Mitochondrial haplogroup G is associated with nonalcoholic fatty liver disease, while haplogroup A mitigates the effects of PNPLA3

Abstract Objectives Mitochondrial dysfunction plays a pivotal role in the pathogenesis of nonalcoholic fatty liver disease (NAFLD). We hypothesized that mitochondrial DNA (mtDNA) haplogroups affect the risk of NAFLD in Han Chinese patients and interact with PNPLA3 genotypes. Design NAFLD and control patients were recruited from a tertiary care centre. The mitochondrial genome was amplified in overlapping segments and sequenced. Mitochondrial haplogroups were determined using Mitomaster. PNPLA3 rs738409 genotyping was performed using restriction fragment length polymorphism analysis. Patients We enrolled 655 NAFLD patients and 504 controls. Results More NAFLD patients encoded haplogroup G; odds ratio (OR) 1.85 (95% confidence interval [CI] 1.16, 2.80). Subhaplogroup G3 was present more frequently in NAFLD patients (25.8% vs 6.5%). The PNPLA3 CG genotype resulted in an OR of 1.66 (95% CI 1.25, 2.21), and the GG genotype resulted in an OR of 2.33 (95% CI 1.72, 3.17) for NAFLD. Patients with mitochondrial haplogroup A had a significantly higher frequency of genotype GG. Among patients with haplogroup A, no PNPLA3 genotype was associated with increased NAFLD risk (CG: OR 1.17, 95% CI 0.55, 2.34; GG: OR 1.04 95% CI 0.66, 2.65). Excluding haplogroup A, the OR for CG was 1.58 (95% CI 1.18, 2.12), and the OR for GG was 1.81 (95% CI 1.30, 2.51). Conclusion Haplogroup G was associated with an increased risk of NAFLD PNPLA3 GG genotype was overrepresented among patients encoding haplogroup A and was not associated with NAFLD risk among haplogroup A patients. Mitochondrial genetics influence NAFLD risk and interact with PNPLA3 genotypes.


| INTRODUC TI ON
Nonalcoholic fatty liver disease (NAFLD) encompasses a spectrum of disorders beginning with hepatic steatosis, which can progress to steatohepatitis, cirrhosis and hepatocellular carcinoma in a subset of patients. 1 The prevalence of NAFLD has risen worldwide along with increases in type 2 diabetes, obesity and metabolic syndrome and is estimated at approximately 25%. 2 The prevalence of NAFLD in East Asian is particularly high (estimated at 27%). ROS and oxidative stress can disrupt function of the ETC, alter mitochondrial membrane potential and trigger mitochondrial permeability transition pore formation. 5 Indeed, these patients exhibit enlarged and swollen mitochondria and have impairment in the enzymatic activity of the ETC. 6,7 In additional to affecting enzymatic function, ROS can also damage mitochondrial DNA (mtDNA). NAFLD patients have been shown to have decreased mtDNA copy numbers and epigenetic modifications impairing expression of genes encoding ETC components. 8 Furthermore, liver mtDNA in patients with NAFLD has been shown to have a higher mutational rate and degree of heteroplasmy. 8 Importantly, defining groups of mtDNA polymorphisms termed haplogroups have been shown to influence susceptibility to a variety of disorders. 9 In one population, haplogroup L was found to be protective against the development of steatohepatitis. 10 Nuclear genes associated with NAFLD have been extensively investigated, with an isoleucine to methionine substitution in PNPLA3 being the most robust variant linked to the development of NAFLD. 11 PNPLA3 encodes a protein with hydrolase activity acting on triglycerides and retinyl esters. Loss of function results in accumulation of triglycerides and retinyl esters in lipid droplets of hepatocytes and hepatic stellate cells. 12 Complex interactions between genetic and environmental factors contribute to the development and progression of NAFLD. The nuclear and mitochondrial genomes are intricately associated and have co-evolved. 13 Interactions between the nuclear and mitochondrial genomes have been demonstrated in several disease processes [14][15][16][17] and have also been shown to modulate whole body metabolism. 18 Nuclear and mitochondrial genomes are known to interact in NAFLD, with varying degrees of liver steatosis and mitochondrial function on different nuclear DNA backgrounds. 14 Importantly, the mitochondrial genome differs considerably comparing Eastern and Western populations, and mtDNA is known to modulate the nuclear genome. 19 Mitochondria play a key role in the pathogenesis of NAFLD. 5,8,20 Mitochondrial haplogroups affect mitochondrial function and have co-evolved with the nuclear genome. 13, 17 We therefore hypothesized that East Asian mitochondrial DNA haplogroups will play a role in the development of NAFLD and affect the penetrance of PNPLA3 polymorphisms.

