key: cord-0733391-fpm0qk03
authors: Shi, Fushan; Lv, Qian; Wang, Tingjun; Xu, Jidong; Xu, Wei; Shi, Yuhua; Fu, Xinyu; Yang, Tianming; Yang, Yang; Zhuang, Lenan; Fang, Weihuan; Gu, Jinyan; Li, Xiaoliang
title: The nonstructural protein 5 of coronaviruses antagonizes GSDMD-mediated pyroptosis by cleaving and inactivating its pore-forming p30 fragment
date: 2021-02-24
journal: bioRxiv
DOI: 10.1101/2021.02.23.432418
sha: e8e3ac0dcdbd6d4547b74a5f0c386c052f12e66d
doc_id: 733391
cord_uid: fpm0qk03

Coronaviruses (CoV) are a family of RNA viruses that typically cause respiratory, enteric and hepatic diseases in animals and humans. Here, we used porcine epidemic diarrhea virus (PEDV) as a model of coronaviruses (CoVs) to illustrate the reciprocal regulation between CoVs infection and pyroptosis. For the first time, we clarified the molecular mechanism of porcine Gasdermin D (pGSDMD)-mediated pyroptosis and demonstrated that amino acids T239 and F240 within pGSDMD-p30 are critical for pyroptosis. Furthermore, 3C-like protease Nsp5 from SARS-CoV-2, MERS-CoV, PDCoV and PEDV can cleave human/porcine GSDMD at the Q193-G194 junction upstream of the caspase-1 cleavage site to produce two fragments which fail to trigger pyroptosis or inhibit viral replication. Thus, we provide clear evidence that coronoviruses may utilize viral Nsp5-GSDMD pathway to help their host cells escaping from pyroptosis, protecting the replication of the virus during the initial period, which suggest an important strategy for coronoviruses infection and sustain.

Coronaviruses (CoV) are a family of RNA viruses that typically cause respiratory, enteric and 24 hepatic diseases in animals and humans. Here, we used porcine epidemic diarrhea virus (PEDV) 25 as a model of coronaviruses (CoVs) to illustrate the reciprocal regulation between CoVs infection 26 and pyroptosis. For the first time, we clarified the molecular mechanism of porcine Gasdermin D 27 (pGSDMD)-mediated pyroptosis and demonstrated that amino acids T239 and F240 within 28 pGSDMD-p30 are critical for pyroptosis. Furthermore, 3C-like protease Nsp5 from SARS-CoV-2, 29 MERS-CoV, PDCoV and PEDV can cleave human/porcine GSDMD at the Q193-G194 junction 30 upstream of the caspase-1 cleavage site to produce two fragments which fail to trigger pyroptosis 31 or inhibit viral replication. Thus, we provide clear evidence that coronoviruses may utilize viral 32 the oligomerization of pGSDMD-p30 following treatment of NSC and NSA (two specific 141 inhibitors for oligomerization of hGSDMD-p30) 30, 31 . As shown in Fig. 3B , the results showed that 142 both NSC and NSA inhibited the pyroptosis induced by porcine/human GSDMD-p30 (Fig. 3B) , 143

suggesting that there are residues which impact the oligomerization of pGSDMD-p30 and still 144 required further exploration in the future. 145

Earlier reports have demonstrated that full length of hGSDMD and mGSDMD have an 146 autoinhibitory structure, in which GSDMD-C terminal inhibits the activity of GSDMD-N terminal 147 to induce pyroptosis 23, 30 . Based on the multiple-sequence alignment, L295, Y378 and A382 of 148 pGSDMD formed a pocket which associated with GSDMD-N terminal according to the homology 149 modeling results (Fig. 3C) . Thus, the three residues were tested as the potential sites in pGSDMD. showed that all of the mutants had the activity to induce pyroptosis (Fig. 3D) . However, there 155 were no statistic differences between 2D and 3D (Fig. 3E ). The aforementioned results suggest 156 that L295, Y378 and A382 are the critical sites for autoinhibitory structure of full length of 157 pGSDMD. 158 159 PEDV Nsp5 associates with and cleaves pGSDMD. To investigate the relationship between 160 PEDV infection and pyroptosis, Vero cells were transfected with plasmids encoding pGSDMD 161 full-length (pGSDMD-FL) and GSDMD N-terminal (pGSDMD-p30) and then infected with 162 PEDV. The LDH release assays results showed that PEDV infection had an inhibition effect on 163 pyroptosis induced by pGSDMD-p30 (Fig. 4A) . Meanwhile, the replication of PEDV was also 164 inhibited significantly by pyroptosis induced by pGSDMD-p30 (Fig. 4B) . These results suggest 165 that there might be reciprocal regulation between PEDV and the pGSDMD-p30-mediated 166 pyroptosis. 167

