ALDH2 deficiency promotes alcohol-associated liver cancer by activating oncogenic pathways via oxidized DNA enriched extracellular vesicles

Abstract

Background & Aims:

Excessive alcohol drinking is one of the major causes of hepatocellular carcinoma (HCC). Approximately 30–40% Asian population are deficient for aldehyde dehydrogenase 2 (ALDH2), a key enzyme to detoxify the ethanol metabolite acetaldehyde. However, how ALDH2 deficiency affects alcohol-related HCC remains obscure.

Methods:

ALDH2 polymorphism in 646 patients with viral hepatitis B (HBV) infection with or without alcohol drinking was studied. A new model of HCC induced by chronic carbon tetrachloride (CCl₄) and alcohol administration was developed and studied in three lines of Aldh2-deficient mice: including Aldh2 global knockout (KO) mice, Aldh2^*1/^*2 knock-in mutant mice, and liver-specific Aldh2 KO mice.

Results:

We demonstrated that ALDH2 deficiency was not associated with liver disease progression but was associated with an increased risk of HCC development in cirrhotic HBV patients with excessive alcohol consumption. The mechanisms underlying HCC development associated with cirrhosis and alcohol consumption were studied in Aldh2-deficient mice. We found that all three lines of Aldh2-deficient mice were more susceptible to CCl₄ plus alcohol-associated liver fibrosis and HCC development. Furthermore, our results from in vivo and in vitro mechanistic studies revealed that after CCl₄ plus ethanol exposure, Aldh2-deficient hepatocytes produced a large amount of harmful oxidized mtDNA via extracellular vesicles (EVs), which were then transferred into neighboring HCC cells and together with acetaldehyde activated multiple oncogenic pathways (JNK, STAT3, BCL-2, and TAZ), thereby promoting HCC.

Conclusions:

ALDH2 deficiency is associated with an increased risk of alcohol related-HCC development from fibrosis in patients and in mice. Mechanistic studies reveal a novel mechanism that Aldh2-deficient hepatocytes promote alcohol-associated HCC by transferring harmful oxidized mtDNA-enriched EVs into HCC and subsequently activating multiple oncogenic pathways in HCC.

Lay Summary

Alcoholics with ALDH2 polymorphism have an increased risk of digestive track cancer development, however, the link between ALDH2 deficiency and HCC development has not been well established. In this study, we show that ALDH2 deficiency exacerbates alcohol-associated HCC development both in patients and mouse models. Mechanistic studies revealed that after chronic alcohol exposure, Aldh2-deficient hepatocytes produce a large amount of harmful oxidized mitochondrial DNA via extracellular vesicles, which can be delivered into neighboring HCC cells and subsequently activate multiple oncogenic pathways, thereby promoting HCC.

Graphical Abstract

Introduction

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related death worldwide. Chronic alcohol abuse is a major cause of HCC development; its metabolite, acetaldehyde is believed to play an important role in inducing HCC.¹ Mitochondrial aldehyde dehydrogenase (ALDH2) is a major enzyme for acetaldehyde elimination. The Glu487Lys polymorphism (also named rs671, with the Glutamate corresponding to *1 allele, and Lysine corresponding to *2 allele) at codon 487 in the ALDH2 gene causes the substitution of glutamate (Glu) by lysine (Lys), which exists in approximately 40% east Asian populations.² Such a polymorphism (Glu to Lys, or G to A, or *1 to *2) disrupts ALDH2 activity, causing high blood acetaldehyde concentration and “alcohol flush reactions” after alcohol consumption.¹ Alcoholics with heterozygous ALDH2^*1/^*2 or homozygous ALDH2^*2/^*2 polymorphism have ALDH2 deficiency and have an increased risk of developing digestive track cancers,³ however, the association of this polymorphism with HCC development and how acetaldehyde affects HCC still remain obscure.

A case-control study of a small number of patients (78 cases) suggests that the frequency of ALDH2^*2/^*2 allele had a correlation with an increased risk of HCC among heavy drinkers,⁴ but it is not clear whether ALDH2 polymorphism is associated with HCC caused by other etiologies other than alcohol. In the current study, we examined 646 patients with viral hepatitis B (HBV) infection and found that ALDH2 deficiency was not associated with an increased risk of HCC development in HBV patients without alcohol consumption but was a risk factor for HCC in cirrhotic HBV patients with excessive alcohol consumption, suggesting cirrhosis, alcohol consumption, and ALDH2 deficiency synergistically promote HCC. To model this clinical condition, we developed a mouse model of HCC induced by chronic CCl₄ administration (fibrosis) and alcohol feeding, and tested this model in three lines of Aldh2-deficient mice: including Aldh2 global knockout (KO), Aldh2^*1/^*2 knock-in, and liver-specific Aldh2 KO (Aldh2^Hep−/−) mice. Our data revealed that all of these Aldh2-deficient mice were more susceptible to CCl₄ plus alcohol-associated HCC development compared to their wild-type mice.

