Foot-and-Mouth Disease Virus 3A Protein Causes Upregulation of Autophagy-Related Protein LRRC25 To Inhibit the G3BP1-Mediated RIG-Like Helicase-Signaling Pathway

We show that foot-and-mouth disease virus (FMDV) 3A inhibits retinoic acid-inducible gene I (RIG-I)-like helicase signaling by degrading G3BP1 protein. Furthermore, FMDV 3A reduces G3BP1 by upregulating the expression of autophagy-related protein LRRC25. Additionally, other picornavirus 3A proteins, such as Seneca Valley virus (SVV) 3A, enterovirus 71 (EV71) 3A, and encephalomyocarditis virus (EMCV) 3A, also degrade G3BP1 by upregulating LRRC25 expression. This study will help us improve the design of current vaccines and aid the development of novel control strategies to combat FMD.

infection, viral RNA is detected by cytosolic sensors. In most cell types, the cytoplasmic retinoic acid-inducible gene I (RIG-I)-like receptors (RLRs), including RIG-I and melanoma differentiation-associated gene 5 (MDA5), play key roles in sensing RNA virus invasion (11,12). RIG-I and MDA5, which are composed of two caspase recruitment domains (CARDs) and an RNA helicase domain, have been shown to participate in antiviral innate immunity (13).
Signaling mediated by RIG-I and MDA5 is delicately controlled by many host proteins. For example, hemoglobin subunit beta (HB) directly inhibits MDA5-mediated signaling by reducing MDA5-double-stranded RNA (dsRNA) affinity while promoting the RIG-I-mediated signaling through enhancing K63-linked RIG-I ubiquitination (14); knockdown of insulin-like growth factor 1 receptor (IGF-1R) triggers viral RNA sensor MDA5-and RIG-I-mediated mitochondrial apoptosis in cancer cells (15). Tripartite motif 38 (TRIM38) positively regulates MDA5-and RIG-I-mediated induction of downstream genes and acts as a small ubiquitin-like modifier (SUMO) E3 ligase for their dynamic sumoylation (16). Zinc finger CCHC domain-containing protein 3 (ZCCHC3) binds to dsRNA and enhances the binding of RIG-I and MDA5 to dsRNA. ZCCHC3 also recruits the E3 ubiquitin ligase TRIM25 to RIG-I and MDA5 complexes, which facilitates RIG-I and MDA5 K63-linked polyubiquitination and subsequent activation. (17). It has been reported that Ras-GAP SH3-binding protein 1 (G3BP1) is localized with RIG-I (18). Therefore, in this study, we investigated whether porcine G3BP1 mediates the RLH signaling pathway and how FMDV proteins regulate the G3BP1-mediated signaling pathway.
G3BP1, also known as G3BP or HDH-VIII, is a ubiquitously expressed protein and functions as a sequence-specific, phosphorylation-dependent helicase, a cofactor, an endoribonuclease, and more (19). Recently, it has been reported that FMDV L pro and 3C pro may cleave G3BP1 to inhibit stress granule (SG) formation; however, FMDV L pro and 3C pro do not interact with G3BP1 (20,21). This finding prompted us to determine the FMDV proteins that interact with G3BP1 and the mechanism by which it regulates G3BP1 to facilitate FMDV replication and growth. Here, we found that FMDV 3A interacted with G3BP1 and inhibited the G3BP1-mediated RLH signaling pathway. In addition, FMDV 3A degraded G3BP1 by upregulating autophagy-related protein LRRC25 to inhibit the RLH signaling pathway, which, in turn, increased FMDV replication and growth. These findings reveal a mechanism of critical importance that allows FMDV to evade the immune system.

