Experiments involving live animals were conducted in accordance with the ‘‘Regulations for the Administration of Affairs Concerning Experimental Animals’’ promulgated by the State Science and Technology Commission of Henan Province. This study was approved by the Animal Care and Use Ethics Committee of the Henan Normal University.

The healthy common carp (23 ± 2 g, 11 ± 1 cm) were from the aquaculture base in Zhengzhou, Henan province. Before the experiment, the fish were randomly grouped with 50 fish in each group, and then acclimated in 200 L tanks for two weeks. During acclimation, the fish were maintained at 25 ± 2 °C, and pH 7.0 ± 0.2, with the dissolved oxygen concentration of 6.0 ± 0.2 mg/L and natural light/dark cycle, and 1/3 water in tank was exchanged with the fresh aerated water twice a week. The fish were fed with pellets at 5% body weight. The common carp (4% of stock) were randomly sampled for bacterial or virus detection, and no pathogen was detected in the sampled fish, the experimental fish were considered to be healthy (36).

Mycoplasma-free Human Embryonic Kidney (HEK) 293T cells were cultured in DMEM medium (Hyclone, USA) supplemented with 10% (v/v) Fetal bovine serum (FBS, Gibco, USA), 2 mM L-Glutamine, and 4.5 g Glucose, at 37 °C in a 5% CO₂ incubator. For mycoplasma-free Epithelioma papulosum cyprini (EPC) cell, the M199 medium (Hyclone, USA) supplemented with 10% FBS, 1000 mg D-Glucose, 2200 mg NaHCO₃, and 100 mg L-Glutamine was used, and cultured at 25˚C in an incubator with 5% CO₂. The cells were passaged every 2-3 d at a split ratio of 1:3 when the cells reached 90% confluency.

A. hydrophila, which was stored in our laboratory and isolated from common carp (15), cultured on Luria-Bertani broth (LB) plates at 28°C overnight, and single colony of each strain was then inoculated into 5 mL LB medium at 28 °C with constant shaking (180 rpm), and then sub-cultured in 100 mL LB medium with the ratio of 1:100 at 28 °C at 180 rpm until the absorbance values (OD₆₀₀) was 0.6 (37). The half lethal concentration (LC₅₀) of bacteria was 7 × 10⁷ CFU/mL. The bacteria were harvested by centrifugation at 8000 rpm at 4 °C for 10 min, washed three times with 0.75% physiological saline (0.75% NaCl), and then resuspended in 0.75% NaCl with the concentration of 7 × 10⁶ CFU/mL.

Total RNA was extracted from the samples using RNA extraction reagent RNAiso (TaKaRa, Japan) on the basis of the protocol of manufacture. Total RNA was dissolved in the RNase free water (Sangon, China). The concentration was determined by NanoDrop 2000 spectrophotometer (Thermo company, USA), and the purity was detected through 1% agarose gel electrophoresis. First-strand cDNA used for gene cloning was synthesized using the PrimeScript™ II 1st Strand cDNA Synthesis Kit (TaKaRa, Japan), and the PrimeScript™ RT Master Mix was used to synthesize cDNA template for quantitative Real Time PCR (qRT-PCR) (Vazyme, China). The reverse transcription products were stored at -20°C until used.

The open reading frame (ORF) amplification of CcGSDMEa was performed in a 20 μL reaction system, which was made up of 1 μL cDNA template, 1 μL forward primer, 1 μL reverse primer, 10 μL 2 × Taq Master Mix, and 7 μL ddH₂O. The PCR procedure was set as follows: 5 min at 94 °C for pre-denaturation, 34 cycles (30 s at 94°C for denaturation, 30 s at 58 °C for annealing, and 120 s at 72 °C for extension), and 10 min at 72°C for final extension. The products were analyzed on 1.5% agarose electrophoresis and purified by DNA gel extraction kit (Omega, USA). The purified DNA was ligated into the pMD-19T vector (Takara, Japan), and then transformed into Escherichia coli DH5α (Biomed, China). Positive clones were sequenced in Sangon Biotech Company (Shanghai, China).