ALDH2 deficiency is known to induce excessive acetaldehyde accumulation and oxidative stress during alcohol consumption, leading to mitochondrial DNA damage and base modifications such as oxidation of deoxyguanosine to 8-hydroxy-2′-deoxyguanosine (8-OhDG) leading to mitochondrial dysfunction.⁵ Consequently, mitochondrial dysfunction-mediated DNA damage further accelerates cellular senescence, and contributes to aging-associated phenotypes and pathologies in various types of diseases and cancer including HCC.⁶ Here we provided evidence suggesting that extracellular vesicles (EVs) can transfer these damaged DNA from hepatocytes into HCC cells, thereby promoting HCC progression, suggesting a cross-talk between damaged hepatocytes and HCC cells via the transfer of oxidized mtDNA-enriched EVs that promote HCC.

EVs are well-known as a mean of intercellular communications by delivering proteins, miRNAs, DNAs, and glycolipids. They play important roles in physiology and pathology of the liver.^7,8 Interestingly, EVs are also commonly detected in tumor microenvironment and play a role in facilitating tumorigenesis by regulating angiogenesis, immunity, and metastasis.⁹ In the current study, we demonstrated that after chronic CCl₄ and ethanol exposure, Aldh2-deficient hepatocytes produced oxidized mtDNA-enriched EVs, which were transferred into HCC cells and activated multiple oncogenic pathways that have previously been implicated in promoting HCC. These pathways include signal transducer and activator of transcription 3 (STAT3),^10,11 C-Jun N-terminal kinase (JNK),^12,13 transcriptional co-activator with PDZ-binding motif (TAZ),^14,15 and BCL-2 protein.^16,17

Materials and Methods

Human subject cohort

A total of 929 subjects were included in the study, including 102 cases with HBV-related chronic hepatitis, 264 cases with HBV-related cirrhosis, 280 cases with HBV-related HCC, and 283 healthy individuals as controls. Among HCC patients, there were 116 subjects who drink alcohol (male >40 g/day, female >20 g/day, drinking period >5 years), and these patients are defined as HCC with excessive alcohol drinking. The remaining of the HCC patients is defined as HCC without alcohol drinking. The baseline demographic and clinical characteristics of this cohort were described in supporting materials and Table S1. ALDH2 rs671 includes three genotypes, GG, AA, and GA (Table S2). The study was approved by the First Affiliated Hospital of Jilin University Institutional Review Board and the Research and Development Committee. All participants provided written informed consent.

Animals

Aldh2 knockout (Aldh2^−/−) mice on a C57BL/6N background were described previously¹⁸ and C57BL/6N mice used as control. Heterozygous Aldh2^*1/^*2 knock-in mice were kindly provided by Dr. Mochly-Rosen.¹⁹ Aldh2^*1/^*2 mice were crossed with C57BL/6N mice to generate Aldh2^*1/^*2 knock-in mice and wild-type littermate controls. The Aldh2 floxed mice were generated by activating the Aldh2 gene in Aldh2^{tm1a(EUCOMM)Wtsi} mice by crossing of Aldh2^{tm1a(EUCOMM)Wtsi} mice (kindly provided by Dr. Patel)²⁰ with homozygous FLPcR mice (The Jackson Laboratory), which express Flippase (flip) in germline cells. Liver-specific Aldh2 knockout (Aldh2^Hep−/−) mice were generated via several steps of crossing Aldh2 floxed mice with Albumin Cre mice (The Jackson Laboratory). Mice were housed in polycarbonate cages (4 mice per cage) and maintained in a temperature and light controlled facility (12:12 light-dark cycle) under standard food and water ad libitum. All experiments were approved by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) Animal Care and Use Committee.

Induction of alcohol-associated HCC

Three different lines of Aldh2-deficient mice and their corresponding controls (10–12-week-old male mice) were injected intraperitoneally with CCl₄ (Sigma) (0.2 ml/kg in olive oil; 2 times per week) for 28 weeks and fed 4% v/v ethanol-containing Lieber-DeCarli diet (Bio-Serv, Flemington, NJ) for the last 10 weeks out of 28 weeks.

Statistical analysis

Distributions of allele and genotype frequencies were analyzed using Chi-squared test. Logistic regression analyses to calculate an odds ratio (ORs) and 95% confidence interval were performed. For animal experiments, data are presented as the means ±SEM. Significance was evaluated by using Student’s t-test between two groups and One-way ANOVA for multiple comparisons. All P<0.05 values were considered to be significant.

See other materials and methods in the Supporting materials.

Results

ALDH2 rs671 polymorphism is associated with an increased risk of HCC in HBV patients with excessive alcohol consumption, but not in those without alcohol drinking

To study the association of ALDH2 rs671 polymorphism with liver disease progression and cancer, we examined 646 HBV patients with chronic liver disease, cirrhosis, and HCC with or without excessive alcohol consumption (Table S1). Because those who carried an A allele are associated with low ALDH2 activity, and the number of patients with homozygous AA genotype was small, we thus combined patients with GA and AA genotypes into one group GA/AA for ALDH2 deficiency in our analyses. Among these genotypes, there were no differences in age, gender, anti-viral drug treatment, and viral load (Table S1). As illustrated in Table 1A, among patients with excessive alcohol drinking, the allele frequencies for GA/AA in patients with HBV-related HCC were statistically higher than those with HBV-related cirrhosis. In contrast, no distribution in the allele frequency for those with HBV-related cirrhosis and HBV-related HCC were observed among patients without alcohol drinking (Table 1A and Fig. S1A).

