MLKL (designated PoMLKL) and RIPK1/3 (designated PoRIPK1/3) homologs were identified in teleost Paralichthys olivaceus. PoMLKL contains an N-terminal four-helix bundle (4HB) domain and a C-terminal pseudokinase domain (PsKD) connected by an interdomain brace region (BR) ( Figure 1A ). The protein architecture of PoMLKL is similar to that of human and mouse MLKL (21). PoRIPK1 and PoRIPK3 exhibit similar protein architectures, both harboring an N-terminal protein kinase C (PKC) domain and a C-terminal RIP homotypic interaction motif (RHIM) ( Figure 1A ). PoMLKL, PoRIPK1 and PoRIPK3 were expressed abundantly in immune related tissues, including spleen, blood, head kidney and gill, in P. olivaceus ( Figure 1B ). When the fish were infected with Edwardsiella tarda, a serious pathogen causing haemorrhagic septicaemia and gastroenteritis in fish, increasing bacterial load was detected in spleen and head kidney in a time dependent manner ( Figure 1C ). In the infected fish, the expressions of PoMLKL, PoRIPK1, and PoRIPK3 in spleen were significantly upregulated at 24 hpi, 12 hpi and 24 hpi, 6 hpi and 12 hpi, respectively ( Figure 1D ). The expressions of PoMLKL, PoRIPK1, and PoRIPK3 in head kidney were significantly upregulated at 12 hpi and 24 hpi, 24 hpi, and 6 hpi to 24 hpi, respectively ( Figure 1E ).

When ectopically expressed in HEK293T cells, PoRIPK1 distributed evenly in the cytoplasm, while PoRIPK3 and PoMLKL formed small puncta and larger aggregates, respectively, in the cytoplasm ( Figures 2A–C ). Compared with the control cells, the cells expressing PoRIPK1, PoRIPK3, and PoMLKL showed normal morphologies with little LDH release ( Figures 2A–C ). To examine whether PoMLKL possessed necroptosis-inducing capacity, two PoMLKL truncates were constructed, one consisting of the 4HB domain and the other consisting of 4HB plus the BR region (4HB-BR) ( Figure 2D ). Compared with the cells expressing full length (FL) PoMLKL, the cells expressing 4HB or 4HB-BR underwent necroptosis, accompanying with osmotic swelling and massive release of LDH ( Figures 2E, F ). When expressed in Escherichia coli, 4HB and 4HB-BR inhibited bacterial growth, whereas PoMLKL had no effect on bacterial growth ( Figures 2G, H ).

To determine the activation of PoMLKL-mediated necroptosis, we co-expressed PoMLKL with PoRIPK1 and/or PoRIPK3 in HEK293T cells. Confocal microscopic analysis showed that co-expression of PoRIPK1 and PoMLKL did not lead to cell death, but co-expression of PoRIPK3 and PoMLKL induced necroptosis and massive release of LDH ( Figures 3A, B ). PoMLKL was colocalized with PoRIPK3 but not with PoRIPK1 ( Figure 3A ). When PoMLKL, PoRIPK1, and PoRIPK3 were co-expressed, PoRIPK1 colocalized with PoRIPK3 and PoMLKL, and necroptosis occurred with higher LDH release ( Figures 3A, B ). The sequence alignment showed that there exist two potential phosphorylation residues, T360 and S361, in PoMLKL, which correspond to the phosphorylation residues in human MLKL (HsMLKL) (T357 and S358) and mouse MLKL (MmMLKL) (S345 and S347) ( Figure 3C ). Previous studies reported that phosphorylation of these residues in HsMLKL and MmMLKL incurred the conformation switch from inert to active state (22). In order to determine the role of these residues in PoMLKL-mediated necroptosis, we mutated either or both residues to Glu, which mimicked the negatively charged phosphate group. When expressed in HEK293T cells, PoMLKL bearing T360E or S361E mutation induced necroptosis and LDH release, and PoMLKL bearing T360E-S361E double mutation induced necroptosis and more LDH release ( Figures 3D, E ). We next mutated PoMLKL T360 and/or S361 to Ala, which could not be phosphorylated by PoRIPK3. When the mutant T360A or S361A was co-expressed with PoRIPK3, LDH release was significantly decreased compared with cells co-expressing PoMLKL and PoRIPK3 ( Figure 3F ). When the double mutant T360A-S361A was co-expressed with PoRIPK3, little LDH release was observed ( Figure 3F ). Furthermore, unlike PoMLKL, which was phosphorylated by the co-expressed PoRIPK3, the T360A-S361A mutant was not phosphorylated by PoRIPK3 ( Figure 3G ). Besides T360 and S361, other residues of PoMLKL, including N105 and D106 in the 4HB domain and L293, K231, and G318 in the CT domain, were also mutated to examine their effect on necroptosis. When co-expressed with PoRIPK3, the mutants of L293P, K231R, G318D, and N105A-D106A induced significantly reduced LDH release ( Figure 3H ). Moreover, the 4HB-BR domain bearing N105A-D106A mutation lost the ability to kill bacteria ( Figures 3I, J ).