The rapid amplification of cDNA end (RACE) method was used to amplify the 3’ and 5’ end sequences of CcGSDMEa. According to the instruction of PrimeScript™ II 1st Strand cDNA Synthesis Kit, the Oligo dT Primer was replaced by the 3’ RACE Olig(T)-adaptor for reverse transcription to obtain the 3’ RACE template. With regard to 5’ RACE amplification template, the obtained cDNA product was purified with E.Z.N.A. Cycle-pure Kit (Omega, USA), and Poly (A) tail was added at the 3’ end after purification. A total of 50 μL reaction system was as follows: cDNA 10 μL, 5 × TdT buffer 10 μL, 1% BSA 5 μL, dATP (10 mmol/L) 2.5 μL, TdT enzyme 1 μL, and ddH₂O 21.5 μL. Reaction condition was at 37 °C for 30 min and at 80 °C for 3 min. The obtained final product was 5’ RACE template. The PCR amplification was performed using the primers of 3’/5’ -Out and 3’ RACE Adaptor/5’ RACE Olig(T)-Adaptor for the first round PCR, and then followed by the nested PCR with the primers of 3’/5’ -In and 3’/5’ RACE Adaptor using the first round PCR products (diluted 50 times) as the template. The reaction system and PCR procedure were same as ORF amplification, except only 20 cycles for first round PCR. The primers used for gene cloning were shown in Table 1 .

The amplified ORF was overlapped with the 3’/5’ terminal sequences to obtain the full length of CcGSDMEa cDNA sequence, implemented in software DNAMAN. CLC Main Workbench 6.8 was used to translate the amino acid sequences. Theoretical isoelectric point (pI) and molecular weight (MW) were respectively calculated on the basis of deduced amino acid sequences by the Compute pI/MW software (https://web.expasy.org/compute_pi/). The domains were predicted and annotated with Simple Modular Architecture Research Tool-SMART (http://smart.embl-heidelberg.de/), and the three-dimensional structure of protein was predicted by I-TASSER (https://zhanggroup.org/I-TASSER). Homologous sequences were searched and analyzed by the BLAST algorithm of NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Multiple sequence alignments were carried out using the DNAMAN 8.0, and phylogenetic tree was constructed using the neighbor-joining method, implemented in MEGA 7.0 program with a bootstrap test of 1000 replicates.

Various tissues including gill, liver, spleen, head kidney, trunk kidney, heart, intestine, brain, muscle, and skin were taken aseptically from six healthy fish (23 ± 2 g), respectively. Total RNA extraction and cDNA synthesis were performed as described above. The AceQ qPCR SYBR Green Master Mix (Vazyme, China) and LightCycler^® 480 II instrument (Roche, Switzerland) were used for qRT-PCR to detect gene expression at mRNA level. The qRT-PCR was performed in a 20 μL reaction system (10 μL 2 × AceQ qPCR SYBR Green Master Mix, 1 μL specific primers, 3 μL cDNA template and 6 μL ddH₂O), and cycling program (180 s at 95 °C for pre-incubation, followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s for amplification). The amplification efficiency of qRT-PCR primers for CcGSDMEa and CcIL-1β gene was 97.34% and 99.07%, respectively. The 18S was used as the reference gene, and the amplification efficiency was 100.8% ( Table 1 ). The results were analyzed by the 2^-ΔΔCt method.

The healthy common carp (23 ± 2 g) were randomly divided into two groups (the treated group and the control group), with 50 fish each group. Fish were injected intraperitoneally with 100 μL A. hydrophila suspension at a concentration of 7 × 10⁶ CFU/mL in treated group. In the control group, the fish were injected with 0.75% NaCl in the same volume. The experiments were carried out in triplicate. The fish were not fed during the experiment period and maintained under the same conditions with those during the acclimation period. The mortality of fish was observed and recorded every day. After A. hydrophila challenge, six fish from each group were randomly selected at 0 h, 3 h, 6 h, 12 h, 24 h, 48 h and 96 h, respectively, which were anesthetized with ethyl 3-aminobenzoate (MS-222, Sigma, USA), and then sampled. Blood was collected from the caudal vein and centrifuged at 3000 rpm at 4 °C for 10 min to collect serum, which was used for Enzyme-Linked Immunosorbent Assay (ELISA). The tissues (gill, liver, spleen, head kidney, intestine and skin) were sampled immediately and stored in liquid nitrogen. Total RNA extraction, cDNA synthesis, and qRT-PCR were carried out as described above.