Table 1.

Association of ALDH2 polymorphisms and progression of HCC in patients

Table 1A
Genotypes	HBV-related disease progression with excessive alcohol drinking			P-values	Significant difference between groups
	aChronic HBV without cirrhosis (n=27)	bHBV-associated cirrhosis (n=94)	cPatients with HBV-HCC(n=116)
GG	24 (12.50%)	83 (43.23%)	85 (44.27%)	0.012*	c vs. b
GA/AA	3 (6.67%)	11 (24.44%)	31 (68.89%)
ORs and 95% Cl for the GA/AA between c vs. b: 2.752 (95% Cl: 1.30–5.83, =P0.007)
Genotypes	HBV-related disease progression without excessive alcohol drinking			P-values	Significant difference between groups
	aChronic HBV without cirrhosis (n=75)	bHBV-associated cirrhosis (n=170)	cPatients with HBV-HCC(n=164)
GG	42 (17.72%)	107 (45.15%)	88 (37.13%)	0.213*	N/A
GA/AA	33 (19.18%)	63 (36.63%)	76 (44.19%)

Table 1B
Genotypes	HCC staging with excessive alcohol drinking					P-values	Significant difference b etween groups
	0 (n=2)	A(n=18)	B (n=23)	C (n=49)	D (n=24)
GG	2 (2.35%)	15 (17.65%)	22 (25.88%)	32 (37.65%)	14 (16.47%)	0.01*	C vs. B, D vs. B,
GA/AA	0 (0%)	3 (9.68%)	1 (3.23%)	17 (54.83%)	10 (32.26%)
ORs and 95% Cl for the GA/AA genotype between stage C vs. B: 11.69 (95% Cl: 1.45–94.36, P=0.007) ORs and 95% Cl for the GA/AA genotype between stage D vs. B: 15.71 (95% Cl: 1.81–136.55, P=0.003)
Genotypes	HCC staging without excessive alcohol drinking					P-values	Significant difference between groups
	0 (n=5)	A (n=25)	B (n=37)	C (n=62)	D (n=35)
GG	2 (2.27%)	16 (18.18%)	21 (23.86%)	34 (38.64%)	15 (17.05%)	0.51*	N/A
GA/AA	3 (3.95%)	9 (11.84%)	16 (21.05%)	28 (36.84%)	20 (26.32%)

^*Chi-squared test was used for statistical evaluatation

We next determined the effects of ALDH2 rs671 polymorphism on HCC staging among patients with HBV-related HCC stratified by excessive alcohol consumption. As illustrated in Table 1B and Fig. S1B, we observed the association between those with A allele (GA/AA) and advanced stages of HCC only in those with alcohol drinking. The allele frequencies for GA/AA in HCC patients with alcohol drinking who had Barcelona Clinic Liver Cancer (BCLC) stage C (54.83%) or D (32.26%) were significantly higher than those with BCLC stage B (3.23%, P=0.007). We also used the logistic regression analysis to determine the association between those with A allele and HCC staging and found the ORs of 11.69 (P=0.01) and 15.71 (P=0.003), in patients with GA/AA genotypes presented with BCLC stage C and D, respectively, compared to those with stage B. Taken together, our data suggested that the risk of HCC development in HBV patients with underlying cirrhosis only occurred in ALDH2 deficient patients who consumed alcohol. Additionally, ALDH2 deficiency is also associated with advanced stage of HCC among excessive alcohol drinkers.

Aldh2-deficient mice are more susceptible to CCl₄+EtOH-induced HCC

To model our clinical observation on HCC development in drinkers with underlying cirrhosis, we thus developed a mouse model of liver cancer by chronic CCl₄ administration for 18 weeks to induce severe liver fibrosis and followed by feeding 4% alcohol diet along with the injection of CCl₄ for an additional 10 weeks (denoted CCl₄+EtOH) (Fig. 1A). Hepatic histopathology from CCl +EtOH-treated mice showed fibrosis, steatosis, inflammation, hepatocyte ballooning degeneration, and tumor nodules (data not shown). Immunohistochemistry analyses detected the expression of HepPar-1, an immunohistochemical marker of HCC (Fig. S2A), suggesting CCl₄+EtOH treatment induces HCC, however, no tumor metastasis was found in other organs (Fig. S2B). These results suggest that the induction of CCl₄-mediated liver fibrosis followed by alcohol administration is a mouse model for alcohol-associated HCC.

Fig. 1. Aldh2 deficiency accelerates HCC development in a model of alcohol-associated HCC induced by chronic CCl₄+EtOH.

(A) Three lines of Aldh2-deficient mice and their WT control mice were subjected to chronic CCl₄ administration for 28 weeks and ethanol feeding for the last 10 weeks. (B) Serum levels of ALT. (C) Liver tissues were stained with Sirius red and α-SMA antibody or tissues were subjected to Western blotting analysis. (D) Representative gross findings of livers and HCC occurrence in Aldh2^−/− mice. (E) The number and size of tumor after chronic CCl₄+EtOH. Student’s t-test was used for statistical evaluation (*P<0.05; **P<0.01).