To explore the role of RHIM in PoRIPK1/3 activation of PoMLKL, the mutants PoRIPK1ΔCT and PoRIPK3ΔCT were constructed, which lacked the CT region containing RHIM ( Figure 4A ). When PoRIPK1ΔCT was co-expressed with PoRIPK3 and PoMLKL in HEK293T cells, PoRIPK1ΔCT was not able to colocalize with PoRIPK3 and PoMLKL, but PoRIPK1ΔCT had no effect on PoRIPK3 colocalization with PoMLKL and necroptosis induction ( Figures 4B, C ). It should be noted that, compared with cells co-expressing PoRIPK1, PoRIPK3, and PoMLKL, cells co-expressing PoRIPK1ΔCT, PoRIPK3, and PoMLKL released significantly lower LDH ( Figure 4B ). When co-expressed with PoRIPK3ΔCT, neither PoRIPK1 nor PoMLKL could interact with PoRIPK3ΔCT, and no cell death or LDH release was observed ( Figures 4B, C ). The tetrapeptide motif of VQVG (human) and VQIG (mouse) in the RIPK3 RHIM domain are proved to be essential for homo-oligomerization and MLKL activation (3, 23, 24). Mutation of the corresponding tetrapeptide (VQSG) in PoRIPK3 to AAAA significantly decreased the induction of PoMLKL-mediated necroptosis ( Figures 4D, E ). In addition to VQSG, the importance of four residues (K46, D146, S231 and S232) in the N-terminal PKC domain of PoRIPK3 was also analyzed. Compared with PoRIPK3, the mutants of K46A, D146N, and S231A-S232A exhibited attenuated ability to activate PoMLKL, indicated by significantly decreased LDH release ( Figure 4E ).

In order to investigate the effect of necroptosis on bacterial infection, MLKL-mediated necroptosis in flounder cells (FG-9307) was inhibited by the MLKL specific inhibitor GW806742X or NSA ( Supplementary Figure S1 ). In the presence of the inhibitor, E. tarda infection induced cell death was significantly decreased, and the intracellular bacterial load was significantly increased ( Figures 5A, B ). We further knocked down the expression of PoMLKL by using siRNA ( Supplementary Figure S2 ), and found that E. tarda infection-induced cell death significantly decreased, and the intracellular bacterial number was significantly increased ( Figures 5C, D ). When flounder were infected with E. tarda in the presence of MLKL inhibitors, the bacterial load in spleen and head kidney significantly increased, and host mortality occurred early ( Figures 5E–G ). These results collectively indicated that necroptosis exerted anti-microbial effects in fish immunity.

