Ninjurin-1 mediates cell lysis and detrimental inflammation of PANoptosis during influenza A virus infection

Categories: Disease & Virus

September 23, 2025

NINJ1 is upregulated upon IAV infection

Since little is known about the expression pattern of NINJ1 during viral infection, we first determined the transcript level in IAV-infected murine lungs. The expression of Ninj1 tended to increase after infection (Fig. 1a), and the increase in expression was more pronounced when infected with a lethal dose (Fig. 1b), which indicated its possible association with disease severity. To further investigate its temporal expression pattern in different cell types, we analyzed the single-cell RNA sequence (scRNA-seq) data from murine lungs infected with a lethal dose or a non-lethal dose of IAV at different dpi produced by our group recently²⁷ (CNCB, PRJCA034049; Supplementary Table 1). The cells were clustered into 6 major types and 15 subtypes (Fig. 1c–g and Supplementary Fig. 1a–d). In line with the results shown in Fig. 1b, the Ninj1 level was elevated more after a lethal dose challenge (Fig. 1e). Ninj1 was expressed mainly in myeloid cells (MYEs) rather than alveolar epithelial cells (AECs) (Fig. 1f), and the proportion of MYEs expanded upon IAV infection, especially in macrophages (MAs) (Fig. 1d and Supplementary Fig. 1c). Specifically, among MYEs, Ninj1 was most highly expressed in neutrophils (NEUs), followed by MAs (Fig. 1g). Although the expression of Ninj1 increased in most cell types, the expression peak did not synchronize. The Ninj1 expression peaked at 1 to 2 dpi in MAs and at ~5 dpi in NEUs (Fig. 1g). Generally, a high dose of IAV resulted in higher expression of NINJ1 in most cell types (Fig. 1g). Notably, the expression patterns of numerous inflammation- and PANoptosis-associated genes (e.g., Il1b, Il18, Zbp1, Gsdmd, and Casp8) were similar to those of Ninj1, indicating Ninj1’s potential role in IAV-induced, ZBP1-mediated PANoptosis (Fig. 1e, f). Considering that MAs presented the second highest expression of Ninj1 and that ZBP1-mediated PANoptosis has been well depicted in MAs, we aimed to confirm this phenomenon in BMDMs in vitro. In line with our scRNA-seq results, RNA-seq analysis of IAV-infected BMDMs (CNCB, PRJCA034048) revealed that Ninj1 and Zbp1 were upregulated (Fig. 1h). Additionally, this upregulation was confirmed at the protein level (Fig. 2b). Overall, our in vivo and in vitro transcriptional analyses preliminarily revealed a potential association between Ninj1 and disease progression as well as macrophage PANoptosis during IAV infection.

Oligomerization of NINJ1 synchronizes with PANoptosis during IAV infection

Upregulation of NINJ1 does not necessarily indicate PMR, whereas oligomerization does. NINJ1 undergoes oligomerization on the plasma membrane once activated during lytic cell death, though the exact trigger is unclear.^17,19 Apparent NINJ1 oligomerization was detected at 12 hours post infection (hpi) (Fig. 2a). We also measured classical indicators of PANoptosis, including the activation of PANoptosis initiators and executioners, the release of LDH and inflammatory cytokines, and the percentage of PI⁺ cells. Coincidentally, the pyroptosis (cleavage of GSDMD and GSDME), apoptosis (cleavage of Caspase-3, -7, and -8), and necroptosis (phosphorylation of MLKL) pathways were also activated at 12 hpi, except for the upregulation of ZBP1 (Fig. 2b), which occurred as early as 6 hpi. At the same time, the percentage of PI⁺ cells and LDH release increased significantly as cell viability decreased (Fig. 2c–f), which are indicators of PMR and lytic cell death. Silver staining of the released proteins in the supernatants revealed no obvious selectivity for molecular weight (Fig. 2g). We noted that obvious elevation of TNF-α preceded elevation of IL-1β, which occurred at 9 hpi and 12 hpi, respectively (Fig. 2h, i), in line with the dependency of IL-1β release on cell lysis. In addition to LDH, another well-known DAMP, HMGB1, was also released in abundance (Fig. 2b). Overall, oligomerization of NINJ1 and occurrence of PANoptosis are highly synchronous. This finding prompted us to hypothesize that NINJ1 might play an important role in the course of IAV-induced PANoptosis.