The different forms of CcGSDMEa were cloned into His-tagged pET-32a vector or GFP-tagged pEGFP-C1 vector through Hind III and KpnI sites to generate rCcGSDMEa-FL (1-532 aa) or GFP-CcGSDMEa-FL, rCcGSDMEa-NT (1-252 aa) or GFP-CcGSDMEa-NT and rCcGSDMEa-CT (257-532 aa) or GFP-CcGSDMEa-CT recombinant plasmids. CcCaspase-1a/1b/3a/3b/7a/7b sequences were inserted into N-DmrB-pcDNA3.1-3HA expression vector through EcoR I and KpnI sites, respectively. The CcGSDMEa, CcCaspase-1a/1b/3a/3b/7a/7b, pET-32a (+) vector, pEGFP-C1 vector and N-DmrB-pcDNA3.1-3HA vector were digested by restriction enzymes (New England Biolabs, USA), and then they were purified and ligated together by the T4 ligation enzyme (TakaRa, Japan). The recombinant plasmids were transferred into E. coli strain DH5α, and the positive clones were verified by sequencing. The primers used to construct plasmids were listed in Table 2 . To remove the endotoxin, the constructed recombinant eukaryotic plasmids were extracted with an EndoFree Plasmid Kit following the instruction of manufacturer (Omega, USA).

A total of 5 × 10⁶ EPC cells were seeded per well in six-well plates for 16 h, then co-transfected with GFP-CcGSDMEa-FL (2.5 μg) and CcCaspase-1a/1b/3a/3b/7a/7b (2.5 μg) plasmids using Lipofectamine 3000 Reagent (Thermo Fisher Scientific, USA) according to the instructions when the cells reached 90% confluency. 36 h later, the EPC cells were exposed to 1 μM AP20187 (a chemical inducer of dimerization, which could penetrate the cell membrane and induce the DmrB fusion protein to dimerize. Selleck, China) for 6 h to mediate CcCaspase-1a/1b/3a/3b/7a/7b dimerization (38, 39). Next, the cells were collected, and the total proteins were extracted using the protein extraction kit (Solarbio, China) according to the specification. In detail, the EPC cells were lysed in lysis buffer supplemented with 1% PMSF, protease inhibitor and phosphatase inhibitor on ice for 15 min, and then centrifuged at 12000 rpm at 4 °C for 10 min. Protein concentration was detected by BCA assay kit (Solarbio, China). Subsequently, the protein was mixed with 5 × protein loading buffer (Solarbio, China) and boiled for 10 min, and then used for western blotting.

The HEK293T cells (5 × 10⁵ cells/well) were seeded on coverslips into 6-well cell culture plates overnight, and the plasmids expressing GFP-CcGSDMEa-FL, GFP-CcGSDMEa-NT and GFP-CcGSDMEa-CT (1000 ng/well) were transfected with Lipofectamine 3000 Reagent when the cells reached 70% confluency, and the pEGFP-C1 vector was used as the control.

At 24 h after transfection, the cells were collected, and washed three times with sterile 1 × PBS buffer (Solarbio, China), and then fixed with 4% paraformaldehyde at room temperature for 1 h, washed three times. The cells were incubated with 0.1% Hoechst 33342 (Invitrogen, USA) for ten minutes for the sake of nucleus staining, or incubated with 0.1% DiI (Beyotime, China) for 10 min for the sake of cell plasma membrane staining, and then washed three times. The cell coverslips were mounted on glass slides with Antifade Mounting Medium (Beyotime, China) and air-dried at room temperature. The fluorescence images of cells were observed under a fluorescence microscope (Zeiss, Germany). In addition, the HEK293T cells were fixed with 2.5% glutaraldehyde at 4 °C overnight, and dehydrated using series gradient alcohol and tertiary butanol, and then the integrity of cytomembrane was observed under a biotypic scanning electron microscope (JEOL, Japan).