Next, we tested this mouse model in three different lines of Aldh2-deficient mice, including Aldh2^−/−, Aldh2^*1/^*2 knock-in, Aldh2^Hep−/− mice, and their corresponding controls. Aldh2^*1/^*2 mice were developed by replacing the mouse wild-type Aldh2 allele with a mouse E487K mutant Aldh2 allele, and resulted in marked loss of ALDH2 activity.¹⁹ The body weight was decreased in Aldh2^−/− and Aldh2^*1/^*2 mice but not in Aldh2^Hep−/− mice after alcohol feeding (Fig. S2C–D). Interestingly, serum ALT levels were lower in all Aldh2-deficient mice after CCl₄+EtOH versus WT mice (Fig. 1B). Collagen deposition and α-SMA expression were greater in CCl₄+EtOH-treated Aldh2^−/− versus WT mice, however, no differences were observed between these two groups after treatment with only CCl₄ (Fig. 1C). Greater degree of liver fibrosis was also observed in CCl₄+EtOH-treated Aldh2^*1/^*2 and Aldh2^Hep−/− mice compared to their controls (Fig. S3A–C). Furthermore, the number and size of tumors were greater in CCl₄+EtOH treated Aldh2^−/− mice versus WT mice (Fig. 1D). In CCl₄ treated group, the number, but not the size of tumor was also higher in Aldh2^−/− mice compared to WT mice (Fig. 1D). Moreover, CCl₄+EtOH-treated Aldh2^*1/^*2 and Aldh2^Hep−/− mice also had greater degree of liver fibrosis and number/size of tumors compared to their WT mice (Fig. 1E, Fig. S3A–C).

Liver progenitor cells (LPCs) with self-renewing abilities have been implicated in liver carcinogenesis through their transformation toward cancer stem cells and high ALDH activity is considered a functional marker of LPC.²¹ Thus, we hypothesized whether Aldh2 deficiency promotes LPC proliferation, which accelerates HCC progression. Because very few LPCs are detected in our CCl₄+EtOH model, we used two other models that induce significantly LPCs in the liver by feeding mice with cholinedeficient ethionine supplemented (CDE) or 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) diet. Our data revealed that deletion of Aldh2 did not affect LPC proliferation in CDE- or DDC-fed mice (Fig. S4A–C). Deletion of Aldh1a1 neither affected LPC proliferation in these models nor affected HCC progression induced by DEN+CCl₄ (Fig. S4A–C). Interestingly, the number of LPCs was lower in Aldh2^−/− mice than in WT mice after alcohol feeding in both models (Fig. S4B). These data suggest that ALDH deficiency-associated HCC progression is unlikely mediated by stimulating LPCs.

Aldh2-deficient mice are more susceptible to CCl₄+EtOH-induced oxidative stress and activation of various oncogenic pathways

To understand the mechanisms underlying ALDH2 deficiency-associated HCC acceleration, we measured reactive oxygen species (ROS) in alcohol-associated hepatic tumor regions. As illustrated in Fig. 2A, immunohistochemistry analyses revealed that CCl₄ +EtOH-treated Aldh2^−/− mice expressed greater hepatic levels of lipid peroxidation markers including malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE) than those of WT mice, which was also confirmed by western blot analyses. In contrast, the expression of FOXO3A, an oxidative stress resistance protein, was lower in Aldh2^−/− mice versus WT mice. We further evaluated the expression of NRF2 and CTNNB1, which complexes with FOXO3A for antioxidant gene transcription.²² Interestingly, expression of NRF2 protein and mRNA was not altered while expression of β-catenin (encoded by CTNNB1) was upregulated in CCl₄+EtOH-treated Aldh2^−/− mice versus WT mice (Fig. S5A). In addition, levels of hepatic glutathione (GSH), an important anti-oxidant, and the expression of ROS detoxification genes such including Sod1, Sod2 and Catalase 1 was lower in CCl₄ +EtOH-treated Aldh2^−/− mice versus WT mice (Fig. 2B). Similar to these data, hepatic mRNA levels of Sod1, Sod2 and Catalase 1 were also down-regulated in CCl₄+EtOH-treated Aldh2^Hep−/− mice versus WT mice (Fig. S5B). The mechanisms underlying downregulation of anti-oxidant genes in Aldh2-deficient mice are not clear, but it may be due to extensively produced oxidative stress in these mice as described above.

Fig. 2. Chronic CCl₄+EtOH activates oxidative stress and multiple oncogenic pathways in Aldh2-deficient mice.

Aldh2^−/− and WT mice were subjected to CCl₄+EtOH for 28 weeks as described in Figure 1. (A) Liver tumor sections were subjected to immunohistochemistry and Western blotting analyses. (B) Liver tissues were subjected to Glutathione (GSH) assay and RT-qPCR analyses. (C) Liver non-tumor and tumor regions were subjected to Western blotting to evaluate indicated proteins involved in oxidative stress and cancer development. (D) Liver sections were subjected to TUNEL assay. Student’s t-test was used for statistical evaluation (*P<0.05; **P<0.01; ***P<0.001).

