Ninjurin-1 executes plasma membrane rupture and impairs anti-microbial immunity in an early jawed vertebrate Authors: Xiaotong Jin, Liming Yuan, Chunyan Sun, Shuai Jiang Published: 2025 ABSTRACT Plasma membrane rupture (PMR), once considered a passive event in necrosis, is actively executed by ninjurin-1 (NINJ1). Although PMR is a well-documented phenomenon across animal species, its molecular mechanism and immunological function in early diverging jawed vertebrates remain unclear. Herein, we identify a functional homolog of NINJ1 in turbot. Ectopic expression of turbot NINJ1 effectively lyses the plasma membrane, leading to rapid PMR and massive release of cellular contents. Given that turbot NINJ1 has two transmembrane helices, it could distribute within membrane-bound organelles and the plasma membrane as PMR progresses. Co-expression with turbot gasdermin significantly amplifies NINJ1-mediated PMR, even when both proteins are present at non-cytotoxic levels. Structural analysis reveals the importance of its extracellular and transmembrane helices, with specific functional residues crucial for its PMR-inducing activity. Bacterial infection upregulates NINJ1 and activates pyroptosis signaling pathway, and provokes strong proinflammatory responses. Inhibiting NINJ1 improves fish survival, whereas blocking pyroptosis actually dampens host antimicrobial immunity. Collectively, the present findings delineate a mechanistically distinct role of NINJ1 in regulating programmed cell death in turbot, highlighting the complex and bifunctional nature of cell death in teleost immunity.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00018-025-06020-0. Introduction Programmed cell death represents a genetically encoded process crucial for immune homeostasis by precisely removing infected or damaged cells [1]. Apoptosis, a classical form of programmed cell death, is featured by the formation of membrane-enveloped apoptotic bodies [2, 3]. These cellular structures are recognized and efficiently phagocytosed by macrophages, minimizing the leakage of intracellular proinflammatory contents and thereby preventing the induction of inflammation [4, 5]. Different from apoptosis, other forms of programmed cell death, e.g., pyroptosis, are characterized by membrane perforation and subsequent cell lysis, leading to massive release of intracellular proinflammatory molecules [6, 7]. When pyroptosis occurs, the executioner pore-forming protein, gasdermin, is typically activated by proteolytic cleavage [8, 9]. The unleashed N-terminal fragments are then recruited to the plasma membrane, where they oligomerize to form transmembrane channels and induce membrane perforation [10–12]. The perforated cell membrane results in osmotic cell swelling, featured by membrane blebbing and formation of balloon-like morphological structure, ultimately culminating in plasma membrane rupture (PMR) [13, 14]. Ninjurin-1 (NINJ1) is an approximately 16 kDa plasma membrane protein belonging to the NINJ family, which were initially identified as adhesion molecules critical for promoting axonal growth following injury [15]. Most recent studies reported the pivotal role of NINJ1 as the key executioner of PMR in lytic cell death, a process that has long been thought to be a passive event driven by osmotic swelling [16, 17]. Upon receiving diverse death stimuli, such as gasdermin-mediated membrane perforation, NINJ1 transitions from an inactive homodimeric state to large, fence-like filaments embedded within the plasma membrane [18]. These filaments actively wrap around and solubilize membrane fragments, leading to the formation of large holes in the plasma membrane. Unlike the smaller pores (~ 20 nm) formed by gasdermin oligomerization, the NINJ1 filaments effectively “cut” and release membrane disks [19, 20]. This membrane targeted disruption leads to the formation of large holes (~ 45 nm) in the plasma membrane, leading to rapid PMR and massive release of intracellular proinflammatory factors [21]. In addition to gasdermin-mediated pyroptosis, some other types of programmed cell death, e.g., necroptosis and even non-lytic apoptosis, are also capable of activating NINJ1-mediated PMR [22]. For instance, oligomerization of mixed lineage kinase domain-like protein (MLKL), a key effector of necroptosis, can perforate the cytoplasmic membrane and activate NINJ1-dependent PMR in a way similar to that of gasdermin [23, 24]. Furthermore, in chemotherapy drug- or Fas ligand-induced apoptosis, NINJ1, but not gasdermin D (GSDMD), plays the key role in mediating the release of large intracellular molecules [25]. Similar results were also observed in lytic cell death induced by bacterial pore-forming toxins like listeriolysin or streptolysin [16, 26]. By contrast, cell lysis caused by non-biological stressors such as freezing and thawing is independent of NINJ1, as knockout of NINJ1 has little effect on releasing the intracellular contents [16]. These findings collectively demonstrate that NINJ1-mediated PMR is an active and highly-regulated cellular process. Teleost fish represent early-diverging jawed vertebrates, having diverged from human lineages approximately 400 mya [27, 28]. While gasdermin-executed pyroptosis and its related cell lysis have been investigated in teleost species [29–32], the precise role and mechanism of PMR in teleost immunity remain elusive. In the present study, we explore the immunological function and molecular mechanism of NINJ1-mediated PMR in a teleost fish, the turbot Scophthalmus maximus. When expressed in cells, turbot NINJ1 exhibits efficient membrane lysis capacity, leading to rapid PMR and substantial release of intracellular molecules. Co-expression of turbot gasdermin Ea (GSDMEa) enhances NINJ1-mediated cell lysis, even when both proteins are expressed at sublethal levels. Structural analysis reveals the critical role of the amphipathic helical region and the transmembrane helices for NINJ1 functionality. Infection with the serious fish pathogen Vibrio anguillarum up-regulates NINJ1 and the pyroptotic signaling pathway, and provokes a proinflammatory immune response. This immune response, however, appears to have dual roles in host defense: inhibition of NINJ1 prevents mortality from bacterial infection, while blocking pyroptosis increases lethality. The paradoxical roles of NINJ1 and GSDMEa highlight the antagonistic yet fine-tuned regulation machinery of programmed cell death in turbot, providing new insights into the proinflammatory immunity against bacterial infection. Results Turbot NINJ1 induces rapid plasma membrane rupture and lytic cell death Plasma membrane rupture (PMR), a hallmark of necrotic cell death, has long been considered a passive osmotic lysis event. In the present study, we identified a homolog of NINJ1, a recently recognized executioner of PMR, from a teleost fish, the turbot (Scophthalmus maximus). When expressed in HEK293T cells, turbot NINJ1 induced plasma membrane swelling, characterized by a prominent bubble-like morphological structure (Fig. 1A), indicating damages to the cytoplasmic membrane. Concurrently, large intracellular contents, i.e., lactate dehydrogenase (LDH, ~ 140 kDa), were massively released from NINJ1-expressing HEK293T cells (Fig. 1B), confirming the occurrence of PMR. Notably, NINJ1 was highly recruited to the cell membrane, particularly at sites of membrane protrusions and bubble-like structures (Fig. 1C). Using time-lapse live-cell imaging, we observed that NINJ1 was initially highly enriched at the cell membrane (Fig. 1D; Supplemental Movie 1, 2). At this early stage, the cells were still alive, featured by the normal morphologies and resistance to nuclear dye Sytox Green staining. In a very short time (16 min after the first imaging), NINJ1 induced rapid membrane blebbing and swelling, and the cells were still negative to Sytox Green and Annexin V staining (Fig. 1D). Soon after the membrane blebbing (26 min after the first imaging), these membrane-associated small bubbles quickly fused, forming an entire large plasma membrane bubble. At this moment, Sytox Green infiltrated and stained the nucleus of NINJ1-expressing cells. After about 3 h, these cells were observed with heavy staining of the inner membrane leaflet probe Annexin V (Fig. 1D). Given the rapid nature of NINJ1-mediated PMR, Annexin V staining may not reflect phosphatidylserine flipping but rather binding of