FUN14 Domain-Containing 1–Mediated Mitophagy Suppresses Hepatocarcinogenesis by Inhibition of Inflammasome Activation in Mice

Mice and Liver Tumorigenesis

Mice were maintained in the Center for Experimental Animals at the Institute of Zoology, Chinese Academy of Sciences (Beijing, China). All experimental mice were of C57BL/6 genetic background and maintained in the same conditions. FUNDC1 whole-body knockout mice were generated as described.29

Albulin-Cre mice were crossed with FUNDC1^fl/fl to generate the hepatocyte-specific knockout mice (FUNDC1^Δhep). To confirm FUNDC1 deletion in hepatocytes, tail genomic DNA was extracted and genotyping was done by PCR analysis using the following primers: F: 5′-GGAACAGCTCCAGATGGCAA-3′; R: 5′-AGCATGTTTAGCTGGCCCAA-3′. FUNDC1 transgenic (TG⁺) mice were established in Peking University (Beijing, China) and bred in the Animal Center of the Institute of Zoology, Chinese Academy of Sciences. TG⁺ mice and TG^– mice were crossed to generate TG⁺ and TG^– mice. Genotype of mice was confirmed by PCR using the following primers: F: 5′-GAGTGTTTGGCCACAG TTCGG-3′; R: 5′-CCAGAAGTCAGATGCTCAAGGG-3′. Mito-keima mice were crossed with FUNDC1^fl/fl or FUNDC1^Δhep mice to generate Mito-Keima; FUNDC1^fl/fl or mito-Keima; FUNDC1^Δhep mice.

For chemical induction of HCC, 15-day-old male mice and littermates were given a single intraperitoneal injection of DEN (25 mg/kg, N0258-1G; Sigma-Aldrich, St. Louis, MO) and then fed with regular chow food. Mice were sacrificed after 8 months, and livers were removed, then imaged and fixed with paraformaldehyde.

For acute liver injury, 4-week-old FUNDC1^Δhepand FUNDC1^fl/fl male mice were given a single intraperitoneal injection of DEN (100 mg/kg, N0258-1G; Sigma-Aldrich). These mice were sacrificed at the indicated time points (24 and 48 hours). All animals were housed in a temperature- and light-controlled animal facility, and all animal experiments were performed according to protocols approved by the Animal Care and Use Committee at the Institute of Zoology, Chinese Academy of Sciences.

Statistical Analysis

All data are presented as means ± standard error of mean (SEM). Statistical comparisons were performed with a two-tailed unpaired Student t test. The survival analysis was performed using the log-rank test. For all experiments, P values <0.05 were considered as significant (not significant [N.S.], P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001).

Detailed methodology can be found in the Supporting Materials and Methods.

FUNDC1 Accumulates In Liver Tissue of Human HCC Patients

We analyzed human HCC tissue microarrays, and found that expression of FUNDC1 was significantly higher in 78% (64 of 82) of HCC samples compared to 12.5% (10 of 80) of matching peritumoral tissues (Fig. 1A). Immunoblotting analysis of HCC and peritumoral tissues from the same patients revealed that FUNDC1 expression was elevated in the tumor group compared to controls (Fig. 1B). Real-time quantitative PCR analysis showed that Fundc1 was transcriptionally up-regulated in HCC tissues (Fig. 1C). Biochemical and immunohistochemistry (IHC) analysis revealed a decrease in mitochondrial protein levels in HCC tissues compared to peritumoral tissues (Fig. 1D). mtDNA copy number, as measured by real-time PCR, was also reduced, although expression levels of the mitochondrial biogenesis factors, peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1) and nuclear respiratory factor (NRF) 1, were not significantly changed (Fig. 1E,F). It thus appeared that level of FUNDC1 was inversely correlated with mitochondrial mass, which is likely attributed to the known role of FUNDC1 in enhancing mitochondrial clearance and/or mitochondrial turnover. Three other autophagy-related proteins, Beclin1, autophagy related 5 (ATG5), and Unc-51 like autophagy activating kinase 1, were also assayed, and they showed a similar increase in HCC tissues (Supporting Fig. S1A-D).