Nonstructural protein 5 (Nsp5), the 3C-like protease, which mediates the cleavage of viral 168 polyproteins, has reported be able to cleave a number of host proteins, such as DCP1A and NEMO, 169 to suppress antiviral host responses 10, 11, 15, 16 . Therefore, we speculated that PEDV Nsp5 can cleave 170 pGSDMD to suppress pyroptosis. As shown in Fig. 4C , HEK293T cells were transfected with 171 PEDV Nsp5 or pGSDMD-p30, or co-transfected with these two recombinant plasmids. The 172 supernatants were collected at different time points and tested for LDH release. The results 173

showed that the expression of PEDV Nsp5 inhibited pyroptosis induced by pGSDMD-p30 (Fig. 174 4C). For further validation, HEK293T cells were transfected with plasmids as indicated in Fig. 4D , 175 and the Western blotting results showed that there was a faster-migrating protein band (about 25 176 kDa) in samples co-transfected with PEDV Nsp5 and p3×Flag-N-GSDMD-FL (Fig. 4D, lane 6) , 177

and there were two cleavage protein bands, of 35 kDa (p30) and 25 kDa respectively, in samples 178 co-transfected with HA-caspase-1, PEDV Nsp5 and p3×Flag-N-GSDMD-FL (Fig. 4D, lane 7) . 179

These results implied that pGSDMD was a target cleaved by PEDV Nsp5. To further confirm the 180 pGSDMD cleavage mediated by PEDV Nsp5, p3×Flag-N-GSDMD-FL was co-transfected with an 181 increasing dose of PEDV Nsp5 into HEK293T cells. Western blotting results showed that 182 pGSDMD cleavage progressively increased in a PEDV Nsp5-dose-dependent manner (Fig. 4E) . 183

We next investigated the colocalization of pGSDMD and PEDV Nsp5 with confocal microscopy. 184

As shown in Fig. 4F , HEK293T cells were transfected with plasmids as shown and the protein 185 localization were examined after 24 h. An indirect immunofluorescence assay showed that 186 pGSDMD and Nsp5 colocalized in the cytoplasm (Fig. 4F) . The CoIP experiments also 187 demonstrated that PEDV Nsp5 interacted with and cleaved pGSDMD (Fig. 4G) . 188

To further investigate whether PEDV Nsp5 cleaves pGSDMD by means of its protease activity, 189 two Nsp5 mutants, H41A and C144A, which disrupted the protease activity of Nsp5 8,32-34 , were 190 constructed and co-transfected with p3×Flag-N-GSDMD-FL into HEK293T cells. As shown in 191 Nsp5 did not impact its interaction with pGSDMD (Fig. 4I) . Hence, the protease activity of PEDV 194

Nsp5 is essential for pGSDMD cleavage but not interaction. Q197-G198 pairs were tested as the potential cleaved sites 35, 36 . Therefore, these three Q residues 201 were replaced with A and these three mutants pGSDMD-Q193A, pGSDMD-Q195A, 202 pGSDMD-Q197A were co-transfected with vector or PEDV Nsp5. As shown in Fig. 5B , Western 203 blotting results showed that pGSDMD-Q193A was resistant to PEDV Nsp5-mediated cleavage, 204 while pGSDMD-Q195A and pGSDMD-Q197A were not (Fig. 5B) , suggesting that PEDV Nsp5 205 cleaves pGSDMD at residue Q193-G194 junction (Fig. 5C) . 206 PEDV Nsp5 cleaves pGSDMD to generate pGSDMD 1-193 and pGSDMD 194-488 , and pCaspase-1 207 cleaves pGSDMD at residue D279. Thus, we next investigated whether these cleaved fragments of 208 pGSDMD 1-193 , pGSDMD 194-279 and pGSDMD 194-488 can induce pyroptosis. As shown in Fig. 5D , 209 these three truncated mutants were separately transfected into HEK293T cells and results showed 210 that none of them induced pyroptosis (Fig. 5D ). Meanwhile, since the protein band of 211 pGSDMD 194-279 was too small to be visualized, we subsequently cloned them into EGFP-tagged 212 vectors and then transfected them into HEK293T cells. The results further confirmed that these 213 three truncated mutants cannot induce pyroptosis (Extended Data Fig. 4 ). Next, we further 214 examined whether pCaspase-1 could associate with and cleave these three truncated mutants. As 215 shown in Fig. 5E and F, HEK293T cells were co-transfected with plasmids as indicated, and the 216 results of CoIP assay and Western blotting assays showed that pCaspase-1 could associate with 217 and cleave full-length of pGSDMD, but had no interaction with pGSDMD 1-193 , pGSDMD 194-279 218 and pGSDMD 194-488 . 219