phosphatidylserine after infiltration through NINJ1-cutting holes on the plasma membrane. In NINJ1-transfected HEK293T cells, membrane damage, indicated by Sytox Green staining, occurred in a large proportion at 12 h post-transfection, whereas intracellular LDH was not released at this time point (Fig. 1E, F). The kinetics of Sytox Green staining peaked at approximately 24 h, preceding the peak of LDH release at 48 h, suggesting that NINJ1-mediated membrane leakage occurs prior to catastrophic membrane disruption. It is noteworthy that following membrane osmotic swelling, NINJ1 was highly recruited to the nucleus (180 min after the first imaging), and the nucleus remained intact at this time moment (Fig. 1D). However, at the late stage (48 h after expression) of NINJ1-mediated membrane lysis, the nucleus was severely disrupted and wrapped by the disintegrated plasma membrane, ultimately leaving a shrunken cellular corpse (Fig. 1G). Besides its cytotoxic effects on eukaryotic cells, we also investigated NINJ1’s cytotoxicity to prokaryotic cells by expressing it in Escherichia coli. Compared with non-induced controls, E. coli cells expressing NINJ1 yielded much fewer bacterial colonies, indicating a bactericidal activity for turbot NINJ1 (Fig. 1H). Considering that the N-terminal amphipathic helical region of NINJ1 can insert into the plasma membrane, and promotes NINJ1 oligomerization and PMR [19, 33], we thus explore whether it has cytotoxicity to bacterial cells. When extracellularly incubated with various bacteria, two artificially synthesized polypeptides (α-helix 1 and 2) corresponding to the amphipathic region of turbot NINJ1 did not exhibit obvious bactericidal activity (Fig. 1I), indicating that the N-terminal amphipathic helix is necessary but not sufficient for the execution of prokaryotic PMR.Fig. 1Turbot NINJ1 induces rapid plasma membrane rupture. (A) C-terminally mCherry tagged turbot NINJ1 was expressed in HEK293T cells for 24 h. Cells expressing mCherry backbone vector were taken as the control. Images were captured with a confocal microscope. Scale bar, 10 μm. (B) The release of lactate dehydrogenase (LDH) was measured from cells expressing mCherry or mCherry tagged NINJ1, with non-transfected cells as the control. Values are the means ± SD. n = 3. ****p < 0.0001. (C) HEK293T cells were transfected with NINJ1 for 24 h, and then stained with nuclear dye Hoechst 33,342 (H33342) and membrane dye DiO. NINJ1-mediated plasma membrane blebbing was observed with a confocal microscope. Scale bar, 20 μm. (D) HEK293T cells expressing mCherry tagged NINJ1 was stained with nuclear dye Sytox Green and the phosphatidylserine binding Annexin V. Time-lapse live-cell images, including red fluorescence (mCherry), green fluorescence (Sytox green), magenta fluorescence (Alexa Fluor 647 labeled Annexin V) and bright field, were captured using a confocal microscope. The black, white and blue arrows indicate the swelling membrane, NINJ1 membrane localization and the nucleus respectively. Scale bar, 10 μm. (E and F) HEK293T cells were transfected with backbone vector or vector expressing NINJ1, and the relative fluorescence units of Sytox Green (E) and release of LDH (F) were determined at 12, 24, 36, and 48 h post-transfection. Values are the means ± SD. n = 3. ***p < 0.001, ****p < 0.0001. (G) Cells expressing mCherry tagged NINJ1 was observed with a confocal microscope at 48 h after transfection. Sytox Green and Alexa Fluor 647 labeled Annexin V were used to indicate nucleus and membrane respectively. Scale bar, 10 μm. (H) Vectors expressing NINJ1 were transformed into E. coli cells, and the cells were grown on LB plates with or without IPTG overnight. The number of colony-forming units (CFU) was counted and statistically analyzed. Values are the means ± SD. n = 3. *p < 0.05. (I) Bacterial strains, including Gram-positive bacteria B. subtilis, M. luteus and S. iniae, and Gram-negative bacteria V. harveyi, P. fluorescens and E. tarda, were cultured in the presence or absence of NINJ1 amphipathic α-helix 1 (41–52 aa) and 2 (55–71 aa) polypeptides for 12 h. The OD600 of bacteria were measured, and the survival rate of each bacterium was statistically calculated. Values are the means ± SD. n = 3 Turbot NINJ1 induces rapid plasma membrane rupture. (A) C-terminally mCherry tagged turbot NINJ1 was expressed in HEK293T cells for 24 h. Cells expressing mCherry backbone vector were taken as the control. Images were captured with a confocal microscope. Scale bar, 10 μm. (B) The release of lactate dehydrogenase (LDH) was measured from cells expressing mCherry or mCherry tagged NINJ1, with non-transfected cells as the control. Values are the means ± SD. n = 3. ****p < 0.0001. (C) HEK293T cells were transfected with NINJ1 for 24 h, and then stained with nuclear dye Hoechst 33,342 (H33342) and membrane dye DiO. NINJ1-mediated plasma membrane blebbing was observed with a confocal microscope. Scale bar, 20 μm. (D) HEK293T cells expressing mCherry tagged NINJ1 was stained with nuclear dye Sytox Green and the phosphatidylserine binding Annexin V. Time-lapse live-cell images, including red fluorescence (mCherry), green fluorescence (Sytox green), magenta fluorescence (Alexa Fluor 647 labeled Annexin V) and bright field, were captured using a confocal microscope. The black, white and blue arrows indicate the swelling membrane, NINJ1 membrane localization and the nucleus respectively. Scale bar, 10 μm. (E and F) HEK293T cells were transfected with backbone vector or vector expressing NINJ1, and the relative fluorescence units of Sytox Green (E) and release of LDH (F) were determined at 12, 24, 36, and 48 h post-transfection. Values are the means ± SD. n = 3. ***p < 0.001, ****p < 0.0001. (G) Cells expressing mCherry tagged NINJ1 was observed with a confocal microscope at 48 h after transfection. Sytox Green and Alexa Fluor 647 labeled Annexin V were used to indicate nucleus and membrane respectively. Scale bar, 10 μm. (H) Vectors expressing NINJ1 were transformed into E. coli cells, and the cells were grown on LB plates with or without IPTG overnight. The number of colony-forming units (CFU) was counted and statistically analyzed. Values are the means ± SD. n = 3. *p < 0.05. (I) Bacterial strains, including Gram-positive bacteria B. subtilis, M. luteus and S. iniae, and Gram-negative bacteria V. harveyi, P. fluorescens and E. tarda, were cultured in the presence or absence of NINJ1 amphipathic α-helix 1 (41–52 aa) and 2 (55–71 aa) polypeptides for 12 h. The OD600 of bacteria were measured, and the survival rate of each bacterium was statistically calculated. Values are the means ± SD. n = 3 Turbot NINJ1 trafficking to membrane-bound organelles and accumulating at the plasma membrane Given that turbot NINJ1 is capable of localizing and disintegrating the plasma membrane, we explored the potential transport routes for NINJ1. During the initial stage of NINJ1 expression, we observed that NINJ1 mainly distributed within the cytoplasm rather than at the plasma membrane in HEK293T cells (Figs. 2A, B). Because NINJ1 possesses conserved transmembrane helices, this intracellular distribution implies its localization to the cytosolic membrane-bound organelles. Consistently, we observed strong colocalization of NINJ1 with the Golgi apparatus (Fig. 2C, D), endoplasmic reticulum (ER, Fig. 2E, F), and lysosomes (Fig. 2G, H). Notably, cells at this stage remained viable and exhibited normal cellular morphology (Figs. 2C, E, and G), suggesting that NINJ1, might be in transit, distributed to these membrane-bound organelles prior to cell lytic death. As PMR progressed, turbot NINJ1 almost entirely translocated to the cellular membrane, evidenced by its strong colocalization with the inner membrane leaflet probe Annexin V (Figs. 2I, J). This translocation pattern suggests an intracellular trafficking and ultimately accumulation of NINJ1 at the plasma membrane, likely mediated by the conventional ER-Golgi trafficking route within cellular membrane transport systems.Fig. 2Intracellular trafficking of NINJ1 to membrane-bound organelles and accumulation at the cytoplasm membrane. HEK293T cells were expressed with mCherry-tagged NINJ1 for 12 h, and stained with nuclear dyes Hoechst 33,342 (A, C, E, G) or Sytox Green (I). For the subcellular localization, the above cells were further stained with membrane dye DiO (A), Golgi apparatus Tracker Green (C), endoplasmic