Introduction

ALF is a severe consequence of abruptly massive hepatocyte death in the absence of existing liver disease, and can evolve over days or weeks to a lethal outcome without liver transplantation.1 Multiple insults, such as infectious, immune, and drugs, can lead to ALF and present with a consistent pattern of rapid-onset aminotransferase elevation, coagulation insufficiency, and hepatic encephalopathy.2,3 Although liver transplantation is the only treatment that can improve the outcome of ALF, it is critically limited due to a shortage of donors, lifelong immunosuppression, and comparatively high mortality.4,5 Therefore, there is an urgent need to explore novel therapy strategies for ALF.

Immune dysregulation is central to the pathogenesis of ALF, and it is widely accepted that profound activation of systemic inflammatory response rather than the primary etiologic insults leads to most mortality.6 Macrophages play a crucial role in driving the initiation, propagation, and resolution of ALF.7,8 After primary insults, the liver resident macrophages, KCs, detect hepatocyte death through DAMP/TLR signaling and initiate a pro-inflammatory response. The pro-inflammatory cytokines and chemokines released from activated KCs recruit monocyte-derived macrophages and other immune cells, which drive further cytokines and vasoactive mediator production, and consequently provoke systemic inflammation.6,9 The NLR family pyrin domain containing 3 (NLRP3) inflammasome activation in macrophages contributes to the inflammatory response during ALF10,11,12,13,14. NLRP3 is the most representative NOD-like pattern recognition receptor (PRR) which can recognize conserved microbial motifs (also known as pathogen-associated molecular patterns, PAMPs) and endogenous danger signals (also known as damage-associated molecular patterns, DAMPs). Activation of NLRP3 triggers the assembling of the cytosolic multi­protein complex called NLRP3 inflammasome, which serves as a proteolytic activation platform for caspase-1 leading to the GSDMD cleaved and pro-cytokine converted to mature forms. The N-terminal of the GSDMD pore allows proinflammatory cytokines (IL-1β and IL-18) release and drives pyroptotic cell death (called pyroptosis).15 Regulation of the expression of NLRP3 or the activation of its complex protects mice against lethal experimental ALF, establishing NLRP3 as a potential therapeutic target for liver injury and ALF10,11,12,13,14,16.

Pyruvate kinase (PK) is a key enzyme of glycolysis, catalyzing the conversion of phosphoenolpyruvate to pyruvate. There are four isoforms of PK in mammals with unique tissue expression patterns and regulatory properties.17 Particularly, the M1- and M2-type isozymes of PK (PKM1 and PKM2) are encoded from the same Pkm gene by alternative RNA splicing.18 PKM1 is mainly expressed in differentiated tissues such as muscle and brain, while PKM2 is predominantly expressed during development and in tumors. Functionally, PKM1 has high glycolytic activity due to its stable tetrameric conformation, whereas PKM2 has two states of low or high glycolytic activity because of its allosteric modulation between dimeric and tetrameric conformations.17,19 Besides its low enzymatic activity in the cytoplasm, the dimeric PKM2 has the ability to translocate into the nucleus and act as a transcriptional regulator, regulating gene expression via interaction with some transcription factors, such as STAT3.20,21 Administration of Shikonin (a PKM2 enzyme activity and nuclear translocation inhibitor) or knockdown PKM2 decrease aerobic glycolysis and the inflammasome activation in macrophages.22 Recent studies suggest that PKM2-dependent glycolysis regulates inflammasome activation and production of inflammatory cytokines in macrophages.22,23,24,25 Our previous study shows that targeting PKM2 to inhibit macrophage activation-mediated inflammation can improve mortality and alleviate hepatic injury in D-galactosamine (D-GalN)/LPS-induced ALF mice.26 However, whether PKM2 can regulate NLRP3 inflammasome activation in macrophages via transcriptional regulation is unknown.

In this study, we demonstrate that PKM2 contributes to ALF by increasing NLRP3 expression and promoting NLRP3 inflammasome activation in KCs. The dimeric PKM2 translocates into the nucleus and combines with STAT3, enhancing its phosphorylation and recruitment to the NLRP3 promoter region. Myeloid cell-specific Pkm gene knockout and pharmacological inhibition of PKM2 nuclear translocation can reduce mortality and alleviate hepatic injury in D-GalN/LPS-induced ALF mice by decreasing NLRP3-mediated pyroptosis. These findings improve our understanding of the emerging role of PKM2-nonmetabolic function in pyroptosis and suggest that PKM2 represents a potential therapeutic target for ALF.

