MLKL promotes hepatocarcinogenesis through inhibition of AMPK-mediated autophagy

Abstract

The pseudokinase mixed lineage kinase domain-like (MLKL) is an essential component of the activation of the necroptotic pathway. Emerging evidence suggests that MLKL plays a key role in liver disease. However, how MLKL contributes to hepatocarcinogenesis has not been fully elucidated. Herein, we report that MLKL is upregulated in a diethylnitrosamine (DEN)-induced murine HCC model and is associated with human hepatocellular carcinomas. Hepatocyte-specific MLKL knockout suppresses the progression of hepatocarcinogenesis. Conversely, MLKL overexpression aggravates the initiation and progression of DEN-induced HCC. Mechanistic study reveals that deletion of MLKL significantly increases the activation of autophagy, thereby protecting against hepatocarcinogenesis. MLKL directly interacts with AMPKα1 and inhibits its activity independent of its necroptotic function. Mechanistically, MLKL serves as a bridging molecule between AMPKα1 and protein phosphatase 1B (PPM1B), thus enhancing the dephosphorylation of AMPKα1. Consistently, MLKL expression correlates negatively with AMPKα1 phosphorylation in HCC patients. Taken together, our findings highlight MLKL as a novel AMPK gatekeeper that plays key roles in inhibiting autophagy and driving hepatocarcinogenesis, suggesting that the MLKL-AMPKα1 axis is a potential therapeutic target for HCC.

Introduction

Hepatocellular carcinoma (HCC) is a common cause of cancer mortality worldwide, particularly in Asia and Africa [1]. The poor prognosis of HCC patients is due to recurrence, metastasis and lack of treatment, especially when patients are diagnosed at advanced stages. The molecular pathogenesis of HCC is related to chronic inflammation, viral infection, toxic exposure, and metabolic factors [2]. However, the regulatory mechanisms involved in HCC development and progression are poorly understood. Therefore, novel therapeutic targets and strategies for the prognosis prediction and clinical individualized treatment of HCC patients need to be identified.

Autophagy is a conserved eukaryotic process in which damaged intracellular components and protein aggregates are cleared to maintain body homeostasis. Dysregulation of autophagy is tightly associated with liver diseases, including liver injury, fatty liver disease and HCC [3, 4]. Moreover, enhancing autophagy via genetic modification or pharmacological treatment alleviates HCC progression [5, 6]. Thus, modulating autophagy has gained increased attention as a therapeutic strategy for HCC. Several critical upstream molecules, such as AMPK and mTOR, have been shown to regulate autophagy [7]. In particular, AMPK regulates autophagy through phosphorylating ULK1 and Raptor, leading to increased expression of autophagy-related genes [8]. In addition, low AMPK activity has been shown to correlate with poorer prognosis and drug resistance in HCC [9]. Interestingly, compounds that activate AMPK may produce good effects against HCC. However, the molecular mechanisms underlying the activation of AMPK have not yet been elucidated.

Mixed lineage kinase domain-like protein (MLKL) is widely known to be a key regulator and effector of necroptosis that leads to oligomerization in the plasma membrane and subsequent cell death [10]. We and others have previously shown that MLKL plays an important role in development, homeostasis, inflammation, injury and cancer [11–15]. Increasing evidence indicates that MLKL is associated with liver inflammation and injury. MLKL ablation alleviates CCl₄-induced liver fibrosis [16]. MLKL deficiency prevents hepatic insulin resistance and glucose intolerance [17]. Knockout of MLKL ameliorates western diet-induced liver injury [18]. In addition, MLKL plays important roles in modulating parthanatos and the immune response in HCC [19]. However, despite these efforts, the understanding of the mechanism underlying hepatic MLKL in hepatocarcinogenesis is limited.

In this study, we showed that MLKL in hepatocytes plays a critical role in HCC progression. Mechanistic investigations indicated that MLKL suppresses autophagy by blocking AMPK activation. Interestingly, we demonstrated that MLKL serves as a bridging factor between AMPKα1 and the phosphatase PPM1B, which leads to AMPKα1 dephosphorylation. Additionally, we observed that the expression of MLKL and AMPKα1 phosphorylation were inversely correlated in HCC samples. These findings establish that MLKL serves as a precise therapeutic target for HCC by regulating the AMPK/autophagy axis.

Results

MLKL is overexpressed in DEN-induced murine HCC model and is associated with poor prognosis in HCC

To investigate the genetic alterations in HCC, we established a mouse model of HCC by injecting DEN (Supplementary Fig. 1a). At 10 months, the mice exhibited increased liver/body weight ratios and accelerated tumor formation (Fig. 1a, b, Supplementary Fig. 1b). To gain insight into the molecular aspects of hepatocellular carcinoma, we performed a 4-dimensional label-free (4D label-free) quantitative proteomics method via mass spectrometry (Fig. 1c). In total, 4859 proteins were quantified in the mouse livers of the DEN group compared to those of the control group, of which 461 were significantly differentially expressed. Notably, the expression of 162 proteins increased, while that of 299 proteins decreased (Supplementary Fig. 1c, d). Intriguingly, volcano plot analysis revealed that MLKL was upregulated in the livers of the DEN-induced mice (Fig. 1d). To further verify these findings, we performed immunohistochemistry (IHC) staining, and compared with those in the control group, the tumor regions in the DEN group exhibited substantially increased MLKL levels in the mouse liver (Fig. 1e). Western blotting analysis also revealed that MLKL upregulation was concomitant with YAP activation (Fig. 1f, Supplementary Fig. 1e).

Fig. 1. MLKL is overexpressed in DEN-induced murine HCC model and is associated with poor prognosis in HCC.

a Representative images of livers from the control and DEN groups. b Representative histopathology image of livers from the control and DEN groups (scale bars, 20 μm). c Schematic diagram of the 4-dimensional label-free quantitative proteomic analysis of mouse liver tissues (n = 3 per group). d Volcano plot (DEN vs. control) of differentially expressed proteins; upregulated (red) and downregulated (blue) proteins are labeled. e Representative images of liver sections from the control and DEN groups of mice stained for MLKL, AFP, and YAP (scale bars, 20 μm). f Western blotting analysis of MLKL and YAP expression in control and DEN-induced liver tissues (n = 3 per group). g Levels of MLKL proteins assessed in HCC tumor (n = 91) and adjacent tissues (n = 82). Statistical significance was determined by Mann-Whitney’s U test. h Expression of low and high MLKL was assessed in HCC tumors and adjacent tissues. Statistical significance was determined by chi-square test. i Kaplan–Meier analysis of overall survival in HCC patients with low or high MLKL expression. Statistical significance was determined by log-rank test. Data is showed as the mean ± SD.

To substantiate this observation, we analyzed MLKL expression in human HCC tissues and adjacent tissues. We verified that the MLKL level was significantly greater in HCC tumors than in adjacent tissues (Fig. 1g). As summarized in Fig. 1h, 57.14% of HCC tumors exhibited higher expression of MLKL, whereas only 23.17% of the adjacent tissues exhibited this pattern (Fig. 1h). We also performed IHC staining of a tumor tissue microarray. Kaplan-Meier (KM) analysis revealed that HCC patients with higher MLKL expression exhibited shorter overall survival than those with low MLKL expression (Fig. 1i). These results highlight the clinical importance of MLKL in HCC and suggest a causal role for MLKL in hepatocarcinogenesis.