NINJ1 mediates IAV-induced cell lysis along with the release of DAMPs and IL-1β without compromising cell death

To elucidate how NINJ1 affects the cell death modality and inflammatory responses during IAV infection, we compared WT and Ninj1^-/- BMDMs in several respects. Morphologically, Ninj1^-/- BMDMs exhibited a persistent balloon-like morphology, whereas WT BMDMs eventually dissociated only with their debris left (Fig. 3a). Plasma membrane leakage occurred in both (Fig. 3a, b), but PMR and the release of cellular contents, including LDH and HMGB1, were profoundly reduced by Ninj1 knockout or mutation of one of the residues (K45Q) critical for NINJ1 oligomerization^17,18,19 (Fig. 3c–e and Supplementary Fig. 2a–c). Unexpectedly, NINJ1 deficiency did not rescue BMDMs from death (Fig. 3f), which contrasts with previous observations of comorbid heat stress and infection, another PANoptosis model.²⁴ Similar results were observed in THP-1 cells, although the percentage of PI⁺ cells decreased somewhat after NINJ1 ablation (Supplementary Fig. 3j–m). In pyroptosis, NINJ1 reportedly has no effect on TNF-α or IL-1β production or release,¹⁷ which also applies to comorbid heat stress and infection-induced PANoptosis.²⁴ As expected, equal amounts of TNF-α were released (Fig. 3g). However, IL-1β release was reduced after NINJ1 ablation, and this decrease was not due to transcriptional regulation (Fig. 3h and Supplementary Fig. 2d). In comparison, such a reduction was not observed in Gsdmd^-/- BMDMs (Fig. 3h). This unexpected contrast reveals the heterogeneity of NINJ1 and GSDMD in regulating IL-1β release among different inflammatory cell death models, even if similar PANoptosis occurs. Infection with two other IAV strains yielded similar results, indicating that the impact of NINJ1 on IL-1β release is not strain specific but rather a universal phenomenon (Supplementary Fig. 2e, f). These phenotypic changes were not the result of changes in the viral load (Fig. 3d, i).

Glycine can prevent PMR by inhibiting oligomerization of NINJ1.²³ We aimed to verify whether it also applied in IAV-induced PANoptosis. Indeed, glycine treatment prominently restrained NINJ1 oligomerization without affecting its expression (Fig. 3j, k). Like Ninj1 knockout, glycine treatment strongly inhibited the release of LDH, HMGB1, and IL-1β, whereas cell viability was merely affected (Fig. 3l, m and Supplementary Fig. 3a–d). Glycine inhibited NINJ1 oligomerization and LDH release in a dose-dependent manner (Supplementary Fig. 3e–i) but did not have a superimposed effect on LDH release in Ninj1^-/-BMDMs (Fig. 3l), which confirmed that NINJ1 is the exact target of glycine. Additionally, all of these results were similarly observed in THP-1 cells (Supplementary Fig. 3j–m) and infection with other strains of influenza (Supplementary Fig. 3e–i).

NINJ1-mediated cell lysis is completely dependent on ZBP1 during IAV infection

IAV-induced PANoptosis is type I interferon and ZBP1 dependent.^14,28 Indeed, both PMR and cell death were reversed and all programmed cell death pathways were silenced by Zbp1 or Ifnar1 deficiency (Fig. 4a–f). As expected, NINJ1 oligomerization diminished simultaneously (Fig. 4g). Zbp1^-/- and Ninj1^-/- BMDMs exhibited comparable LDH release (Fig. 4h), and glycine treatment did not further reduce LDH release in Zbp1^-/- or Ifnar1^-/- BMDMs (Fig. 4e). These findings indicate that NINJ1 oligomerization and NINJ1-mediated LDH release in IAV-infected macrophages are completely dependent on ZBP1 and type I IFN responses. Although NINJ1 deficiency reduced LDH release to the same extent as ZBP1 deficiency did (Fig. 4h), Ninj1^-/- BMDMs died normally, whereas Zbp1^-/- BMDMs did not (Fig. 4d). The inhibition of IL-1β release was more thorough in Zbp1^-/- BMDMs than in Ninj1^-/- BMDMs (Fig. 4i). Together, our results indicate that NINJ1 oligomerization and cell lysis mediated by it are dependent on ZBP1 in IAV-induced PANoptosis and that NINJ1 differs from ZBP1 in its role in cell viability and inflammatory responses.

