Gasdermin D Inhibits Coronavirus Infection by Promoting the Noncanonical Secretion of Beta Interferon

ABSTRACT Pyroptosis, a programmed cell death, functions as an innate immune effector mechanism and plays a crucial role against microbial invasion. Gasdermin D (GSDMD), as the main pyroptosis effector, mediates pyroptosis and promotes releasing proinflammatory molecules into the extracellular environment through pore-forming activity, modifying inflammation and immune responses. While the substantial importance of GSDMD in microbial infection and cancer has been widely investigated, the role of GSDMD in virus infection, including coronaviruses, remains unclear. Enteric coronavirus transmissible gastroenteritis virus (TGEV) and porcine deltacoronavirus (PDCoV) are the major agents for lethal watery diarrhea in neonatal pigs and pose the potential for spillover from pigs to humans. In this study, we found that alphacoronavirus TGEV upregulated and activated GSDMD, resulting in pyroptosis after infection. Furthermore, the fragment of swine GSDMD from amino acids 242 to 279 (242-279 fragment) was required to induce pyroptosis. Notably, GSDMD strongly inhibited both TGEV and PDCoV infection. Mechanistically, the antiviral activity of GSDMD was mediated through promoting the nonclassical release of antiviral beta interferon (IFN-β) and then enhancing the interferon-stimulated gene (ISG) responses. These findings showed that GSDMD dampens coronavirus infection by an uncovered GSDMD-mediated IFN secretion, which may present a novel target of coronavirus antiviral therapeutics.

potential and potential spillover transmission from pigs to humans (5)(6)(7). As severe acute respiratory syndrome coronavirus (SARS-CoV) and SARS-CoV-2 do, TGEV and PDCoV infect both respiratory epithelia and intestinal epithelia, which provide a good model for human highly pathogenic coronaviruses (8,9). Therefore, a better understanding of the cellular responses following swine enteric coronavirus (SECoV) infection will elucidate the innate immunity of human coronaviruses and identify the novel therapeutic targets for reemerging coronavirus diseases.
Type I IFN (IFN-I) (alpha/beta interferon [IFN-a/b]) is the initial host innate cytokine in response to virus infection and is critical for host defense against virus infection (32)(33)(34)(35). IFN-I binds to the ubiquitously expressed IFN-I receptor (IFNAR) through autocrine and paracrine methods and induces a wide range of interferon-stimulated gene (ISG) expression, promoting an antiviral state in bystander cells and restricting viral replication (36)(37)(38)(39)(40). Studies from our and other groups have demonstrated that infection with TGEV and PDCoV can induce the transcriptional expression of IFN-I and increase extracellular IFN-b levels (41)(42)(43). The plasma membrane pores formed by N-GSDMD also serve as conduits for transporting inflammatory cytokines lacking signal peptide, including interleukin 1b (IL-1b) and IL-18, across intact membrane lipid bilayers and contribute to the host inflammatory responses (31)(32)(33)(34)(35). Previous studies also demonstrate that GSDMD plays a crucial role in the unconventional secretion of inflammatory cytokines tumor necrosis factor alpha (TNF-a), IL-6, and CCL2 in macrophages (36). However, the importance of GSDMD in modifying IFN-I production remains elusive. Recent studies demonstrated that GSDMD antagonizes IFN-I production by dampening the cyclic GMP-AMP synthase (cGAS)-dependent signaling pathway in macrophages infected by intracellular bacteria (44,45). Whether and how GSDMD manipulates IFN-b responses in the context of virus infection are not clear.
Given that GSDMD plays a pivotal role in pyroptosis and the secretion of inflammatory cytokines, we explored the antiviral effect of GSDMD on coronavirus replication in the porcine enteric coronavirus infection model. We found that TGEV infection can upregulate and activate GSDMD. Furthermore, we demonstrated that GSDMD could promote the unconventional pathway secretion of IFN-b, thereby playing an anticoronavirus effect. Our findings highlight the regulatory effect of GSDMD on IFN-b release after coronavirus infection and provide new insights into the antiviral role of GSDMD in the scenario of coronavirus infection.