MLKL in hepatocytes promotes hepatocarcinogenesis

To investigate the function of MLKL in hepatocarcinogenesis, we evaluated DEN-induced HCC in MLKL global knockout (Mlkl^−/−) mice (Fig. 2a, Supplementary Fig. 2a). Mlkl^−/− mice had significantly lower tumor/body weight ratios, fewer tumors and smaller tumor sizes than WT mice (Fig. 2b–e). A incidence of tumors larger than 5 mm was also dramatically lower in Mlkl^−/− mice (Fig. 2f). Furthermore, we detected fewer Ki67-positive cells and lower AFP and YAP levels in Mlkl^−/− tumors (Fig. 2g, Supplementary Fig. 2b). These results suggest that MLKL plays an important role in promoting carcinogen-induced HCC in mice.

Fig. 2. MLKL in hepatocytes promotes hepatocarcinogenesis.

a Design and establishment of a DEN-induced HCC mouse model of global MLKL deficiency (Mlkl^−/−). b Gross morphology of the livers of WT and Mlkl^−/− mice 10 months after DEN treatment. c–e Statistical analysis of the liver-to-body weight ratio (c), number of tumors (d) and maximum tumor volume (e) in the livers of the WT and Mlkl^−/− mice at 10 months after DEN treatment (n = 6 per group). Statistical significance was determined by two-tailed unpaired Student’s t test. f Statistical analysis of the incidence of tumors >5 mm in length in WT and Mlkl^−/− mice 10 months after DEN treatment (n = 6 per group). g Representative images of liver sections from WT and Mlkl^−/− mice (n = 6 per group) stained for AFP, YAP, and Ki67 (scale bars, 20 μm). h Gross morphology of Mlkl^F/F, Mlkl^ΔHep and Mlkl^ΔMye livers 10 months after DEN treatment. i–k Statistical analysis of the liver-to-body weight ratio (i), number of tumors (j) and maximum tumor volume (k) in the Mlkl^F/F, Mlkl^ΔHep and Mlkl^ΔMye livers at 10 months after DEN treatment (n = 6 per group). Statistical significance was determined by Mann–Whitney’s U test (i, k) and one-way ANOVA test (j). l Representative images of liver sections from Mlkl^F/F, Mlkl^ΔHep and Mlkl^ΔMye mice (n = 6 per group) stained for AFP, YAP, and Ki67 at 10 months after DEN treatment (scale bars, 20 μm). m Gross morphology of the liver in Mlkl^F/F + AAV8 control (black), Mlkl^ΔHep + AAV8 control (brown), and Mlkl^ΔHep + MLKL AAV8 (blue) mice after DEN treatment. n–p Statistical analysis of the liver-to-body weight ratio (n), number of tumors (o) and maximum tumor volume (p) in the indicated groups of mice (n = 6 per group) after DEN treatment. Statistical significance was determined by Mann–Whitney’s U test (n, p) and one-way ANOVA test (o). Data is showed as the mean ± SD.

To determine which cell types are critical for the inhibition of hepatocarcinogenesis by MLKL ablation, we generated conditional knockouts of MLKL in hepatocytes (Mlkl^F/F albumin-Cre, Mlkl^ΔHep) and myeloid cells (Mlkl^F/F LysM-Cre, Mlkl^ΔMye). As revealed by PCR and western blotting analysis, MLKL was specifically ablated in hepatocytes and myeloid cells, respectively (Supplementary Fig. 2c, d). At 10 months, Mlkl^ΔHep mice exhibited significantly fewer tumors and smaller tumor sizes than did Mlkl^F/F mice; however, the number and size of tumors from Mlkl^ΔMye and Mlkl^F/F mice were similar (Fig. 2h–k, Supplementary Fig. 2e). Compared to those in Mlkl^F/F mice, the Ki67, AFP and YAP levels were also markedly lower in the tumors of Mlkl^ΔHep mice but not in those of Mlkl^ΔMye mice (Fig. 2l, Supplementary Fig. 2f). These data demonstrated that MLKL in hepatocytes has a distinct role in DEN-induced hepatocarcinogenesis.

To further examine whether re-expression of MLKL in the livers of Mlkl^ΔHep mice rescues this phenotype, we evaluated DEN-induced hepatocarcinogenesis in Mlkl^ΔHep mice with or without MLKL reintroduction by injecting AAV8-TBG-MLKL virus or control virus (Supplementary Fig. 2g, h). Notably, AAV-mediated reintroduction of MLKL increased the number of tumors and reversed the phenotypes caused by MLKL ablation (Fig. 2m–p, Supplementary Fig. 2i). As expected, compared with those in control mice, the Ki-67-, AFP- and YAP-positive areas in Mlkl^ΔHep mice were dramatically greater than those in control mice (Supplementary Fig. 2j, k), which confirmed the promotive effect of MLKL on HCC. Taken together, these findings suggest that hepatic MLKL plays a tumor-promoting role in hepatocellular carcinoma progression.

MLKL potentiates hepatocarcinogenesis through inhibiting autophagy

To determine the mechanism through which MLKL influences hepatocarcinogenesis, we isolated primary hepatocytes from Mlkl^F/F and Mlkl^ΔHep tumors and further performed gene expression analysis via RNA-seq. The expression levels of autophagy-regulatory genes were significantly increased in Mlkl^ΔHep tumor cells (Fig. 3a, Supplementary Fig. 3a), suggesting that autophagy may be induced by MLKL ablation; thus, we investigate whether MLKL regulates the autophagy process during hepatocarcinogenesis.

Fig. 3. MLKL regulates autophagy in hepatocytes to control hepatocarcinogenesis.

a Heatmap summarizing the differential gene expression of RNA-seq data related to the autophagy process (n = 4 per group). b Western blot analysis of LC-3, phosphorylated ULK1 (p-ULK1) and total ULK1 expression in the livers of Mlkl^F/F and Mlkl^ΔHep mice with DEN-induced HCC (n = 4 per group). c Western blotting analysis of LC-3, p-ULK1, and ULK1 expression in liver tumors from Mlkl^F/F + AAV8 control, Mlkl^ΔHep + AAV8 control, and Mlkl^ΔHep + MLKL AAV8 mice in DEN-induced HCC models (n = 4 per group). d Western blotting analysis of LC-3, p-ULK1, and ULK1 expression in MLKL-silenced and control HepG2 and Huh7 cells. e Western blotting analysis of LC-3, p-ULK1, and ULK1 expression in MLKL-KD HepG2 and Huh7 cells transfected with FLAG-MLKL. f GFP-LC3 fluorescence assay in MLKL-silenced and control HepG2 and Huh7 cells. Statistical significance was determined by two-tailed unpaired Student’s t test. g Autophagic vacuoles were measured in MLKL-silenced and control HepG2 cells by transmission electron microscopy. Statistical significance was determined by Mann–Whitney’s U test. h Gross morphology of the liver in Mlkl^F/F + solvent mice (black), Mlkl^ΔHep+ solvent mice (brown), Mlkl^F/F + CQ mice (orange), and Mlkl^ΔHep + CQ mice (blue) after DEN treatment. i-k Statistical analysis of the liver-to-body weight ratio (i), number of tumors (j) and maximum tumor volume (k) in the indicated groups of mice after DEN treatment (n = 6 per group). Statistical significance was determined by one-way ANOVA test (i, j, k). l Representative images of liver sections of AFP, YAP, and Ki67-stained Mlkl^F/F + solvent mice, Mlkl^ΔHep+ solvent mice, Mlkl^F/F + CQ mice, and Mlkl^ΔHep + CQ mice (scale bars, 20 μm) after DEN treatment (n = 6 per group). Data is showed as the mean ± SD.