Oxidative stress is known to activate mitogen-activated protein kinase (MAPK) (e.g. JNK, p38) and STAT3 phosphorylation, which is associated with HCC development.^10,11,13 Thus, we evaluated these kinase proteins in adjacent non-tumor and tumor regions in CCl₄+EtOH-treated HCC mouse model. As illustrated in Fig. 2C and Fig. S5C, pJNK, pP38 and pSTAT3 levels were significantly higher in Aldh2-deficient tumor tissues compared to those of WT. Furthermore, the number of TUNEL⁺apoptotic cells was lower, whereas Bcl-2 mRNA levels were higher in CCl₄+EtOH-treated Aldh2^−/− mice versus WT mice (Fig. 2C–D, Fig. S5D).

TAZ and its target CTGF levels are higher in HCC from CCl₄+EtOH-treated Aldh2-deficient mice versus WT mice

Hyperactivation of YAP/TAZ, two transcription factors mediating Hippo signaling, promotes cell proliferation while inhibits cell death, ultimately leading to tumorigenesis.^14,15 To test whether YAP/TAZ is involved in HCC development, we measured the expression of YAP/TAZ and found that the levels of TAZ protein and mRNA were higher in HCC from CCl₄+EtOH-treated Aldh2^−/− mice versus WT mice; while YAP levels were comparable between these two groups (Fig. 3A–B). Greater TAZ protein levels in Aldh2^−/− mouse HCC was further confirmed by immunohistochemistry analyses (Fig. 3C).

Fig. 3. The Hippo signal transducer TAZ and its target gene Ctgf expression are upregulated in CCl₄+EtOH-treated Aldh2^−/− mice versus WT mice.

Aldh2^−/− and WT mice were subjected to CCl₄+EtOH for 28 weeks. Liver tissues were collected for experiments. (A) Western blotting analysis. (B) RT-qPCR analysis. (C) Immunohistochemistry analysis. (D) Cytoplasmic/nucleus fractions were extracted and subjected to Western blotting. (E) Liver tissue sections were subjected to RT-qPCR analysis and CTGF immunohistochemistry. Student’s t-test was used for statistical evaluation (*P<0.05; **P<0.01).

YAP and TAZ transcription factors regulate downstream gene expression when they are translocated from cytoplasm to nuclei.^14,15 Here we found that nuclear TAZ protein levels but not nuclear YAP protein were much higher in CCl₄+EtOH-treated Aldh2^−/− HCC compared to WT HCC (Fig. 3D), suggesting that TAZ but not YAP is activated in HCC from CCl₄+EtOH-treated Aldh2^−/− mice. In agreement with TAZ activation, hepatic expression of the TAZ targeted gene Ctgf mRNA and protein levels was higher in tumor regions from CCl₄+EtOH Aldh2^−/− mice versus WT mice (Fig. 3E).

Aldh2 deficiency exacerbates DNA damage, oxidized mtDNA formation, hepatocyte senescence after CCl₄+EtOH treatment

To address the mechanism between Aldh2 deficiency and alcohol-associated HCC, we further studied oxidative stress-induced DNA damage response by measuring the expression of γH2AX, a sensitive marker for damaged DNAs.²³ As illustrated in Fig. 4A, in non-tumor region, γH2AX protein levels were greater in Aldh2-deficient mice than those in WT mice; γH2AX protein levels in tumor regions were higher than in WT non-tumor region although no differences between Aldh2^−/− and WT groups.Immunohistochemistry analyses further confirmed greater γH2AX foci number in non-tumor regions from Aldh2^−/− mice versus WT mice. Moreover, the increased γH2AX expression in CCl₄+EtOH-treated Aldh2-deficient mice was mainly detected in mitochondrial fraction (Fig. 4A), but not in CCl₄ alone-treated mice (Fig. S6A). Moreover, the expression of ataxia telangiectasia mutated (Atm), which encodes a protein that phosphorylates γH2AX in response to DNA double-strand breaks, was greater in Aldh2-deficient non-tumor regions; whereas the expression of oxoguanine DNA glycosylase 1 (Ogg1), excised base repairing gene, was down-regulated (Fig. 4B).

Fig. 4. DNA damage and cellular senescence are elevated in CCl₄ +EtOH treated Aldh2^−/− mice versus WT mice.

Aldh2^−/− and WT mice were subjected to CCl₄+EtOH treatment for 28 weeks. (A) Tissue lysates and liver tissue sections were subjected to Western blot and immunohistochemistry analysis of γH2AX. Mitochondrial fractions were also purified and subjected to Western blot analysis for γ H2AX. (B) Liver tissues were subjected to RT-qPCR analysis. (C) Heat map of the mRNA expression changes of the cellular senescence pathways in the liver from vehicle-treated and chronic CCl₄+EtOH-treated mice. (D) Liver tissues were subjected to SA-β-gal staining (red arrow head indicates positive staining) and RT-qPCR analyses. (E, F) Primary hepatocytes (WT and Aldh2^−/−) were treated with CCl₄±EtOH as indicated to induce hepatocyte damage and subjected to Western blotting and RT-qPCR analyses. Student’s t-test was used for statistical evaluation (*P<0.05; **P<0.01; ***P<0.001).

