA (Clontech), p3xFlag-CMV-10 (Sigma-Aldrich), or pFUGW (Addgene), creating pHA-pOASL, pFlag-pOASL, or pFUGW-pOASL. The primers for amplification of the pOASL gene are listed in Table 1. HEK293T cells seeded in six-well plates (Corning) were transfected with 2 g of pHA-pOASL and 2 l of X-tremeGene HP DNA transfection reagent (catalog no. 06366236001; Roche) according to the manufacturer's instructions. At 6 h posttransfection (hpt), the medium was changed to fresh DMEM containing 10% FCS, and the cells were incubated for another 48 h for further assays.
Establishment and identification of a cell line overexpressing pOASL. To construct a stable cell line overexpressing pOASL, HEK293T cells grown in 10-cm cell dishes were cotransfected with 10.5 g of pFUGW-pOASL or pFUGW-EGFP harboring the egfp gene (serving as a control), in combination with 7 g of psPAX2 and 3.5 g of pMD2.G, and the resulting recombinant lentiviruses, Lenti-pOASL and Lenti-EGFP, were harvested at 48 hpt from the supernatants of the transfected cells. Subsequently, to generate the stable cell lines PK-pOASL and PK-EGFP, PK-15 cells were transduced with the abovementioned lentiviruses at an MOI of 1. The expression of EGFP-pOASL and EGFP in the cells was immunoblotted with a mouse anti-EGFP monoclonal antibody (MAb) (1:1,000; catalog no. A00185; GeneScript).
Cell viability assay. To evaluate the effects of ectopic expression on the growth of the cell lines, a cell viability assay was carried out using cell counting kit 8 (CCK-8) (catalog no. CK04; Dojindo) according to the manufacturer's instructions.
IFA and virus titration. CSFV was titrated based on indirect immunofluorescence assay (IFA) as described previously (19). Viral titers were calculated according to a previous report (43) and expressed as median tissue culture infective doses (TCID 50 )/ml.
Real-time RT-PCR. Total RNA was isolated from the CSFV-infected PK-15 cells or swine organs using TRIzol reagent (catalog no. 15596026; Invitrogen) and reverse transcribed into the cDNA using reverse transcriptase XL (avian myeloblastosis virus [AMV]) (catalog no. 2621; TaKaRa) according to the manufacturer's instructions. A qRT-PCR assay was performed to quantify the genomic RNA copy numbers of CSFV (44).
RNA interference assay. Three specific siRNAs targeting different regions of the pOASL, pRNaseL, or pMDA5 gene and a nontargeting control siRNA (siNC) were commercially synthesized by GenePharma; the siRNAs are listed in Table 2. The siRNA transfection assay was performed as described previously (19). PK-15 cells (5 ϫ 10 4 cells/well) were plated in 24-well plates and transfected with 0.2 M sipOASLs or siNC using X-tremeGene siRNA transfection reagent (catalog no. 4476093001; Roche) according to the manufacturer's instructions. At 36 hpt, the transfected cells were inoculated with rCSFV-Fluc or Shimen at an MOI of 0.1. After 2 h, the cell medium was replaced with fresh DMEM containing 2% FCS and incubated at 37°C. At 48 hpi, the supernatants and lysates of the cells were collected for further analysis.
Animal infection experiment. To test the expression level of pOASL in CSFV-infected pigs, three pigs were left uninfected or infected with 10 5 TCID 50 Shimen. Various organs (heart, liver, spleen, lung, kidney, tonsils, and lymph nodes) were collected at 3 days postinoculation (45). The mRNA expression level of pOASL was evaluated by qRT-PCR as described previously (19).
Dual luciferase reporter assay. HEK293T cells in 24-well plates were transfected with pHA-pOASL or pCMV-HA (0.5 g each), together with the promoter reporter plasmid pIFN-␤-Fluc, pISRE-Fluc, or pNF-B-Fluc (0.2 g each) and the internal reference reporter plasmid expressing TK-Renilla luciferase (Rluc) (10 ng). At 24 hpt, the transfected cells were stimulated with SeV at 20 hemagglutinin units (HAU)/ml or left untreated for 24 h and assayed for reporter activities using the Dual-Luci assay kit (Promega). The luciferase activities are presented as Fluc/Rluc relative expression levels.
Co-IP and Western blotting. HEK293T cells were cotransfected with pFlag-pOASL and pMyc-pMDA5 or pMyc-pRIG-I expressing pMDA5 or pRIG-I (2 g each). At 48 hpt, the cells were lysed with NP-40 lysis buffer (catalog no. P0013F; Beyotime) containing phenylmethanesulfonyl fluoride (PMSF) to a final concentration of 1 mM for 1 h at 4°C. To remove cellular debris, the lysates were centrifuged at 12,000 ϫ g at 4°C for 25 min, and protein G-agarose (catalog no. 11243233001; Roche) was used to preclear the supernatants for 4 h at 4°C and then incubated with an anti-Flag MAb (catalog no. F1804; Sigma-Aldrich) or with an irrelevant isotype-matched MAb (catalog no. A00185; GeneScript) acting as a negative control for 5 h at 4°C. The immunoprecipitates were washed five times with the NP-40 lysis buffer and dissolved in phosphate-buffered saline (PBS) for Western blotting, as mentioned above.