First, gene dysregulation was validated in liver tumors from Mlkl^F/F and Mlkl^ΔHep mice. The levels of the autophagy pathway regulators LC3 II and ULK1 phosphorylation were significantly greater in liver tumors from Mlkl^ΔHep mice than in those from Mlkl^F/F mice (Fig. 3b, Supplementary Fig. 3b). Importantly, re-expression of MLKL rescued the increase in LC3 II and p-ULK1 in tumor tissues mediated by MLKL ablation (Fig. 3c, Supplementary Fig. 3c). We next investigated whether MLKL regulates autophagy in hepatocellular carcinoma cells. In parallel, knockdown of MLKL (MLKL KD) in HepG2 and Huh7 hepatocellular carcinoma cells markedly induced the conversion of LC3 I to LC3 II and increased the level of p-ULK1 (Fig. 3d, Supplementary Fig. 3d). Conversely, downregulation of LC3 II and p-ULK1 was observed in MLKL-overexpressing cells (Supplementary Fig. 3e). Moreover, re-expression of MLKL in MLKL-KD HepG2 and Huh7 cells restored the protein levels of LC3 II and p-ULK1 to those of the control cells (Fig. 3e, Supplementary Fig. 3f). MLKL KD caused a significant increase in the number of GFP-LC3 puncta (Fig. 3f). In addition, more autophagosomes were observed in MLKL KD cells (Fig. 3g). Consistent with these findings, re-expression of MLKL markedly abrogated the accumulation of GFP-LC3 puncta in MLKL KD cells (Supplementary Fig. 3g). Collectively, these results demonstrate that MLKL regulates autophagy in liver tumor cells.

We examined whether the loss of hepatic MLKL suppressed hepatocarcinogenesis through autophagy induction using the autophagy inhibitor chloroquine (CQ) (Supplementary Fig. 3h). As expected, CQ treatment in Mlkl^ΔHep mice (Mlkl^ΔHep + CQ) significantly increased tumor numbers and tumor burdens compared to the control (Mlkl^ΔHep + solvent) (Fig. 3h–k). Consistently, the levels of Ki-67, AFP and YAP were dramatically increased in Mlkl^ΔHep mice after CQ treatment (Fig. 3l, Supplementary Fig. 3i). Taken together, these data demonstrate that MLKL deficiency promotes autophagy and thereby suppresses hepatocarcinogenesis.

MLKL directly interacts with AMPKα and negatively regulates its activity

To explore the mechanism underlying MLKL-regulated autophagy, we expressed Flag-tagged MLKL in HEK293T cells and then purified MLKL-bound protein complexes. Potential MLKL-interacting proteins were further identified by LC‒MS/MS analysis. The results indicated that AMPKα1 is a potential MLKL-interacting protein (Fig. 4a, b, Supplementary Fig. 4a, b). We thus validated the interaction between MLKL and AMPKα1 by coexpressing MLKL and AMPKα1 in HEK293T cells. Co-IP analysis indicated that MLKL could interact with AMPKα1 (Fig. 4c), and vice versa (Fig. 4d). We also determined that MLKL and AMPKα1 could interact with each antibody in primary hepatocytes (Fig. 4e). The GST pull-down assay further confirmed the direct interaction between MLKL and AMPKα1 (Fig. 4f). Additionally, confocal microscopy analysis revealed that MLKL and AMPKα1 could be colocalized (Supplementary Fig. 4c). Taken together, these data indicate that MLKL physically interacts with AMPKα1.

Fig. 4. MLKL directly interacts with AMPKα and negatively regulates its activity.

a MLKL-overexpressing cells were lysed and purified with anti-FLAG affinity beads. The FLAG-MLKL-associated proteins were analyzed by silver staining. b Mass spectrometry analysis of MLKL-associated proteins indicated that AMPKα1 is a candidate protein that interacts with MLKL. The peptide coverage of MLKL and AMPKα1 is shown. c Western blotting analysis of immunoprecipitates from HEK293T cells transfected with HA-MLKL or Flag-AMPKα1 plasmids. d Western blotting analysis of immunoprecipitates from HEK293T cells transfected with FLAG-MLKL or HA-AMPKα1 plasmids. e Western blotting analysis of lysates and immunoprecipitates from HepG2 cells. f GST, GST-AMPKα, and His-MLKL were produced from bacteria, and western blotting of GST was used to evaluate the interaction between His-tagged MLKL and GST-tagged AMPKα1. g Schematic diagram of fragments of AMPKα1. h Western blotting analysis of immunoprecipitates from HEK293T cells transfected with the indicated plasmids. i Schematic diagram of fragments of MLKL. j Western blot analysis of immunoprecipitates from HEK293T cells transfected with the indicated plasmids. k Graphical representation showing the predicted binding sites of the MLKL protein with AMPKα1, including residues Val385/Phe386/Gln388 (mut 1), Lys331/Arg333/Arg365 (mut 2), Leu377/Glu381 (mut 3), and Ile242/Gln245 (mut 4). l Western blot analysis of lysates and immunoprecipitates from HepG2 cells transfected with the indicated plasmids. m Western blotting analysis of immunoprecipitates from HEK293T cells transfected with the indicated plasmids. n Western blot analysis of T172-phosphorylated AMPKα1 (p-AMPKα1) and total AMPKα1 expression in MLKL-silenced and control HepG2 and Huh7 cells. o Western blotting analysis of p-AMPKα1 and AMPKα1 expression in MLKL-silenced HepG2 and Huh7 cells transfected with FLAG-MLKL. p Western blotting of p-AMPKα1 and AMPKα1 in the livers of Mlkl^F/F and Mlkl^ΔHep model mice (n = 4 per group) with DEN-induced HCC. q Western blot analysis of p-AMPKα1 and AMPKα1 expression in liver tumors from DEN-induced HCC model mice treated with Mlkl^F/F + AAV8 control, Mlkl^ΔHep + AAV8 control, or Mlkl^ΔHep + MLKL AAV8 (n = 4 per group). r Western blotting analysis of p-AMPKα1 and AMPKα1 expression in MLKL-KD HepG2 cells transfected with full-length MLKL or R333A MLKL.