NINJ1 oligomerization is not regulated by the NLRP3 inflammasome

Numerous studies have explored the essential role of the NLRP3 inflammasome in both innate and adaptive immunity during IAV infection.^11,12,13 We found that knocking out NLRP3 inflammasome components did not abrogate IAV-induced lytic cell death (Supplementary Fig. 4a–d), similar to previous studies.²⁸ Accordingly, NINJ1 oligomerization was still retained (Supplementary Fig. 4e), and PANoptosis pathways remained intact, including GSDMD activation (Supplementary Fig. 4f). Additionally, IL-1β release was only partially reduced by NLRP3 inflammasome deficiency (Supplementary Fig. 4f). This is not contradictory since GSDMD and pro-IL-1β can be cleaved by Caspase-8.^29,30,31 Overall, the NLRP3 inflammasome does not participate much in IAV-induced macrophage PANoptosis or NINJ1 oligomerization except for a minor role in IL-1β maturation.

Activation of any cell death pathway in IAV-induced PANoptosis is sufficient to induce NINJ1 oligomerization

Considering that GSDMD cleavage, GSDME cleavage, and MLKL phosphorylation, together with NINJ1 oligomerization, were all abrogated in Zbp1^-/- BMDMs (Fig. 4c, g), we wondered whether there is a certain association between these pore-forming executioners and NINJ1. We first knocked out each of these pore-forming proteins. NINJ1 oligomerization and lytic cell death were not affected at all (Fig. 5a–c and Supplementary Fig. 5a–c). This isn’t beyond expectation since functional redundancy is quite common among cell death pathways. We then proposed that deficiency of all lytic cell death pathways, namely, pyroptosis and necroptosis, would lead to failure of NINJ1 oligomerization. Surprisingly, not only Gsdmd^-/-Gsdme^-/-, Gsdmd^-/-Mlkl^-/-, and Gsdme^-/-Mlkl^-/- BMDMs but also Gsdmd^-/-Gsdme^-/-Mlkl^-/- BMDMs underwent robust cell lysis, reaching a level similar to that of their WT counterparts (Fig. 5b, c and Supplementary Fig. 5a, b, d, e). Comparable NINJ1 oligomerization was observed in the macrophages of all these genotypes (Fig. 5d). The inhibition of LDH release by glycine treatment (Fig. 5c) further confirmed that NINJ1 oligomerized normally. Only the apoptosis pathway was intact in Gsdmd^-/-Gsdme^-/-Mlkl^-/- BMDMs (Fig. 5e), which is conventionally thought to be immunologically silent. However, we failed to convert the inflammatory cell death to a relatively immunologically silent one, namely, apoptosis in a traditional sense.

Then what is the trigger of NINJ1 oligomerization in the absence of the pyroptosis and necroptosis pathway? Considering that the apoptosis pathway is the only activated PANoptosis pathway underlying the difference between Zbp1^-/- and Gsdmd^-/-Gsdme^-/-Mlkl^-/- BMDMs, one possibility is that the apoptosis pathway alone is sufficient to activate NINJ1 to induce cell lysis. The other is that NINJ1 functions independently of the pyroptosis, necroptosis and apoptosis pathways. Therefore, we treated Gsdmd^-/-Gsdme^-/-Mlkl^-/- BMDMs with the pan-caspase inhibitor Z-VAD-FMK to inhibit all apoptosis pathways. Since GSDMD and GSDME are already deficient, there is no need to consider the effects of Z-VAD-FMK’s effect on pyroptosis-inducing caspases. After treatment, all programmed cell death pathways were blocked (Fig. 5e), followed by decreased NINJ1 oligomerization and lytic cell death (Fig. 5f–i and Supplementary Fig. 5f–h). Moreover, supplementation with glycine did not further reduce LDH release (Fig. 5i). Therefore, activation of the apoptosis pathway could induce NINJ1 oligomerization to rupture the plasma membrane, even in the absence of the pyroptosis and necroptosis pathway. In comparison, NINJ1 oligomerization and LDH release were not decreased but were even increased in WT BMDMs after Z-VAD-FMK treatment (Fig. 5g, i). So were in Gsdmd^-/-Gsdme^-/- BMDMs, whose MLKL was phosphorylated as usual (Fig. 5e, g, i). Glycine treatment also partially inhibited NINJ1 oligomerization and LDH release in Z-VAD-FMK-treated Gsdmd^-/-Gsdme^-/- BMDMs (Fig. 5i and Supplementary Fig. 5i). NINJ1 was formerly reported to function little in necroptosis.^17,22,23 However, here we showed that in the absence of the apoptosis and pyroptosis pathways, necroptosis could also induce NINJ1 oligomerization, and inhibiting NINJ1 oligomerization could partially alleviate necroptosis-induced cell lysis during IAV infection. Treatment with the caspase-8 inhibitor Z-IETD-FMK yielded results showing similar trends, though NINJ1 oligomerization and LDH release were only partially inhibited (Fig. 5e–i and Supplementary Fig. 5f–h). The remaining apoptosis pathway was not thoroughly inhibited, as the cleavage of Caspase-3 and Caspase-7, which are possibly cleaved by Caspase-9, could still be detected. The activation of Caspase-9 disappeared when we substituted Z-IETD-FMK with Z-VAD-FMK (Fig. 5e), which inhibited NINJ1 oligomerization more thoroughly (Fig. 5g). Together, our results suggest that NINJ1 responds to the activation of any of the three classical cell death pathways of PANoptosis and that even the apoptosis pathway alone is able to induce robust cell lysis by inducing NINJ1 oligomerization (Supplementary Fig. 6).