reticulum (ER) Tracker Green (E), lysosome Tracker Green (G), and Alexa Fluor 647 labeled Annexin V (I). All images were captured with a confocal microscope. Scale bar, 10 μm. (B, D, F, H, J) Normalized intensity profiles were generated from the white lines in A, C, E, G and I. These profiles display the relative pixel intensity along the arrowheads. Colocalization was assessed by analyzing the coincidence of relative fluorescence intensity between mCherry-tagged NINJ1 and organelle markers Intracellular trafficking of NINJ1 to membrane-bound organelles and accumulation at the cytoplasm membrane. HEK293T cells were expressed with mCherry-tagged NINJ1 for 12 h, and stained with nuclear dyes Hoechst 33,342 (A, C, E, G) or Sytox Green (I). For the subcellular localization, the above cells were further stained with membrane dye DiO (A), Golgi apparatus Tracker Green (C), endoplasmic reticulum (ER) Tracker Green (E), lysosome Tracker Green (G), and Alexa Fluor 647 labeled Annexin V (I). All images were captured with a confocal microscope. Scale bar, 10 μm. (B, D, F, H, J) Normalized intensity profiles were generated from the white lines in A, C, E, G and I. These profiles display the relative pixel intensity along the arrowheads. Colocalization was assessed by analyzing the coincidence of relative fluorescence intensity between mCherry-tagged NINJ1 and organelle markers NINJ1 promotes gasdermin-mediated cell Lysis Our previous research established that turbot gasdermin Ea (GSDMEa) was capable of perforating the plasma membrane via its N-terminal (NT) fragment, and consequently inducing pyroptosis [32]. Based on this finding, we examined the role of NINJ1 in GSDMEa-mediated cell death. As shown in Fig. 3A and B, single expression of either NINJ1 or GSDMEa NT in HEK293T cells could induce membrane damage and release of LDH in a dose-dependent manner. At low transfection levels (e.g., 10 ng plasmid/well), individual expression of GSDMEa-NT or NINJ1 failed to induce LDH release, indicating the absence of PMR (Fig. 3A and B). In contrast to the expression of single protein, co-expression of NINJ1 and GSDMEa NT at the same non-cytotoxic dose (10 ng each/well) led to PMR, featured by substantial releases of LDH at 36 h (~ 45%) and 48 h (~ 80%) (Fig. 3C). Scanning electron microscopy further validated that, unlike single protein expression, this co-expression of low-level GSDMEa NT plus NINJ1 induced severe membrane damage (Fig. 3D). The cytotoxicity became apparent when cells were transfected with a single recombinant vector at a higher dose (e.g., 20 ng plasmid/well). At this concentration, co-expression of both proteins triggered a much higher amount of LDH release than that of single protein expression, suggesting a synergistic effect of NINJ1 on GSDMEa-mediated cell death (Fig. 3E). Examination of membrane integrity further validated that co-expression of NINJ1 and GSDMEa incurred substantially more severe membrane injury than either protein alone (Fig. 3F). When cells were transfected with NINJ1 or GSDMEa at a high dose (e.g., 120 ng/well), approximately 70% LDH was released from NINJ1-expressing cells, whereas about 35% LDH was released from cells transfected with the same dose of GSDMEa-expressing vectors, suggesting that NINJ1 is more efficient than GSDMEa NT in disrupting the plasma membrane.Fig. 3NINJ1 promotes gasdermin-mediated cell death. (A, B) HEK293T cells were transfected with different doses of recombinant vectors expressing NINJ1 (A) or N-terminal (NT) domain of GSDMEa (B). After 36 h post-transfection, the amount of LDH released from the above cells was determined. Values are the means ± SD. n = 3. (C) HEK293T cells were transfected with vectors expressing NINJ1 or GSDMEa-NT at low dosage (10 ng/well), herein referred to NINJ1low or GSDMEa-NTlow, or co-transfected with both vectors expressing NINJ1low plus GSDMEa-NTlow (each vector 10 ng/well). The release of LDH from the above cells was measured at different time point. Values are the means ± SD. n = 3. ****p < 0.0001. (D) HEK293T cells were transfected with vectors expressing NINJ1 and/or GSDMEa-NT as described in (C). After 36 h post-transfection, the cell membrane integrity was determined by scanning electron microscopy. Scale bar, 10 μm. (E) HEK293T cells were transfected with vectors expressing NINJ1 or GSDMEa-NT at a higher dosage (20 ng/well), herein referred to NINJ1high or GSDMEa-NThigh, or co-transfected with both vectors expressing NINJ1high and GSDMEa-NThigh (each vector 20 ng/well). The release of LDH from the above cells was measured at different time point. Values are the means ± SD. n = 3. ****p < 0.0001. (F) HEK293T cells were transfected with vectors expressing NINJ1 and/or GSDMEa-NT as described in (E). After 36 h post-transfection, the cell membrane integrity was determined by scanning electron microscopy. Scale bar, 10 μm (left two images) and 5 μm (right two images). *p < 0.05, and ****p < 0.0001 NINJ1 promotes gasdermin-mediated cell death. (A, B) HEK293T cells were transfected with different doses of recombinant vectors expressing NINJ1 (A) or N-terminal (NT) domain of GSDMEa (B). After 36 h post-transfection, the amount of LDH released from the above cells was determined. Values are the means ± SD. n = 3. (C) HEK293T cells were transfected with vectors expressing NINJ1 or GSDMEa-NT at low dosage (10 ng/well), herein referred to NINJ1low or GSDMEa-NTlow, or co-transfected with both vectors expressing NINJ1low plus GSDMEa-NTlow (each vector 10 ng/well). The release of LDH from the above cells was measured at different time point. Values are the means ± SD. n = 3. ****p < 0.0001. (D) HEK293T cells were transfected with vectors expressing NINJ1 and/or GSDMEa-NT as described in (C). After 36 h post-transfection, the cell membrane integrity was determined by scanning electron microscopy. Scale bar, 10 μm. (E) HEK293T cells were transfected with vectors expressing NINJ1 or GSDMEa-NT at a higher dosage (20 ng/well), herein referred to NINJ1high or GSDMEa-NThigh, or co-transfected with both vectors expressing NINJ1high and GSDMEa-NThigh (each vector 20 ng/well). The release of LDH from the above cells was measured at different time point. Values are the means ± SD. n = 3. ****p < 0.0001. (F) HEK293T cells were transfected with vectors expressing NINJ1 and/or GSDMEa-NT as described in (E). After 36 h post-transfection, the cell membrane integrity was determined by scanning electron microscopy. Scale bar, 10 μm (left two images) and 5 μm (right two images). *p < 0.05, and ****p < 0.0001 Extracellular and transmembrane helices are essential for NINJ1 cytotoxicity Protein sequence alignment shows that turbot NINJ1 is similar to its counterparts in human and mouse (Fig. 4A). The N-terminal coiled region, intracellular connection loop, and extracellular C-terminal region exhibit low conservedness, while residues within the extracellular and transmembrane helices are highly conserved (Fig. 4A, B). In accordance with secondary structural analysis, three-dimensional structural modeling shows that turbot NINJ1 consists of three anti-parallel helices: an N-terminal amphipathic helix and two transmembrane helical regions (Fig. 4C, left panel). This structural configuration is similar to that observed in mouse NINJ1. Oligomerization into filaments is a prerequisite for NINJ1 membrane lysis function. Structural analysis further shows that, when activated, the N-terminal amphipathic helix accommodates a kinked conformation, resulting into two helices (α1 and α2). The α-helix 1 protruding forward to connect with the neighboring NINJ1 protomer, forming highly self-oligomerized filaments (Fig. 4C,right panel). To gain insight into the structural basis of membrane lysis functionality, we performed site-directed mutagenesis, substituting a series of residues with alanine, and examined PMR of HEK293T cells expressing these NINJ1 mutants. We initially introduced consecutive alanine substitutions into the conserved tetrapeptide motif 7EMNG10 in the N-terminal coiled region. The resulting mutant, 7AAAA10 (designated Mut7-10), exhibited minimal impact on cellular cytotoxicity, with cells undergoing PMR and massive LDH release similar to that of wild-type NINJ1 (Fig. 4D). In contrast, mutation of another conserved motif, 38PLNMN42, in the same region, completely abrogated NINJ1 cytotoxic activity (Fig. 4D). We next mutated residues within the N-terminal