Oncogenic/oxidative stress-derived DNA damage is a common mediator for cellular senescence²⁴ and is known to promote the generation of neoplastic cells.⁶ Thus, we examined cellular senescence and found that after CCl₄+EtOH, cellular senescence was significantly induced compared to normal liver tissues as illustrated by the alterations of several genes involved in cellular senescence pathways (Fig. 4C). In addition, the number of senescence-associated β-galactosidase (SA-β-gal) positive cells, and mRNA levels of P21 and P53 were higher in liver tissues from Aldh2^−/− mice versus WT mice after CCl₄+EtOH Fig. 4D. Furthermore, mRNA expression of ll1b, ll6 and Ccl2, known as common pro-inflammatory cytokines of senescence associated secretory phenotypes (SASP),²⁵ were also higher in CCl₄+EtOH-treated Aldh2^−/− mice versus WT mice (Fig. 4D).

To model the effect of CCl₄+EtOH treatment in vivo, hepatocyte damage was induced in vitro by incubating with CCl₄+EtOH. Such treatment induced higher levels of γH2AX in Aldh2^−/− hepatocytes than in WT hepatocytes; but no differences were observed in CCl₄ alone group (Fig. 4E). In addition, the expression of anti-oxidant mRNA levels (Sod1, Sod2, and catalase 1) was lower whereas expression of senescence associated genes (P21, P53, Ccl2, and II6) was higher in CCl₄+EtOH-treated Aldh2^−/− hepatocytes than in WT hepatocytes (Fig. 4F, Fig. S6B).

Finally, CYP2E1 has been implicated in tumorigenesis by processing pro-carcinogen into carcinogen.²⁶ However, hepatic CYP2E1 protein levels were comparable between WT and Aldh2^−/− mice (Fig. S6C), suggesting that the increased susceptibility of Aldh2-deficient mice to CCl₄+EtOH-induced HCC is not due to the CYP2E1 alternations.

Aldh2-deficient hepatocytes produce higher levels of oxidized mtDNA-enriched EVs than WT hepatocytes after CCl₄+EtOH

To further confirm that Aldh2-deficient hepatocytes are more susceptible to CCl₄+EtOH-induced DNA damage, we measured the major oxidative DNA-damage product 8-OhDG.²⁷ As illustrated in Fig. 5A, hepatic 8-OhDG levels in non-tumor tissues as well as serum 8-OhDG levels were significantly higher in CCl₄+EtOH-treated Aldh2^−/− mice versus WT mice.

Fig. 5. Aldh2^−/− mice produce a greater number of 8-OhDG enriched EVs, which can be engulfed by HCC cells, than WT mice after CCl₄+EtOH.

Aldh2^−/− and WT mice were subjected to CCl₄+EtOH for 28 weeks. (A) Liver tissues were subjected to immunofluorescence staining with 8-OhDG antibody. Hepatic and serum levels of 8-OhDG were assessed. (B) Concentration of serum-derived EVs was quantified using BCA assay and the characteristic of EVs was identified by Nanosight, TEM, and Western blotting. (C) DNA fraction was extracted from serum-derived EVs and subjected to PCR amplification, and were loaded on 2% Agarose gel. (D) Isolated EVs were further subjected to flow cytometric analysis. (E) Schematic overview of the experimental design. Fluorescent labelled-EVs were introduced to C57BL/6N mice and CCl₄+EtOH-associated-HCC bearing mice. The mice were sacrificed after 16 hours, and livers were observed under fluorescent stereo microscope (White arrow heads or circles indicate tumor regions). (F) RT-qPCR analyses of EV engulfment related genes in the liver. Student’s t-test was used for statistical evaluation (*P<0.05; **P<0.01).

EVs are known to play an important role in maintaining cellular homeostasis by excreting harmful mtDNA from cells,²⁸ and senescent cells are known to produce more EV.²⁹ Therefore, we hypothesized whether Aldh2^−/− hepatocytes remove these harmful oxidized mtDNA via EVs. First, we detected mtDNA in EVs and found that the EVs from CCl₄+EtOH-treated Aldh2^−/− mice showed greater mtDNA copy numbers than those from WT mice (Fig. S7A). Second, the concentration of EVs from Aldh2^−/− mouse serum or Aldh2^−/− hepatocyte cultured medium was greater than that of WT (Fig. 5B, Fig. S7B). In addition, the size and morphology of isolated EVs were confirmed by NanoSight tracking and electron microscopy and western blot analysis of EV marker proteins (Hsp70 and TSG101) (Fig. 5B). Furthermore, the serum derived-EVs from CCl₄+EtOH-treated mice were enriched with mtDNA (Cyto C ox, D-loop and Nd1), but contained low levels of nuclear DNA (Gapdh) (Fig. 5C). Hepatocyte-specific protein CYP2E1 was also detected in EVs from CCl₄+EtOH-treated mice (Fig. S7B), suggesting that these EVs are mainly derived from damaged hepatocytes. Finally, flow cytometric analyses revealed that EVs from CCl₄+EtOH-treated Aldh2^−/− mice or from Aldh2^−/− hepatocytes contained much higher levels of 8-OhDG⁺ EV fraction than that of WT mice or WT hepatocytes (Fig. 5D, Fig. S7C).