NINJ1 functions poorly in IAV-infected alveolar epithelial cells

While the expression of NINJ1 in AECs was much lower than that in myeloid cells (Fig. 1f), its potential role in AECs could not be ruled out since NINJ1 was also upregulated in AECs (Fig. 1g and Supplementary Fig. 7b). In contrast to that in macrophages, GSDME-mediated pyroptosis is the major mode of cell death in IAV-infected AECs.^32,33,34 Therefore, we compared the death patterns of IAV-infected WT, Ninj1^-/- and Gsdme^-/- A549 cells. Improved cell viability, attenuated cell lysis, and decreased leakage of intracellular components were observed only in Gsdme^-/- A549 cells but not in Ninj1^-/- A549 cells (Supplementary Fig. 7a–d). Accordingly, glycine treatment had no effect on LDH release (Supplementary Fig. 7e). Since respiratory epithelial cells provide niches for IAV replication and the production of progeny virions, we determined the viral titers in the culture supernatants and found no significant difference between WT and Ninj1^-/- A549 cells (Supplementary Fig. 7f). Although direct evidence of NINJ1 oligomerization in A549 cells is temporarily unavailable because of a lack of appropriate detection antibodies, we may conclude that NINJ1 plays little role in IAV-infected AECs.

NINJ1 drives IAV-induced lung pathology and excessive inflammation

Several in vivo studies have revealed the potential role of NINJ1 in different diseases, but no studies have investigated viral pneumonia. Unlike Zbp1 deficiency, Ninj1 deficiency improved survival after IAV challenge (Fig. 6a, b). In line with the in vitro results, Gsdmd^-/-, Gsdme^-/-, and Mlkl^-/- mice did not benefit from gene deficiency (Fig. 6c). Pathological injuries were apparently alleviated in Ninj1^-/- mice (Fig. 6d, e). The protection conferred by Ninj1 deficiency was not a result of changes in the viral load (Fig. 6f). Compared with that from WT mice, the BALF (bronchoalveolar lavage fluid) from Ninj1^-/- mice contained fewer cellular pellets after centrifugation and lower protein concentrations, and significantly lower levels of HMGB1 and proinflammatory cytokines (IL-1β, IL-6, and TNF-α) were detected (Fig. 6g–l). These differences were most pronounced during the severe infection phase (6 or 8 dpi) rather than during the early infection phase (3 dpi), although the exact time points of significant contrast differed slightly between individual markers. In contrast, more proteins and HMGB1, as well as comparable IL-1β and TNF-α levels, were observed in the BALF from Gsdmd^-/- mice (Supplementary Fig. 8a–d). Zbp1^-/- mice bore a greater viral burden and more proteins in the BALF (Fig. 6f and Supplementary Fig. 8a), possibly due to failure to eradicate IAV infection. Although excessive IL-1β mediates the immunopathology of hyperinflammation,^35,36 IL-1β is fundamental to host defense against flu challenge.^12,37 Both Il1r^-/- and Il1b^-/- mice were more susceptible to IAV-induced mortality (Supplementary Fig. 8e, f), and additional ablation of Il1b weakened the protection conferred by Ninj1 deficiency (Supplementary Fig. 8g). Therefore, it seems that NINJ1 decreases IL-1β release to an extent that does not sacrifice its anti-viral immunity. Since inflammatory cell infiltration was reduced (Fig. 6d), to better understand the immunological changes caused by Ninj1 knockout, we analyzed BALF cells via flow cytometry. The immune cell subset composition showed no difference at 3 dpi (Fig. 6m, n and Supplementary Fig. 8h, 9a–c), ensuring normal inflammatory responses to clear the virus. Reduced immune infiltration primarily occurred at 6 dpi, specifically affecting Ly6C⁺ monocytes (Fig. 6m, n and Supplementary Fig. 8h), which are closely linked to IAV-induced lung injury.^38,39 Significant differences were not detected in the type I interferon responses (Fig. 6o) or anti-IAV CD8⁺ T cell responses (Supplementary Fig. 8i, 9c). The impaired epithelial barrier integrity was ameliorated after Ninj1 knockout, as shown by less reduced expression of the tight junction protein ZO-1 (Supplementary Fig. 8j). This is likely an indirect consequence of Ninj1 knockout (reduced immune inflammation) rather than a direct result, since in vitro experiments (Supplementary Fig. 7) did not provide direct evidence that NINJ1 plays a significant role in AECs. Notably, we excluded Ninj1^-/- and Ninj1^-/-Il1b^-/- mice from studies that developed abnormally in appearance, i.e., stunted growth, hydrocephaly, and ataxia (Supplementary Fig. 8k), as reported in the Mouse Genome Informatics (MGI) database.⁴⁰