amphipathic helical region, a highly conserved segment between turbot and human NINJ1. The result demonstrated that mutations in this region, including every five consecutive alanine substitutions, were sufficient to abolish the membrane lysis activity of NINJ1 (Fig. 4E), underscoring the evolutionarily critical role of this extracellular helix for NINJ1 functionality. For the two transmembrane helical domains, consecutive mutations in either region ablated the PMR-inducing activity of turbot NINJ1, leading to negligible release of LDH (Fig. 4F, G). These findings highlight the crucial roles of the extracellular and transmembrane helices in NINJ1’s ability to induce PMR.Fig. 4The extracellular amphipathic helix and two transmembrane helices are essential for NINJ1-mediated plasma membrane rupture. (A) Multiple sequence alignment of NINJ1 from human, mouse and turbot. The N-terminal amphipathic helix and two transmembrane helical regions are highlighted with blue rectangle beneath the sequences. TM1 and TM2 represent transmembrane helical region 1 and 2 respectively. (B) The topological model of turbot NINJ1. NINJ1 possesses an extracellular N-terminal coiled region, an amphipathic α-helix, and two conserved hydrophobic transmembrane helical domains connected with an intracellular loop. C-terminus locates outside of the cell. (C) The three-dimensional structural model of turbot NINJ1 protomer (left, inactive state) and oligomer (right, active state). The N-terminal amphipathic helix is kinked and divided into two helices, named α-helix 1 and 2. (D-G) Wild-type NINJ1 or consecutive alanine substituted mutants were overexpressed in HEK293T cells. After 24 h post-expression, the cytotoxicity of mutated NINJ1 was normalized to the wild-type NINJ1 control, as measured by LDH release (right panels). Left panels: the selected regions for mutagenesis in N-terminus (D), extracellular amphipathic α-helix (E), transmembrane domain 1 (F) and transmembrane domain 2 (G). Values are the means ± SD. n = 3 The extracellular amphipathic helix and two transmembrane helices are essential for NINJ1-mediated plasma membrane rupture. (A) Multiple sequence alignment of NINJ1 from human, mouse and turbot. The N-terminal amphipathic helix and two transmembrane helical regions are highlighted with blue rectangle beneath the sequences. TM1 and TM2 represent transmembrane helical region 1 and 2 respectively. (B) The topological model of turbot NINJ1. NINJ1 possesses an extracellular N-terminal coiled region, an amphipathic α-helix, and two conserved hydrophobic transmembrane helical domains connected with an intracellular loop. C-terminus locates outside of the cell. (C) The three-dimensional structural model of turbot NINJ1 protomer (left, inactive state) and oligomer (right, active state). The N-terminal amphipathic helix is kinked and divided into two helices, named α-helix 1 and 2. (D-G) Wild-type NINJ1 or consecutive alanine substituted mutants were overexpressed in HEK293T cells. After 24 h post-expression, the cytotoxicity of mutated NINJ1 was normalized to the wild-type NINJ1 control, as measured by LDH release (right panels). Left panels: the selected regions for mutagenesis in N-terminus (D), extracellular amphipathic α-helix (E), transmembrane domain 1 (F) and transmembrane domain 2 (G). Values are the means ± SD. n = 3 Inter- and intra-molecular interactions are functionally essential to NINJ1-mediated PMR Since, as shown above, mutations within the extracellular amphipathic and transmembrane helices have significant effects on NINJ1-mediated PMR, we thus further examined the functional residues crucial for turbot NINJ1 functionality. Residues K41, K42, and D50 in the extracellular amphipathic helix, along with G92, T120, and V131 in the transmembrane helices, are predicted to participate in NINJ1 inter-molecular interactions, facilitating the formation of oligomeric filaments (Fig. 4A, C). Among these residues, K41/K42 are conserved residues located near the N-terminal end of the α-helix. When expressed in HEK293T cells, NINJ1 bearing the K41A/K42A double mutation had a much reduced capability to mediate PMR, and further evidenced by lower LDH release and Sytox Green staining (Fig. 5A-C). Besides K41/K42, D50 is another conserved residue in the extracellular amphipathic helix; mutation of the corresponding residue D53 in human NINJ1 was reported to impair the formation of an inter-molecular salt bridge [33]. Our findings showed that mutation of D50 in turbot NINJ1 completely abolished its cell lysis function, confirming its critical role in mediating cell lysis (Fig. 5A-C). In the transmembrane helical regions, G92 and T120 are predicted to be involved in the inter-molecular contact of turbot NINJ1. Mutations of these two residues (G92A and T120A) impaired the PMR-inducing activity of turbot NINJ1, leaving viable cells resistant to LDH release and Sytox Green staining (Fig. 5A-C). I134 is another residue reported to be involved in inter-molecular contact of human NINJ1, while this residue is not conserved and replaced with valine (V131) in turbot NINJ1 (Fig. 4A). Since both Val and Ile possess aromatic side chains and share similar hydrophobic nature, mutation of V131 in turbot NINJ1 reduced its ability to induce PMR (Fig. 5A-C), similar to its human counterpart mutation. Dynamic analysis of wild-type and mutated NINJ1-mediated cell death revealed that K41A/K42A, G92A, and T120A mutations compromised membrane lysis activity of turbot NINJ1, albeit to a lesser extent. In contrast, D50A and V131A mutants completely abrogated PMR-inducing capacity even after a prolonged overexpression period (Fig. 5D).Fig. 5Mutations disrupting inter- and intra-molecular interactions severely impair NINJ1-mediate cell lysis. (A) HEK293T cells were transfected with mCherry-tagged WT or mutated NINJ1 for 24 h. Cell morphology and Sytox Green staining were examined by fluorescence microscopy. Scale bar, 30 μm. The mutated residues, including K41A, K42A, D50A, G92A, T120A and V131A, are predicted to impair inter-molecular contact. (B, C) The cells transfected as described above were measured for LDH release (B) and relative fluorescence units of Sytox Green (C). Values are the means ± SD. n = 3. ****p < 0.0001. (D) Sytox Green fluorescence of cells expressing WT or mutated NINJ1 were determined at 12, 24, 36, 48 and 60 h. (E) HEK293T cells were transfected with mCherry-tagged WT or mutated NINJ1 for 24 h. Cell morphology and Sytox Green staining were examined by fluorescence microscopy. Scale bar, 30 μm. The mutated residues, including I81A and Q88A, are predicted to abolish intra-molecular interactions. (F, G) The cells transfected as described above were measured for LDH release (F) and relative fluorescence units of Sytox Green (G). Values are the means ± SD. n = 3. ****p < 0.0001. (H) Sytox Green fluorescence of cells expressing WT or mutated NINJ1 were determined at 12, 24, 36, 48 and 60 h Mutations disrupting inter- and intra-molecular interactions severely impair NINJ1-mediate cell lysis. (A) HEK293T cells were transfected with mCherry-tagged WT or mutated NINJ1 for 24 h. Cell morphology and Sytox Green staining were examined by fluorescence microscopy. Scale bar, 30 μm. The mutated residues, including K41A, K42A, D50A, G92A, T120A and V131A, are predicted to impair inter-molecular contact. (B, C) The cells transfected as described above were measured for LDH release (B) and relative fluorescence units of Sytox Green (C). Values are the means ± SD. n = 3. ****p < 0.0001. (D) Sytox Green fluorescence of cells expressing WT or mutated NINJ1 were determined at 12, 24, 36, 48 and 60 h. (E) HEK293T cells were transfected with mCherry-tagged WT or mutated NINJ1 for 24 h. Cell morphology and Sytox Green staining were examined by fluorescence microscopy. Scale bar, 30 μm. The mutated residues, including I81A and Q88A, are predicted to abolish intra-molecular interactions. (F, G) The cells transfected as described above were measured for LDH release (F) and relative fluorescence units of Sytox Green (G). Values are the means ± SD. n = 3. ****p < 0.0001. (H) Sytox Green fluorescence of cells expressing WT or mutated NINJ1 were determined at 12, 24, 36, 48 and 60 h In addition to exploring inter-molecular interactions, we also examined the functional significance of residues potentially involved in intra-molecular contact. In human NINJ1, I84 