Results

Knockout of Pkm in myeloid cells prevents D-GalN/LPS-induced ALF in mice

In adult mouse liver, PKM2 is mostly expressed in KCs, the liver resident macrophages (Fig. 1A). To confirm the role of PKM2 in the pathogenesis and progression of ALF in vivo, we bred myeloid cell-specific Pkm gene knockout mice (Pkmflox/flox-Lyz2cre, PkmΔ) by crossing Pkmflox/flox and Lyz2-cre transgenic mice as Supplemental Methods. Pkmflox/flox mice carrying loxP sites flanking exon 4, 5, and 6 of the Pkm gene. The knockout of Pkm in myeloid cells was confirmed by Immuno-fluorescence analysis of PKM2 protein expression in WT and PkmΔ mouse liver (Fig. S1A), qPCR analysis of PKM1 and PKM2 mRNA (Fig. S1C) and western blot analysis of PKM1 and PKM2 protein (Fig. S1D) expression in KCs. Littermates Pkmflox/flox mice were used as WT control. We performed the classic D-GalN/LPS ALF model in 8-week-old healthy male WT and PkmΔ mice 27,28, liver/blood samples collecting and survival analysis according to the time points on the schematic (Fig. 1B). As expected, myeloid cell-specific knockout of Pkm gene significantly improved the gross liver appearance in ALF mice model with the absence of typical hemorrhage sign (Fig. 1C). Hematoxylin-eosin (HE) staining of liver sections showed the hepatic architecture in WT-ALF group was disrupted with the extensive hepatocyte necrosis and sinusoid hemorrhage, and accompanied by massive inflammatory cell infiltration. While those classically histological changes of ALF were ameliorated in the PkmΔ-ALF group (Fig. 1C). Additionally, serum indexes including AST, ALT, and LDH were significantly increased after D-GalN/LPS administration. Pkm-specific knockout in myeloid cells markedly reduced the elevation levels of serum ALT, AST, and LDH (Fig. 1E–G). The above results indicated that myeloid cell-specific knockout of the Pkm gene attenuated hepatic inflammatory injury in ALF model mice. Moreover, Kaplan–Meier analysis of survival rates between PkmΔ mice and control WT mice revealed that myeloid cell-specific Pkm gene knockout significantly reduced the mortality of mice with D-GalN/LPS induced ALF (P = 0.007, log-rank test; Fig. 1D). Taken together, these findings hint that the expression of Pkm gene in KCs is required for pathogenesis and progression of ALF in model mice.

Fig. 1: Conditional knockout of Pkm in myeloid cells prevents D-GalN/LPS-induced ALF in mice.
figure 1

A Fluorescence microscopy analysis of PKM2 (green), F4/80 (red), and nuclei labeled with DAPI (blue) in liver tissues; the scale bar indicates 20 µm. B Schematic of D-GalN/LPS-induced ALF mouse model and the analysis time points. C Macroscopic analysis of the liver and hematoxylin-eosin (HE) staining of liver sections; scale bar indicates 100 µm. D Kaplan–Meier analysis of survival rates between PkmΔ mice and WT mice. E–G Serum markers, including AST, ALT, and LDH, were measured. **P < 0.01, ***P < 0.001.

Full size image

PKM2 knockout protects ALF model mice via suppressing NLRP3-mediated pyroptosis

To explore the role of the Pkm gene in D-GalN/LPS-induced ALF, we performed single-cell RNA-sequencing (scRNA-seq) on hepatic parenchymal and nonparenchymal cells isolated from healthy and ALF model liver tissues of WT and PkmΔ mice (n = 2 male mice per group), and 53,481 cells were included in the analysis. The study has not yet been published. In this study, we found that comparing with the negative control (NC) group, the transcriptomic profile of the ALF group was significantly different. In liver macrophages, KEGG enrichment analysis revealed that most genes upregulated after ALF belong to pathways related to inflammatory response, chemokine, and cytokine. (Fig. 2B). Meanwhile, myeloid cell-specific knockout of Pkm gene can significantly change the gene expression profile, especially downregulating those genes that upregulated in ALF and related to inflammatory pathways (Fig. 2C). Based on the above results, we selected the relatively well-ranked and less studied NOD-like receptor signaling pathway for further investigation. The NLRP3 inflammasome pathway is the most extensively studied and representative of NOD-like receptor signaling pathways. Violin plots showed that the expression of NLRP3, IL-1β, and IL-18 were upregulated upon induction of ALF, and downregulated in the PkmΔ group (Fig. 2D). Serum IL-1β and IL-18 levels were also consistent with the aforementioned changes (Fig. 2E). To confirm the function of Pkm gene in NLRP3 inflammasome activation, WT and Pkm-deficient KCs were isolated and treated with LPS then stimulated with or without NLRP3 inflammasome activator, such as nigericin and ATP. Western blot analysis demonstrated that Pkm gene knockout reduced the level of cleaved GSDMD (p31) in KCs primed by LPS and subsequently stimulated with nigericin (Fig. 2F) or ATP (Fig. S2A). Similarly, ELISA results showed that knockout of the Pkm gene inhibited IL-1β and IL-18 release in LPS-primed KCs following treatment with nigericin (Fig. 2H) or ATP (Fig. S2D). Collectively, those findings suggest that the knockout of the Pkm gene suppresses NLRP3-mediated pyroptosis activation. To further clarify whether PKM1 or PKM2 was required in KCs pyroptosis activation, we suppressed the expression of PKM1 or PKM2 by lentiviral transduction with PKM1 or PKM2-specific shRNA (Figs. 2G and S1E). Knockdown of PKM2, but not PKM1, significantly inhibited cleavage of GSDMD (Fig. 2G) and reduced IL-1β and IL-18 release by KCs (Fig. 2I).