EV number was reported to be positively correlated with serum ALT levels.^30,31 However, our above data show that serum ALT levels were lower but EV concentration was higher in Aldh2^−/− compared to WT mice after CCl₄+EtOH. The lower ALT levels in Aldh2^−/− mice may be due to greater levels of BCL-2 expression in these mice versus WT mice (as shown in Fig. 2C), the higher levels of EVs may be due to greater levels of ROS in Aldh2^−/− mice versus WT mice because anti-oxidant treatment reduced EV production in CCl₄+EtOH-treated hepatocytes (Fig. S7D). The above data suggest that damage hepatocytes can secrete damage DNA via EVs. Then we asked whether HCC can engulf these EVs. To validate the bio-distribution of EVs, we injected mice with fluorescent-labelled EVs and found that tumor regions had much higher fluorescent intensity than normal/non-tumor liver regions (Fig. 5E). Furthermore, EV engulfment related genes were significantly higher in Aldh2^−/− mouse liver tumor tissues versus non-tumor regions (Fig. 5F). In addition, after in vitro incubation with fluorescent-labelled EVs, HCC cells engulfed more EVs than primary hepatocytes, as demonstrated by greater reduction of the fluorescence-intensity in cultured medium (Fig. S7E). Collectively, these findings suggest that HCC cells engulf more EVs than primary hepatocytes.

Acetaldehyde and oxidized mtDNA enriched EVs synergistically activate multiple oncogenic pathways in HCC

The above data suggest that Aldh2-deficient hepatocytes produce higher levels of oxidized mtDNA-enriched EVs, which can be engulfed by HCC cells. To examine the effects of these EVs on HCC, we treated HepG2 cells (with or without acetaldehyde) with serum EVs from CCl₄+EtOH-treated WT or Aldh2^−/− mice. As illustrated in Fig. 6A, treatment with Aldh2^−/− EVs and acetaldehyde synergistically induced higher levels of ROS production, H₂O₂ generation, and greater gene expression levels of TAZ, CTGF, and BCL-2 compared to treatment with EVs from WT mice; whereas treatment with EVs or acetaldehyde alone had mild or no effects on these values.

Fig. 6. Oxidized mtDNA-enriched EVs and acetaldehyde synergistically activate oxidative stress and multiple oncogenic pathways in HCC cells.

(A, B) HepG2 cells were treated with acetaldehyde (100 μM) and serum-derived EVs (panel A) or purified 8-OhDG DNA from these EVs (panel B), and then subjected to OxiSelect intracellular ROS assay for ROS and H₂O₂ measurement, or RT-qPCR analysis. (C, E) HepG2 cells were treated with acetaldehyde (100 μM) and synthetic 8-OhDG (varying concentrations), and then subjected to OxiSelect intracellular ROS assay (panel C), RT-qPCR (panel D), and Western blot analyses (panel E). One-way ANOVA was used for statistical evaluation (*P<0.05; **P<0.01; ***P<0.001).

Because of oxidized-mtDNA enriched in these EVs, we next determined if oxidized DNA such as 8-OhDG DNA contributed to these effects by EVs observed in Fig. 6A. First, we extracted 8-OhDG⁺ DNA from EVs and incubated with these DNA in HepG2 cells, and found 8-OhDG⁺ DNA and acetaldehyde synergistically upregulated TAZ and CTGF mRNA levels in HepG2 cells (Fig. 6B). Second, we treated HepG2 with synthetic 8-OhDG DNA with or without acetaldehyde and found that synthetic 8-OhDG DNA treatment elevated the production of ROS and H₂O₂ levels in HepG2 cells in a dose-dependent manner in the presence of acetaldehyde (Fig. 6C). Such effects were not seen in primary hepatocytes (Fig. 6C) or murine hepatocyte cell line AML12 (Fig. S8). Finally, 8-OhDG and acetaldehyde synergistically upregulated the expression of TAZ, CTGF and BCL-2 mRNAs (Fig. 6D) as well as the expression of TAZ and phosphorylated JNK, p38, and STAT3 proteins in HepG2 cells (Fig. 6E).

Discussion

In the current study, by analyzing our human data of ALDH2 polymorphisms in 646 patients with chronic liver diseases from HBV and HCC, we found that ALDH2 deficiency had an increased risk of HCC development in cirrhotic HBV patients with alcohol drinking but not in those without alcohol drinking. To model this condition, we induced liver fibrosis in mice using CCl₄ followed by alcohol treatment and tested this model in three lines of Aldh2-deficient mice. Our data revealed that Aldh2-deficient mice were more susceptible to HCC development induced by CCl₄+EtOH. Such increased risk of HCC development in Aldh2-deficient mice after alcohol consumption is likely due to elevated acetaldehyde given acetaldehyde is a carcinogen that can cause DNA damage and mutagenesis, resulting in activation of the oncogenic pathway IL-6/STA3¹⁸ and tumor transformation.¹ Both chronic CCl₄ administration and acetaldehyde have been shown to induce genomic DNA damage,^32,33 which may induce tumor initiation. Therefore, tumor initiation via genomic DNA damage induced by both CCl₄ and acetaldehyde likely contribute to the HCC development in this CCl₄+EtOH model in Aldh2^−/− mice. In addition to these mechanisms, we also identified the novel mechanism by demonstrating the crosstalk between Aldh2-deficient hepatocytes and HCC via the oxidized mtDNA transferring. Aldh2-deficient hepatocytes produce a greater number of oxidized mtDNA enriched EVs than WT hepatocytes, which was likely due to CCl₄+EtOH induction of greater levels of DNA damage in these cells than in WT cells. Another mechanism may be because of greater levels of cellular senescence from Aldh2-deficient hepatocytes compared to WT hepatocytes given cellular senescence is known to accelerate EV secretion.²⁹