and Q91 locate at intra-molecular interfaces, and mutations of these residues disrupt NINJ1 filament formation [33]. In turbot NINJ1, we identified conserved corresponding residues, I81 and Q88, in the transmembrane helical region. When I81 and Q88 residues were substituted with alanine, the membrane lysis activity was significantly reduced at 24 h after transfection. HEK293T cells expressing I81A and Q88A remained viable and exhibited healthy morphology, concurrent with little LDH release and resistance to Sytox Green staining (Fig. 5E-G). However, after prolonged overexpression (e.g., 48 and 60 h), the I81A mutant eventually exhibited apparent cytotoxicity, leading to high rate of cell death akin to wild-type NINJ1 (Fig. 5H). Different from I81A, the Q88A mutant completely abolished cytotoxicity, failing to induce LDH release and Sytox Green infiltration even after overexpression for 60 h. Hydrophobicity and electrostatic potential are critical determinants of NINJ1 functionality The hydrophobic nature of transmembrane helices is crucial for NINJ1 membrane localization and lysis activity [34]. Human NINJ1 harbors V82 and L121 in its hydrophobic interface; however, the counterpart residues in turbot NINJ1 are replaced with residues T79 and V118 respectively (Fig. 4A). Despite the non-conservedness, mutation of these residues to alanine (T79A and V118A) decreased membrane lysis activity of turbot NINJ1, as HEK293T cells expressing these mutants largely retained normal morphology compared to those expressing wild-type NINJ1 (Fig. 6A). This observation was further evidenced by a reduction in membrane lysis-induced LDH release and Sytox Green infiltration in cells expressing T79A or V118A mutants (Fig. 6B, C). In spite of their initially limited cytotoxicity, these mutants were capable of inducing cell death after prolonged expression, reaching levels comparable to wild-type NINJ1 after 36 h expression (Fig. 6D). In addition to the hydrophobicity, we further explore the role of electrostatic potential for turbot NINJ1 functionality. Given the high abundance of negatively charged phosphatidic acid, phosphatidylserine, and phosphatidylinositol species in the cytoplasmic membrane, positively charged residues are thought to be critically important for protein-membrane interaction. In turbot NINJ1, we identified and mutated three positively charged residues: K62 (in the extracellular helical region), K100 and K111 (both in the transmembrane region), to assess their contribution to membrane lysis activity. As shown in Fig. 6E, expression of the K62A mutant resulted in a reduced capability to mediate PMR, evidenced by lower killing scores based on LDH release and decreased Sytox Green uptake (Fig. 6F, G), while expression of K100A and K111A mutants showed limited membrane disrupting activity. All these mutants ultimately achieved comparable PMR to wild-type NINJ1 after 48 h of expression (Fig. 6H).Fig. 6Hydrophobicity and positive electrostatic potential are critical determinants of NINJ1 functionality. (A) HEK293T cells were transfected with mCherry-tagged WT or mutated (T79A and V118A) NINJ1 for 36 h. Cell morphology and Sytox Green staining were examined by fluorescence microscopy. Scale bar, 30 μm. (B, C) The cells transfected as described above were measured for LDH release (B) and relative fluorescence units of Sytox Green (C). Values are the means ± SD. n = 3. **p < 0.01, and ***p < 0.001. (D) Sytox Green fluorescence of cells expressing mCherry-tagged WT or mutated NINJ1 were determined at 12, 24, 36, 48 and 60 h. (E) HEK293T cells were transfected with mCherry-tagged WT or mutated (K62A, K100A and K111A) NINJ1 for 36 h. Cell morphology and Sytox Green staining were examined by fluorescence microscopy. Scale bar, 30 μm. (F, G) The cells transfected as described above were measured for LDH release (F) and relative fluorescence units of Sytox Green (G). Values are the means ± SD. n = 3. *p < 0.05, **p < 0.01, and ****p < 0.0001. (H) Sytox Green fluorescence of cells expressing mCherry-tagged WT or mutated NINJ1 were determined at 12, 24, 36, 48 and 60 h. (I) HEK293T cells expressing NINJ1 were lysed, and treated with or without crosslinker disuccinimidyl suberate (DSS). Western blot analysis was performed to determine NINJ1 oligomerization. (J) WT or mutated NINJ1 were expressed in HEK293T cells, and crosslinked by DSS. Western blot analysis was performed to determine oligomerization of wild type and mutated NINJ1 Hydrophobicity and positive electrostatic potential are critical determinants of NINJ1 functionality. (A) HEK293T cells were transfected with mCherry-tagged WT or mutated (T79A and V118A) NINJ1 for 36 h. Cell morphology and Sytox Green staining were examined by fluorescence microscopy. Scale bar, 30 μm. (B, C) The cells transfected as described above were measured for LDH release (B) and relative fluorescence units of Sytox Green (C). Values are the means ± SD. n = 3. **p < 0.01, and ***p < 0.001. (D) Sytox Green fluorescence of cells expressing mCherry-tagged WT or mutated NINJ1 were determined at 12, 24, 36, 48 and 60 h. (E) HEK293T cells were transfected with mCherry-tagged WT or mutated (K62A, K100A and K111A) NINJ1 for 36 h. Cell morphology and Sytox Green staining were examined by fluorescence microscopy. Scale bar, 30 μm. (F, G) The cells transfected as described above were measured for LDH release (F) and relative fluorescence units of Sytox Green (G). Values are the means ± SD. n = 3. *p < 0.05, **p < 0.01, and ****p < 0.0001. (H) Sytox Green fluorescence of cells expressing mCherry-tagged WT or mutated NINJ1 were determined at 12, 24, 36, 48 and 60 h. (I) HEK293T cells expressing NINJ1 were lysed, and treated with or without crosslinker disuccinimidyl suberate (DSS). Western blot analysis was performed to determine NINJ1 oligomerization. (J) WT or mutated NINJ1 were expressed in HEK293T cells, and crosslinked by DSS. Western blot analysis was performed to determine oligomerization of wild type and mutated NINJ1 Oligomerization is a prerequisite for NINJ1-mediated PMR. We herein performed crosslinking experiments to detect the oligomeric state of wild type and mutated NINJ1. Without the crosslinker disuccinimidyl suberate, as shown in Fig. 6I, NINJ1 was found primarily in a monomeric state in both cytoplasm and cell membrane fractions. However, upon incubation with the crosslinker, a substantial proportion of NINJ1 was highly oligomerized in both membrane and cytoplasmic fractions. Compared with the wild-type NINJ1, mutants bearing K41/42A, D50A, T120A, and V131A substitutions exhibited much higher proportions of oligomerized NINJ1 in the cell membrane fractions (Fig. 6J). Since these mutations exhibited much lower cytotoxicity as described above, this might explain their accumulation at the membrane without causing widespread cell death. Consistent with this finding, K41/42A, D50A and T120A mutants also existed in a higher oligomeric state than wild-type NINJ1 in the cytoplasm. Conversely, other NINJ1 variants designed to alter intra-molecular contacts (I81A and Q88A), hydrophobicity (T79A and V118A), and positive electrostatic potential (K62A, K100A and K111A) displayed markedly reduced abundance in the cellular membrane fractions (Fig. 6J), implying a primary defect in membrane association. Inhibition of NINJ1-mediated PMR promotes antimicrobial immunity in turbot To identify potential factors that affect NINJ1-mediated PMR, we treated NINJ1-expressing HEK293T cells with different agents, including a cytoprotectant (glycine) [35], a membrane rupture inhibitor (muscimol) [36], a membrane stabilizer (punicalagin) [37], and an antioxidant (N-acetylcysteine) [38]. The result showed that muscimol and punicalagin significantly inhibited NINJ1-mediated cell lysis, evidenced by a substantial reduction in Sytox Green staining (Fig. 7A, B). Whereas glycine and N-acetylcysteine had little effect on NINJ1-mediated cell death. We next examined the role of NINJ1-mediated cell death to fish immunity. Vibrio anguillarum, a significant opportunistic bacterial pathogen causing lethal vibriosis, was used for infection studies [39]. Fish head kidney acts as a major lymphoid organ, primarily composing of hemopoietic immune cells that produce cytokines to activate the immune response [40, 41]. When turbot head kidney cells were infected at a multiplicity of infection (MOI) of 1, a remarkable cohort of cells