Fig. 2: PKM2 knockout protects ALF model mice via suppressing pyroptosis in liver macrophages.
figure 2

A Schematic of D-GalN/LPS-induced ALF mouse model and the analysis time points. B–D Single-cell transcriptome analysis of mouse liver (n = 2 mice per group): B, C KEGG enrichment analysis of genes upregulated/downregulated in liver macrophages: B genes upregulated in WT-ALF group compared to negative control (WT-NC) group; C genes downregulated in PkmΔ-ALF group compared to WT-ALF group; and D violin plots showed the expression of selected genes in each group of liver macrophages. E Serum IL-1β and IL-18 were assayed with ELISA. F, G PKM1, PKM2, GSDMD (P53), and N-terminal (P31) in lysates were assayed using western blot, and ACTB was used as a loading control. F WT or Pkm-deficient KCs were treated with LPS and then stimulated with or without nigericin. G PKM1 or PKM2 knockdown KCs (specific shRNA, 48 h) were treated with LPS then stimulated with or without nigericin. H, I The levels of IL-1β and IL-18 in the cell supernatant were measured by ELISA. **P < 0.01, ***P < 0.001.

Full size image

Macrophages PKM2 deficiency suppresses pyroptosis activation

The morphologies of the KC's pyroptotic death were observed by light microscopy. As expected, most KCs in the WT group emerged swelling (indicated by white arrow) upon induction of pyroptosis, whereas significantly fewer KCs underwent swelling in the PkmΔ group (Fig. 3A). Immunofluorescence analysis showed ASC colocalized with NLRP3 indicating the formation of NLRP3 inflammasome (indicated by white arrow) in KCs after LPS/nigericin treatment. Moreover, the formation of NLRP3 inflammasome was markedly reduced in PKM2 deficiency KCs (Figs. 3B and  S1H). Western blot and ELISA analysis demonstrated that after 5 min of nigericin administration in LPS-primed KCs, GSDMD had been cleaved, while IL-1β and IL-18 release occurred mainly after 30 min. PKM2 knockdown inhibited the GSDMD cleavage and the release of IL-1β and IL-18 (Fig. 3C, D). The same results were observed in experiments using bone marrow-derived macrophages (BMDMs) (Fig. S2B–E). In addition, the Western blot results also showed that the expression level of PKM2 in KCs was increased after LPS administration (Fig. S1B). These findings suggest that PKM2 regulates the activation of NLRP3-mediated pyroptosis in KCs.

Fig. 3: Macrophages PKM2 deficiency suppresses pyroptosis activation.
figure 3

A Microscope images of mouse WT or Pkm-deficient KCs treated with LPS (500 ng ml−1, 3 h) then stimulated with or without nigericin (20 μM, 40 min), scale bar indicates 100 µm. B Fluorescence microscopy imaging of NLRP3 inflammasomes in KCs treated with LPS (500 ng ml−1, 3 h) then stimulated with or without nigericin (20 μM, 10 min), to visualize the subcellular localization of NLRP3 and ASC. Scale bars indicate 10 μm. C, D LPS-primed KCs were treated with or without nigericin (20 μM, 5 min and 30 min): C PKM2, GSDMD (P53), and N-terminal (P31) in lysates of KCs were assayed using western blot; D The levels of IL-1β and IL-18 in the cell supernatant were measured by ELISA. *P < 0.05, **P < 0.01, ***P < 0.001.

Full size image

Inhibition of PKM2 nuclear translocation suppresses pyroptosis activation

TEPP-46 is a selectively enzymatic activator of PKM2 that promotes tetramer formation and inhibits nuclear translocation.29 Compound 3k is a potent and selective PKM2 inhibitor (PKM2i) that locks PKM2 into a low enzymatic activity conformation.30 To explore the mechanism by which PKM2 regulates NLRP3-mediated pyroptosis activation in KCs, we treated mice with TEPP-46 or PKM2i 30 min after D-GalN/LPS administration (Fig. 4A). Survival analysis showed that TEPP-46 significantly increased the survival of ALF-modeled mice (P = 0.020, log-rank test), whereas PKM2i did not (Fig. 4B). The promotion of glycolytic activity with TEPP-46 and the inhibition of glycolytic activity with PKM2i or Pkm knockdown were confirmed by measuring the PK activity, lactate production and glucose consumption in KCs (Fig. S3). Western blot and ELISA analysis revealed that administration of TEPP-46, but not PKM2i, significantly reduced cleavage of GSDMD and release of IL-1β and IL-18 by KCs (Fig. 4C, D) or BMDMs (Fig. S2C–F). Moreover, immunofluorescence results showed that TEPP-46, rather than PKM2i, can inhibit NLRP3 inflammasome (which includes ASC and NLRP3 and indicated by white arrow in Fig. 4E) formation in KCs after LPS/nigericin administration (Figs. 4E and  S1I). Similarly, light microscopy revealed that TEPP-46 inhibited LPS/nigericin-induced pyroptotic death (indicated by white arrow) of KCs, while PKM2i did not (Fig. 4F). Collectively, the above results indicate that PKM2 is supposed to be involved in the NLRP3-mediated pyroptosis activation of KCs through nuclear translocation, but not enzymatic activity.

Fig. 4: Inhibition of PKM2 nuclear translocation suppresses pyroptosis activation.
figure 4

A Schematic of TEPP-46 or PKMi treatment in D-GalN/LPS-induced ALF mouse model and the analysis time points. B Survival rates of TEPP-46 and PKM2i groups were respectively compared to the vehicle group using Kaplan–Meier analysis. C, D KCs were treated with LPS (500 ng ml−1) for 3 h, while TEPP-46 and PKM2i were respectively added into the corresponding subgroups, then stimulated with or without nigericin and assayed PKM2, GSDMD (P53) and N-terminal (P31) using western blot (C) or measured IL-1β and IL-18 levels in the medium using ELISA (D). E Fluorescence microscopy imaging of NLRP3 inflammasomes (visualized by the colocalization of NLRP3 and ASC) in KCs treated with LPS (500 ng ml−1, 3 h) and nigericin (20 μM, 10 min) in the presence and absence of TEPP-46 or PKM2i. Scale bars indicate 10 μm. F Microscope images of KCs treated with LPS (500 ng ml−1, 3 h) then stimulated with nigericin (20 μM, 40 min) in the presence and absence of TEPP-46 or PKM2i. Scale bar indicates 100 µm. *P < 0.05, **P < 0.01.