Initiation of HCG is Promoted in FUNDC1^Δhep Mice

The above results prompted us to address the role of FUNDC1 in HCG. For this purpose, liver-specific FUNDC1 knockout mice were generated by crossing FUNDC1^fl/fl with Albumin-Cre transgenic mice. As revealed by western blotting and PCR analysis, FUNDC1 was specifically ablated in hepatocytes, but was present in nonparenchymal cells (Fig. 2A and Supporting Fig. S2A). We injected a single intraperitoneal dose of DEN (25 mg/kg) into 15-day-old male mice to induce HCC. This is a widely used HCC mouse model, which is considered analogous to human HCC. Most of the male mice developed typical HCC 8-10 months after DEN treatment. However, control mice, which were injected with saline, did not develop spontaneous tumors in liver even until 2 years (Supporting Fig. S2B). Analysis of serum samples collected from FUNDC1^fl/fl and FUNDC1^Δhep male mice 8 months after DEN treatment revealed that levels of alanine transaminase (ALT) and aspartate transaminase (AST), which are markers of hepatocellular injury or liver inflammation in most liver diseases,30 were dramatically increased in FUNDC1^Δhep mice compared to controls (Fig. 2B). The number of tumors was markedly increased in FUNDC1^Δhep mice compared to controls (Fig. 2C). However, maximal tumor size, which is an indicator of tumor development, was not significantly different between FUNDC1^fl/fl and FUNDC1^Δhep mice (Fig. 2D). These results suggest that FUNDC1 plays a critical role in the early stage of liver tumorigenesis. Furthermore, the tumor-occupied area and liver-to-body weight ratio were not significantly different in FUNDC1^Δhep mice and control littermates (Fig. 2E,F). -fetoprotein and glypican 3, markers for HCC, were positive and similar between FUNDC1-deficient tumors and tumors from FUNDC1^fl/fl mice (Supporting Fig. S2C). There was no significant difference in the level of apoptotic hepatocytes, assessed by terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end-labeling (TUNEL) assay, in FUNDC1^fl/fl and FUNDC1^Δhep 8 months after DEN treatment (Supporting Fig. S2D). Similarly, we also used a single intraperitoneal dose of DEN (25 mg/kg) to inject 15-day-old male FUNDC1 whole-body knockout mice (FUNDC1^–/–) and littermates. FUNDC1^–/– mice had a higher number of tumors than controls, whereas maximal tumor size was lower than in controls. Ki67 staining revealed a lower number of proliferating cells in tumor tissues of FUNDC1-null mice compared to controls, whereas the number of proliferating hepatocytes in the adjacent peritumoral tissues was higher than in control mice (Supporting Fig. S2E). Hence, ablation of FUNDC1 in hepatocytes increased the susceptibility to DEN-induced HCG compared to controls and resulted in a higher mortality rate (Supporting Fig. S2F).