We further investigated the binding between MLKL and AMPKα1, and recombinant fragments were constructed. To map the region at which MLKL binds to AMPKα1, a series of AMPKα1 fragments were generated in HEK293T cells (Fig. 4g). Co-IP assays revealed that the N-terminal region (amino acids 1–312) of AMPKα1 was critical for the interaction between MLKL and AMPKα1 (Fig. 4h). Conversely, mapping the MLKL fragments required for AMPKα1 binding demonstrated that the C-terminus (amino acids 179–471) was responsible for the interaction with AMPKα1 (Fig. 4i, j), suggesting that the C-terminus of MLKL is directly associated with the N-terminus of AMPKα1. Based on the structural features of MLKL and AMPKα1, ten key amino acid residues could be important for the interaction of MLKL with AMPKα1 (Fig. 4k, Supplementary Fig. 4d). We next mutated these amino acid residues of MLKL (mut-1: Val385/Phe386/Gln388, mut-2: Lys331/Arg333/Arg365, mut-3: Leu377/Glu381, mut-4: Ile242/Gln245) to alanine to investigate the critical mutations involved in the interaction with AMPKα1. Among these mutant groups, MLKL mut-2 markedly attenuated the MLKL-AMPKα1 interaction, whereas the other mutations had no significant influence on the interaction (Fig. 4l, Supplementary Fig. 4e). These results suggest that the Lys331/Arg333/Arg365 amino acids are likely to be critical sites of the AMPKα1 interaction. We further showed that R333A strongly attenuated MLKL and AMPKα1 binding, whereas K331A and R365A did not disrupt the MLKL-AMPKα1 interaction (Fig. 4m, Supplementary Fig. 4f). Taken together, these results indicate that the R333 residue of MLKL is critical for the interaction of MLKL with AMPKα1.

Having demonstrated that MLKL interacts with AMPKα1, we next investigated whether MLKL modulates AMPKα1 activity by examining AMPKα1 phosphorylation (Fig. 4n, Supplementary Fig. 4g). The expression of p-AMPKα1 was also increased in MLKL KD HepG2 and Huh7 cells. In contrast, overexpression of MLKL significantly reduced the level of p-AMPKα1 (Supplementary Fig. 4h). Similarly, reintroduction of MLKL decreased the level of p-AMPKα1 in MLKL KD cells (Fig. 4o, Supplementary Fig. 4i). To further determine the influence of MLKL on AMPKα1 activity in vivo, we examined the activation of AMPKα1 in Mlkl^ΔHep tumors. The level of p-AMPKα was greater in Mlkl^ΔHep tumors than in their counterparts (Fig. 4p, Supplementary Fig. 4j). Importantly, re-expression of MLKL in Mlkl^ΔHep mice significantly restored the level of p-AMPKα1 in liver tumors (Fig. 4q, Supplementary Fig. 4k). These results suggest that MLKL serves as a universal negative regulator of AMPKα1 activity. To clarify whether MLKL modulates AMPKα1 activity via the MLKL-AMPKα1 interaction, we overexpressed WT or R333A MLKL in MLKL KD cells. We showed that WT but not R333A MLKL inhibited AMPKα1 activity (Fig. 4r, Supplementary Fig. 4l). Taken together, these results demonstrate that the MLKL-AMPKα1 interaction is required for the inhibition of AMPKα1 activation by MLKL.

Given that MLKL plays an important role in necroptosis, we next tested whether MLKL regulates AMPK activity and autophagy through necroptosis. Because the T357/S358 phosphorylation site on MLKL is essential for canonical necroptotic cell death [20], we ectopically expressed WT or T357A/S358A MLKL and analyzed the interaction between MLKL and AMPKα1. T357A/S358A MLKL had no significant influence on the interaction between MLKL and AMPKα1 (Supplementary Fig. 4m). Moreover, MLKL-KD HepG2 cells were transfected with WT or T357A/S358A MLKL. The results revealed that both WT and T357A/S358A MLKL inhibited AMPK activity and autophagy (Supplementary Fig. 4n). Thus, these data indicate that the necroptotic-independent function of MLKL negatively regulates AMPK activity and autophagy.

MLKL negatively regulates autophagy and promotes hepatocarcinogenesis through AMPK

To address whether the inhibition of AMPKα1 activity in MLKL indeed caused defects in autophagy, we examined autophagy in MLKL KD cells treated with or without the AMPK inhibitor Compound C (C.C.). The results showed that the MLKL KD-induced increase in LC3 II and p-ULK1 expression and increase in GFP-LC3 puncta formation were attenuated by C.C. (Fig. 5a, b, Supplementary Fig. 5a, b), suggesting that MLKL deficiency promotes autophagy induction by enhancing AMPKα1 activity.

Fig. 5. MLKL negatively regulates autophagy and promotes hepatocarcinogenesis through AMPK.

a Western blotting analysis of p-AMPKα1, AMPKα1, p-ULK1, ULK1 and LC3 expression in MLKL-silenced and control HepG2 and Huh7 cells treated with and without Compound C (C.C.). b Confocal fluorescence assay of GFP-LC3 puncta in MLKL-silenced and control HepG2 cells with and without C.C. Statistical significance was determined by one-way ANOVA test. c Gross morphology of the liver in Mlkl^F/F + solvent mice (black), Mlkl^ΔHep+ solvent mice (brown), and Mlkl^F/F + C. C mice (green); Mlkl^ΔHep + C. C mice (purple) after DEN treatment. d–f Statistical analysis of the liver-to-body weight ratio (d), number of tumors (e) and maximum tumor volume (f) in the four groups of mice (n = 6 per group) after DEN treatment. Statistical significance was determined by one-way ANOVA test (d, f) and Mann–Whitney’s U test (e). g Representative images of liver sections showing AFP, YAP, and Ki67 staining in Mlkl^F/F + solvent mice, Mlkl^ΔHep+ solvent mice, and Mlkl^F/F + C. C mice and Mlkl^ΔHep + C. C mice (n = 6 per group) after DEN treatment. (scale bars, 20 μm). h Western blotting analysis of p-AMPKα1, AMPKα1, p-ULK1, ULK1 and LC3 expression in liver tumors from the indicated groups of mice (n = 6 per group) in the DEN-induced HCC model. Data is showed as the mean ± SD.

To support our hypothesis that MLKL deficiency suppresses hepatocarcinogenesis by activating AMPKα1 to induce autophagy, we examined tumor formation through pharmacological inhibition of AMPKα1 in Mlkl^F/F and Mlkl^ΔHep mice. C.C. treatment significantly increased the number and size of tumors in Mlkl^ΔHep mice (Fig. 5c–f, Supplementary Fig. 5c). Consistently, C.C. dramatically increased Ki-67, AFP and YAP expression in Mlkl^ΔHep mice (Fig. 5g, Supplementary Fig. 5d). Notably, C.C. also suppressed p-AMPKα1, p-ULK1 and LC3 II in Mlkl^ΔHep mice (Fig. 5h, Supplementary Fig. 5e). Together, these results suggest that MLKL deficiency-induced AMPKα1 activation is required to induce autophagy and subsequently suppress hepatocarcinogenesis.

MLKL suppresses AMPKα1 activation by promoting PPM1B-dependent dephosphorylation

We next investigated the mechanism through which MLKL possibly influences AMPKα1 activation. Liver kinase B1 (LKB1) and calcium-calmodulin kinase kinase-2 (CaMKK2) are known as the predominant AMPKα1 upstream kinases [21, 22]. To determine whether upstream kinase availability contributes to the effects of MLKL on AMPKα1 activation, we knocked down LKB1 or CaMKK2. However, it failed to influence AMPKα1 activity in MLKL KD cells (Supplementary Fig. 6a, b). The ATM kinase was also identified as the critical upstream kinase that mediates AMPK activation by phosphorylating AMPKα1 [23]. We also used the ATM inhibitor KU-55933 and observed that it did not alter AMPKα1 activation in MLKL KD cells (Supplementary Fig. 6c). Taken together, these results indicate that MLKL suppresses AMPKα1 activation independent of upstream kinases.