Full size image

PKM2 nuclear translocation promotes NLRP3 expression

Previous results of scRNA-seq analysis showed that the expression of NLRP3 was upregulated upon induction of ALF, and downregulated in the PkmΔ group (Fig. 2D). To further investigate the functional importance of PKM2 nuclear translocation in NLRP3 expression, genetic knockout and pharmacological inhibition were used. Immunofluorescence analysis of mouse liver sections confirmed that myeloid cell-specific knockout of the Pkm gene significantly reduced the expression of NLRP3 in KCs (Fig. 5A). qPCR and western blot results showed that both PKM2 knockout and TEPP-46, rather than PKM2i, can significantly decrease the expression levels of NLRP3 mRNA (Fig. 5B, C) and protein (Fig. 5D, E). Immunofluorescence results showed that PKM2 translocated into nuclear after LPS or LPS/nigericin administration in KCs. TEPP-46, rather than PKM2i, can inhibit the PKM2 nuclear translocation. Pkm gene knockout can decrease the expression of PKM2 in both cytoplasm and nucleus (Fig. 5F). Moreover, PKM2 knockout and TEPP-46 administration substantially reduced the levels of PKM2 in the nucleus, while PKM2i did not (Fig. 5G). Taken together, these results suggest that PKM2 nuclear translocation regulates KCs pyroptosis via promoting the NLRP3 expression.

Fig. 5: Knockout or inhibit nuclear translocation of PKM2 decreases NLRP3 expression.
figure 5

A Fluorescence microscopy analysis of F4/80 (green), NLRP3 (red), and nuclei labeled with DAPI (blue) in WT and PkmΔ mouse liver tissues; scale bars indicate 20 µm. B, E LPS-primed WT or PkmΔ KCs were treated with or without nigericin, expression level of NLRP3 was assayed with qPCR (B) and western blot (E). C, D KCs treated with LPS and TEPP-46 or PKM2i for 3 h, then stimulated with or without nigericin, NLRP3 expression level was assayed with qPCR (C) and western blot (D). F Fluorescence microscopy imaging of PKM2 and STAT3 in WT and PkmΔ KCs were treated as above described. Scale bars indicate 10 μm. G Cytoplasmic and nuclear protein samples extracted from KCs were analyzed by western blot to determine PKM2 levels in those compartments. GAPDH and Lamin were used as cytoplasm and nuclear loading controls, respectively. *P < 0.05, **P < 0.01, ***P < 0.001.

Full size image

PKM2 promotes STAT3 phosphorylation in Kupffer cells

The phosphorylation of STAT3 at Y705 residue can upregulate the expression of NLRP3 and contribute to bortezomib-induced neuropathy.31 Interestingly, the dimeric PKM2 has protein kinase activity to phosphorylate STAT3 at Y705 in the nucleus, enhancing its transcriptional activity and promoting the differentiation of Th17 cells.32 We then examined whether PKM2 can interact with STAT3 and promote the expression of NLRP3 in KCs. Western blot analysis revealed that PKM2-deficient KCs have significantly lower levels of Y705-phosphorylated STAT3 (p-STAT3) (Fig. 6A). TEPP-46 also significantly decreased the levels of p-STAT3 in KCs, while the PKM2i did not (Fig. 6B). To confirm the direct and specific interaction between PKM2 and STAT3, immunoprecipitation analysis was performed using antibodies against PKM2 and STAT3. Results of western blot showed that STAT3 immunoprecipitated with PKM2, and likewise PKM2 coimmunoprecipitated with STAT3 in KCs (Fig. 6C). According to a prediction by the JASPAR database, five potential STAT3 binding sites were identified in the promoter of NLRP3 (2000 bp upstream of the transcription start site). To clarify the STAT3 binding motif in NLRP3 promoter, five potential STAT3 binding motifs were respectively cloned into luciferase reporter vectors to construct luciferase reporter plasmids named BS1 to BS5 (Fig. 6D). Results of luciferase reporter analysis showed that, only BS3 and BS4 exhibited increasing of luciferase activity upon LPS/nigericin administration, and the level of luciferase activity elevated higher by BS3 (Fig. 6F). Further experiments demonstrated that the upregulation of luciferase activity by BS3 and BS4 could be blocked by corresponding mutant plasmids (BS3M and BS4M) and diminished by PKM2 knockdown using lentiviral transduction of PKM2 shRNA (Figs. 6E, G and S4A, B). The above results indicated that the BS3 and BS4 binding sites, mainly BS3, were positive STAT3 binding motifs in the NLRP3 promoter. Further ChIP-qPCR assays verified the direct bind between p-STAT3 and promoter of NLRP3, and the occupation of p-STAT3 on NLRP3 promoter could be reduced by Pkm knockout and TEPP-46 treatment (Figs. 6h and S4C). Additionally, survival analysis showed that JAK2-STAT3 inhibitors, such as AG490 and Ruxolitinib (Rux), can effectively improve the survival rate of ALF-modeled mice by inhibiting STAT3 phosphorylation (Fig. S4D–F). Taken together, these data determined that PKM2 was a nuclear transcriptional coactivator, and promoted NLRP3 expression by interaction with transcriptional factor STAT3 during KCs pyroptosis.