FUNDC1 Deficiency Enhances Inflammatory Responses and LF

We next wished to investigate the underlying mechanism by which FUNDC1 ablation enhanced the propensity for HCG. We first carried out IHC with anti-F4/80 in HCC sections of FUNDC1^fl/fl and FUNDC1^Δhep mice. There was a significant increase in the number of F4/80-positive cells, which may be KCs or infiltrating macrophages in FUNDC1^Δhep tumors (Fig. 3A). This observation was further confirmed by real-time PCR analysis of F4/80 mRNA level (Fig. 3B). In HCC tissues of FUNDC1^Δhep mice, mRNA levels of genes encoding the inflammatory cytokines, tumor necrosis factor alpha (Tnf), Il6, and Il1, were significantly increased. Furthermore, the absence of FUNDC1 resulted in hyperactivation of the JAK/STAT and NF-B pathways in HCC tissues (Fig. 3C), consistent with previous findings.31, 32 Intriguingly, 1 and 2 months after DEN injection of 15-day-old male mice and before onset of cancer, the heightened JAK/STAT and NF-B signature was already detectable in FUNDC1^Δhep liver tissues compared to controls (Fig. 3D). This suggests that the inflammatory cascade has a causal role in DEN-induced HCC. Apart from chronic liver inflammation, we also used an acute liver injury model in which 4-week-old male mice were injected intraperitoneally with DEN (100 mg/kg) for 24 or 48 hours. Deletion of FUNDC1 in hepatocytes augmented the induction of F4/80 mRNA in liver tissues, and this observation was also confirmed at the protein level by IHC detection of F4/80 (Supporting Fig. S3A). mRNA levels of Tnf, Il6, and Il1 were increased in FUNDC1^Δhep liver tissue in the acute model. Consistent with production of the inflammatory cytokines, Tnf and Il6, levels of phosphorylated inhibitor of kappa B (IB) and STAT3 were more strongly increased in FUNDC1^Δhep liver tissues than in controls (Supporting Fig. S3B). When we analyzed HCC and peritumoral tissues from DEN induced 8 months of FUNDC1^–/– and WT mice, we obtained similar evidence of an enhanced inflammatory response (Supporting Fig. S3C). Taken together, these results clearly showed that FUNDC1 deficiency induced a more-severe inflammatory response in liver tissues. Moreover, FUNDC1 ablation in hepatocytes augmented fibrotic events in both tumor and peritumoral tissues from DEN-induced HCC, and this was accompanied by up-regulated expression of genes involved in fibrosis, which suggests that loss of FUNDC1 accelerated fibrosis rate (Fig. 3E). Two months after DEN administration, at an early phase in progression of HCC, degree of fibrosis and expression of fibrotic genes in FUNDC1^Δhep mice were higher than in FUNDC1^fl/fl mice (Fig. 3F). FUNDC1^–/– mice also showed higher levels of LF (Supporting Fig. S3D). Whereas the level of NRF2 was increased in tumor tissues compared to peritumoral tissues, there was no significant difference between tumor tissues harvested from FUNDC1^fl/fl (WT) and FUNDC1^Δhep (FUNDC1^–/–) mice (Supporting Fig. S3E,F).

