SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Insulin resistance (IR) and mitochondrial dysfunction play a central role in the pathophysiology of nonalcoholic fatty liver disease (NAFLD). We hypothesized that genetic factors and epigenetic modifications occurring in the liver contribute to the IR phenotype. We specifically examined whether fatty liver and IR are modified by hepatic DNA methylation of the peroxisome proliferator–activated receptor γ coactivator 1α (PPARGC1A) and mitochondrial transcription factor A (TFAM) promoters, and also evaluated whether liver mitochondrial DNA (mtDNA) content is associated with NAFLD and IR. We studied liver biopsies obtained from NAFLD patients in a case–control design. After bisulfite treatment of DNA, we used methylation-specific polymerase chain reaction (PCR) to assess the putative methylation of three CpG in the PPARGC1A and TFAM promoters. Liver mtDNA quantification using nuclear DNA (nDNA) as a reference was evaluated by way of real-time PCR. Liver PPARGC1A methylated DNA/unmethylated DNA ratio correlated with plasma fasting insulin levels and homeostasis model assessment of insulin resistance (HOMA-IR); TFAM methylated DNA/unmethylated DNA ratio was inversely correlated with insulin levels. PPARGC1A promoter methylation was inversely correlated with the abundance of liver PPARGC1A messenger RNA. The liver mtDNA/nDNA ratio was significantly higher in control livers compared with NAFLD livers. mtDNA/nDNA ratio was inversely correlated with HOMA-IR, fasting glucose, and insulin and was inversely correlated with PPARGC1A promoter methylation. Conclusion: Our data suggest that the IR phenotype and the liver transcriptional activity of PPARGC1A show a tight interaction, probably through epigenetic modifications. Decreased liver mtDNA content concomitantly contributes to peripheral IR. (HEPATOLOGY 2010)

Metabolic syndrome (MS) is strongly associated with several metabolic disturbances, including insulin resistance (IR) in several tissues. Actually, the original description of MS was mainly centered on resistance to insulin-stimulated glucose uptake,1 and the disease was regarded as the IR syndrome. Recent evidence has shown that liver steatosis is positively correlated with all the components of MS2; therefore, nonalcoholic fatty liver disease (NAFLD) is now regarded as the hepatic manifestation of MS. Nevertheless, whereas the epidemiological and clinical evidence are unquestionable, an understanding about the role of the fatty liver in the pathophysiology of the IR phenotype is still elusive. Conversely, several data have implicated IR as a major contributor to the pathogenesis and disease progression of human NAFLD.3

The clustering of traits of MS are highly hereditable, and both genome-wide and candidate gene association studies have identified several loci that influence disease susceptibility. A less explored but remarkable mechanism that is gaining acceptance in our understanding of the pathophysiology of common diseases is the epigenetic regulation of transcriptional control by DNA methylation. Interestingly, the epigenetic mechanisms can affect the regulation of several gene pathways through life and explain the gene–environment interaction, but more importantly, they are potentially modifiable.

Evidence also shows that mitochondrial dysfunction plays a central role in the pathogenesis of IR and its associated complications.4 In this respect, qualitative and quantitative changes in mitochondrial DNA (mtDNA) were involved in the pathogenesis of type 2 diabetes, particularly when the changes occur in the skeletal muscle.5 Moreover, the mtDNA content in peripheral blood leukocytes has been associated with IR in children and adults,6, 7 and is an early event observed in small- and large-for-gestational-age newborns.8 Consistent with these observations, mitochondrial dysfunction also plays a significant role in the pathogenesis of NAFLD,9 suggesting that both IR and NAFLD share several morphological and functional abnormalities. Nevertheless, the evidence is still inconclusive, in part because of limitations to sample human liver tissue.

To propose a unifying hypothesis that may explain how NAFLD and IR are connected, we explored the role of the epigenetic regulation as a potential modifier of both. In particular, we focused on methylation of cytosine at the carbon-5 position in CpG dinucleotides, which is generally associated with gene silencing. Hence, we hypothesized that both genetic factors and epigenetic modifications occurring in the liver tissue contribute to the IR phenotype. To address this hypothesis, we selected two genes: (1) peroxisome proliferative–activated receptor γ coactivator 1 alpha (PPARGC1A), because its altered signaling contributes to glucose intolerance, IR, and type 2 diabetes, and it controls mitochondrial biogenesis; and (2) mitochondrial transcription factor A (TFAM), because it is involved in the maintenance of the mitochondrial genome. We examined specifically whether fatty liver and IR are associated with hepatic DNA methylation of the promoter of these two genes. In addition, we further evaluated whether the liver mtDNA copy number is associated with NAFLD and IR, as well as the status of PPARGC1A and TFAM promoter methylation.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Patients.

We studied liver biopsies obtained from adult NAFLD patients in a case–control design. A total of 63 unrelated NAFLD patients and 11 control subjects were included in the study.

Patients were considered for inclusion if they had histopathological evidence of fatty liver disease, simple steatosis, or nonalcoholic steatohepatitis (NASH) on liver biopsy specimens obtained during the study period. Secondary causes of steatosis, including alcohol abuse (≥30 g alcohol daily for men and ≥20 g for women), total parenteral nutrition, hepatitis B and hepatitis C virus infection, and the use of drugs known to precipitate steatosis were excluded. By using standard clinical and laboratory evaluation as well as liver biopsy features when applicable, autoimmune liver disease, metabolic liver disease, Wilson's disease, and α-1–antitrypsin deficiency were likewise ruled out in all patients.