Fig. 6: Nuclear PKM2 increases NLRP3 expression via STAT3 phosphorylation in Kupffer cells.
figure 6

A, B Western blot was performed to identify total and phosphorylated (Y705) levels of STAT3 in A WT and PkmΔ KCs treated with LPS followed with or without nigericin, B KCs treated with LPS and nigericin in the presence and absence of TEPP-46 or PKM2i. C The interaction between STAT3 and PKM2 was examined by immunoprecipitation (IP). WT or Pkm-deficient KCs lysates were subjected to IP with anti-PKM2 or anti-STAT3 antibody, and anti-IgG was used as control, followed by western blot (WB) analysis. D Five potential STAT3 binding sites occupied in the promoter of NLRP3 were respectively cloned into luciferase reporter vectors (BS1–BS5). E The luciferase reporter vector containing the third wild-type (BS3) or mutant-form (BS3M) binding site. F Promoter-luciferase reporter activity changes of the five vectors in RAW264.7 cells after LPS (500 ng/ml, 3 h) and nigericin (20 µM, 10 min) treatments. G Promoter luciferase reporter activity in RAW264.7 cells treated with LPS (500 ng/ml, 3 h) and nigericin (20 µM, 10 min). H ChIP-qPCR analysis using anti-pY705-STAT3 antibody to determine the occupation of pY705-STAT3 on the BS3 of NLRP3 promoter in bone marrow-derived macrophages (BMDMs) treated with LPS (500 ng/ml, 3 h) and nigericin (20 µM, 10 min). (I) Schematic depicting nuclear translocated PKM2-mediated STAT3 phosphorylation promoting NLRP3 expression and pyroptosis activation. *P < 0.05, **P < 0.01, ***P < 0.001.

Full size image

Discussion

Macrophages, especially KCs (liver resident macrophages), play a crucial role in the defense against liver infection and the initial immune response after drug injury; however, excessive activation of macrophages and overproduction of cytokines contributes to uncontrolled systemic inflammation during ALF.33 Recent studies about the relationship between cellular metabolism and immunology revealed that PKM2-dependent glycolysis promotes inflammasome activation and cytokines production in macrophages.22,23,24,25 Our previous study suggested PKM2 is a potential therapeutic target for ALF.26 Notably, PKM2 can not only act as a glycolytic enzyme in the cytoplasm but also translocate to the nucleus to regulate gene expression,20,21 suggesting that it may be involved in the ALF via other non-metabolic functions unrelated to glycolysis.

In the present study, we demonstrated that the transcriptional coactivator PKM2 promotes NLRP3 expression and pyroptosis activation in macrophages by enhancing STAT3 phosphorylation in the nucleus. Specific knockout or pharmacological inhibit nuclear translocation of PKM2 protected mice against lethal ALF and reduced the cleavage of GSDMD and the release of IL-1β and IL-18 in macrophages, while inhibition of glycolytic activity did not. Thus, the current study suggests that PKM2 mainly mediates non-metabolic pathways such as combining and enhancing the phosphorylation level of STAT3 to promote NLRP3 inflammasome activation in macrophages during ALF (Fig. 6I).

PKM2-dependent glycolysis plays a critical role in NLRP3 inflammasome activation and subsequent proinflammatory cytokines release in macrophages through lactate-mediated EIF2AK2 phosphorylation.22,23 In the present study, we have demonstrated that PKM2 is mainly expressed in KCs in adult mice liver, and LPS treatment significantly increases its expression. We therefore hypothesized that PKM2-dependent glycolysis would also be required for the NLRP3-mediated pyroptosis activation of liver macrophages in ALF. As hypothesized, PKM2 deficiency did impair the NLRP3 inflammasome activation and cytokines (IL-1β and IL-18) production of KCs and BMDMs, consequently protecting mice from GalN/LPS-induced ALF. Interestingly, activators but not inhibitors of PKM2 glycolytic activity also reduced the NLRP3-mediated pyroptosis activation of KCs and the mortality of ALF-modeled mice.

As described above, PKM2 tetramers have high metabolic activity while dimers have low metabolic activity but are able to translocate into the nucleus and regulate gene expression as a transcriptional coactivator.21 TEPP-46, the PKM2 enzymatic activator we used in this study, promotes glycolytic activity and inhibits nuclear translocation of PKM2 by stabilizing it in the tetrameric form. Therefore, we hypothesized that PKM2 nuclear translocation, but not PKM2-dependent glycolysis, plays a major role in the regulation of NLRP3-mediated pyroptosis in liver macrophages. However, it was reported that inhibition of PKM2-dependent glycolysis with PKM2 knockdown or shikonin impairs NLRP3 inflammasome activation of macrophages22, which contrasts with our findings. This difference might be explained by nuclear PKM2 knockout or off-target effects of shikonin in the inhibition of PKM2 nuclear translocation. Shikonin, a pharmacological inhibitor of PKM2, decreased both dimerization and tetramerization of PKM2 by bounding to PKM2 protein and promoting PKM2 macromolecular polymer aggregation.34,35 A limitation to be noted is that the effect of glycolysis inhibition (PKM2i group) on NLRP3 inflammasome activation is not remarkable in this study, which does not exclude the possibility of false-negative results due to different time points for assay. So, our results are not sufficient to exclude the involvement of PKM2-dependent glycolysis in regulating NLRP3-mediated pyroptosis.