Knockout of FUNDC1 Causes Accumulation of Dysfunctional Mitochondria

FUNDC1 was characterized as a receptor that mediates mitophagy, a process for eliminating dysfunctional mitochondria. In DEN-treated FUNDC1^fl/fl mice, FUNDC1 protein levels were higher in tumor tissues than in matching peritumoral tissues. This is consistent with HCC patient samples. Correspondingly, mitochondrial protein levels were lower in FUNDC1^fl/fl tumor tissues compared to peritumoral tissues. However, mitochondrial protein levels were higher in FUNDC1^Δhep tumor tissue than in FUNDC1^fl/fl tumor tissue, whereas PGC1 levels were similar in tumors from FUNDC1^fl/fl and FUNDC1^Δhep mice. IHC analysis of mitochondrial heat shock protein 60 (Hsp60) and real-time PCR of mtDNA further corroborated the increased mitochondrial mass in tumor tissue of FUNDC1^Δhep mice (Fig. 4A and Supporting Fig. S4A). In contrast, adenosine triphosphate (ATP) levels were reduced in FUNDC1^Δhep tumor tissues compared to tumor tissues of FUNDC1^fl/fl mice (Fig. 4B and Supporting S4B). We selected the 1- and 2-month time points in the process of DEN-induced HCG for analysis of mitochondrial markers and functions. Mitochondrial proteins and mtDNA levels were reduced in liver tissues from FUNDC1^fl/fl mice before the occurrence of HCC, and this reduction was largely blocked in liver tissues from FUNDC1^Δhep mice (Fig. 4C). It is possible that DEN-induced mitochondrial defects are among the early events in HCG. To test this notion, 4-week-old FUNDC1^fl/fl and FUNDC1^Δhep mice were injected with a single intraperitoneal dose of DEN (100 mg/kg) for 24 or 48 hours. Immunoblotting of DEN-treated, FUNDC1-depleted liver tissues revealed higher levels of mitochondrial proteins, indicative of greater accumulation of mitochondria, compared to controls (Supporting Fig. S4C). This result was also confirmed by real-time PCR to detect mtDNA and by IHC to detect Hsp60 (Supporting Fig. S4D). Mitochondrial functions, as measured by ATP production and oxygen consumption rate, were assessed in livers of FUNDC1^Δhep and FUNDC1^fl/fl mice. Indeed, there was a 20% decrease in levels of ATP and respiration in FUNDC1^Δhep liver tissues following DEN treatment (Supporting Fig. S4E). Surprisingly, loss of FUNDC1 reduced the production of ATP, even though mitochondrial protein levels were much higher in liver tissues. We further investigated whether the mitochondrial defects are attributed to impair FUNDC1-mediated mitophagy in hepatocytes. Mitochondrial marker proteins were decreased in isolated hepatocytes from FUNDC1^fl/fl mice treated with acute DEN (100 mg/kg), whereas the protein degradation was inhibited to a certain extent in isolated hepatocytes from FUNDC1^Δhep liver tissue. However, there was no difference in levels of mitochondrial proteins in isolated resident KCs from acute DEN-treated WT and FUNDC1^–/– mice (Supporting Fig. S4F). To further demonstrate that FUNDC1 mediates mitophagy in hepatocytes in response to DEN treatment, we bred mito-keima mice with FUNDC1^fl/fl and FUNDC1^Δhep mice and isolated and cultured hepatocytes from these mito-keima–labeled primary hepatocytes in the collagen-coated, glass-bottom cell-culture dish. DEN treatment significantly increased the appearance of red puncta, the marker of the mitophagy,33 in Hep WT (wild-type hepatocytes), but not in FUNDC1-depleted hepatocytes (Fig. 4D). We further compared levels of LC3 and p62, which are biochemical hallmarks of autophagy, between mitochondria isolated from hepatocytes from acute DEN-treated FUNDC1^fl/fl and FUNDC1^Δhep mice. We found that there were higher levels of LC3 and lower levels of p62 in FUNDC1^fl/fl hepatocellular mitochondria than in FUNDC1^Δhep hepatocellular mitochondria (Fig. 4E). Electron microscopy analysis confirmed that acute DEN treatment induced the appearance of mitophagosomes, and FUNDC1 ablation blocked mitophagosome formation (Fig. 4F).