Among the oxidized DNAs, 8-OhDG is one of the major products of DNA oxidation and has been widely used as a biomarker for oxidative stress and carcinogenesis.²⁷ Although 8-OhDG as a biomarker and risk factor for cancer development has been well documented, whether and how 8-OhDG contributes to alcohol-associated HCC development remain obscure. In the current study, we demonstrated that damaged Aldh2-deficient hepatocytes from CCl₄+EtOH exposure excreted much higher levels of 8-OhDG via EVs than WT hepatocytes. These harmful 8-OhDG-enriched EVs can be engulfed by HCC cells as demonstrated by in vivo and in vitro experiments with labelled EVs. The mechanisms by which HCCs engulfed more EVs than primary hepatocytes are not clear but may be due to the increased expression of EVs engulfment related genes in HCCs as demonstrated in this study. Importantly, we also demonstrated that treatment of HCC cells with 8-OhDG together with acetaldehyde induced activation of multiple oncogenic pathways that are known to promote HCC, including STAT3,^10,11 JNK,^12,13 TAZ,^14,15 and BCL-2.^16,17 Such activation is likely induced, at least in part, by oxidative stress induced by 8-OhDG and acetaldehyde given oxidative stress is well-known to activate these oncogenic pathways.³⁴ Interestingly, 8-OhDG plus acetaldehyde-mediated activation of oxidative stress and multiple oncogenic pathways was only observed in HCCs but not in primary hepatocytes. The underlying mechanisms for this are not clear and it may be related to the fact that compared to HCC, primary hepatocytes express higher levels of anti-oxidant genes and ALDH2, which ameliorate ethanol-induced oxidative stress.

Another interesting finding from this study was that TAZ but not YAP was activated in Aldh2^−/− tumors from CCl₄+EtOH-treated mice and in 8-OhDG+acetaldehyde-treated hepatocytes although both TAZ and YAP are downstream transcription factors in Hippo pathway with largely overlapping functions including promoting HCC progression.^14,15 A recent study revealed that oxidative stress induced reversible S-glutathionylation at conserved cysteine residues within TAZ but not YAP, which subsequently increases TAZ protein stability and promotes TAZ activation.³⁵ This may explain why TAZ not YAP is markedly elevated in CCl₄+EtOH-treated Aldh2^−/− HCC versus WT HCC given greater ROS in Aldh2^−/− than in WT HCC.

In summary, our study demonstrated that oxidized DNA-enriched EVs and acetaldehyde synergistically contribute to the development of alcohol-associated HCC, suggesting that inhibition of oxidized mtDNA-enriched EV production could be a novel therapeutic strategy for ameliorating alcohol-associated HCC in ALDH2-deficient individuals.

Highlights.

• ALDH2 deficiency is associated with an increased risk of HCC from cirrhosis with alcohol drinking.

• Chronic CCl₄+EtOH treatment induces greater hepatic mitochondrial DNA damage in Aldh2-deficient mice than WT mice.

• Oxidized mitochondrial DNA is delivered to HCC cells via hepatocyte-derived extracellular vesicles.

• Oxidized mitochondrial DNA and acetaldehyde synergistically promotes ROS production and multiple oncogenic pathways in HCC.

Abbreviations:

ACE

acetaldehyde

ALDH2

Aldehyde dehydrogenase 2

ALT

alanine aminotransferase

ARF6

ADP-ribosylation factor 6

ATM

Ataxia telangiectasia mutated

BCLC

Barcelona Clinic Liver Cancer

BCL-2

B cell lymphoma 2

CCl₄

carbon tetrachloride

CDC42

cell division control protein 42 homolog

CDE

cholinedeficient ethionine supplemented

CM

cultured medium

CTGF

connective tissue growth factor

EtOH

ethanol

CYP2E1

Cytochrome P450 2E1

DDC

3,5-diethoxycarbonyl-1,4-dihydrocollidine

DDR

DNA damage response

DEN

diethylnitrosamine

EVs

extracellular vesicles

FOXO3

forkhead box O3

GSH

glutathione

HCC

hepatocellular carcinoma

HNE

hydroxynonenal

H2AX

H2A histone family member X

JNK

c-Jun N-terminal kinase

LAMP-1

lysosomal associated membrane protein 1

LPCs

Liver progenitor cells

MDA

malondialdehyde

MP

microparticle

mtDNA

mitochondrial DNA

NAC

N-acetyl-L-cysteine

nDNA

nuclear DNA

Ogg-1

8-oxoguanine glycosylase

ROS

reactive oxigen species

SOD

superoxide dismutase

STAT3

signal transducer and activator of transcription 3

TAZ

PDZ-binding domain

TUNEL

Terminal deoxynucleotidyl transferase dUTP nick end labeling

WT

wild type

WWTR1

WW domain-containing transcription regulator protein 1

YAP

yes-associated protein 1

α-SMA

α-smooth muscle actin

8-OhDG

8-hydroxy-2’-deoxyguanosine