were dead as indicated by Sytox Green staining (Fig. 7C). This cell death increased further at a higher MOI of 10. Compared to uninfected cells, the dying infected cells exhibited a ballooned membrane structure, indicative of a necrotic form of cell death (Fig. 7D). During infection, both GSDMEa and caspase-3 were upregulated and activated (Fig. 7E, Figure S1A, B), suggesting the occurrence of pyroptosis in response to V. anguillarum infection [32]. By contrast, the necroptosis executioner mixed-lineage kinase domain-like (MLKL) showed no apparent change during infection. Concomitant with infection-induced cell death, proinflammatory cytokines such as IL-1β, TNF-α, and INF-γ were significantly elevated after bacterial infection, whereas IL-18 was downregulated (Fig. 7F). In vivo infection further revealed that inhibition of pyroptosis by specific inhibitors of gasdermin (dimethyl fumarate) and caspase-3 (Z-VAD-FMK) significantly increased bacterial load in blood, spleen, and liver (Fig. 7G-I). This inhibition consequently decreased the survival of infected fish (Fig. 7J). Of note, inhibition of NINJ1 by muscimol had no obvious effect on bacterial load after V. anguillarum infection (Fig. 7G-I), but promoted the antimicrobial immunity of infected fish, contributing to a higher survival rate (Fig. 7J).Fig. 7Inhibition of NINJ1-mediated PMR promotes antimicrobial immunity in turbot. (A and B) HEK293T cells were overexpressed with turbot NINJ1 in the presence or absence of Glycine, muscimol, Punicalagin, and N-acetyl-Cysteine. The cell morphology was visualized by bright-field imaging, and cell death was monitored by Sytox Green staining. Scale bar, 200 μm. Values are the means ± SD. n = 3. *p < 0.05, and **p < 0.01. (C) Turbot head kidney cells were infected with Vibrio anguillarum at a multiplicity of infection (MOI) of 0.1, 1 and 10. At 24 h post-infection, cells were treated with Sytox Green and subjected to fluorescence imaging. Scale bar, 200 μm. (D) Turbot head kidney cells were infected with or without V. anguillarum as described above, and visualized using scanning electron microscopy. Scale bar, 10 μm. (E and F) Turbots were infected with V. anguillarum for different hours, and the expression of cell death-related genes, including NINJ1, GSDMEa, caspase-3 and MLKL (E), and pro-inflammatory cytokines, including IL-1β, TNF-α, IL-18 and INF-γ (F), in the spleen were determined by qRT-PCR. Values are the means ± SD. n = 3. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. (G, H and I) Turbots were infected with V. anguillarum in the presence or absence (control) of indicated inhibitors, and bacterial loads (shown as colony-forming units, CFU) in blood (G), spleen (H), and liver (I) were determined at different time points. Values are the means ± SD. n = 3. **p < 0.01, ***p < 0.001, and ****p < 0.0001. (J) Turbots were infected with V. anguillarum in the presence or absence (control) of the indicated inhibitors and monitored the daily mortality. The survival significance between the inhibitor-treated and control groups was determined by log-rank test. n = 20. *p < 0.05, and **p < 0.01 Inhibition of NINJ1-mediated PMR promotes antimicrobial immunity in turbot. (A and B) HEK293T cells were overexpressed with turbot NINJ1 in the presence or absence of Glycine, muscimol, Punicalagin, and N-acetyl-Cysteine. The cell morphology was visualized by bright-field imaging, and cell death was monitored by Sytox Green staining. Scale bar, 200 μm. Values are the means ± SD. n = 3. *p < 0.05, and **p < 0.01. (C) Turbot head kidney cells were infected with Vibrio anguillarum at a multiplicity of infection (MOI) of 0.1, 1 and 10. At 24 h post-infection, cells were treated with Sytox Green and subjected to fluorescence imaging. Scale bar, 200 μm. (D) Turbot head kidney cells were infected with or without V. anguillarum as described above, and visualized using scanning electron microscopy. Scale bar, 10 μm. (E and F) Turbots were infected with V. anguillarum for different hours, and the expression of cell death-related genes, including NINJ1, GSDMEa, caspase-3 and MLKL (E), and pro-inflammatory cytokines, including IL-1β, TNF-α, IL-18 and INF-γ (F), in the spleen were determined by qRT-PCR. Values are the means ± SD. n = 3. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. (G, H and I) Turbots were infected with V. anguillarum in the presence or absence (control) of indicated inhibitors, and bacterial loads (shown as colony-forming units, CFU) in blood (G), spleen (H), and liver (I) were determined at different time points. Values are the means ± SD. n = 3. **p < 0.01, ***p < 0.001, and ****p < 0.0001. (J) Turbots were infected with V. anguillarum in the presence or absence (control) of the indicated inhibitors and monitored the daily mortality. The survival significance between the inhibitor-treated and control groups was determined by log-rank test. n = 20. *p < 0.05, and **p < 0.01 Discussion Recent studies provide strong evidence that NINJ1 serves as a bona fide “membrane cutter” [42]. Upon NINJ1-mediated PMR, NINJ1 oligomerizes to form irregular rings and curved filaments, which subsequently cut and solubilize membrane fragments [19, 20]. NINJ1 can be activated downstream of various cell death stimuli, including pyroptosis, apoptosis, necroptosis, and even bacterial toxin-induced cell lysis [16, 43, 44]. In teleost, pyroptosis-induced PMR has been observed in response to immune stimulation, bacterial infection and environmental stresses [45–48], suggesting the involvement of NINJ1 in teleost lytic cell death. In the present study, we identified a functional NINJ1 in turbot, which adopted conserved protein architecture, mainly consisting of an amphipathic helix at the N-terminal region and two conserved hydrophobic transmembrane helices at the C-terminal region. Consistently, turbot NINJ1 exhibited rapid PMR-inducing activity, similar to its counterpart in human and mouse. When overexpressed in cellular systems, turbot NINJ1 distributed mainly in the cellular compartments, including cytoplasmic membrane as well as intracellular membrane-bound organelles (e.g., Golgi apparatus, endoplasmic reticulum and lysosome). It should be noted that NINJ1 was highly enriched at the cellular membrane prior to cell swelling and lysis, strongly supported the intracellular transition from membrane-localized inactive protomer to activated, destructive chain-like filaments within the plasma membrane. Furthermore, the sequential nature of membrane damage, where nuclear dye staining by Sytox Green (indicating membrane permeabilization) precedes inner leaflet staining by Annexin V, illustrated that NINJ1-mediated membrane injury is an early event, initiating lesions before proceeding to full membrane rupture. In addition to cellular cytotoxicity, turbot NINJ1 exerted direct antimicrobial activity against E. coli. Given that amphipathic helical structures are characteristic features of antimicrobial peptides and are essential for their membrane-lytic function [49, 50], we hypothesized that the N-terminal amphipathic helix of NINJ1 might possess bactericidal properties independently. However, the synthetic amphipathic peptides had little effect on bacterial growth, indicating that while the N-terminal amphipathic helical region is required, it is not solely sufficient for turbot NINJ1-mediated cell killing. NINJ1-mediated PMR is an actively regulated process driven by its ability to transition from an inactive state to highly oligomerized filament [16, 18]. The activation is fundamentally dependent on the interplay of both intramolecular and intermolecular contacts. K41/K42 are two positively charged residues located in the N-terminal amphipathic helix of turbot NINJ1. Mutations of K41/K42 dysregulated the amphipathy of NINJ1, leading to a marked reduction in membrane lysis capacity. The functional significance of these residues is underscored by their high conservation in human and mouse NINJ1 (K44/K45), mutation of these residues yielded similar reductions in PMR [19]. This emphasizes the crucial role of positively charged residues in the N-terminal helical region for overall NINJ1 functionality, likely facilitating initial electrostatic interactions with the negatively charged headgroups of membrane phospholipids. Apart from the amphipathic effect of the N-terminal helix, K45 residue of human NINJ1 was also reported to form salt