NINJ1 is a potential bioindicator of hyperinflammation and poor outcome in patients with severe viral pneumonia

Although we revealed a potential association between elevated NINJ1 and disease severity in IAV-infected murine lungs (Fig. 1), whether this relationship also applies in the clinic remains unknown. Since BAL is not a routine procedure for influenza unless clinically indicated, BALF samples from severe IAV-infected patients were temporarily unavailable to us in the short term, and we did not find appropriate scRNA-seq data on clinical respiratory samples in public databases. We noted that PANoptosis is involved not only in IAV infection but also in SARS-CoV-2 infection.^41,42,43,44 Like IAV, SARS-CoV-2 is a single-stranded RNA virus that causes severe pneumonia as well as mild upper respiratory infection. Since influenza and COVID-19 share many similarities in terms of virology and pathophysiology, we sought to verify the role of NINJ1 in COVID-19 patients. We analyzed scRNA-seq data from BALF samples collected by our group from 10 hospitalized COVID-19 patients and 5 controls (CNCB, PRJCA033941; Supplementary Table 2). All 10 patients were severely/critically ill, among whom 5 died eventually and 5 survived. We examined NINJ1 expression globally in the three groups and found that NINJ1 was profoundly elevated after infection and was most abundantly expressed in the deceased group (Supplementary Fig. 10a). The cells were clustered into 13 major types (Supplementary Fig. 10b, c). The proportions of monocytes/macrophages (Mo/Mas) and Neus increased, whereas the proportion of alveolar macrophages (AMs) decreased after SARS-CoV-2 infection (Supplementary Fig. 10d). This change in cell fraction was more significant in patients who died, indicating that Mo/Mas and Neus might play important roles in COVID-19 disease progression. We further assessed NINJ1 expression patterns in different cell types and found that NINJ1 was highly expressed specifically in Mo/Ma and Neu and exhibited a progressive increase among the three groups (Supplementary Fig. 10e). None of the other PANoptosis-associated genes exhibited trends similar to those of NINJ1 in Mo/Mas (Supplementary Fig. 10e). The expression of NINJ1 was positively correlated with inflammatory responses but not with antiviral responses both globally and in Mo/Mas, with a correlation coefficient that was not much lower than that of IL1B (Supplementary Fig. 10f–h). Although NINJ1 was also highly expressed and showed a stepwise increase in Neus, a positive correlation was not detected in Neus or AMs (Supplementary Fig. 10f). Owing to the limited sample size of our study, we re-analyzed another single-nuclei RNA sequencing data from a published dataset (GSE171524⁴⁵) comprising data directly from lung specimens from COVID-19 patients within hours of death. These samples had advantages over our samples since lung specimens could provide a more comprehensive view of SARS-CoV-2-induced immunopathology than BALF samples. After cell clustering (Supplementary Fig. 11a, b), we found that NINJ1 was specifically highly expressed similarly in increased Mo/Mas (Supplementary Fig. 11c, d). In line with our results, the expression of NINJ1 was aberrantly increased in Mo/Mas from deceased COVID-19 patients (Supplementary Fig. 11e). Collectively, our scRNA-seq analysis reveals the potential of NINJ1 in Mo/Mas to be a biomarker of hyperinflammation status and poor prognosis in patients with severe COVID-19. Further analysis of other viral pneumonia based on a larger sample size is warranted.