Clinically healthy Paralichthys olivaceus were obtained from a fish farm. The fish were kept at 20°C in aerated seawater as reported previously (43). To obtain tissue samples, the fish were put to death using an excessive amount of tricaine methane sulfonate (Sigma, St. Louis, MO). All live animal experiments were approved by the Ethics Committee at the Institute of Oceanology, Chinese Academy of Sciences.

The GenBank ID of PoRIPK1, PoRIPK3, and PoMLKL are 109624097, 109645932 and 10964593, respectively. PCR was employed to amplify the codon-optimized protein coding sequences (CDS) of PoMLKL, PoRIPK1, and PoRIPK3. Site-directed mutations of PoRIPK3 K46A, D146N, S231N/S232N and ⁴⁵⁵VQSG⁴⁵⁸−⁴⁵⁵AAAA⁴⁵⁸, and PoMLKL N105A-D106A, K231R, L293P, G318D, S360A, S361A, S360A-S361A, S360E, S361E, and S360E-S361E were performed using the Hieff Mut Site-Directed Mutagenesis Kit (Yeasen, Shanghai, China). The accuracy of the mutations was confirmed through sequencing analysis. The primers used are listed in Supplementary Table S1 .

To detect the expression or phosphorylation of PoMLKL, immunoblotting was conducted as reported previously (44, 45). In brief, protein samples were subjected to SDS-PAGE and subsequently transferred onto nitrocellulose membranes. Following a 5% skim milk blocking step, the membranes were probed with primary antibody and then with a secondary antibody labeled with horseradish peroxidase (HRP). The primary antibodies utilized were mouse antibodies targeting HsMLKL (phospho S358) (1:3000 dilution) and β-actin (1:5000 dilution). The secondary antibody was goat anti-mouse IgG-HRP antibody (1:5000 dilution). All antibodies were obtained from Abcam in Cambridge, MA, USA.

HEK293T cells (American Type Culture Collection) were cultivated in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (Gibco, Renfrewshire, UK) at 37°C in a 5% CO₂ environment. Edwardsiella tarda were cultured in Luria Bertani broth (LB) medium with shaking at 28°C (180 rpm). Transient transfection of plasmid into HEK293T cells was performed using Polyjet Transfection Reagent (Signagen, USA). For gene overexpression experiments, pmCherry-N1 vectors (Clontech, Mountain View, CA, USA) expressing PoMLKL variants, PoRIPK1, or PoRIPK1ΔCT were constructed, pEGFP-C1 vectors (Clontech, Mountain View, CA, USA) expressing PoRIPK3 variants were constructed, and pCMV-N-BFP vector (Beyotime, CN) expressing PoMLKL was constructed. HEK293T cells were transfected with these constructs, and the control cells were transfected with the backbone vectors. All plasmids were extracted using endotoxin-free plasmid kit (Sparkjade Biotechnology Co. Ltd., Shandong, China).

To assess cell death, the activity of lactate dehydrogenase (LDH) released into the cell culture supernatant was measured using the CytoTox96 LDH release kit (Promega, Leiden, Netherlands) as reported previously (46). The percentage of cytotoxicity was calculated using the formula: percent cytotoxicity = 100 × (experimental sample - culture medium background)/(maximum LDH release - culture medium background).

Microscopy display was performed as previously reported (47). To examine cell morphology, the cells were plated on 24-well plate (Costar, Corning, NY, USA) at about 60% confluency and subjected to the indicated treatment. The bright-field and fluorescent views of the cells were recorded using a Carl Zeiss LSM 710 confocal microscope (Carl Zeiss, Jena, Germany).