Loss of FUNDC1 Aggravates Activation of Caspase-1

Our findings indicate that DEN elicits both chronic inflammatory responses and acute mitochondrial stress responses, including loss of mitochondrial functions and enhanced mitophagy. We were interested to understand how the acute mitochondrial stresses lead to chronic inflammatory response. Because mitophagy is critical for removal of dysfunctional or damaged mitochondria, which are implicated in the activation of inflammatory responses,34 we examined the effect of FUNDC1 deficiency on activation of caspase-1 in hepatocytes upon DEN treatment. FUNDC1^Δhep and FUNDC1^fl/fl mice were injected with DEN (100 mg/kg), which were sacrificed 24 or 48 hours later. FUNDC1-deficient hepatocytes had higher levels of the cleaved form of caspase-1 than Hep WT in response to DEN treatment in vivo. Additionally, mRNA levels of Nlrp3 and Il1 were increased in FUNDC1-depleted hepatocytes, and treatment of FUNDC1^Δhep mice with DEN also enhanced the level of IL1 in serum samples collected at 24 and 48 hours postinjection (Supporting Fig. S5A). To investigate the inflammatory phenotype in the early stages of HCG, we analyzed mice 1 month after DEN-treated (25 mg/kg) 15-day-old male mice. As expected, hepatocytes isolated from liver tissues of FUNDC1^Δhep mice 1 month after DEN injection showed evidence of mitochondrial accumulation (Supporting Fig. S5B). mRNA levels of Nlrp3 and Il1 in DEN-treated FUNDC1^Δhep liver tissues were also increased, which is indicative of inflammasome activation (Fig. 5A). We further assessed the cleaved form of caspase-1 in 1-month DEN treated FUNDC1^Δhep and FUNDC1^fl/fl mice. Interestingly, levels of cleavage of caspase-1 were higher in FUNDC1-deficient hepatocytes; however, caspase-1 activation was not different in both WT and FUNDC1-depleted KCs from WT and FUNDC1^–/– animals (Fig. 5B). This result is also confirmed by using the FLICA caspase-1 assay25 to detect active caspase-1 in isolated hepatocytes and KCs from DEN-treated 1-month WT and FUNDC1^–/– mice (Fig. 5C). DEN specifically activates the inflammasome, as measured by caspase-1 cleavage; similar results were confirmed in primary hepatocytes isolated from mice 2 and 5 months after DEN (25 mg/kg) treatment (Supporting Fig. S5C). However, the stimulator of interferon genes (STING) pathway was not strongly affected because the phosphorylated TANK binding kinase 1 (TBK1) and interferon regulatory factor 3 (IRF3), the downstream of STING, were not induced in either hepatocytes or KCs isolated from mice 1 month after DEN treatment (Supporting Fig. S5D). DEN treatment enhanced caspase-1/NLRP3 or caspase-1/absent in melanoma 2 (AIM2) complex formation in vivo, which was further augmented in FUNDC1-depleted hepatocytes (Fig. 5D). This is further substantiated by the more-pronounced cleavage of caspase-1 and matured IL1 in cultured primary hepatocytes isolated from FUNDC1^Δhep treated with lipopolysaccharide (LPS) and DEN (100 or 500 µg/mL) compared to FUNDC1^fl/fl mice (Fig. 5E,F). In summary, FUNDC1 ablation promoted accumulation of dysfunctional mitochondria, and mitophagy reduced, but did not completely prevent, inflammasome activation.

Previous findings have reported that IL1 released from activated hepatocytes or KCs is able to promote hepatocellular proliferation.35, 36 We found that mRNAs of Tnf and Il6 were increased in KCs following incubation with the conditioned medium from ATP-treated, LPS-primed hepatocytes. As expected, such induction is higher when KCs are incubated with medium from FUNDC1-depleted hepatocytes. On the other hand, when the conditioned medium from aforementional KCs was used to treat isolated hepatocytes, levels of phosphorylated (p)-Stat3 and p-IB were significantly enhanced, and such activation is more pronounced in FUNDC1-depleted hepatocytes (Supporting Fig. S5E). Importantly, addition of IL1 antibody into the conditioned medium blocked induction of mRNAs of Tnf and Il6 in KCs and prevented hepatocellular proliferation, as revealed by the 5-bromo-2-deoxyuridine incorporation analysis (Supporting Fig. S5F). It would be logical to assume that FUNDC1 depletion in hepatocytes results in greater production of mature IL1. It can stimulate KCs and produces more IL6 and TNF, which, in turn, up-regulates the signaling pathways in hepatocytes.