bridge with negatively charged residue D53 of a neighboring NINJ1 protomer [33]. Disrupting either residue abrogates cell lysis, indicating its essential role in stabilizing the intermolecular interfaces necessary for functional oligomerization. Our study reveals that this salt-bridge forming residue (K45 and D53 in human) is also conserved in the amphipathic helical region of turbot NINJ1 (K42 and D50). Mutation of D50 almost completely abolished cytotoxic capacity, emphasizing that this specific salt-bridge contact is an evolutionary conserved inter-protomer interaction indispensable for NINJ1 lytic function across species. Apart from K41, K42 and D50, our findings demonstrate that G92 and T120 are also important for PMR, mutation of these two residues abrogated cell lysis and reduced both membrane translocation and oligomerization formation. In human NINJ1, there exist conserved corresponding residues G95 and T123, they are involved in intermolecular interactions between neighboring protomers [33]. Mutations such as G95L and T123L result in the failure of forming NINJ1 filament structures, indicating their importance in the oligomerization process. In addition to residues forming inter-molecular contacts, I81 and Q88 are another two essential residues presented at the intramolecular interfaces of turbot NINJ1. When mutated, I81A and Q88A completely abolish the ability to induce cell lysis. Sequence alignment confirms that these residues are conserved at corresponding positions (I84 and Q91) in human and mouse NINJ1. The cryo-electron microscopy analysis further reveals that I84 and Q91 are vital for intra-molecular interaction [33], confirming by the fact that mutation of each residue disrupts the oligomerization of NINJ1 and inhibit the formation of filament structure. Different from the highly conserved residues involved in intra- or inter-molecular contacts, we found that residues in the hydrophobic interface of turbot NINJ1 (T79 and V118) are not well conserved in its human homolog (V82 and L121) [19]. However, the functional consequences of mutating these residues in turbot NINJ1 (T79A and V118A) were strikingly similar to those observed for their human counterparts: reduced oligomerization, impaired membrane translocation and reduced cell lysis. The hydrophobic nature facilitates the partitioning of NINJ1 into the membrane core, promoting the protein to transition from loosely membrane-associated state to efficiently oligomerized filament, which enabled the membrane-disrupting activity. Beyond the critical contribution of hydrophobicity, positively charged residues, specifically K62, K100, and K111 in turbot NINJ1, play indispensable roles in PMR. Mutations at these positions (K62A, K100A, and K111A) resulted in decreased membrane translocation and attenuated oligomerization compared to wild-type NINJ1. This impairment is directly attributable to the reduced net positive charge, which significantly compromises the anionic phospholipids binding affinity of NINJ1. The corresponding positively charged residues, especially K103 and K114 in human NINJ1, are crucial for mediating electrostatic interactions with the negatively charged headgroups of phospholipids, particularly anionic species such as phosphatidic acid and phosphatidylinositol [18, 51]. This finding supports the vital role of these positively charged residues in mediating the initial membrane binding and ensuring stable localization of NINJ1, which are prerequisites for subsequent functional oligomerization. Given its critical role in PMR, inhibition of NINJ1 activation by muscimol significantly inhibited cell death, and concurrently increased the survival rate of V. anguillarum infected turbot. This observation suggested that excessive or dysregulated NINJ1-mediated membrane lysis might contribute to aberrant cell death with catastrophic consequences for host innate immunity. Similar results have been reported in human cells, where inhibiting NINJ1-mediated PMR ameliorated hepatocellular necrosis, which in turn reduced the release of intracellular damage-associated molecular patterns and proinflammatory cytokines [52]. In turbot, our previous study revealed that GSDMEa could be activated by caspase-3 to induce pyroptosis [32]. Herein, we observed the upregulation of GSDMEa and caspase-3 following V. anguillarum infection, suggesting the involvement of pyroptosis in the antimicrobial immunity. However, in contrast to NINJ1-mediated cell death, pyroptosis exhibited a protective role against bacterial infection. Inhibition of pyroptosis led to a significant increase in bacterial load within immune organs (spleen, liver, and blood), resulting in substantially higher mortality rates compared to control fishes. This finding strongly suggests a crucial protective role of pyroptosis in fish antimicrobial immunity. Indeed, pyroptosis has been established as an essential component of immune defense against pathogenic infections, particularly for viruses and cytosolic bacteria [53–56]. For instance, the ubiquitin ligase effector of Shigella flexneri could block pyroptosis and thus facilitated bacterial evasion from immune surveillance in human cells, and eliminating NLRC4 and GSDMD in mice further enhanced bacterial infection and exacerbated infection induced death [57]. Similarly, for another cytosolic bacterium Burkholderia thailandensis, GSDMD-mediated pyroptosis protected mice from infection-induced disease severity, while inhibition of caspase-11-GSDMD axis led to severe mortality of infected mice [58]. In these cases, pyroptosis contributes to host defense by inducing lysis of infected cells, thereby eliminating the replicative cellular niche and facilitating bacterial clearance [13]. Although pyroptosis exerted an antimicrobial role in turbot against infection, the protective effect can be largely compromised by NINJ1-mediated cell lysis. This contrasting outcome suggests a dual-edged sword effect of NINJ1 and GSDMEa in regulating turbot cell death and immune responses. Collectively, our study identifies a functional NINJ1 in turbot fish, demonstrating its conserved role in PMR during lytic cell death. Turbot NINJ1 could synergistically cooperate with GSDMEa and amplifies pyroptosis-induced PMR. Structural analysis revealed that both conserved and non-conserved functional residues, primarily distributed in three helical regions, are essential for its molecular interactions, hydrophobicity, and electrostatic potential. V. anguillarum infection increased NINJ1 expression and triggered pyroptosis. While NINJ1 can contribute to detrimental, potentially overwhelming, cell lysis during infection, pyroptosis represents a distinct, highly regulated lytic cell death that is critical for effective pathogen clearance and host survival. This contrasting outcome suggests distinct and complex roles for NINJ1 and GSDMEa in mediating programmed cell death and antimicrobial immunity in turbot. Materials and methods Animals, cell lines and bacteria Healthy turbots (Scophthalmus maximus) were obtained from a local farm in Qingdao and maintained at 24 ℃ in aerated seawater under laboratory conditions. HEK293T cells (obtained from the American Type Culture Collection) were cultured in Dulbecco’s Modified Eagle’s Medium (Thermo Fisher Scientific, Waltham, MA) supplemented with 10% fetal bovine serum (Sigma-Aldrich, Madrid, Spain) at 37 ℃ and 5% CO2. Vibrio anguillarum was grown in Luria broth (LB) at 28 ℃. Confocal microscopy and live cell time-lapse imaging HEK293T cells were plated on 35-mm glass-bottomed culture dishes (Nest Scientific Rahway, USA) overnight, and transfected with pmCherry-N1 vector expressing the indicated plasmids (1 µg/well) using Polyjet transfection reagent (SignaGen Laboratories, Ijamsville, USA) according to the manufacturer’s instruction. After transfection, cells were stained with Sytox Green, Hoechst 33,342, DiO or Annexin V-Alexa Fluor 647. For colocalization analysis, cells expressing NINJ1 were stained with lysosome Tracker Green, Golgi apparatus-Tracker Green, and endoplasmic reticulum-Tracker Green. All these fluorescent probes were purchased from Thermo Fisher Scientific (Eugene, USA). Cells were then subjected to the aforementioned treatment and imaged using a Carl Zeiss LSM 900 confocal microscope (Carl Zeiss, Jena, Germany). For live cell time-lapse imaging, cells expressing mCherry-tagged NINJ1 were incubated with Sytox Green