RESULTS
TGEV infection upregulates GSDMD and triggers pyroptosis. To determine whether pyroptosis occurs in TGEV infection, we initially monitored the mRNA expression of GSDMD after TGEV infection at different multiplicities of infection (MOIs) in swine testis (ST) cells. TGEV infection significantly upregulated the mRNA levels of GSDMD and exhibited a dose-dependent manner at 24 h postinfection (hpi) (Fig. 1A). The kinetics of GSDMD mRNA in ST cells infected with TGEV at an MOI of 1 showed a substantial upregulation of GSDMD expression from 24 hpi (Fig. 1B). To assess endogenous GSDMD activation after TGEV infection, the cleaved N-GSDMD was detected by Western blotting. The protein levels of N-GSDMD in TGEV-infected ST cells increased and displayed a dose-dependent response to TGEV infection (Fig. 1C), indicating that TGEV infection resulted in the activation of GSDMD. This membrane rupture by the N-GSDMD-formed pores results in the release of cytosolic lactate dehydrogenase (LDH) into the extracellular space, and the presence of extracellular LDH is widely recognized as a marker of pyroptosis (46). LDH release from the TGEV-infected cells was gradually elevated after TGEV infection and coincided with the increased GSDMD expression ( Fig. 1B and D). The increased LDH release was consistent with the cell activity after TGEV infection measured by ATP (Fig. 1E). Pyroptosis is characterized by swelling followed by rupture of the plasma membrane, resulting in the release of small cytoplasmic contents, distinct from the other two lytic programmed cell deaths (apoptosis and necroptosis) by their morphology (47). The typical morphological features of pyroptosis, such as rupture of the cell membrane and contents released, were also observed by electron microscopy, suggesting that TGEV infection causes pyroptosis (Fig. 1F). Together, these data show that TGEV infection promotes GSDMD expression and induces the activation of GSDMD, which eventually leads to pyroptosis.
Amino acids 242 to 279 are required for pyroptosis induced by GSDMD. Swine GSDMD, a protein composed of 488 amino acids, contains two defined domains (N-GSDMD and C-GSDMD) separated by a linker region, just like human and mouse GSDMD ( Fig. 2A). The GSDMD junction area is cleaved by caspase-1 at the conserved residue D275 in humans (9,13,48) and by caspase-4/11 at D276 in mice (27,49). The cleavage site of swine GSDMD was predicted at D279 based on the alignment with human and murine GSDMD sequences ( Fig. 2A). To verify the functional N-terminal fragment of swine GSDMD, we constructed a GSDMD N-terminal fragment from amino  (Fig. 2D). The cytosolic cleaved N-GSDMD fragment has to translocate to the plasma membrane and bind to phospholipids to form pores in the plasma membrane (46). Next, we determined the localization of N-GSDMD_1-279 in the transient expressing cells. The result showed that N-GSDMD_1-279 was mainly located on the cell membrane, further confirming the ability of swine N-GSDMD_1-279 to form pores in the plasma membrane and induce pyroptosis (Fig. 2E).
To further clarify the required fragments of the GSDMD N terminus to mediate and 2H), accompanied by a substantial increase in the number of PI-staining cells (Fig. 2I). However, N-GSDMD_1-222, N-GSDMD_1-232, and full-length GSDMD did not induce significant cell death. These results suggest that the region of GSDMD from amino acids 242 to 279 is critical for GSDMD pore-forming activity. GSDMD inhibits the replication of TGEV. Next, we explored whether GSDMD affects the replication of TGEV. To this end, we initially constructed a Gsdmd knockout (Gsdmd 2/2 ) ST cell line by CRISPR-Cas9. The Gsdmd knockout was confirmed by sequencing and protein Western blotting (data not shown). Compared with that of wild-type (WT) ST cells, TGEV infection was largely increased in the Gsdmd 2/2 ST cells as measured by the TGEV replication kinetics curve of viral genomes (Fig. 3A) and infectious particles (Fig. 3B). The increased TGEV infection in the Gsdmd 2/2 ST cells relative to wild-type ST cells was observed starting at 6 hpi ( Fig. 3A and B), indicating that GSDMD manipulates TGEV infection not through affecting the viral entry. As expected, the pyroptosis of TGEV-infected Gsdmd 2/2 ST cells measured by LDH release was significantly reduced compared with that of wild-type ST cells, while the viability of Gsdmd 2/2 ST cells increased (Fig. 3C). The enhanced TGEV infection was confirmed by TGEV nucleocapsid (N) protein indirect immunofluorescence (IFA) (Fig. 3D), which was in line with the result of transient overexpression of GSDMD, i.e., that the replication of TGEV was significantly inhibited in ST cells transfected with GSDMD vector compared with mock vector control as measured by the quantification of viral RNA, virus titration, and N protein IFA ( Fig. 3E to G, respectively). These results demonstrated that GSDMD dampens TGEV infection.