Dephosphorylation of AMPK plays an important role in AMPK activity [24]. We therefore investigated whether MLKL suppresses AMPKα1 activity by influencing AMPKα1 dephosphorylation. HepG2 cells without or with ectopic MLKL overexpression were incubated in medium without glucose to induce AMPKα1 activation, after which the medium was changed to glucose. Indeed, compared with the control, MLKL overexpression significantly impaired the level of p-AMPKα1 (Fig. 6a, b), suggesting that MLKL is able to promote AMPKα1 dephosphorylation. The AMPKα1 phosphorylation level is regulated by several phosphatases, such as PP1, PP2A, PP2C (PPM1A, PPM1B), PPM1D, and PPM1E. Mass spectrometry revealed that PPM1B was an MLKL-associated protein (Supplementary Fig. 6d). Thus, we validated the interaction between MLKL and phosphatases by co-IP and found that PPM1B could interact with MLKL (Fig. 6c), and vice versa (Fig. 6d). However, MLKL did not interact with other phosphatases (Supplementary Fig. 6e, f). Furthermore, an interaction between endogenous MLKL and PPM1B was also found in HepG2 cells (Fig. 6e).

Fig. 6. MLKL suppresses AMPKα1 activation by promoting PPM1B-dependent dephosphorylation.

a MLKL-overexpressing and control HepG2 cells were cultured in nonglucose medium to induce AMPKα1 activation for 6 h. Then, the cells were cultured in medium supplemented with glucose for different durations. Western blotting analysis of p-AMPKα1 was performed. b The level of p-AMPKα1 was quantitated after three independent experiments in Fig. 6a. Statistical significance was determined by two-way repeated-measures ANCOVA. c, d Western blotting analysis of immunoprecipitates from HEK293T cells transfected with the MLKL and PPM1B plasmids. e Western blotting analysis of lysates and immunoprecipitates from HepG2 cells. f Western blotting analysis of MLKL-silenced and control HepG2 cells transfected with PPM1B siRNA or PPM1E siRNA. g Western blotting analysis of MLKL-overexpressing and control HepG2 cells transfected with PPM1B siRNA or PPM1E siRNA. h Western blotting analysis of MLKL-silenced and control HepG2 cells transfected with PPM1B siRNA. i Confocal fluorescence assay of GFP-LC3 puncta from MLKL-silenced and control HepG2 cells transfected with PPM1B siRNA. Statistical significance was determined by one-way ANOVA test. j Gross morphology of the liver in Mlkl^F/F + shPPM1B and Mlkl^ΔHep+ shPPM1B mice in the DEN-induced HCC mouse model at 10 months. k–m Statistical analysis of the liver-to-body weight ratio (k), number of tumors (l) and maximum tumor volume (m) in the Mlkl^F/F + shPPM1B and Mlkl^ΔHep+ shPPM1B mice after DEN treatment (n = 6 per group). Statistical significance was determined by two-tailed unpaired Student’s t test. n Western blotting analysis of lysates and immunoprecipitates from HepG2 cells. o–p Western blot analysis of lysates and immunoprecipitates from MLKL-overexpressing (o) and MLKL-knockdown (p) HepG2 cells. q Western blot analysis of lysates and immunoprecipitates of Mlkl^F/F and Mlkl^ΔHep tumors. r Working model showing that MLKL acts as a bridging factor between AMPKα1 and the phosphatase PPM1B to suppress AMPKα1 activity. Data is showed as the mean ± SD. ns indicates not significant.

To determine whether MLKL regulates AMPKα1 activity through PPM1B, PPM1B and PPM1E expression was knocked down via siRNA transfection. PPM1B or PPM1E knockdown significantly increased AMPKα1 activation; however, knocking down PPM1B but not PPM1E abolished MLKL KD-induced AMPKα1 activation (Fig. 6f, Supplementary Fig. 6g). Moreover, PPM1B knockdown also abrogated the MLKL overexpression-induced inhibition of AMPKα1 activity (Fig. 6g, Supplementary Fig. 6h). These results suggest that PPM1B might be involved in the MLKL-mediated inhibition of AMPKα1 activity. To clarify whether PPM1B contributes to MLKL-mediated autophagy inactivation, we depleted PPM1B in MLKL KD cells. PPM1B knockdown significantly blocked the increases in LC3 II expression and GFP-LC3 puncta formation in MLKL-knockdown cells (Fig. 6h, i, Supplementary Fig. 6i, j). Moreover, we depleted mPPM1B in Mlkl^F/F mice and Mlkl^ΔHep mice via AAV8-TBG-sh PPM1B virus in DEN-induced hepatocarcinogenesis model. No difference in tumor burden was observed between Mlkl^F/F and Mlkl^ΔHep mice after PPM1B depletion (Fig. 6j–m, Supplementary Fig. 6k, l), suggesting that MLKL ablation restricts hepatocarcinogenesis via PPM1B.

We next investigated the mechanism by which MLKL influences PPM1B-mediated AMPKα1 dephosphorylation. Given that PPM1B binds to AMPKα1, which also interacts with MLKL, we hypothesized that these three proteins form a protein complex. Notably, co-IP assays indicated that MLKL, AMPKα1 and PPM1B interact with each other (Fig. 6n). Importantly, MLKL overexpression dramatically enhanced the AMPKα1-PPM1B interaction (Fig. 6o, Supplementary Fig. 6m). Conversely, the interaction between AMPKα1 and PPM1B was decreased by the knockdown of MLKL (Fig. 6p, Supplementary Fig. 6n), suggesting that MLKL serves as a bridging partner between AMPKα1 and PPM1B. To further verify this phenomenon, we examined the interaction between AMPKα1 and PPM1B in Mlkl^F/F mice and Mlkl^ΔHep tumors. Similarly, MLKL deficiency attenuated the interaction between AMPKα1 and PPM1B in vivo (Fig. 6q, Supplementary Fig. 6n).

Taken together, these results indicate that MLKL inhibits AMPKα1 activity by facilitating PPM1B-dependent dephosphorylation and that the loss of MLKL causes destabilization of the AMPKα1-PPM1B complex, leading to activation of the AMPKα1/autophagy axis to inhibit hepatocarcinogenesis (Fig. 6r).

MLKL overexpression increased DEN-induced hepatocarcinogenesis

To further confirm the effect of MLKL in hepatocarcinogenesis, we investigated tumor formation through the overexpression of MLKL by the AAV8-TBG-MLKL virus in WT mice (Supplementary Fig. 7a, b). Strikingly, compared with those in AAV8-TBG-control mice, the liver/body ratios and tumor burdens were greater in AAV8-MLKL mice (Fig. 7a–d). In agreement with these findings, the levels of AFP, YAP and Ki67 were significantly increased in the tumors of the AAV8-TBG-MLKL mice (Fig. 7e, f). In addition, MLKL overexpression effectively decreased the levels of p-AMPKα1, p-ULK1 and LC3 II (Fig. 7g, h, Supplementary Fig. 7c, d). Consistent with the notion that MLKL deficiency restricts the interaction between AMPKα1 and PPM1B, we clearly observed that the interaction between AMPKα1 and PPM1B was greater in AAV8-TBG-MLKL tumors than in control tumors (Fig. 7i, Supplementary Fig. 7e). Together, these results suggest that MLKL has a protumorigenic effect on hepatocarcinogenesis.