The different forms of rCcGSDMEa recombinant plasmids were transferred into E. coli strain BL21 (DE3), and the transformants were cultured in LB with 100 μg/mL Ampicillin (Amp), shaken at 200 rpm at 37 °C until OD₆₀₀ to 0.6. The bacteria diluted at 10⁴ times, were evenly spread with 50 μL on LB agar plates with or without 0.1 mM IPTG, and then cultured at 37 °C overnight. The CFU on the plates was counted and statistically calculated. At the same time, the growth curve of E. coli was tested by measuring the OD₆₀₀ after induction by IPTG.

The cytotoxicity was monitored by evaluating the release of LDH from cells. The cell culture medium was centrifuged at 400 g for 5 min at 4°C, and the supernatant was collected. The LDH content released from the cells was detected according to the instruction of the Lactate Dehydrogenase Cytotoxicity Detection Kit (Beyotime, China). The untransfected cells were used as the blank background control, and the LDH released from the cells treated with LDH release agent was used as the positive control with maximum enzyme activity. Each sample was tested in triplicate. Cytotoxicity (%) = (treated sample A₄₉₀ - blank background control A₄₉₀)/(maximum enzyme activity A₄₉₀ - blank background control A₄₉₀) × 100.

The healthy fish (9 ± 0.5 g) were randomly divided into five groups (i.e., Control group: PBS; Knockdown groups: siControl group and siCcGSDMEa group; Overexpression groups: pEGFP-C1 group and GFP-CcGSDMEa-FL group). The primers of siRNAs (siCcGSDMEa-F/R, 5’-GGA UCU AUA GAG UUG GGA ATT-3’, 5’-UUC CCA ACU CUA UAG AUC CTT-3’; siControl-F/R, 5’-UUC UCC GAA CGU GUC ACG UTT-3’, 5’-ACG UGA CAC GUU CGG AGA ATT-3’) were designed and synthesized by Sangon Biotech Company (Shanghai, China). For the different knockdown groups, the fish were intramuscularly injected with 100 μL (10 μg) siControl or siCcGSDMEa, respectively. The fish were intramuscularly injected with 100 μL (10 μg) GFP-CcGSDMEa-FL or pEGFP-C1 plasmid in different overexpression groups, respectively (29, 40). In the control group, the fish were injected with 100 μL PBS by the same method. The experiments were carried out in triplicate. After 3 and 7 d, the tissues (gill, liver, spleen, trunk kidney and head kidney) were sampled in the aseptic condition and used for analyzing the efficiency of knockdown or overexpression by RT-qPCR.

In knockdown groups, overexpression groups and the control group, the fish were intraperitoneally injected with 100 μL A. hydrophila suspension (7× 10⁶ CFU/mL) after knockdown or overexpression 3 d. The experiments were carried out in triplicate. After challenge for 24 h, the tissues were sampled respectively in the aseptic condition, and used to count the bacterial number in samples by plate count and detect the expression of CcIL-1β by RT-qPCR. Blood was collected from the caudal vein and centrifuged at 3000 rpm at 4 °C for 10 min to collect the serum, which was used for ELISA to measure the content of CcIL-1β.

EPC cells (5 × 10⁶ cells/well) were seeded into 6-well cell culture plates overnight, and then the siControl, siCcGSDMEa, pEGFP-C1 and GFP-CcGSDMEa-FL plasmids (1000 ng/well) were transfected when the cells reached 90% confluency, respectively. The untransfected cells were used as the control. The efficiency of knockdown or overexpression in vitro was analyzed after 24 h and 48 h, respectively. The untransfected cells and the cells transfected successfully with the above plasmids were divided into two subgroups, respectively. One group was stimulated by 10 μL A. hydrophila (1×10⁴ CFU/well), and the other group was added with isometric PBS. After 3 h, the content of LDH and CcIL-1β in cell culture medium was assessed to determinate the role of CcGSDMEa in regulating IL-1β secretion.