and Annexin V-Alexa Fluor 647 at 37 ℃ for 15 min. Following pre-warming of the laser source, the cells were placed into the environmental control chamber maintained at 37 ℃ and 5% CO2. Time-lapse images acquired every 2 min over a total duration of 6 h [59]. Measurement of releasing lactate dehydrogenase (LDH) HEK293T cells were cultured in a 96-well plate (Costar, Corning, NY, USA) overnight and transfected with mCherry-tagged NINJ1 wild-type and mutants (0.1 µg per well) using Polyjet according to the manufacturer’s instruction. After 6 h, the culture medium was replaced with fresh Opti-MEM (Thermo Fisher Scientific, Waltham, MA). At different time points, the cell supernatant was collected and measured for released LDH using the CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega, Leiden, Netherlands) according to the manufacturer’s instruction. The percentage LDH release was calculated as (LDHsample-medium background)/(LDH100% lysis-medium background) × 100%. Cell death was expressed as a percentage cytotoxicity, termed the “killing score”. The killing score was calculated to normalize the LDH release from experimental samples against the spontaneous and maximum LDH release controls. The formula was: Killing score (%) = (LDHsample - LDHnegative control)/(LDHWT- LDHnegative control) × 100%. Bactericidal assay Escherichia coli Transetta (DE3) was transformed with pET-30a expressing NINJ1 and cultured in LB medium untill exponential phase (OD600 0.4–0.6). Then the bacteria were plated on LB agar plates with or without 0.5 mM IPTG and incubated overnight at 37 ℃. The number of bacterial colonies was then counted and statistically analyzed. Antimicrobial activity The six selected bacteria, including Bacillus subtilis, Micrococcus luteus, Streptococcus iniae, Vibrio harveyi, Pseudomonas fluorescens, and Edwardsiella tarda, were cultured as described before until the exponential phase (OD600 0.4–0.6). Bacteria were added to 96-well plates (100 µL per well) and incubated with peptides under continuously shaking. The OD600 value was measured every 2 h to assess bacterial growth. Sequence alignment and structural modelling The NINJ1 sequence was obtained from the National Center for Biotechnology Information Reference Sequence database (www.ncbi.nlm.nih.gov/RefSeq). The MEGA11 (http://megasoftware.net/) software was used to align the sequences, and the multiple sequence alignment was then created using ESPript 3.0 (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi). The accession numbers of NINJ1 from the National Center for Biotechnology Information Reference Sequence are as follows: human NINJ1, NP_004139.2; mouse NINJ1, NP_038638.1; turbot NINJ1, XP_035488868.1. The three-dimensional models were generated using the SWISS-MODEL software (https://swissmodel.expasy.org/). Crosslinking assay HEK293T cells were transfected with pCS2-N-Myc vector expressing NINJ1 wild-type and mutants using Polyjet transfection reagent (SignaGen Laboratories, Ijamsville, USA) according to the manufacturer’s instruction. After 6 h, the medium was replaced with fresh Opti-MEM (Thermo Fisher Scientific, Waltham, MA), and the cells were grown for an additional 42 h. The cells were then pelleted by centrifugation at 600 g for 5 min and subjected to three freeze-thaw cycles by alternating between room temperature and liquid nitrogen. The supernatants were collected after centrifugation at 20,000 g at 4 ℃ for 30 min. The pellets were resuspended in ice-cold PBS (pH 7.4), centrifuged under the same conditions to discard the supernatant, and the resulting membrane pellets were dissolved in ice-cold PBS. Next, Disuccinimidyl suberate (DSS) crosslinker (MedChem Express, NJ, USA) was added to the cytoplasmic and membrane samples to a final concentration of 300 µM and incubated for 30 min at room temperature. The reaction was stopped by adding Tris-HCl to a final concentration of 5 mM and incubating for 15 min at room temperature. The cross-linked proteins were analyzed by immunoblotting. Immunoblotting The cytoplasmic and membrane samples were separated by SDS-PAGE and transferred onto nitrocellulose membranes by using Trans-Blot Turbo (Bio-Rad, California, USA). The membranes were then blocked with 5% skim milk for 1 h at room temperature and blotted with mouse anti-Myc tag IgG (Abcam, Cambridge, USA, 1:5000 dilution). After three washes, the membrane was blotted with horseradish peroxidase (HRP) labeled goat anti-mouse IgG (Abcam, Cambridge, USA, 1:5000 dilution). The protein bands were visualized using an ECL kit (Thermo Fisher, Waltham, USA). Quantitative reverse transcription PCR (qRT-PCR) To examine gene expression levels after bacterial infection, turbots were injected intramuscularly with V. anguillarum (1.0 × 106 CFU/fish). At 0, 6, 12, and 24 h after injection, spleens were taken aseptically from turbots and used for total RNA extraction using Trizol reagent (Invitrogen, Carlsbad, CA, USA). cDNA synthesis was performed with the First Strand cDNA Synthesis Kit (ToYoBo, Osaka, Japan) according to the manufacturer’s instruction. qRT-PCR was carried out in a QuantStudio 3 (Thermo Fisher Scientific, Waltham, MA) using the SYBR ExScript qRT-PCR Kit (Vazyme Biotechnology, Nanjing, China) according to the manufacturers’ instruction. The mRNA levels of the target genes in each sample were normalized to that of β-actin. Bacterial dissemination in fish tissues Turbots were randomly divided into three groups (15 fish/group). Group 1 was intraperitoneally injected with muscimol (3 mg/kg). Group 2 was intraperitoneally injected with a combination of Z-VAD-FMK (5 mg/kg), GW806742 × (0.5 mg/kg), and Dimethyl fumarate (20 mg/kg). Group 3, serving as the control, was injected with PBS. At 8 h post-injection, all groups were intramuscularly infected with V. anguillarum as described above. At 6 and 12 h after infection, blood, spleen, and liver were collected and homogenized in sterile PBS. The homogenized tissues were diluted and then plated on LB plates. After incubation at 28 ℃, the colony-forming units (CFU) on the plates were statistically calculated. Fish survival after bacterial infection Turbots were randomly divided into five groups (20 fish/group), and treated with inhibitors against apoptosis, pyroptosis, ncroptosis and NINJ1 [29, 36, 60, 61]. Group 1 was intraperitoneally injected with PBS; group 2 was intraperitoneally injected with muscimol (3 mg/kg); group 3 was intraperitoneally injected with Z-VAD-FMK (5 mg/kg), GW806742 × (0.5 mg/kg), and Dimethyl fumarate (20 mg/kg); and group 4 was intraperitoneally injected with a combination of Z-VAD-FMK (5 mg/kg), GW806742 × (0.5 mg/kg), Dimethyl fumarate (20 mg/kg), and muscimol (3 mg/kg). Eight hours after injection, groups 1–4 were intramuscularly injected with V. anguillarum (1.0 × 106 CFU/fish). As the control, group 5 received intraperitoneal and intramuscular injections of PBS. The turbots were monitored every day for mortality. Statistical analysis Data analysis was performed with GraphPad Prism 9 software (https://www.graphpad.com/). Data were analyzed with the Student’s t-test, one-way analysis of variance (ANOVA), and Log rank test (for the analysis of turbot survival). Data are shown as the means ± SD. Statistical significance was defined as p < 0.05. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Table S1. Primers used in the current study Supplementary Material 1. Table S1. Primers used in the current study Supplementary Material 2. Supplementary Movie 1. Time-lapse live-cell imaging of plasma membrane rupture. HEK293T cells were ectopically expressed mCherry-tagged turbot NINJ1 for 24 h, and then stained with AF647-labeled Annexin V and Sytox Green before imaging. Supplementary Movie 2. Time-lapse live-cell imaging of cells expressing mCherry tag. HEK293T cells were ectopically expressed mCherry tag for 24 h, and then stained with AF647-labeled Annexin V and Sytox Green before imaging Supplementary Material 2. Supplementary Movie 1. Time-lapse live-cell imaging of plasma membrane rupture. HEK293T cells were ectopically expressed mCherry-tagged turbot NINJ1 for 24 h, and then stained with AF647-labeled Annexin V and Sytox Green before imaging. Supplementary Movie 2. Time-lapse live-cell imaging of cells expressing mCherry tag. HEK293T cells were ectopically expressed mCherry tag for 24 h, and then stained with AF647-labeled Annexin V and Sytox Green before imaging