Fig. 7. MLKL overexpression increased DEN-induced hepatocarcinogenesis.

a Gross morphology of the liver in 8-month-old AAV8-con and AAV8-MLKL mice in the DEN-induced HCC mouse model. b–d Statistical analysis of the liver-to-body weight ratio (b), number of tumors (c) and maximum tumor volume (d) in the AAV8-con and AAV8-MLKL mice (n = 6 per group). Statistical significance was determined by two-tailed unpaired Student’s t test. e Representative images of liver sections of AFP, YAP, and Ki67-stained tissues from AAV8-con and AAV8-MLKL mice (n = 6 per group) (scale bars, 20 μm). f Quantification of AFP, YAP and Ki67 staining in the AAV8-con and AAV8-MLKL mice at 8 months. Statistical significance was determined by two-tailed unpaired Student’s t test (Ki67, YAP) and Mann-Whitney’s U test (AFP). g Western blotting of p-AMPKα1 and AMPKα1 in liver tumors from 8-month-old AAV8-con and AAV8-MLKL mice in the DEN-induced HCC mouse model. h Western blotting of p-ULK1, ULK1 and LC3 in AAV8-con and AAV8-MLKL tumors. i Western blotting analysis of immunoprecipitates from AAV8-con and AAV8-MLKL tumors. Data is showed as the mean ± SD.

High MLKL expression is related to low pAMPKα1 expression in HCC

To validate the relevance and clinical significance of MLKL/pAMPKα1 expression, we investigated the association between the expression of MLKL and p-AMPKα1 in HCC patients. IHC analysis revealed that MLKL expression was negatively associated with p-AMPKα1 levels in tumors (Fig. 8a). Notably, approximately 63.63% of the samples with higher MLKL expression presented weaker p-AMPKα1 staining, while nearly 67.24% of the samples with lower MLKL expression showed greater p-AMPKα1 staining (Fig. 8b). Moreover, patients with MLKL^highpAMPKα1^low expression had poorer prognoses than patients with MLKL^lowpAMPKα1^high expression (Fig. 8c). Taken together, these clinical results further suggest that the inverse association between MLKL and p-AMPKα1, illustrating the coexpression pattern of these two proteins, is a potential prognostic factor in HCC patients.

Fig. 8. High MLKL expression is related to low pAMPKα1 expression in HCC.

a Representative images of HCC tumor tissues stained with MLKL or p-AMPKα1. b IHC staining analysis indicating the inverse correlation between MLKL and p-AMPKα1. Statistical significance was determined by chi-square test. c KM analysis of HCC specimens with MLKL or p-AMPKα1 expression. Statistical significance was determined by log-rank test. d A schematic model showing that MLKL depletion suppresses hepatocarcinogenesis by promoting AMPK activity. MLKL stabilizes the MLKL-AMPKα1-PPM1B complex and enhances PPM1B-mediated dephosphorylation of AMPKα1 to inhibit AMPK activity; loss of MLKL causes destabilization of the AMPKα1-PPM1B complex and increases AMPK activity, leading to activation of autophagy to inhibit hepatocarcinogenesis.

Discussion

Although a significant decrease in the HCC mortality rate results from surgical resection, liver transplantation and immunotherapy, the survival outcome of HCC patients is still unsatisfactory [25, 26]. Thus, understanding the mechanisms underlying the aberrant pathogenesis of HCC is urgently needed for the development of novel therapeutic biomarkers and strategies to reduce the mortality rate. In the present study, we found upregulation of MLKL in HCC patients and demonstrated that hepatic MLKL deficiency suppressed hepatocarcinogenesis by inducing autophagy. Mechanistically, MLKL performed a new nonnecroptotic function by acting as a novel gatekeeper of AMPKα1 activity by promoting the interaction of AMPKα1 and PPM1B to dephosphorylate AMPKα1 (Fig. 8d).

MLKL is best known for its key role in necroptosis. Emerging evidence indicates that MLKL is involved in various pathological processes and plays a critical role in inflammation and tumorigenesis [27, 28]. Although necroptosis is required for inflammation and cell growth, the functions of MLKL other than necroptosis have been gradually recognized [29]. For instance, the expression of MLKL was increased in obese and diabetic patients. In addition, MLKL modulates nonalcoholic fatty liver disease and liver hepatitis. Given that diabetes, obesity, inflammation and liver disease are risk factors for HCC [30, 31], it is conceivable that MLKL may play an important role in the progression of HCC. A recent report showed that MLKL is closely associated with parthanatos and immune evasion in hepatocellular carcinoma [19]. However, the function of MLKL in hepatocarcinogenesis has largely not been determined. Here, we performed label-free quantitative mass spectrometry analysis and revealed that MLKL was upregulated in a DEN-induced HCC mouse model. Importantly, high MLKL expression is positively correlated with poor prognosis in patients with HCC. Furthermore, we demonstrated that MLKL deficiency in hepatocytes, but not in myeloid cells, suppressed hepatocarcinogenesis. Reintroduction of MLKL into Mlkl^ΔHep mice effectively abolished the tumor growth inhibition in DEN-treated Mlkl^ΔHep mice. Moreover, MLKL overexpression significantly increased DEN-induced HCC. These data indicate that MLKL is a potential biomarker for HCC and plays a causal role in the progression of hepatocarcinogenesis.

Autophagy has dual roles in the pathophysiology of cancer development. Autophagy is essential for the clearance of damaged organelles or harmful proteins. Deletion of autophagy-related genes such as BECN1, ATG5, and ATG7 accelerates tumorigenesis [32, 33]. Conversely, the induction of autophagy could overcome metabolic stress and restrict tumor growth and tumor formation [34]. Dynamic regulation of autophagy plays an important role in liver diseases. Autophagy deficiency in hepatocytes is associated with liver injury. Wu et al. reported that MLKL drove western diet-induced liver injury by regulating autophagy and demonstrated regulatory interactions between MLKL and autophagy [18]. Our results are consistent with the results of this study. Herein, we reported that MLKL deficiency in hepatocytes contributed to the accumulation of multifaceted autophagy-component genes. Moreover, our study revealed the universal role of MLKL in autophagy during hepatocarcinogenesis. Treatment with the autophagy inhibitor CQ accelerated tumor formation in mice with hepatic MLKL deficiency. These findings indicate that MLKL regulates hepatocarcinogenesis via a mechanism dependent on autophagy. Together these results suggest that MLKL functions as an important negative regulator of autophagy and plays a crucial role in the progression of liver diseases.