The levels of CcGSDMEa and CcIL-1β in serum were measured according to the method of ELISA (6, 40). Briefly, 96-well Immuno-Maxisorp plates were coated with the serum (1:1) in 2× coating buffer (0.05 M carbonate buffer, pH 9.6) overnight at 4 °C. Next, 10% skimmed milk powder was used as the blocking agent at 37 °C for 2 h, and then were subsequently washed three times (wash buffer: 20 mM PBST, 274 mM NaCl, 5.4 mM KCl, 20 mM Na₂HPO₄, 3.6 mM KH₂PO₄, 0.5 mL Tween-20, PH 7.4). Polyclonal antibodies against CcGSDMEa or CcIL-1β (1: 500, prepared in our laboratory) were added to the wells to incubation at 37 °C for 2 h and an addition wash. After incubation with horseradish peroxidase labeled Goat anti-Mouse IgG/HRP (Solarbio, China) at 37 °C for 1 h, the plates were washed again. 3,3’,5,5’ -tetramethylbenzidine (TMB, Sigma, USA) was used as the substrate for color development at 37 °C for 30 min. The reaction was terminated by adding 50 μL of termination solution (2M H₂SO₄). The OD₄₅₀ absorbance value was detected in a microplate reader (PerkinElmer, USA) within 10 min, and the amount of excretive CcGSDMEa and CcIL-1β (pg/mL) was estimated by comparing to reference curves constructed using the corresponding standards.

The protein samples (30 μg) were separated by SDS-PAGE before being transferred to a PVDF membrane (Millipore, USA) according to traditional protocol. The membranes were blocked in 5% w/v nonfat dry milk in TBST and then washed three times using TBST for 10 min every time. The mouse anti-GFP Ab (1:10,000, GenScript, China) or mouse anti-β-actin Ab (1:10,000, Beyotime, China) was incubated overnight at 4°C and washed another three times, then followed by incubation with the appropriate secondary goat anti-mouse IgG-HRP Ab (1:5000; Solarbio, China) and washed three times again, whereafter detected by super ECL detection reagent (Millipore, USA) using gel imaging system (Biorad, USA).

All values in this study were presented as means ± standard deviation (M ± SD) (n=3). The data statistical analysis was performed by SPSS 20.0 and GraphPad Prism 5.0. The statistical significance was determined using Student’s t test and one-way ANOVA followed by Tukey’s test. Significant difference was set as P <0.05 (*), P <0.01 (**), or P <0.001 (***).

In this study, CcGSDMEa (GenBank accession number: ON981707), a homologous GSDME gene, was obtained from common carp, which is distributed within 7320 bp genomic fragment on chromosome A19 with 9 exons and 8 introns. The location and transcription direction of the adjacent genes to CcGSDMEa were consistent with those of DrGSDMEa and HsGSDME, and it was suggested that genes adjacent to CcGSDMEa loci shared an overall conserved chromosome synteny with those of humans and zebrafish ( Figure 1A ). The full length of CcGSDMEa cDNA sequence is 2736 bp with a 1599 bp ORF, a 177 bp 5’-untranslated region and a 960 bp 3’-untranslated region ( Figure S1 ). The ORF encoded 532 amino acids was predicted with molecular weight of 58.57 kDa and theoretical isoelectric point of 4.92. It displayed high identities (58.9%) and similarities (76.0%) to zebrafish DrGSDMEa, but low identities (29.3% and 24.2%) and similarities (51.2% and 45.8%) to zebrafish DrGSDMEb and human HsGSDME, respectively ( Table 3 ).

To better understand the function of CcGSDMEa, the features of domain and tertiary architecture were characterized. The CcGSDMEa contained one gasdermin domain, which is consistent with the other GSDM family members except for DrGSDMEa ( Figure 1B ). In addition, the CcGSDMEa shared highly conserved tertiary architectures with zebrafish DrGSDMEa, which had a conserved N-terminal pore-forming domain, C-terminal auto-inhibitory domain and a flexible and pliable hinge region ( Figure 1C ). In linker regions, the predicted tetrapeptide motif, which was specifically recognized by caspases, was ₂₅₂SEVD₂₅₆. This was the same as the cleavage site of DrGSDMEa. Phylogenetic tree revealed that CcGSDMEa was grouped with zebrafish DrGSDMEa with a high score of bootstrap tests. Interestingly, CcGSDMEa was not grouped with GSDMEb of common carp, but was clustered into a separate clade ( Figure 1D ), which provided us a better clue for future analysis of GSDME evolutionary function in teleost. In addition, both members of the deafness-associated gene (i.e., GSDME and PJVK) were clustered together, and they were distant from the other members (GSDMA, GSDMB, GSDMC and GSDMD).