The known primary function of GSDMD is to form pores in the plasma membrane by N-GSDMD after activation, which contributes to the nonlytic release of inflammatory cytokines and pyroptosis depending on the degree and timing of GSDMD poreforming activity (28)(29)(30). To clarify the underlying mechanisms of the anti-TGEV activity of GSDMD, we initially assessed the effect of N-GSDMD_1-279 with pore-forming activity on TGEV infection. The overexpression of N-GSDMD_1-279 reduced the replication of TGEV as measured by TGEV replication kinetics quantified by viral RNA and titers ( Fig. 3H and I, respectively). The suppression of TGEV infection by N-GSDMD_1-279 was confirmed by TGEV N protein IFA (Fig. 3J). Together, these results demonstrated that GSDMD inhibits the replication of TGEV through GSDMD pore activity.
GSDMD promotes RNA-elicited IFN-b release. Next, we further explored the underlying mechanisms through which GSDMD suppresses TGEV infection. The results of TGEV replication kinetics in Gsdmd knockout or overexpression ST cells showed that GSDMD-mediated viral suppression occurred starting at 6 hpi and was not observed at 0 h, indicating that GSDMD does not affect TGEV entry. Given the critical roles of the IFN-elicited ISG response against virus infection, we hypothesized that GSDMD suppresses TGEV infection by enhancing the IFN-mediated antiviral ISG responses induced by TGEV infection. We initially monitored the expression of ISGs in TGEV-infected ST cells with N-GSDMD_1-279 transient expression after TGEV infection. As we expected, unlike the TGEV infection result that the overexpression of N-GSDMD_1-279 reduced TGEV infection, the overexpression of N-GSDMD_1-279 promoted the transcriptional mRNA expression of ISG15, ISG54, ISG56, and OASL in ST cells after TGEV infection compared with the mock vector control (Fig. 3K to N). These indicate that GSDMD enhances the IFN response in TGEV-infected ST cells.
Previous studies showed that TGEV infection strongly elicits IFN-b production (50). IFN-I can function through paracrine forms to exert antiviral activity in bystander cells (51,52). Several studies have found that the N-GSDMD-formed pores in the plasma membrane result in the release of small cytosolic contents, including inflammatory cytokines (53)(54)(55). We assumed that the pores formed by N-GSDMD could also promote cytosolic IFN-b release into extracellular space to suppress TGEV infection. To test this, we initially monitored IFN-b production in Gsdmd 2/2 ST cells and WT ST cells after stimulation with poly(IÁC), an analog of double-stranded RNA that can robustly induce IFN-I. Surprisingly, the supernatant levels of IFN-b protein notably showed a substantial decrease in Gsdmd 2/2 ST cells within 48 h after poly(IÁC) stimulation compared with WT ST cell supernatant (Fig. 4A). Furthermore, the decreased supernatant IFN-b by poly(IÁC)-primed Gsdmd 2/2 ST cells was sustained throughout the study period. Similarly, consistent with previous studies showing that GSDMD was critical for IL-6 nonclassical secretion (44), the decreased supernatant IL-6 protein in response to poly(IÁC) stimulation was also observed in Gsdmd 2/2 ST cells compared with WT ST cells (Fig. 4B). Consistent with supernatant IFN-b protein levels, reduced ISG15, ISG54, and ISG56 induction were observed in Gsdmd 2/2 ST cells compared with WT ST cells after poly(IÁC) treatment (Fig. 4C).