As described in the preceding text, pyroptosis of cells induced by pGSDMD-p30 had an 220 inhibition effect on PEDV replication. Therefore, we next investigated whether PEDV cannot. Based on this, we conjectured that the active motif of pGSDMD to induce pyroptosis 235 located at the amino acids between 193 and 279. Thus, we constructed a series of pGSDMD 236 truncated mutants which encoding pGSDMD 1-254 , pGSDMD 1-244 , pGSDMD 1-234 , pGSDMD 1-224 , 237

and pGSDMD 1-214 , and transfected them into HEK293T cells. As shown in Figure 6A The results showed that all of the point mutants, except T239D and F240D (Fig. 6C) , can induce 243 pyroptosis, suggesting that T239 and F240 are the essential sites for pGSDMD-p30 to induce 244 pyroptosis. The results were further proved by PI staining assay (Extended Data Fig. 5) . Notably, 245 the point mutant R238D can inhibit the release of LDH but cannot inhibit the intake of PI, 246

suggesting that the mutation of R238 led to smaller pores on cell membrane than wild type 247 pGSDMD-p30. 248

To further investigate the effects of T239D and F240D on viral replication, the two mutants 249 were transfected into Vero cells along with vector, pGSDMD-FL and pGSDMD-p30, and 24 h 250 after transfection, cells were infected with PEDV at an MOI of 0.5 and then the replication of 251 virus were tested by quantitative real-time PCR. As presented in Fig. 6D , in contrast to 252 pGSDMD-p30, neither T239D nor F240D could inhibit the replication of PEDV (Fig. 6D ), further 253 confirming that inhibition of pyroptosis induced by pGSDMD-p30 is essential for PEDV to 254 Recent studies have identified that human/murine GSDMD is a direct substrate of 287 caspase-1/4/5/11 and serves as the executioner for pyroptosis. However, the amino acids sequence 288 and molecular characterization of pGSDMD have not been illustrated. In order to investigate the 289 role of pGSDMD-mediated pyroptosis in PEDV infection, we first clarified the molecular 290 characterization of pGSDMD. Porcine GSDMD has 488 aa and can be cleaved by porcine 291 caspase-1 at D279 to produce GSDMD-NT (p30, 1-279 aa), which lead to pyroptosis. 292

Site-directed mutagenesis studies revealed that C38/C39 and C191/C192 (human/murine) 293 mutations impaired hGSDMD-p30/mGSDMD-p30 oligomerization, which is critical for 294 hGSDMD-p30/mGSDMD-p30 during pore formation 22,30 . However, our results indicated that 295 mutation of porcine C38 or S191 (corresponding to human C38 and C191) had no effect on 296 p30-induced pyroptosis. Interestingly, inhibitors of hGSDMD-p30 oligomerization could also 297 abrogate pGSDMD-p30-induced pyroptosis. The results suggest that other critical site(s) 298 determine(s) pGSDMD-p30 oligomerization. Furthermore, our results demonstrated that T239 and 299 F240 within pGSDMD-p30 are critical for inducing pyroptosis. 300

Generally, the 3C-like protease of CoVs is critical for viral replication by cleaving polyprotein 301 precursors to produce mature nonstructural proteins. However, the 3C-like protease has also 302 acquired mechanisms to evade host innate immune responses. It is reported that CoVs Nsp5 can 303 antagonize innate immune signaling pathways by disrupting one or more components of the 304 IFN-inducing pathways 10-12 . Our present study first demonstrated that CoVs Nsp5 can subvert 305 innate immune responses by cleaving and inactivating GSDMD. Thus, GSDMD represents a 306 novel target of CoVs Nsp5. PEDV Nsp5 not only interacted with and cleaved full length of 307 pGSDMD, but also abrogated pGSDMD-p30-induced pyroptosis by cleaving the p30 fragment. 308

Conversely, protease-dead mutants of the four CoVs Nsp5 were unable to cleave human/porcine 309 GSDMD. Thus, these results suggest a reciprocal regulation between CoVs Nsp5 and pyroptosis. 310

It is noteworthy that CoVs Nsp5 cleaves human/porcine GSDMD at the Q193-G194 junction. 311

Our results suggest that amino acids T239 and F240 within pGSDMD-p30 are critical for 312 pyroptosis. These two sites within pGSDMD-p30 determine CoVs replication. Upon cleavage by 313

CoVs Nsp5, the truncated N-terminal fragment without T239 and F240 sites failed to induce 314 pyroptosis or inhibit viral replication. Interestingly, a newly published study demonstrated that 315

Zika virus (ZIKV) protease directly cleaved the hGSDMD into N-terminal fragment (1-249), 316 which contains T239 and F240. ZIKV NS2B3 protease cleaves hGSDMD at residue R249 to 317 produce hGSDMD 1-249 fragment, which lead to pyroptosis in a caspase-independent manner 39 . 318