RESULTS
G3BP1 is involved in the defense response against FMDV. Assuming that FMDV L pro and 3C pro cleave G3BP1 to inhibit SG formation (20,21), we investigated the effects of G3BP1 on FMDV replication and growth. Previous studies have shown that FMDV 3C pro cleaves G3BP1 at glutamic acid-284 (E284) (21). G3BP1 or G3BP1E284A overexpression inhibits FMDV genomic copies and titer in porcine kidney 15 (PK-15) cells, but G3BP1E284A is more potent than G3BP1 in inhibiting FMDV genomic copies and titer ( Fig. 1A and B). Furthermore, FMDV infection inhibited G3BP1 and G3BP1E284A expression (Fig. 1C), suggesting that other FMDV proteins mediated G3BP1 expression inhibition, other than FMDV 3C pro . Next, the function of endogenous G3BP1 in FMDV genomic copies and titer was investigated. An RNA interference (RNAi) plasmid was constructed for G3BP1, and RNAi plasmids markedly reduced the expression of endogenous G3BP1 in PK-15 cells (Fig. 1F). In reverse transcriptase PCR (RT-PCR) and titer experiments, FMDV genomic copies and titer were increased in G3BP1-knockdown PK-15 cells (Fig. 1D and E). In addition, there was an increased expression of 3A and VP3 in G3BP1-knockdown PK-15 cells compared with control cells (Fig. 1F). Collectively, these results suggest that G3BP1 negatively regulates FMDV replication and growth.
Porcine G3BP1 promotes the RLH-mediated signaling pathway. G3BP1 has been shown to bind to viral dsRNA and RIG-I to enhance interferon-␤ (IFN-␤) responses (18). It has also been shown to share 92% sequence identity with its porcine ortholog, from which observation suspicion of porcine G3BP1 participation in RIG-I-mediated signaling arose. In reporter assays, overexpression of porcine G3BP1 increased Sendai virus (SeV)-triggered activation of the IFN-␤ promoter and interferon-stimulated response element (ISRE) in human embryonic kidney 293T (HEK293T) cells ( Fig. 2A and B). Further experiments indicated that overexpression of G3BP1 inhibited SeV-triggered activation of the IFN-␤ promoter and ISRE in a dose-dependent manner in HEK293T cells ( Fig. 2C and D). In addition, we also found that G3BP1 and G3BP1E284A increased the SeVinduced IFN-␤ promoter and ISRE activation, not including G3BP1 (1 to 284 amino acid [aa]) and G3BP1 (285 to 467 aa) ( Fig. 2E and F). In an RT-PCR experiment, we observed that SeV-triggered transcription of the Ifnb1, Rantes, Ip10, Tnfa, Isg56, Il6, and Il8 genes were increased in G3BP1-overexpressed HEK293T cells compared with control cells (Fig.  2G). Additionally, enzyme-linked immunosorbent assay (ELISA) experiments indicated that the levels of secreted IFN-␤ induced by SeV infection increased in G3BP1overexpressed HEK293T cells compared with wild-type cells (Fig. 2H). These results suggest that porcine G3BP1 also positively regulates a virus-triggered signaling pathway.
Various components are involved in virus-triggered signaling pathways. As shown in Fig. 2I and J, G3BP1 potentiates the IFN-␤ promoter and ISRE activation, mediated by RIG-I and MDA5 but not virus-induced signaling adapter (VISA). These results suggest that G3BP1 targets RIG-I and MDA5.
FMDV 3A interacts with G3BP1. Next, the FMDV proteins that interact with G3BP1 were determined. In transient-transfection and coimmunoprecipitation experiments, G3BP1 interacted with FMDV 3A but not VP0, VP1, VP2, VP3, 2B, 3B, 3C pro , 3D, and L pro proteins (Fig. 3A). Endogenous coimmunoprecipitation experiments indicated that FMDV 3A was associated with G3BP1 in PK-15 cells and porcine alveolar macrophages (PAMs) following FMDV infection ( Fig. 3B and C). To examine the colocalization of the FMDV 3A protein with G3BP1, HEK293T cells were cotransfected with plasmids expressing Flag-3A and Myc-G3BP1, and the subcellular localization of 3A protein and G3BP1 was examined by confocal microscopy (Fig. 3D). To confirm that endogenous G3BP1 colocalizes with the 3A protein, PK-15 cells were infected with FMDV and analyzed by confocal microscopy. Confocal images of the cells immunostained with anti-3A and anti-G3BP1 antibodies showed colocalization of G3BP1 with the FMDV 3A protein (Fig.  3E). Collectively, these findings confirm that FMDV 3A interacts with G3BP1.