To confirm the enhanced IFN-b release by GSDMD pore-formed activity, we monitored the supernatant IFN-b in ST cells following poly(IÁC) stimulation in the presence of disulfiram, an inhibitor of pore formation by GSDMD. Treatment of cells with disulfiram reduced the release of IFN-b induced by poly(IÁC) (Fig. 4D), suggesting that the pore-forming effect of GSDMD is involved in the IFN release. To further verify the GSDMD-mediated noncanonical secretion of IFN, we monitored the supernatant IFN-b of ST cells transfected with N-GSDMD_1-279 after stimulation with poly(IÁC) in the presence of brefeldin A (BFA), a specific inhibitor of cytokine classical secretion pathway. We found that BFA significantly reduced the poly(IÁC)-induced canonical IFN-b release but did not abolish the enhanced supernatant IFN-b by the N-GSDMD_1-279 overexpression (Fig. 4E) GSDMD inhibits TGEV replication by promoting IFN-b release. We demonstrated above that GSDMD could promote IFN/ISG responses by increasing the release of IFNb after poly(IÁC) poly(IÁC)stimulation. Therefore, it is rational to hypothesize that GSDMD dampens TGEV infection by enhancing the IFN-I-mediated ISG response through paracrine forms. We initially explored whether GSDMD could regulate the IFNb release in the context of TGEV infection. Compared with the findings with WT ST cells, Gsdmd knockout substantially reduced the protein levels of supernatant IFN-b after TGEV infection (Fig. 5A). As expected, Gsdmd knockout undermined the secretion of TGEV-induced IL-6 ( Fig. 5B). Concurrently, the induction of ISGs in TGEV-infected Gsdmd 2/2 ST cells was also less than that of WT ST cells with the same dose of TGEV (Fig. 5C). Inhibition of GSDMD pore formation by disulfiram reduced the supernatant IFN-b induced by TGEV (Fig. 5D). Consistent with the supernatant IFN-b results, the supernatant of TGEV-infected WT ST cells induced higher ISG15, ISG54, ISG56, and OASL expression in the uninfected ST cells than the supernatants of TGEV-infected Gsdmd 2/2 ST cells (Fig. 5E). Furthermore, IFNAR1 knockdown of ST cells relieved the discrepant abilities between the supernatants of TGEV-infected Gsdmd 2/2 ST cells and WT ST cells to elicit ISG15, ISG54, ISG56, and OASL expression in ST cells, indicating that various concentrations of IFN-I largely account for the different capacities of the supernatants to induce ISGs (Fig. 5F to I). Furthermore, exogenous IFN-b pretreatment robustly inhibited TGEV infection in ST cells, suggesting that TGEV is sensitive to IFN-b antiviral activity (Fig. 5J). Together, these results show that GSDMD inhibits TGEV infection by promoting IFN-b release. We and others previously showed that PDCoV infection induces evident IFN-I production, like TGEV infection (43,56). We next explored whether GSDMD undermines PDCoV infection by enhancing the release of IFN-I. PDCoV growth kinetics showed that the Gsdmd knockout promoted PDCoV infection as quantified by viral genomes (Fig. 6A) and titers (Fig. 6B), indicating that GSDMD suppresses PDCoV infection. Moreover, the inhibition with TGEV at an MOI of 1, and cell supernatants were collected at 24 hpi. All supernatants were irradiated with UV for 2 h to ensure that TGEV was inactivated. Uninfected ST cells were treated with the inactivated supernatants for 24 h, and the relative mRNA levels of ISG15, ISG54, ISG56, and OASL were measured. (F to I) ST cells were transfected with IFNAR1 siRNA or control siRNA for 24 h, and the inactivated supernatants obtained before were added. After 24 h, total RNA was isolated, and the relative mRNA expression of ISG15, ISG54, ISG56, and OASL was measured by qPCR. (J) IFN-b suppressed TGEV replication in a dosedependent manner. Viral replication was measured after treatment with porcine IFN-b. The relative mRNA levels of TGEV are presented. The means and SD of the results from three independent experiments are shown. *, P , 0.05; **, P , 0.01; ***, P , 0.001; ****, P , 0.0001. of PDCoV infection by GSDMD was confirmed by the transient overexpression of functional N-GSDMD_1-279 and full-length GSDMD measured by PDCoV S protein IFA (Fig. 6C). In addition, the Gsdmd knockout resulted in a decreased level of supernatant IFN-b and IL-6 protein after PDCoV infection (Fig. 6D and E) and a reduced level of antiviral ISG induction in the PDCoVinfected Gsdmd 2/2 ST cells compared with the infected WT ST cells (Fig. 6F). Meanwhile, the release of IFN-b was inhibited by disulfiram after PDCoV infection (Fig. 6G), suggesting that partial PDCoV-elicited IFN-b is released through GSDMD-formed pores. The reduced supernatant IFN-b in the PDCoV-infected Gsdmd 2/2 ST cells compared with WT ST cells was further confirmed by measuring the ISG (ISG15, ISG54, ISG56, and OASL) induction in the uninfected ST cells after stimulating the supernatant cells harvested from PDCoV-infected Gsdmd 2/2 ST or WT ST cells (Fig. 6H). Moreover, IFNAR1 silencing essentially reduced the ISG expression elicited by the supernatants harvested from the PDCoV-infected Gsdmd 2/2 ST cells and abolished the supernatant' ISG induction disparity between PDCoV-infected Gsdmd 2/2 ST and WT ST cells (Fig. 6I to L). Like TGEV, PDCoV was sensitive to the antiviral activity of IFN-b and displayed a dose-dependent manner regarding IFN-b inhibition (Fig. 6M). Collectively, these data indicate that GSDMD also inhibits PDCoV infection by promoting the release of IFN-b.