Consistent with this, a previous study demonstrated that NS5 protein of ZIKA could directly 319 interact with NLRP3 protein and facilitate NLRP3 inflammasome activation 40 , which is an 320 upstream event for hGSDMD-p30-mediated pyroptosis. Therefore, viruses use different strategies 321 to evade host immune responses and facilitate its replication. 322

In summary, we used PEDV as a model of coronaviruses to illustrate the reciprocal regulation 323 between CoVs infection and pyroptosis. For the first time, we clarified the molecular mechanism 324 of pGSDMD-mediated pyroptosis and demonstrated that amino acids T239 and F240 within 325 

At 24 h after transfection, the cells were infected with PEDV at an MOI of 0.5. After 607 24 h, total RNA was extracted and the viral RNA levels of PEDV were evaluated by quantitative 608 real-time PCR using SYBR green. Data were expressed as fold change of the PEDV mRNA level 609 relative to that of the control vector. **, P < 0.01. (C) HEK293T cells were transfected with 610 plasmids encoding PEDV-Nsp5 and pGSDMD-p30, or co-transfected with these two plasmids

At 24 h after transfection, 613 the cells were then processed for Western blotting. (E) HEK293T cells were transfected with 614 plasmids encoding p3×Flag-N-GSDMD-FL and various dose of MYC-Nsp5. After 24 h, cells 615 were lysed for Western blotting

After 24 h, the 649 supernatants were collected and analyzed for LDH levels. **, P < 0.01. (D) Vero cells were 650 transfected with the plasmids encoding pGSDMD-FL, pGSDMD-p30 and its point mutants 651 (pGSDMD-p30-T239D, pGSDMD-p30-F240D)

GSDMD is a common target of Nsp5 of different coronaviruses. (A) Structure alignment 656

Red arrows indicate conserved enzymatic proteolysis residues His41 and Cys144

The 3D structures were derived from the Protein Data Bank with the following accession numbers

B) HEK293T cells were 659 transfected with plasmids encoding p3×Flag-N-GSDMD-FL and Nsp5 encoded by PEDV

After 24 h, cells were lysed and detected by Western blotting

After 24 h, cells were lysed for Western 663 blotting. (D and E) HEK293T cells were transfected with plasmids encoding pGSDMD and 664 wild-type SARS-CoV-2 Nsp5 or its protease-defective mutants (H41A and C145A), hGSDMD 665 and wild-type SARS-CoV-2 Nsp5 or its protease-defective mutants (H41A and C145A)

were lysed for Western blotting. (F and G) HEK293T cells were transfected with plasmids 667 encoding pGSDMD and wild-type MERS-CoV Nsp5 or its protease-defective mutants

C148A), hGSDMD and wild-type MERS-CoV Nsp5 or its protease-defective mutants (H41A and 669 C148A)

Figure 8 Homology modeling of Nsp5 of different CoVs with the cleaved GSDMD peptide 672 substrate. The molding structure of PEDV Nsp5

SARS-CoV-2 (PDB accession number 7BUY) (B and E)

PDCoV (PDV accession number 6JIJ) (D) combined with the cleaved 675 pGSDMD peptide substrate GAVSLQ(193)↓GQGQGH (downward arrows indicates cleavage 676 sites) (A, B, C and D) and hGSDMD peptide substrate GATCLQ(193)↓GEGQGH (downward 677 arrows indicates cleavage sites) (E and F) were analyzed using PyMOL software

Extended Data Figure 1 Alignment of the amino acid sequence of pGSDMD and other GSDMD 681 homologs from human (GenBank accession NP_001159709.1) and mouse (GenBank accession 682 6N9N_A)

Extended Data Figure 2 HEK293T cells were mock transfected or transfected with plasmids 685 encoding p3×Flag-N-GSDMD-FL. At 24 h after transfection

Western blotting with antibodies for Flag, β-actin and the polyclonal antibody directed against 687 pGSDMD prepared in our laboratory

At 24 h after transfection, the cells were processed and staining with PI, and then analyzed 691 with Fluorescence microscopy (A) and Flow cytometry (B)

Extended Data Figure 4 HEK293T cells were mock transfected or transfected with the plasmids 694 encoding EGFP-GSDMD-FL, EGFP-GSDMD-p30

After 48 h, the supernatants were collected 696 and analyzed for LDH levels, and the cells were analyzed with Fluorescence microscopy

Extended Data Figure 5 HEK293T cells were transfected with plasmids encoding pGSDMD-p30

After 24 h, the cells were dyeing with PI and analyzed with Fluorescence 701 microscopy

Extended Data Figure 6 Alignment of the amino acid sequence of Nsp5 of PEDV with Nsp5 of 704

SARS-CoV-2 (GenBank accession NC_045512)

Supplementary Tables Primers used in this study

PDCoV Nsp5 in complex with GAVSLQ(193)tGQGQGH MERS-CoV Nsp5 in complex with GATCLQ(193)tGEGQGH