Of note, we found that PKM2 deficiency or inhibition of PKM2 nuclear translocation significantly reduced the expression of NLRP3 in KCs. It suggested that PKM2 nuclear translocation is involved in the NLRP3-mediated pyroptosis activation of KCs via regulating the expression level of NLRP3. Nuclear PKM2 has been reported to interact with STAT3 and enhance its phosphorylation at Y705, contributing to the differentiation of Th17 cells32 and proinflammatory cytokines production by macrophages.24 Moreover, it was shown that inhibition of STAT3 phosphorylation can alleviate GalN/LPS-induced ALF.36 In the present study, we demonstrated that gene knockdown or pharmacological inhibition of nuclear translocation of PKM2 significantly decreased the phosphorylation level of STAT3 at Y705, and PKM2 could combine with STAT3 in macrophages. Whether the Y705-phosphorylated STAT3 is directly catalyzed by the nuclear dimeric PKM2 as a protein kinase or due to some indirect mechanism via other protein kinases regulated by PKM2 remains to be further investigated.

Recent studies have shown that JAK2/STAT3 pathway is involved in several inflammatory diseases by regulating NLRP3. For example, curcumin could significantly inhibit the occurrence of pyroptosis and fibrosis in aflatoxin B1-exposed duck livers by regulating the activation of the JAK2/STAT3 pathway and NLRP3 inflammasome.37 Inhibition of STAT3 phosphorylation downregulates NLRP3 expression by decreasing the acetylation of histones H3 and H4 on the NLRP3 promoter, consequently alleviates sepsis-induced acute lung injury38 and ischemic stroke-induced neuroinflammation39. Our current data confirm above previous studies showing that Y705-phosphorylated STAT3 can bind with NLRP3 promoter and upregulate NLRP3 expression in macrophages. Moreover, based on the prediction by the JASPAR database, we further determined that the main STAT3 binding motif is a −1329/−1320 region relative to the transcription start site of the mouse Nlrp3 promoter.

In conclusion, our current study demonstrated that PKM2 acts as a critical nonmetabolic regulator of KCs pyroptosis by increasing NLRP3 expression and promoting NLRP3 inflammasome activation during ALF. We also demonstrated that nuclear PKM2-mediated STAT3 phosphorylation is a major event that controls NLRP3 expression and NLRP3 inflammasome activation in macrophages. Therefore, PKM2 may represent a potential therapeutic target for acute liver failure.

Methods

Animals and ALF models

Myeloid cell-specific Pkm gene knockout mice were bred by crossing Pkmflox/flox (Cyagen US Inc) and Lyz2-cre (Cyagen US Inc) transgenic C57BL/6 mice as Supplemental Methods. ALF model was built in WT or PKMflox/flox-Lyz2cre (PkmΔ) C57BL/6 mice (male, 20–25 g, 7–8-weeks-old) by intraperitoneal injection (i.p.) of 350 mg kg−1 D-GalN (Sigma, G1639) and 10 μg kg−1 LPS (Sigma, L4391). WT ALF model mice were randomly divided into TEPP-46 (Selleck, S7302), PKM2i (Selleck, S8616), AG490 (Selleck, S1143), Ruxolitinib (Selleck, S1378) and vehicle (10% DMSO and 90% PBS) groups. TEPP-46 (40 mg kg−1), PKM2i (4 mg kg−1), AG490 (40 mg kg−1), and Ruxolitinib (0.68 mg kg−1) were dissolved in vehicle and i.p. to mice 30 min after the ALF model was built. Organ tissues and blood samples were collected at indicated time points and stored at −80 °C before analysis. The survival rates of the mice were recorded for up to 2 weeks after the model of ALF to ensure that no additional late deaths occurred.

We performed all animal care and experimentation according to the Association for Assessment and Accreditation of Laboratory Animal Care guidelines (http://www.aaalac.org/) and with approval from the Medical Ethics Committee of the Xiangya Hospital.

Liver cell isolation

Liver cells were isolated using a modified protocol of Aleksandra et al. 40 In brief, mouse liver was perfused with perfusion buffer (30 ml, 5 ml/min; 8 g l−1 NaCl, 0.4 g l−1 KCl, 76.8 mg l−1 NaH2PO4, 106.5 mg l−1 Na2HPO4, 2.38 g l−1 HEPES, 0.35 g l−1 NaHCO3, 0.19 g l−1 EGTA, 0.9 g l−1 glucose) followed by digestion buffer (20 ml, 5 ml/min; 8 g l−1 NaCl, 0.4 g l−1 KCl, 76.8 mg l−1 NaH2PO4, 106.5 mg l−1 Na2HPO4, 2.38 g l−1 HEPES, 0.35 g l−1 NaHCO3, 0.42 g l−1 CaCl2, 0.25 g l−1 collagenase type IV (Sigma, C5138)) from the portal vein using an infusion pump. After perfusion, the liver was dissected and removed into cold Dulbecco’s Modified Eagle’s Medium (DMEM) medium and shaken vigorously with forceps then filtered through 70-μm mesh to obtain single cell suspension. For scRNA-seq, the cell suspension was centrifuged at 50 g for 3 min to separate hepatocytes from nonparenchymal cells. After counting and AO/PI staining analysis, hepatocytes and nonparenchymal cells were mixed in a ratio of 2–8.