CcGSDMEa was widely expressed in the tested tissues and exhibited a tissue-specific spatial expression pattern. CcGSDMEa was highly expressed in the skin and gill, followed by the head kidney and trunk kidney, and the lowest expression was in brain ( Figure 2A ). Moreover, the expression levels in immune-related tissues (head kidney, spleen, skin, intestine and gill) were markedly higher than those in heart, liver and brain.

To further analyze the expression profile of CcGSDMEa in response to bacterial infection, A. hydrophila was chosen to stimulate the juvenile common carp. The peaks of CcGSDMEa expression were at 3 h in liver, at 6 h in gill, head kidney and spleen, and at 12 h in intestine and skin ( Figures 2C-H ). Although the peaks of mRNA expression levels were different in various tissues and appeared at the different time points, the expression levels of CcGSDMEa were increased after A. hydrophila infection, compared with the control. The protein expression level of CcGSDMEa showed a trend of increasing at first and then decreasing with the time prolongation in serum, and the expression peak was observed at 12 h ( Figure 2B ). The expression levels among different time points showed significant differences. Meanwhile, the protein expression levels of CcGSDMEa were notably increased at the different time points compared with the control.

To better understand the cleavage mechanism of CcGSDMEa, each of the CcCaspase-1a, CcCaspase-1b, CcCaspase-3a, CcCaspase-3b, CcCaspase-7a and CcCaspase-7b plasmids was co-transfected with CcGSDMEa-FL into EPC cells, and then treated with the dimerizing drug AP20187 for 6 h. The results showed that CcCaspase-1b, CcCaspase-3a, CcCaspase-3b, CcCaspase-7a and CcCaspase-7b could cleave CcGSDMEa to produce CcGSDMEa-NT, except for CcCaspase-1a ( Figure 3A ). The LDH released from cells co-transfected CcCaspase-1b/3a/3b and CcGSDMEa, were dramatically higher than that released from cells only transfected CcGSDMEa ( Figure 3B ). HEK293T cells were co-transfected with CcCaspase-1a/1b/3a/3b/7a/7b and CcGSDMEa, and the morphology of cell membrane was observed. The results showed that, except for the co-transfection of CcCaspase-1a and CcGSDMEa, membrane pores could be observed in the co-transfected cells ( Figure 3C ). It was indicated that any one of CcCaspase-1b/3a/3b/7a/7b could cleave CcGSDMEa to generate CcGSDMEa-NT, which could oligomerize on the cell membrane and penetrate the plasma membrane.

To explore the pore forming function of CcGSDMEa, the CcGSDMEa-FL/NT/CT recombinant plasmids with GFP-tag were constructed ( Figure 4A ). The different recombinant plasmids of CcGSDMEa were transfected into HEK293T cells to observe the morphological changes and LDH release. The results showed that the release of LDH from cells expressing CcGSDMEa-NT was markedly higher than that of cells expressing CcGSDMEa-FL or CcGSDMEa-CT and pEGFP-C1 vector ( Figure 4B ). Cytomembrane pores and rupture phenomenon were also observed only in the transfected cells with CcGSDMEa-NT plasmids ( Figure 4C ). It was suggested that only CcGSDMEa could punch holes in the membrane and disrupt the integrity of cytomembrane through its NT domain.

In addition, the subcellular localizations of CcGSDMEa-FL, NT and CT in HEK293T cells were varied ( Figure 5A ). The CcGSDMEa-FL was localized in the cytoplasm and cytomembrane, the CcGSDMEa-NT was only situated in the cytomembrane, and the CcGSDMEa-CT was distributed in the whole cells. We also found that CcGSDMEa-NT and DiI-labeled cytomembrane were completely colocalized ( Figure 5B ), and it was indicated that CcGSDMEa-NT could accumulate on the cellular membrane.