Cytosolic Mitochondrial DNA is Required For Caspase-1 Activation and IL1β Release

It has been reported that release of mtDNA and mitochondrial ROS activate the inflammasome, which, in turn, promotes the maturation of IL1β.19, 20 Therefore, we further investigated the mechanistic link between cytosolic mtDNA and inflammasome activation in hepatocytes stimulated with LPS and DEN. Indeed, treatment with DEN led to an increase in cytosolic mitochondrial DNA in LPS-primed, FUNDC1-depleted primary hepatocytes (Fig. 6A). Level of secretory mature IL1β was also increased in FUNDC1-depleted primary hepatocytes, and this increase was dramatically reduced by cyclosporine A (CsA; Fig. 6B). In LPS-primed hepatocytes, ATP-driven activation of caspase-1 is a stronger inflammasome activator than DEN treatment. Therefore, we treated primary hepatocytes from FUNDC1^Δhep and FUNDC1^fl/fl mice with LPS and ATP and assayed the cytosolic mtDNA. Level of cytosolic mtDNA was higher in ATP-treated, LPS-primed FUNDC1^Δhep primary hepatocytes than in control hepatocytes. This result was also confirmed by confocal microscopy, using antibodies detecting DNA and the mitochondrial matrix protein, Hsp60 (Fig. 6C). However, it is noteworthy that baseline expression of FUNDC1 was lower in KCs compared to that of hepatocytes. Moreover, treatment of WT and FUNDC1-depleted KCs with LPS and ATP or poly(dA:dT) led to release of statistically indistinguishable levels of mature IL1 β between both (Supporting Fig. S6A,B). Release of mtDNA is likely attributed to mitochondrial damages, given that inhibition of the mitochondrial permeability transition pore (MPTP) by CsA prevented inflammasome activation and reduced cleavage of IL1β in LPS-primed WT and FUNDC1-depleted hepatocytes stimulated with ATP or poly(dA:dT) (Fig. 6D). This finding was further confirmed with ethidium bromide (EtBr) treatment, which diminishes the level of mtDNA in hepatocytes.25 Incubation with EtBr inhibited inflammasome activation in both WT and FUNDC1-depleted hepatocytes (Fig. 6E and Supporting Fig. S6C). Transfection of an FUNDC1-expressing plasmid into FUNDC1-depleted hepatocytes prevented inflammasome activation, as indicated by reduced IL1 cleavage, whereas expression of a mitophagy-defective FUNDC1 ΔLIR mutant, which lacks the LC3-interacting region, only partially rescued activation upon treatment of LPS-primed, FUNDC1-depleted hepatocytes with ATP or poly(dA:dT) (Fig. 6F). These results showed that ablation of FUNDC1 promoted mitochondrial DNA release and activation of the inflammasome in a manner that depended on FUNDC1-mediated mitophagy.

HCG is Decreased in FUNDC1 Transgenic Mice

To further explore the role of FUNDC1 in HCC, we constructed FUNDC1 transgenic mice, which overexpressed FUNDC1 in hepatocytes (Supporting Fig. S6D). We administered a single injection of DEN to 15-day-old male mice to induce HCC. Strikingly, the number of detectable tumor nodes was lower in FUNDC1 transgenic mice than controls (Fig. 7A). Ratio of liver weight and body weight was also decreased in FUNDC1 transgenic mice (Fig. 7B). Surprisingly, mitochondrial proteins and mtDNA were not obviously decreased in HCC tissues from FUNDC1 transgenic mice. However, the mitochondrial biogenesis protein, PGC1, was maintained at a high level in FUNDC1 transgenic HCC tissues (Fig. 7C). As expected, the number of F4/80-positive cells was decreased in livers of FUNDC1 transgenic mice, and mRNA levels of Tnf and Il6 were also decreased in FUNDC1 transgenic HCCs compared to controls. Lower expression of p-Stat3 and a higher level of p-IB also confirmed that FUNDC1 overexpression in transgenic mice reduced the downstream signaling pathways, which was opposite to the effect observed in control mice (Fig. 7D). We also cultured isolated primary hepatocytes from FUNDC1 transgenic mice and littermate mice, and treated them with LPS, ATP, and poly(dA:dT) in vitro. FUNDC1 overexpression reduced the level of cleaved IL1β and inhibited inflammasome activation (Fig. 7E).