AMPK, a metabolic tumor suppressor, can exert its effect on multiple cellular processes, including autophagy. Previous studies have shown that AMPK initiates autophagy by regulating autophagic gene expression and activating ULK [35, 36]. Here, we report that MLKL interacts with AMPKα1 and that the R333 amino acid in MLKL is critical for its interaction with AMPKα1. We also found that MLKL negatively regulates AMPKα1 activity in liver cancer cells and in mouse liver tumor tissues. Interestingly, disrupting the interaction between MLKL and AMPKα1 did not regulate AMPKα1 activity. Notably, MLKL regulates AMPKα1 activity independent of necroptosis, demonstrating that the MLKL-AMPKα1 interaction is required for the regulation of AMPKα1 activity and acts independently of its role in necroptosis. Emerging evidence indicates that AMPK activators such as metformin, aspirin and flavones have been developed for cancer prevention and treatment [37–39]. AMPK is also associated with autophagy by regulating AMPK nuclear export and activity [40]. Our current study demonstrated that MLKL serves as an AMPK binding partner and revealed a novel mechanism through which MLKL regulates autophagy and hepatocarcinogenesis. These findings further indicated that treatment with the AMPK inhibitor Compound C antagonized the increase in the accumulation of LC3 II and GFP-LC3 puncta in MLKL KD cells and that Compound C treatment also accelerated tumor formation in hepatic MLKL deficiency mice. Taken together, these data suggest that MLKL regulates hepatocarcinogenesis through its impact on AMPK-autophagy signaling. In this respect, the detailed function of MLKL as a gatekeeper for AMPK activation needs to be further investigated.

AMPK is activated by the upstream kinases LKB1, CAMKK2 and ATM. However, we found that blocking these upstream kinases did not affect AMPKα1 activity via MLKL, suggesting that the negative regulation of AMPKα1 activity by MLKL does not depend on upstream kinases. Interestingly, we demonstrated that MLKL promotes AMPKα1 dephosphorylation, leading to the inhibition of AMPKα1 activity. Notably, it has been reported that AMPKα1 activity is affected by phosphatases, including PP1, PP2A, PP2C (PPM1A, PPM1B), PPM1D, and PPM1E [24, 41–43]. In our study, we demonstrated that MLKL interacts with PPM1B but not with other phosphatases. PPM1B is a monomeric phosphatase that belongs to the protein phosphatase Mn²⁺/Mg²⁺ (PPM) family and is localized mainly in the cytoplasm [44]. Given that MLKL is primarily localized in the cytoplasm, we speculated that MLKL suppresses AMPKα1 activity by bridging the interaction between PPM1B and AMPKα1. Indeed, among the MLKL proteins, the AMPKα and PPM1B proteins exhibited extensive interactions and could form a complex. MLKL increased the interaction between PPM1B and AMPKα1 to promote PPM1B-dependent dephosphorylation. Furthermore, PPM1B depletion abolished the ability of MLKL to regulate AMPKα1 activity and autophagy and dampened tumor growth inhibition in DEN-treated Mlkl^ΔHep mice. Taken together, these data suggest that MLKL inhibits AMPKα1 activity and consequently suppresses hepatocarcinogenesis by promoting PPM1B-dependent dephosphorylation.

MLKL expression can be regulated via transcriptional mechanisms. A published report suggested that lncRNA-FA2H-2 inhibited the transcription of MLKL [45]. Another study reported that the IFN-γ-dependent STAT1 pathway enhanced MLKL gene transcription [46]. In addition, posttranscriptional regulatory mechanisms affect the levels of MLKL. Several studies have revealed that several E3 ligases regulate MLKL ubiquitylation [47–49]. Notably, the levels of MLKL have been proposed to be regulated by autophagy [18]. Inhibition of autophagy markedly elevated MLKL levels in hepatocytes [18]. Our results demonstrated that the MLKL protein level was significantly increased in HCC tumor tissues and that MLKL in hepatocytes contributes to hepatocarcinogenesis through the inhibition of autophagy. We speculated that the feedback regulation of autophagy and MLKL in hepatocarcinogenesis leads to the accumulation of MLKL in hepatoma cells. Therefore, the mechanisms by which MLKL is upregulated in HCC need to be further investigated.

In conclusion, our study revealed that hepatic MLKL has a protumorigenic role in HCC. MLKL promoted hepatocarcinogenesis through suppressing AMPKα1-mediated autophagy. MLKL facilitated AMPKα1 dephosphorylation via the phosphatase PPM1B to inhibit AMPKα1 activity, thereby suppressing autophagy and promoting hepatocarcinogenesis. Additionally, we observed that the expression of MLKL and p-AMPKα1 was negatively correlated in HCC samples. The results of this study expand our understanding of the role of MLKL in hepatocarcinogenesis and indicate the potential of targeting MLKL-mediated AMPKα1 inactivation for HCC therapy.

Materials and methods

Reagents and antibodies

Diethylnitrosamine, chloroquine and FLAG-beads was obtained from Sigma (St. Louis, MO). Compound C and DAPI were purchased from MedChemExpress (Shanghai, China). Protein A/G agarose was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Fetal Bovine Serum (FBS) was purchased from QmSuero/Tsingmu Biotechnology, Wuhan (Wuhan, China).

Primary antibodies: MLKL (A19685), YAP (A19134) and AFP (A17898) were purchased from ABclonal (Wuhan, China). Phospho-ULK1 (5869T), ULK1 (8054T), phospho-AMPKα1 (2535S), AMPKα1 (2532S), anti-HA (3724S), FLAG (8146S) and anti-Myc (2276S) were obtained from Cell Signaling Technology (Danvers, MA, USA). MLKL (human, ab184718) and PPM1B (ab70804) were obtained from Abcam (Cambridge, MA, USA). LC 3 (L7543) and GST (G1160) were obtained from Sigma-Aldrich (St. Louis, MO). LKB1 (10746-1-AP) was obtained from Proteintech (Chicago, USA). GAPDH (GB11002) and Ki67 (GB121141) antibodies were obtained from Wuhan Servicebio Technology CO., LTD (Wuhan, China). Mice anti-goat (115-035-003) and rabbit anti-goat (111-035-003) HRP conjugated secondary antibodies were from Jackson ImmunoResearch Laboratory (West Grove, PA).

Plasmids

FLAG-tagged MLKL (FL), MLKL-T357/S358, MLKL regions, AMPKα1 (FL), AMPKα1 regions were cloned and inserted into the FLAG vector. HA-tagged MLKL, AMPKα1, PPM1B, PPP2CA were cloned and inserted into the HA vector. Myc-tagged PPP1CC was cloned and inserted into the Myc vector. HA-tagged, PPM1A, PPM1B, PPM1D, PPM1E and FLAG-MLKL plasmids were constructed by MiaoLingBio (Wuhan, China).

Clinical samples

The HCC tissue microarray (HLivH180Su17) was obtained from Shanghai Outdo Biotech in China. The study was approved by the Science and Technology Ethics Committee of Hubei University of Medicine (No. 2023-EER-05).

Animals

Mlkl^−/− mice have been described previously [13]. MLKL-flox mice were generated from GemPharmatech (Nanjing, China). Conditional MLKL knockout mice were generated by homologous recombination, and the same loxP sites were isolated on the sides of exons 3 and 5. MLKL-floxed mice (Mlkl^F/F) were generated by crossing with Cre recombinase. The Mlkl^F/F mice were crossbred with Albumin-cre mice and LysM-Cre mice (Jackson Laboratory), and hepatocyte-specific MLKL knockout mice (Mlkl^ΔHep) and myeloid-specific MLKL knockout mice (Mlkl^ΔMye) were obtained. All mice were maintained in SPF facility conditions, were age matched and were backcrossed onto the C57BL/6J background. To induce hepatocellular carcinoma, male mice were injected with DEN (25 mg/kg, i.p.) on neonatal day 14. The mice were euthanized on the indicated day after DEN treatment. Animal experiments were approved by the Animal Care and Use Committee of Hubei Medicine University (No. 2022-021).