KCs isolation

Single-cell suspension of the liver was obtained as above. Hepatocytes were depleted from cell suspension by centrifugation at 30 g for 3 min twice. The supernatant was centrifuged (500 g × 5 min, 4 °C) to collect the nonparenchymal cells, followed by centrifugation (850 g × 15 min, 4 °C) through a 50–25% two-step Percoll gradient. Carefully collect the middle layers (KCs banding at the interface between the two density layers) into one new 15 ml tube and fill the tube with cold DMEM. Centrifuge (850 g × 15 min, 4 °C) and discard supernatant, resuspend pellet in DMEM (supplemented with 20% fetal bovine serum, 100 IU ml−1 penicillin, and 100 μg ml−1 streptomycin) and plate cells at an appropriate density. After 1 h of incubation, remove nonadherent cells by replacing the culture medium to obtain a pure monolayer culture of KCs based on selective adherence. Culture KCs for an appropriate period (usually 24–48 h) prior to experiments. In each isolation, a small number of obtained cells were seeded in 24-well plates for immunofluorescence. F4/80+ cells were identified as KCs (Fig. S1F), and about 95% of the isolated cells were KCs (Fig. S1G).

Single-cell RNA-sequencing

ALF model and negative control (NC) were established using D-GalN/LPS and the same volume of PBS. Three hours after modeling, hepatic parenchymal and nonparenchymal cells were extracted and mixed at a ratio of 2 to 8 according to the above method. Cells were loaded onto the 10× Chromium Single Cell Platform (10× Genomics) at a concentration of 1000 cells μl−1 as described in the manufacturer’s protocol. Generation of gel beads in emulsion (GEMs), barcoding, GEM-RT clean-up, complementary DNA amplification, and library construction were all performed as per the manufacturer’s protocol. Qubit was used for library quantification before pooling. The 10× Libraries were sequenced on the Illumina NovaSeq6000 sequencer with pair-end reads. Raw scRNA-seq data were analyzed using Cell Ranger 6.0 Single-Cell Software Suite (10× Genomics). Transcripts were aligned to the 10× pre-built mouse reference (refdata-gex-mm10-2020-A). The marker sets used were as in ref. 40.

Cell culture

Mouse KCs and BMDMs were isolated from WT or PkmΔ mice as mentioned above. RAW264.7 and HEK293T cells were obtained from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). BMDMs were cultured in DMEM medium containing 20 ng ml−1 M-CSF (Sino Biological, 51112-MNAH), 10% fetal bovine serum (HyClone, Utah, USA), 100 units ml−1 penicillin and 100 mg ml−1 streptomycin at 37 °C, 5% CO2 and 95% humidity. Other cells were cultured in the same conditions without M-CSF. KCs and BMDMs were primed with LPS (500 ng ml −1) for 3 h, followed by nigericin (20 µM, Selleck, S6653) or ATP (5 Mm, Selleck, S5260) to activate the NLRP3 inflammasome. Cell lysates or total RNA were collected 10 min later for western blot or qPCR, while culture supernatants were collected 30 min later for ELISA. In the corresponding experimental groups, TEPP-46 (100 μM) or PKM2i (3 μM) was added to the medium at the same time with LPS. All cells were mycoplasma-free and determined by qPCR (Beyotime, C0303S).

Cytokine measurements

The concentrations of IL-1β (R&D, MLB00C) and IL-18 (R&D, 7625) in mouse serum and culture medium were measured using commercial enzyme-linked immunosorbent assay (ELISA) kits according to the manufacturer’s instructions.

Biochemical assays

LDH, ALT, and AST levels in mouse serum were respectively measured with the LDH Assay Kit (A020-2), ALT Assay Kit (C009-2-1), and AST Assay Kit (C010-2-1) according to the manufacturer’s instructions (Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Pyruvate kinase (PK) activity, lactate, and glucose levels were respectively measured with the PK Assay Kit (BC0545), Lactate Assay Kit (BC2235), and Glucose Assay Kit (BC2505) according to the manufacturer’s instructions (Solarbio Life Sciences, Beijing, China).

RNAi

All short hairpin RNAs (shRNAs) against PKM1 or PKM2 were designed and cloned into the Lenti-X expression vectors (Clontech, USA). High-titer lentivirus was produced in HEK 293T cells using recombinant pLVX-shRNA vectors and the Lenti-X HTX packaging system (Clontech, USA) according to the manufacturer’s instructions. The PKM1 shRNA, PKM2 shRNA, or control shRNA was transfected into KCs cells using polybrene (Genomeditech, China). The target sequences are shown in Supplementary Table S1.