In order to detect the bactericidal activity of CcGSDMEa, CcGSDMEa-FL, CcGSDMEa-NT and CcGSDMEa-CT were subcloned into pET-32a vector, respectively ( Figure 6A ). The recombinant plasmids rCcGSDMEa-FL/NT/CT were expressed in E. coli by IPTG-induced manner. Of note, the number of E. coli expressing rCcGSDMEa-NT was markedly reduced compared with that of expressing rCcGSDMEa-FL or rCcGSDMEa-CT after IPTG induction ( Figures 6B, C ). The growth curves of E. coli expressing different recombinant plasmids of CcGSDMEa were also been analyzed. The OD₆₀₀ of E. coli expressing rCcGSDMEa-NT indicated little change at different time points after IPTG induction, but the OD₆₀₀ values of E. coli expressing rCcGSDMEa-FL, rCcGSDMEa-CT and pET-32a vector showed an increasing trend with time prolongation after IPTG induction ( Figure 6D ). It was demonstrated that CcGSDMEa-NT could perform bactericidal function.

To better understand the role of CcGSDMEa in regulating CcIL-1β, the CcGSDMEa in fish was knocked down or overexpressed, and the expression of CcIL-1β at the mRNA and protein levels were detected. The knockdown and overexpression of CcGSDMEa were confirmed based on the results of CcGSDMEa mRNA expression levels in vivo. Figure 7 showed that the CcGSDMEa mRNA expression levels in GFP-CcGSDMEa group were markedly increased compared with the levels in pEGFP-C1 group at 3 d and 7 d ( Figures 7A, B ). However, the expression levels of CcGSDMEa in siCcGSDMEa group were lower than those in siControl group ( Figures 7C, D ).

In CcGSDMEa overexpression group, the CcIL-1β expressions at the mRNA and protein levels were notably higher compared with the expression levels in pEGFP-C1 group ( Figures 7E, G ). In CcGSDMEa knockdown group, the expression levels of CcIL-1β decreased compared with the siControl group ( Figures 7F, H ). These results suggested that CcGSDMEa could promote CcIL-1β expression.

As mentioned above, the expression of CcGSDMEa increased in fish when stimulated by A. hydrophila ( Figure 2 ). After CcGSDMEa was overexpressed or knocked down, the bacterial burdens were examined in tissues of fish after A. hydrophila challenge for 24 h. As shown in Figure 8 , the bacterial burdens were obviously lower in head kidney, spleen, gill, liver and trunk kidney in GFP-CcGSDMEa overexpression group than those in the pEGFP-C1 group. While, in siCcGSDMEa group, the bacterial burdens were observably increased in head kidney, gill, liver, spleen and trunk kidney, compared with those in siControl group. It was indicated that the CcGSDMEa could prevent the colonization of A. hydrophila and minish the bacterial burdens in fish.

After EPC cells challenged with A. hydrophila, the expression levels of CcGSDMEa and CcIL-1β significantly increased, compared with the control group ( Figure 9A ), it was suggested that A. hydrophila could trigger the innate immune responses of EPC cells. Similarly, the contents of CcIL-1β protein in the cell culture medium were also increased after A. hydrophila challenge compared with the control group ( Figure 9B ), and it was manifested that A. hydrophila challenge could promote the secretion of CcIL-1β. Moreover, the LDH released from EPC cells treated by A. hydrophila were observably higher than that in the control group ( Figure 9C ), and indicted that A. hydrophila challenge might destroy cytomembrane integrity of EPC cells.

To clarify the effect of CcGSDMEa on CcIL-1β secretion, the CcGSDMEa was overexpressed or knocked down in EPC cells, and then the contents of CcIL-1β protein in cell culture medium were examined after A. hydrophila stimulation. As shown in Figures 9D, E , the overexpression and knockdown of CcGSDMEa were both affirmed. After A. hydrophila challenge, in the CcGSDMEa overexpression group, expression level and secretion content of CcIL-1β and the release of LDH were significantly increased, compared with the pEGFP-C1 group ( Figures 9F-H ). It was illustrated that the overexpression of CcGSDMEa could intensify the effect of A. hydrophila challenge on CcIL-1β secretion and LDH release. However, in siCcGSDMEa group, after A. hydrophila challenge, the expression level and secretion content of CcIL-1β and the release of LDH were notably decreased compared with the siControl group. It was implied that the knockdown of CcGSDMEa could retard the effect of A. hydrophila on CcIL-1β secretion and LDH release. Taken together, these results revealed that CcGSDMEa played an important role in regulating the secretion of CcIL-1β.