DISCUSSION
Pyroptosis functions as an innate immune effector mechanism by removing the replicative niche and enhancing the release of inflammatory factors or other active molecules (57)(58)(59).
Here, we show that alphacoronavirus TGEV infection activated GSDMD and induced pyroptosis. Pyroptosis and cleavage of GSDMD induced by TGEV infection have been recently reported by another group (26). Pyroptosis and GSDMD activation are observed in the context of infection with other coronaviruses, such as SARS-CoV-2 and MHV (25,31,57,60). SARS-CoV-2 infection results in GSDMD cleavage and triggers pyroptosis via inflammasome caspase-1 in human monocytes in vitro and in vivo (31). It seems that it is a common phenomenon for coronaviruses to induce pyroptosis. However, the importance of coronavirusinduced pyroptosis in manipulating virus infection itself is poorly investigated.
The essential roles of GSDMD in intracellular bacterial infection have been widely recognized. It is unknown whether GSDMD serves protective or detrimental functions for virus infection, including coronavirus infection. We demonstrate that GSDMD substantially inhibited infection by swine enteric coronaviruses TGEV and PDCoV by promoting the release of IFN-b and enhancing antiviral ISG responses ( Fig. 5 and 6). The detrimental role of GSDMD in TGEV and PDCoV infection is in line with another recent work showing that inhibition of NLRP3-mediated pyroptosis enhanced the replication of TGEV in intestinal epithelial cells (26). These findings are consistent with the MHV study showing that GSDMD deficiency increased cell death after MHV infection, though the authors did not explore what impact GSDMD deficiency has on MHV infection (25). The viral inhibition of GSDMD was also reported in the murine rotavirus intestinal infection model (61).
Furthermore, we demonstrate that the pore-formed activity of GSDMD is associated with the antiviral activity of GSDMD (Fig. 6), which is why viruses have also evolved multiple mechanisms to avoid the activation of GSDMD. Ma et al. recently showed that SARS-CoV-2 blocks GSDMD cleavage by binding SARS-CoV-2 N protein to the GSDMD linker region (67). Enterovirus 71 protease 3C can counteract GSDMD function by cleaving GSDMD at Q193-G194 in the N terminus to produce a nonfunctional N-terminal fragment (62). There is a pressing need to explore how TGEV and PDCoV counteract GSDMD antiviral activity.