Western blot

Total proteins were extracted from KCs using Cell lysis buffer for Western and IP (Beyotime, P0013J). Nuclear and cytoplasmic proteins were extracted from KCs using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific, 78835) according to the manufacturer’s instructions. Proteins were first resolved by 4–20% SDS-PAGE and then transferred onto polyvinylidene difluoride membranes (0.25 μm, Millipore). After being blocked in block buffer (Beyotime, P0023B) for 1 h, membranes were incubated overnight at 4 °C with primary antibody, and subsequently incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature. The antibodies to PKM1 (#7067; 1:1000), PKM2 (#4053; 1:1000), STAT3 (#12640, 1:1000), and p-STAT3 (#9145, 1:1000) were obtained from Cell Signaling Technology (Danvers, MA, USA). The antibodies to NLRP3 (ab263899, 1:1000) and GSDMD (ab219800, 1:1000) were obtained from Abcam (Cambridge, MA, USA). The antibodies to ACTB (AC026, 1:50,000), GAPDH (A19056, 1:2000) and Lamin B1 (A11495, 1:2000) were obtained from ABclonal Technology (Wuhan, China). The signals were visualized using Clarity Max Western ECL Substrate (Bio-Rad, #1705062). Using Image Lab software to scan the blots and measure the relative intensity of bands.

Immunofluorescence

After stimulation, KCs were fixed in 4% paraformaldehyde for 15 min, permeabilized with 0.25% Triton X-100 in PBST (0.25% Tween 20 in PBS) for 20 min, blocked with 5% BSA in PBST for 10 min, and then incubated with the following antibodies overnight at 4 °C: rabbit anti-ASC (ABclonal, A16672, 1:100), rabbit anti-F4/80 (CST, #30325, 1:100), and rat anti-NLRP3 (R&D systems, MAB7578, 1:100). Next day, cells were incubated with Alexa Fluor-conjugated secondary antibodies, 594-conjugated anti-rabbit IgG (Thermo Scientific, A32754, 1:250) and 488-conjugated anti-rat IgG (Thermo Scientific, A48269, 1:250), for 1 h at room temperature, and finally counterstained with DAPI. The immunofluorescence analysis of liver tissue sections was performed with a commercial multiplex immunofluorescence kit (AiFang Biological, AFIHC024) using tyramide signal amplification technology according to the manufacturer’s instructions. The antibodies to F4/80 (#30325, 1:1000), PKM2 (#4053; 1:1000), and NLRP3 (#15101, 1:1000) were obtained from Cell Signaling Technology (Danvers, MA, USA). Photographs were taken with an immunofluorescence microscope (Axio Imager M2, Zeiss, Germany).

RT-PCR

Total RNA was extracted from KCs using Total RNA kit II (Omega, R6934) and subsequently reverse-transcribed into cDNA using Prime Script RT reagent kit with gDNA Eraser (Takara, RR047A) according to the manufacturer’s instructions. Quantitative RT-PCR was performed using SYBR Premix Ex Taq II Kit (Takara, RR820A) on ABI 7500 Fast real-time PCR platform to measure the mRNA expression levels of PKM1, PKM2, NLRP3, and ACTB (as the internal control). Relative quantification of mRNA was calculated by the 2−ΔΔCt method. All the primer sequences were listed in Supplementary Table S2.

Immunoprecipitation

Immunoprecipitation (IP) experiments were performed using the Pierce Co-Immunoprecipitation (Co-IP) Kit (Thermo Scientific, 26149) according to the manufacturer’s instructions. The antibodies to PKM2 (CST, #4053, 1:100) and STAT3 (CST, #12640, 1:100) were obtained from Cell Signaling Technology (Danvers, MA, USA). Western blot was performed to examine whether PKM2 combined with STAT3.

Chromatin immunoprecipitation

Chromatin immunoprecipitation (ChIP) assays were performed using the Pierce Magnetic ChIP Kit (Thermo Scientific, 26157) according to the manufacturer’s instructions. The anti-p-STAT3 antibody (CST, #9145, 1:100) for the ChIP assay was obtained from Cell Signaling Technology (Danvers, MA, USA). Meanwhile, the anti-IgG antibody (CST, #45262, 1:100) was used as a negative control. The precipitated DNA was measured using quantitative RT-PCR as described above. Primers were designed to amplify regions relative to the transcription start site of mouse Nlrp3 promoter, containing different predicted STAT3-binding sites. The primers were listed in Supplementary Table S2.

Dual luciferase reporter assays

Dual luciferase reporter experiments were performed in RAW264.7 cells using the Dual-Luciferase Reporter Gene Assay Kit (Beyotime, RG027) according to the manufacturer’s instructions. Five potential STAT3 binding sites (BS1 to BS5, predicted using JASPAR, https://jaspar.genereg.net/) occupied in the mouse Nlrp3 promoter, and the mutant-form of two potential STAT3 binding sites (BS3M and BS4M) were respectively inserted into the dual luciferase reporter vectors (pGL4.17[luc2/Neo]). Empty vector (EV), wild-type (BS1 to BS5), or mutant-form (BS3M and BS4M) dual luciferase reporter plasmid were transfected into RAW264.7 cells using Lipo 8000 (Beyotime, C0533). 48 h after transfection, cells were treated with LPS (500 ng ml−1, 3 h) followed by Nigericin (20 µM, 10 min) and then lysed to measure the luciferase activity of firefly and renilla. The sequences of binding sites were listed in Figs. 6D, E and  S4A.

Statistical analysis

Data was analyzed using the SPSS 26.0 and GraphPad Prism 9.4 software. The Kaplan–Meier method and a log-rank test were used to compare the differences in survival rates between groups. Normally distributed data were expressed as mean ± standard deviation (SD) and compared using the ANOVA LSD or t-test. Non-normally distributed variables were presented as median (interquartile range, IQR) and compared using the Mann–Whitney U-test. Each experiment was performed at least three times. P < 0.05 was considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.