FMDV 3A enhances FMDV genomic copies and titer. Next, the effect of FMDV 3A on FMDV genomic copies and titer was examined. RT-PCR and titer experiments demonstrated that the FMDV 3A-overexpressed PK-15 cells increased FMDV genomic copies and titer ( Fig. 4A and B). In accordance with this finding, FMDV 3Aoverexpressed PK-15 cells also increased FMDV VP3 expression (Fig. 4C). These results suggest that FMDV 3A promotes FMDV replication and growth. FMDV 3A represses the G3BP1-mediated RLH antiviral signaling. Previous studies have shown that amino acid deletion at positions 93 to 102 in FMDV 3A (hereafter referred to as 3A D93) and 133 to 143 in FMDV 3A (hereafter referred to as 3A D133) could alter the host range of FMDV (22). In addition, in light of the results of our  previous experiments, we investigated the effect of FMDV 3A on the G3BP1-mediated RLH signaling pathway. This prompted further investigation to clarify whether FMDV 3A and its mutants affected RLH signaling or G3BP1-mediated RLH signaling. In reporter assays, FMDV 3A, 3A D93, and 3A D133 inhibited SeV-triggered activation of the IFN-␤ promoter and ISRE ( Fig. 5A and B). To further confirm the effect of 3A mutants on RLH signaling, PK-15 cells were infected with wild-type FMDV and deletion mutant FMDV. The results indicated that wild-type FMDV and deletion mutant FMDV (3A D93 and 3A D133) inhibited poly(I·C)-triggered IFN-␤ mRNA levels in PK-15 cells (Fig. 5C and D and E). Additionally, wild-type FMDV and deletion mutant FMDV (3A D93 and 3A D133) inhibited poly(I·C)-triggered IFN-␤ expression by ELISAs in PAM cells compared to control cells (Fig. 5F to H). Collectively, these results suggest that FMDV 3A inhibits G3BP1-mediated RLH signaling. FMDV 3A degrades G3BP1 via autophagy. Next, regulation of RLH signaling by porcine G3BP1 was investigated. In transient-transfection experiments, G3BP1 increased RIG-I and MDA5 expression but not VISA expression in a dose-dependent manner in HEK293T cells (Fig. 6A to C). Furthermore, we found that FMDV 3A degraded G3BP1 and G3BP1E284A in a dose-dependent manner in HEK293T cells (Fig. 6D). Moreover, we also observed that FMDV 3A inhibited the expression of G3BP1-mediated RIG-I and MDA5 (Fig. 6E and F). These data collectively demonstrate that FMDV 3A degrades G3BP1.
FMDV 3A degrades G3BP1 through upregulating LRRC25 expression. It has been reported that leucine rich repeat-containing 25 (LRRC25) inhibits IFN-␤ signaling by targeting RIG-I for autophagic degradation (23). This finding prompted us to investigate whether LRRC25 is involved in FMDV 3A-and G3BP1-mediated RLH signaling. In transient-transfection experiments, we found that LRRC25 inhibited RIG-I and MDA5 expression in a dose-dependent manner in HEK293T cells (Fig. 7A and B). Furthermore, FMDV 3A was found to enhance LRRC25 expression in a dose-dependent manner in HEK293T cells (Fig. 7C). In addition, LRRC25 inhibited the expression of G3BP1 in HEK293T cells (Fig. 7D). In transient-transfection and coimmunoprecipitation experiments, FMDV 3A and G3BP1 were shown to interact with LRRC25 in HEK293T cells, but FMDV 3D and LRRC25 showed no interaction ( Fig. 7E and F). Furthermore, FMDV 3A, but not 3D, G3BP1, LRRC25, RIG-I, and MDA5 formed a complex (Fig. 7G to K). Taken together, these results suggest that FMDV 3A degrades G3BP1 by upregulating LRRC25 expression.