qRT-PCR was performed as reported previously (48). To explore of gene expression in fish tissue, aseptic collection of tissues (blood, gill, spleen, head kidney, liver, skin, intestine, heart, brain and muscle) was performed with three fish specimens. Total RNA was extracted from tissues using Trizol reagent (Invitrogen, Carlsbad, CA, USA). cDNA synthesis was carried out using the First Strand cDNA Synthesis Kit (ToYoBo, Japan) as per the manufacturer’s instructions. qRT-PCR was conducted on an Eppendorf Mastercycler (Eppendorf, Hamburg, Germany) with the SYBR ExScript qRT-PCR Kit (Vazyme Biotech Co. Ltd., Nanjing, China). To investigate gene expression in E. tarda-infected fish, P. olivaceus were randomly divided into two groups (35 fish per group) and intramuscularly injected with E. tarda (1 × 10⁶ CFU/fish) or PBS. Tissue samples were collected at 6, 12, and 24 h post-infection (hpi) (9 fish/each time point) and used for qRT-PCR as above. The mRNA levels of the target genes were normalized to the expression of β-actin.

The small interfering RNA (siRNA) sequence targeting PoMLKL (siRNA-PoMLKL) was provided by Sangon (Shanghai, China). The scrambled control (siRNA-NC) served as a negative control and was obtained from the same company. SiRNA was introduced into FG-9307 cells using RNAiMAX transfection reagent (Invitrogen, USA). The interference efficiency was determined by qRT-PCR as above.

FG-9307 cells were plated on 96-well plate (Costar, Corning, NY, USA) at about 60% confluency. E. tarda was cultured in LB broth at 28°C to OD₆₀₀ 0.8. The bacteria were washed with PBS and resuspended in PBS. FG-9307 cells were pretreated with MLKL inhibitors (1 μM GW806742X or 1 μM NSA) (MedChem Express, NJ, USA) for 6 h, or transfected with siRNA as described above for 24 h. The cells were then infected with or without (control) E. tarda (MOI = 10:1) at room temperature for 6 h. Cell death was determined by LDH release assay. To determine bacterial load, cells were lysed with Triton X-100 solution (final concentration of 1%), and the lysate was applied to LB plates containing 50 μg/ml tetracycline. The plate was incubated at 28°C for 24 h, and the number of colonies on the plate was counted.

A total of 90 P. olivaceus specimens were randomly divided into three groups, with 30 fish in each group. The groups received intraperitoneal injections of either 20 mg/kg NSA (MedChem Express, NJ, USA) or 20 mg/kg GW806742X (MedChem Express, NJ, USA), while the control group was injected with equal volume of sterile PBS. At 8 h after injection, all fish were intramuscularly inoculated with 5 × 10⁵ CFU of E. tarda. At 12, 24, and 48 hpi, head kidney and spleen samples were collected (nine fish/each time point). For each time point, the tissues of three fish were mixed and homogenized in sterile PBS. The homogenates were diluted in PBS in a stepwise manner and then spread on LB agar plates. After incubating the plates at 28°C for 18 h, the number of colonies on the plate was counted.

A total of 60 P. olivaceus specimens were divided into three groups (20 fish each group). The three groups were intraperitoneally injected with NSA, GW806742X, or PBS as above. At 8 h after injection, all groups were infected with E. tarda as above. The fish were then monitored daily for mortality.

E. coli Transetta (DE3) was transformed with pET-30a expressing PoMLKL variants (FL, 4HB, 4HB-BR, 4HB-BR, and N105A-D106A). The transformants were grown in LB medium supplemented with 50 mg/ml kanamycin until OD₆₀₀ 0.4-0.6. The cells were then diluted and cultured on LB agar plates containing 50 mg/ml kanamycin with or without 0.2 mM IPTG. After overnight incubation at 37°C, the colony-forming units (CFU) on the plates were counted and statistically analyzed. The transformants were also cultured in LB medium to OD₆₀₀ 0.5, and then 0.2 mM IPTG was added to the culture. Bacterial growth was recorded every hour.

The two-sample Student t test was used for comparisons between groups. Log-rank was used for the analysis of fish survival. Statistical analysis was performed with GraphPad Prism 7 software. (https://www.graphpad.com/). Statistical significance was defined as P < 0.05.