| Patient selection
Patients were enrolled from a tertiary medical centre, Shanghai Tenth People's Hospital affiliated with Tongji University School of Medicine. Informed consent was obtained from each patient. This study conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the institutional review board of Shanghai Tenth People's Hospital. Patients with NAFLD were recruited from the endocrinology clinic. In order to be included, NAFLD patients must have had no history of drinking or for men drink <140 g and for women drink <70 g of alcohol per week. Grams of alcohol was calculated as alcohol consumption (mL) × alcohol content (%) × 0.8 (alcohol-specific gravity). NAFLD patients also must have had no history of other liver-related diseases including hepatitis A, B and D or liver cirrhosis. Serology was performed for each patient in order to rule out viral hepatitis. All NAFLD patients also had ultrasound findings suggestive of fatty liver disease.
Qualitative measures of were assessed including echogenicity, hepatomegaly and intrahepatic vascular blurring as described. 21 Quantitative ultrasonography was performed for each patient to evaluate fat content and fibrosis. Liver fat quantification was performed using the back-scatter coefficient, which measures the returned ultrasound energy from tissues. 22 This method has been shown to be more accurate than conventional ultrasonography. 23 To be included in the current study, patients had to have ≥ grade 1 steatosis. Transient elastography was used to measure hepatic elasticity as described, [24][25][26] with lower values used to exclude advanced fibrosis. 27 Control patients were recruited during routine physical examination. To be included as a control, no history of NAFLD, type 2 diabetes mellitus (T2DM) or type 1 diabetes mellitus (T1DM) could be present based on medical history and laboratory results including blood glucose and haemoglobin A1c. NAFLD was excluded by negative liver ultrasound results. Control patients also had no history of viral hepatitis or cirrhosis based on medical history, serology and liver ultrasound results.
Demographic and laboratory data were collected for each pa- total bilirubin (TBili) were assessed using an automated biochemical analyser.

| Mitochondrial haplogroup determination
The mitochondrial genome was amplified in four segments. Sanger sequencing was performed to obtain the complete mitochondrial genome. Mitochondrial DNA sequences were converted to FASTA files. Mitochondrial DNA haplogroups were determined using Mitomaster (https://www.mitom ap.org/mitom aster/ index.cgi) as described. 28

| PNPLA3 genotyping
The PNPLA3 rs738409 polymorphism was genotyped as described using restriction fragment length polymorphism (RFLP) analysis. 29 Initially, the region of interest was amplified with conventional PCR

| Statistical methods
All statistical analysis was performed using R (R Project for Statistical Computing). Comparisons between each demographic variable were conducted using a Mann-Whitney U test. Comparisons between mitochondrial haplogroup frequencies were performed using Fisher's exact tests as described. 30 Odds ratios and confidence intervals were calculated using the Functions for Medical Statistics Book ('fmsb') package in R P-values were corrected for multiple comparisons using Bonferroni corrections.

| Demographics
A total of 655 patients with NAFLD and 504 control patients were included. Patient demographics are shown in Table 1. Age and gender were not significantly different between control and NAFLD patients. NAFLD patients weighed significantly more and had higher BMI. Triglycerides were higher, and HDL was lower in patients with NAFLD, while there was no significant difference in cholesterol or LDL. NAFLD patients had higher AST, GGT and ALP compared with control, while ALT was not significantly different. Albumin and total protein levels were similar between control and NAFLD patients.

| Mitochondrial haplogroup frequencies
Frequencies of each macrohaplogroup were determined for both control and NAFLD patients. A significantly higher percentage of patients with NAFLD encoded haplogroup G compared with controls ( Table 2). The odds ratio (OR) for NAFLD for haplogroup G was 1.85 (95% confidence interval [95% CI] 1.16, 2.80). The effect of haplogroup G was present after correcting for confounding variables including sex in multivariate models (Table 3). Among those with haplogroup G, subhaplogroup G3 was represented more frequently in patients with NAFLD (N = 16, 25.8%) vs control (N = 4, 6.5%). Of note, there was no significant difference between the percentage of control and NAFLD patients encoding mitochondrial haplogroup D (Table 2). Since haplogroup D and some of its subhaplogroups have been implicated as being protective against a variety of age-related disorders, 31,32 we also looked at the frequency of each of the D subhaplogroups. Among those individuals encoding haplogroup D, there were no significant differences between the percentage of control and NAFLD patients considering each subhaplogroup (Table S1).