LRRC25 increases FMDV genomic copies and titer. We next determined whether endogenous LRRC25 regulates FMDV genomic copies and titer. RT-PCR and titer experiments revealed that LRRC25-overexpressed PK-15 cells increased FMDV genomic copies and titer ( Fig. 8A and B). Consistently, the expression of FMDV 3A and VP3 proteins were increased in LRRC25-overexpressed PK-15 cells (Fig. 8C). A porcine LRRC25-RNAi plasmid was constructed, and immunoblotting analysis indicated that it markedly inhibited the expression of endogenous LRRC25 in PK-15 cells (Fig. 8D). In RT-PCR and titer experiments, we observed that FMDV genomic copies and titer were reduced in LRRC25-knockdown PK-15 cells compared with control cells (Fig. 8D and E). Furthermore, the expression of FMDV 3A and VP3 proteins were decreased in LRRC25knockdown PK-15 cells compared with control cells (Fig. 8F). These results suggest that LRRC25 increases FMDV replication and growth.

DISCUSSION
G3BP1 is a multifunctional protein that participates in many physiological processes, such as SG assembly, immune response, arteriosclerosis, and tumor promotion (24,26,27). In this study, we found that FMDV 3A interacted with G3BP1. G3BP1 enhanced RIG-I and MAD5 expression, but FMDV 3A degraded G3BP1 and inhibited G3BP1-mediated RLH signaling by upregulating LRRC25 expression.
It has been reported that FMDV 3A inhibited RIG-I and MDA5 expression (7) and FMDV could be solely recognized by MDA5 (32), but the mechanism remains unclear. In addition, FMDV 2C inhibited FMDV replication by increasing IFN-␤ production (33). In the study, FMDV 3A increased FMDV replication and degraded G3BP1 by upregulating LRRC25 expression, which inhibited RIG-I and MDA5 expression.
G3BP1 is a key component and a commonly used marker of SG (19). FMDV 3C pro and L pro cleaved G3BP1 to inhibit SG formation (20,21). It has been demonstrated that G3BP1 mediated autophagy pathways (34), but the mechanism remains unclear. In addition, G3BP1 interacts directly with the FMDV IRES and negatively regulates its translation (35). This finding prompted us to determine the mechanism by which G3BP1 may be regulated by FMDV proteins to propagate FMDV replication. In this study, porcine G3BP1 enhanced RLH signaling and FMDV 3A inhibited G3BP1-mediated RLH signaling by upregulating the expression of autophagy-related protein LRRC25. These findings suggest that FMDV or FMDV proteins have a connection with SGs (G3BP1), autophagy, and the RLH signaling pathway, but the mechanism needs to be further studied.
The 3A-deletion FMDV mutants (3A D93 and 3A D133) can alter the host range of FMDV (22), but the mechanism is unclear. An explanation for the above mechanism by natural immunity could not be reached in this study. First, both FMDV 3A D93 and FMDV 3A D133 inhibited RLH signaling and G3BP1-mediated RLH signaling as well as FMDV 3A. Second, the deletion mutants of FMDV also inhibited poly(I·C)-triggered RLH signaling as well as wild-type FMDV.
The FMDV capsid protein VP2 induces autophagy through interaction with heat shock protein beta-1 (HSPB1) and activation of the eukaryotic translation initiation factor 2 subunit alpha-activating transcription factor 4 (EIF2S1-ATF4) pathways (36). In this study, we found that FMDV 3A inhibited G3BP1 expression and G3BP1-mediated RLH signaling by upregulating the autophagy-related protein LRRC25. This study indicates that FMDV 3A mediates the autophagy pathway, but whether FMDV 3A directly or indirectly mediates the autophagy pathway needs to be further studied.
Consistent with previous studies, it was found that G3BP1 and G3BP1E284A inhibit FMDV replication and growth (21). We found that FMDV also inhibited the expression of G3BP1E284A, suggesting that other FMDV proteins mediated G3BP1 degradation. Finally, we confirmed that FMDV 3A degraded G3BP1. First, FMDV 3A degraded G3BP1 and G3BP1E284A in a dose-dependent manner in HEK293T cells. Second, FMDV 3A inhibited endogenous G3BP1 expression.