Mouse adeno-associated virus 8 (AAV8) construction

After overexpressing MLKL in mouse livers, a mouse AAV8 was constructed and injected. Briefly, the MLKL gene was cloned and inserted into AAV-IRES-hrGFP to generate the AAV-IRES-MLKL-hrGFP plasmid, and AAV-IRES-hrGFP served as the empty vector plasmid. The AAV8 virus was generated and isolated by transfecting HEK293T cells with plasmids. AAV8 cells harboring PPM1B interference sequences (AAV8-shPPM1B) and the negative control (AAV8-shNC) were generated by GenePharma (Shanghai, China).

Primary hepatocyte isolation

Primary hepatocytes were isolated from mouse liver tissue. Briefly, liver tumors were cut into small pieces and then digested for 30 min in collagenase I and collagenase IV at 37 °C. After the digestion was terminated, the mixtures of hepatocytes and nonparenchymal cells were separated by centrifugation at 4 °C. Primary hepatocytes were seeded in 6-well plates precoated with collagen and then subjected to further analysis.

RNA-Seq analysis

Primary hepatocytes were isolated from Mlkl^F/F and Mlkl^ΔHep tumors. Each sample contained four independent repeated animals. Total RNA was extracted, and library construction, RNA sequencing and bioinformatics analysis were performed by GeneRead Biotechnology (Wuhan, China). A p value < 0.05 and a fold change ≥1.5 were considered to indicate differentially expressed genes. The raw data can be downloaded from PRJNA1089821.

Cell culture and transfection

Cell lines were obtained from Haixing Bioscience and maintained in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin (Thermo Fisher Scientific). All the cell lines were authenticated by STR profiling, and tested for mycoplasma contamination and results were negative.

Plasmids were transfected into cells using Lipofectamine 3000 (Thermo Fisher Scientific). The plasmid, Lipofectamine 3000 Reagent and P3000 Reagent were mixed with lipid complexes in Opti-MEM. After 15 min, the lipid complexes were added to the cells. siRNAs were transfected using RNAi MAX transfection reagent (Thermo Fisher Scientific). The siRNA oligonucleotides were synthesized by TSINGKE Biology Company (TSINGKE, Beijing, China).

Fluorescence microscopy

Cells were seeded on confocal dishes, fixed and then permeabilized. After blocking with 5% BSA for 1 min, the cells were incubated with primary antibodies, and the proteins were detected with conjugated secondary antibodies and DAPI in the dark. GFP-LC3-expressing HepG2 and Huh7 cells were seeded on confocal dishes, fixed with 4% formaldehyde and then incubated with DAPI. Immunofluorescence images were obtained by confocal microscopy (Olympus, FV3000RS).

Transmission electron microscopy

Cells were fixed with glutaraldehyde, incubated with 1% osmium tetroxide, and dehydrated with alcohol. After the cells were stained with uranyl acetate/lead citrate, transmission electron microscopy (TEM) images were taken with an HT7700 (Hitachi, Japan) transmission electron microscope.

Immunoprecipitation and GST pulldown

For Co-IP experiments, cells and tissue homogenates were lysed in Co-IP buffer supplemented with the protease inhibitors. Lysates were quantified and then incubated with primary antibodies overnight and Protein A/G beads. Affinity beads were collected by centrifugation and washed six times with Co-IP buffer. The subsequent analysis was performed by western blotting. For the FLAG- or HA-tagged IP experiments, the cells were lysed in Co-IP buffer. The lysates were cocultured with anti-FLAG or anti-HA affinity beads overnight. Affinity beads were harvested by centrifugation and then washed six times in cold Co-IP buffer. The subsequent analysis was performed by western blotting.

For the GST pull-down experiments, bacterially expressed GST-AMPKα1 was retained on glutathione Sepharose beads incubated with His-MLKL overnight and then washed six times with cold Co-IP buffer. The subsequent analysis was performed by western blotting.

Proteomic analysis and mass spectrometry

The liver tissues of DEN-induced HCC were separated. The proteins were extracted and subjected to 4D label-free quantitative proteomic analysis (4D-label free) technology by Shanghai Applied Protein Technology (Shanghai, China).

For affinity purification analysis of MLKL-binding proteins, MLKL-overexpressing HEK293T cells were lysed, purified with anti-FLAG affinity beads and eluted with FLAG peptides. The FLAG-MLKL-associated proteins were subjected to SDS‒PAGE and subsequent silver staining. Visible bands were excised and analyzed via mass spectrometry (Shanghai Applied Protein Technology, China).

Tissue microarray and immunohistochemistry

HCC TMAs were subjected to immunohistochemical analyses using anti-MLKL and anti-AMPKα1 antibodies. Protein expression was scored by pathologists, and the results were quantified based on a multiplicative index that included the average staining intensity and extent of staining.

Mouse liver tissues were fixed, dehydrated, and embedded. Tissue slides were deparaffinized in xylene and rehydrated in alcohol, after which antigen retrieval was performed with boiling citrate buffer by microwaving. Next, the slides were blocked in 5% BSA for 1 h and then incubated with the indicated antibodies. Protein staining was performed using a diaminobenzidine (DAB) substrate kit.

Statistics analysis

SPSS version v23.0 (IBM corporation) and Prism 8 (GraphPad Software) were used for statistical analysis. Quantitative data are presented as the mean ± standard deviation (SD) values. Statistical significance was determined by two-tailed unpaired Student’s t test, one-way ANOVA test, Mann–Whitney’s U test, log-rank test and chi-square test, as indicated in figure legends. The experiments were repeated three times independently. P values < 0.05 were considered as significant.

Author contributions

XY and QZ conceived the project and designed the research studies. QZ, XY, MF, JG, and HW performed most of the experiments described. JY performed structure analysis. AZ, and JW provided help with animal and technical assistance in the mouse experiments. YH, ZS and YL provided conceptual advice and helpful discussion. XY, QZ and MF analyzed data. XY and QZ prepared the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (81902852, 82003237), Natural Science Foundation of Hubei Province of China (2022CFB481), Natural Science Foundation of Hubei Provincial Department of Education (T2022021), Projects of International Cooperation and Exchanges (G2022027004L, G2022027012L) and the Advantages Discipline Group (Medicine) Project in Higher Education of Hubei Province (2021-2025) (2024XKQY26, 2024BMXKQY2). The authors appreciate Professor Quentin Qiang Liu and Professor Mian Wu for their helpful suggestions on this study. The authors thank the Biomedical Research Institute of Hubei University of Medicine for instrument support.

Data availability

The authors declare that the data are present in the paper and/or the supplementary information. Additional data supporting the present study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Ethical approval

The study was approved by the Science and Technology Ethics Committee of Hubei University of Medicine (No. 2023-EER-05). Animal experiments were approved by the the Animal Care and Use Committee of Hubei Medicine University (No. 2022-021) and animal care was conducted in accordance with institutional guidelines.

Supplementary information

The online version contains supplementary material available at 10.1038/s41418-024-01314-5.