| D ISCUSS I ON
In this study, we sought to determine whether mtDNA haplogroups affected the development of NAFLD and modulated the effects of the PNPLA3 rs738409 polymorphism. We found that haplogroup G was significantly more common in NAFLD patients compared with controls. We also found that among patients with haplogroup A, there was a higher frequency of the pathogenic rs738409 GG genotype.
Variations in mtDNA can be adaptive or maladaptive, with variations clustering together along discrete branches of the human mtDNA tree to form haplogroups. 9 Each haplogroup is defined by   Of note, we did not find a significant association between haplogroup D or any of its subhaplogroups and NAFLD. Haplogroup D has been associated with various age-related disorders 45 and has been shown to modulated ROS production. 31 Importantly, haplogroup D has not been demonstrated to affect the enzymatic activity of complex I or coupling of the ETC. 46 Therefore, it likely does not affect fatty acid metabolism. Haplogroup D may therefore not be expected to affect the accumulation of hepatic fatty acids. However, our study was not designed to assess progression from steatosis to steatohepatitis. Given its effects on ROS production and ability to modulate beta cell destruction due to autoimmunity in type 1 diabetes, haplogroup D may modulate the development of inflammation in the progression from simple steatosis to steatohepatitis.
The OR in our study for developing NAFLD with the PNPLA3 rs738409 polymorphism is similar to previously published results. 47 Our data suggest that an interaction exists between the PNPLA3  We therefore suspect that the pathogenicity of the PNPLA3 rs738409 polymorphism may be diminished on the background of mitochondrial haplogroup A. Haplogroup A is defined by several polymorphisms including A4824G and C8794T. 40 One study has showed that in Han Chinese patients showed that haplogroup A may decrease the risk of drug addiction. 50 This study also showed that HeLa cells overexpressing 8794T had decreased levels of ROS production compared with those overexpressing 8794C. 50 However, the exact mechanisms of how haplogroup A interacts with the PNPLA3 rs738409 polymorphism and mitigates the risk of NAFLD remain unknown. It is also possible that haplogroup A and its related polymorphism may alter uncoupled respiration. It has recently been demonstrated that modulating adenine nucleotide translocase can increase uncoupled mitochondrial respiration leading to increased F I G U R E 1 Frequency of PNPLA3 rs738409 genotypes. The frequency of each PNPLA3 genotype is shown considering only those patients with haplogroup A (A). The frequency of each PNPLA3 genotype is shown considering patients with all haplogroups except A (B). *GG vs CC and CG, P < .05; **Control vs NAFLD, P < .05; † Control vs NAFLD, P < .05 Sex has an important impact on metabolism, and sex hormones affect mitochondrial metabolism. Females have a higher mitochondrial respiratory rate and lower oxidative stress, with both progesterone 52,53 and oestrogen 54 playing important roles. Female mitochondria may be influenced by a higher expression of uncoupling proteins, which attenuate ROS production. 55 Women have been shown to have a lower risk of NAFLD, 56 potentially due to decreased fatty acid oxidation and increased lipogenesis in males. 57 Consistently, we found that female sex was associated with decreased risk of NAFLD in multivariate regression models (Tables 3 and 4). Haplogroup G was associated with risk of NAFLD after controlling for sex and age >50 (as a surrogate for menopausal status) indicating that its mechanism of action is likely independent of the effect of sex on mitochondrial function and metabolism. Similarly, among patients with haplogroup A, the PNPLA3 GG genotype was not associated with NAFLD regardless of sex ( to the functional effects of mitochondrial haplogroups G and A in our patients, further mechanistic study will need to be carried out in order to assess their effects on mitochondrial function, including ETC enzymatic activity, membrane potential and coupling.
In conclusion, we demonstrate that mitochondrial haplogroup G is associated with an increased risk of NAFLD. We further demonstrate that patients with haplogroup A have a higher frequency of the PNPLA3 rs738409 GG genotype without an increased risk of NAFLD. These results suggest a role for mtDNA in the development of NAFLD and that interactions between mtDNA and PNPLA3 may modulate disease penetrance.