Based on these findings, we propose a working model of FMDV 3A-mediated regulation of G3BP1 in innate response to FMDV. During FMDV infection, FMDV 3A, LCRR25, G3BP1, RIG-I, and MDA5 form a complex. Subsequently, FMDV 3A increases LRRC25 expression, which reduces G3BP1 expression to inhibit RIG-I and MDA5 expression (Fig. 10). These results provide important insights into the molecular mechanisms of FMDV 3A and G3BP1-mediated innate immune response and autophagy pathway for FMDV replication and growth.

MATERIALS AND METHODS
Reagents. Mouse monoclonal antibodies against Flag, Flag-HRP, and Myc (Sigma); hemagglutinin (HA) (Covance); ␤-actin (Sigma); and rabbit polyclonal antibody against LRRC25 (Abcam) were purchased from the corresponding manufacturers. A mouse anti-VP3 and 3A polyclonal antibody was prepared using conventional methods. iQ SYBR green real-time supermix (Bio-Rad), Gammabind G plus Sepharose (Amersham Biosciences), and Moloney murine leukemia virus (M-MLV) reverse transcriptase (Invitrogen) were purchased from the indicated manufacturers. The human IFN-␤ ELISA kit was purchased from Pestka Biomedical Laboratorie. The porcine IFN-␤ ELISA kit was purchased from Solarbio Life Science.
Viruses and cells. HEK293T cells (ATCC) and PK-15 (ATCC) cells were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS). The type-O FMDV and its deletion mutants were prepared in our laboratory and propagated in PK-15 cells, and the supernatants of infected cells were harvested and stored at -80°C for further studies. SeV was provided by Hongbing Shu (Wuhan University).
Constructs. Mammalian expression plasmids encoding VP0, VP1, VP2, 2B, 3B, 3C pro , 3D, L pro , and 3A; its mutants; and VP3 were constructed by PCR amplification of their cDNA from FMDV-infected PK-15 cells. Subsequently, they were cloned into cytomegalovirus (CMV) promoter-based vectors containing a Flag-tag or Myc-tag. The SVV 3A, EV71 3A, and EMCV 3A genes were synthesized and constructed into the plasmid vector containing Myc-tag. Mammalian expression plasmids for porcine HA-tagged or Myc-tagged G3BP1 and Myc-tagged LRRC25 were constructed by standard molecular biology techniques. Mammalian expression plasmids for the HA-tagged porcine G3BP1 mutant were constructed with standard molecular biology techniques. Mammalian expression plasmids encoding HA-RIG-I, Flag-RIG-I, HA-MDA5, and HA-VISA were described previously (25).
TCID 50 detection of FMDV. The 50% tissue culture infective dose (TCID 50 ) in the collected supernatants was determined by virus titration assay, as described previously (37). Briefly, baby hamster syrian kidney 21 (BHK-21) cells were resuspended in Dulbecco's modified Eagle medium (DMEM) with 5% fetal calf serum (FBS) at a concentration of 1.5 ϫ 10 6 cells/ml, which was dispensed at 50 l per well into 96-well flat-bottomed tissue culture plates. The plates were rocked to achieve uniform suspension thickness and were incubated at 37°C for 24 h to 36 h under 5% CO 2 tension to attain 90% confluence. Serial 10-fold dilutions of virus stock prepared in FBS-less DMEM were added in 50-l volumes to all wells. Plates were incubated at 37°C and 5% CO 2 for 72 h, and the presence or absence of cytopathic effect (CPE) was then monitored. TCID 50 was calculated by the Reed-Muench method.
Transfection and reporter gene assays. The HEK293T cells (ϳ1 ϫ 10 5 ) were seeded on 48-well plates and transfected the following day by standard calcium phosphate precipitation. In the same experiment, empty control plasmid was added to ensure that each transfection received the same amount of total DNA. To normalize for transfection efficiency, 10 ng of pRL-TK renilla luciferase reporter plasmid was added to each transfection. Luciferase assays were performed with a dual-specific luciferase assay kit (Promega). Firefly luciferase activities were measured and normalized to renilla luciferase activities.