Our data suggest that pores formed by GSDMD represent an important supplemental mechanism of IFN-I secretion in the context of coronavirus infection, which provides an unknown mechanism by GSDMD to curtail coronavirus infection. Previous studies demonstrated that GSDMD plays a critical role in releasing inflammatory cytokines, including IL-1b, IL-18, and IL-6, in lytic pyroptotic and nonlytic intact cells. No bioactive IL-1b is released from Gsdmd knockout macrophages since proIL-1b lacks a signal peptide that directs them to the secretory pathway (63). Unlike IL-1b and IL-18, both IFN-b and IL-6 have a signal peptide and are secreted via the constitutive (or continuous) secretory pathway after synthesis (64)(65)(66). Consistent with this, we observed that IFN-b protein was  (45). We also observed increased IFN-b mRNA transcription in Gsdmd 2/2 ST cells in response to poly(IÁC) stimulation compared with that in WT ST cells, indicating that GSDMD dampens the transcription of IFN-b (data not shown), which further emphasizes the importance of GSDMD in releasing IFN-b in the context of RNA virus infection, given that GSDMD knockout resulted in more cytosolic IFN-b production in response to poly(IÁC) stimulation. Therefore, our data revealed an undiscovered mechanism employed by the host to resist coronavirus through the GSDMD-mediated modification of the IFN response. However, we recognize our study's limitations. We did not clarify that GSDMD-mediated IFN-b release is through lytic pyroptosis or the pores in the intact plasma membrane of living cells formed by N-GSDMD. Further investigations of the roles of GSDMD in releasing the IFN-I in various cells and virus infection will help us better understand the host's finely tuned IFN-I response in the context of virus infection.

MATERIALS AND METHODS
Cell culture and virus infection. Swine testis (ST) cells and HEK-293T cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco) supplemented with 10% heated-inactivated fetal bovine serum (FBS; Gibco). ST cells were infected with TGEV strain H16 (GenBank accession no. FJ755618) (MOI = 1) or mock infected with DMEM. After 2 h of infection at 37°C, the cells were washed three times with DMEM. Then cells were cultured in maintenance medium (DMEM supplemented with 0.3% trypsin and 1% DMSO) at 37°C. The PDCoV strain NH (GenBank accession no. KU981062.1) was used to infect ST cells at an MOI of 1. After incubation for 2 h at 37°C, cells were washed three times to remove the unbound virus and cultured in maintenance medium (DMEM supplemented with 0.4% trypsin) at 37°C.
Plasmids and antibodies. Plasmids were constructed by homologous recombination using the ClonExpress Ultra one-step cloning kit (Vazyme, China, Nanjing). Pig full-length GSDMD was cloned from ST cell cDNAs, tagged with hemagglutinin (HA) at the N terminus, and then inserted into a pCAGGS vector. The deletion mutants of GSDMD (N-GSDMD_1-222, N-GSDMD_1-232, N-GSDMD_1-242, N-GSDMD_1-252, N-GSDMD_1-262, N-GSDMD_1-272, N-GSDMD_1-279, and C-GSDMD_280-488) were constructed based on the full-length GSDMD plasmid. The sequences of the primers used are listed in Table 1. All plasmids have been confirmed by nucleotide sequence analysis and verified expression on 293T cells.
Antibodies against HA were purchased from Abcam (Cambridge, MA), GSDMDC1 antibody (sc-81868) was purchased from Santa Cruz Biotechnology (Dallas, TX), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mouse monoclonal antibody was purchased from Beyotime (Shanghai, China), and monoclonal antibodies against TGEV N protein and monoclonal antibodies against PDCoV S protein were prepared and stocked by our team.
Real-time quantitative RT-PCR. ST cells were infected with TGEV or PDCoV at 6, 12, 24, 36, and 48 h, and the total RNA was extracted using an RNeasy kit (Qiagen Sciences, Hilden, Germany). cDNA was obtained using a PrimeScript II 1st-strand cDNA synthesis kit (TaKaRa, Dalian, China). Quantitative PCR (qPCR) was performed using a LightCycler 480 reverse transcription PCR (RT-PCR) machine (Roche) with SYBR. All primers are listed in Table 2, and all data were analyzed based on the cycle threshold (DDC T ) method, and GAPDH was used as the internal control. ELISA. IFN-b and IL-6 sandwich ELISAs were performed using an ELISA kit purchased from Bio-Swamp (Wuhan, China) according to the instruction manual.
Statistical analysis. For nonspecial cases, all the results in a graph are displayed as three standard errors of the means (SEMs) from independent experiments and were analyzed by GraphPad Prism (GraphPad software). The two-tailed Student t test was used to compare the two data groups, and P values of ,0.05 were considered significant.

ACKNOWLEDGMENTS
This work was supported by the National Key Research and Development Program of China (2021YFD1801104 and 2021YFD1801105) and the National Natural Science Foundation of China (31802198).
The content is solely the authors' responsibility and does not necessarily represent the official views of the funding resources.