Control subjects were selected from patients attending the Liver Unit, whose age and sex matched the NAFLD patients. In addition to the standard health assessment, a careful ultrasonographic examination of the liver was performed in all the control individuals. Control liver specimens were obtained by way of percutaneous liver biopsy. The reason for performing a liver biopsy in these subjects was based on the presence of persistently mildly elevated serum liver enzyme activity. In all the control subjects, all causes of common liver disease were ruled out, and subjects were included in the study if they did not have histological evidence of fatty change; the histological diagnosis of control livers was minimal changes or mild cholestasis.

The case participants and the controls were selected during the same study period from the same population of patients attending the Liver Unit, and all of them shared the same demographic characteristics (occupation, educational level, place of residence, and ethnicity).

Physical, Anthropometric, and Biochemical Evaluation.

Health examinations included anthropometric measurements, a questionnaire on health-related behaviors, and biochemical determinations. Body mass index was used as an index for relative weight. Additionally, waist and hip circumference were also assessed. Blood was drawn from 12-hour fasting subjects who had rested in a supine position for at least 30 minutes. Serum insulin, total cholesterol, high-density lipoprotein and low-density lipoprotein cholesterol, triglycerides, plasma glucose, and liver function tests were measured using standard clinical laboratory techniques. All biochemical determinations were measured using a Hitachi-912 Autoanalyzer (Roche, Buenos Aires, Argentina) or Immulite 1000 (DPC, Buenos Aires, Argentina). Homeostasis model assessment of IR (HOMA-IR) was used to evaluate an IR index and was calculated as fasting serum insulin (μU/mL) × fasting plasma glucose (mmol/L)/22.5. Elevated blood pressure was defined as systolic arterial blood pressure ≥130 mm Hg and/or diastolic arterial blood pressure ≥85 mm Hg or antihypertensive treatment. Anthropometric measurements and blood sampling were obtained from each patient at the time of liver biopsy and prior to any intervention. Details about physical, anthropometric, and biochemical evaluation are shown in Table 1. All the experiments were conducted in accordance with the guidelines of the 1975 Declaration of Helsinki. Written consent was obtained from all patients in accordance with the procedures approved by our institution's ethical committee.

Table 1. Clinical and Biochemical Characteristics of the Study Population According to Disease Status
VariablesControl Subjects (n = 11)NAFLD Patients (n = 63)P Value*
  • Data are expressed as the mean ± SD.

  • Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; DABP, diastolic arterial blood pressure; GGT, gamma glutamyl transpeptidase; HDL, high-density lipoprotein; LDL, low-density lipoprotein; NS, not significant; SABP, systolic arterial blood pressure.

  • *

    Mann-Whitney U test.

Female/male, n5/632/31NS
Age, years49.4 ± 10.250.3 ± 9.9NS
Smoking habit, cigarettes/day2.5 ± 7.04.3 ± 9.4NS
Physical activity, hours/week2.5 ± 2.71.4 ± 4.5NS
BMI, kg/m227.9 ± 5.931.9 ± 5.60.03
Waist circumference, cm104.5 ± 3.8104.8 ± 11.9NS
Waist/hip ratio0.94 ± 0.080.95 ± 0.09NS
SABP, mm Hg118.3 ± 22.3127.0 ± 15.6NS
DABP, mm Hg73.3 ± 15.079.5 ± 10.6NS
Fasting plasma glucose, mmol/L5.35 ± 1.606.42 ± 2.250.02
Fasting plasma insulin, pmol/L49.3 ± 9.092.4 ± 57.60.02
HOMA-IR index1.7 ± 0.53.5 ± 2.20.03
Total cholesterol, mmol/L5.37 ± 1.115.43 ± 1.18NS
HDL cholesterol, mmol/L1.67 ± 0.281.36 ± 0.540.01
LDL cholesterol, mmol/L3.19 ± 0.923.45 ± 1.00NS
Triglycerides, mmol/L1.40 ± 0.642.04 ± 1.17NS
Uric acid, mmol/L244 ± 107309 ± 101NS
ALT, U/L62.5 ± 26.569.5 ± 47.2NS
AST, U/L38.7 ± 13.744.4 ± 25.4NS
GGT, U/L89.0 ± 28.287.0 ± 80NS
AP, U/L223.6 ± 151.6211.8 ± 95.8NS

Liver Biopsy and Histopathological Evaluation.

Liver biopsy was performed with ultrasound guidance and modified 1.4-mm Menghini needles (Hepafix, Braun, Germany) under local anesthesia on an outpatient basis. Liver biopsy specimens were routinely fixed in 40 g/L formaldehyde (pH 7.4) embedded in paraffin and stained with hematoxylin-eosin, Masson trichrome, and silver impregnation for reticular fibers. The same liver pathologist who was blinded to patient details analyzed all of the biopsy specimens. All specimens were at least 3 cm in length and contained a minimum of eight portal tracts. The degree of steatosis was assessed according to the system developed by Kleiner et al.10 based on the percentage of hepatocytes containing macrovesicular fat droplets. NASH was defined as steatosis plus mixed inflammatory cell infiltration, hepatocyte ballooning and necrosis, glycogen nuclei, Mallory's hyaline, and any stage of fibrosis, including absent fibrosis.11, 12 NAFLD activity score was evaluated as described by Kleiner et al.10

Bisulfite Treatment of DNA and Methylation-Specific Polymerase Chain Reaction.

Genomic DNA was isolated from liver biopsy specimens at the same time as RNA by standard methods.13 DNA bisulfite modification was performed as descibed.14 Briefly, this technique was based on bisulfite treatment of genomic DNA, which converts all the unmethylated cytosines to uracils while conserving the methylated cytosines; we used the EZ DNA Methylation Kit and followed the manufacturer's protocol (Zymo Research Corporation, Orange, CA). The chemically modified DNA was subsequently used as a template for a methylation-specific polymerase chain reaction (PCR) to determine the promoter methylation status of the selected CpG dinucleotides in the PPARGC1A and Tfam promoters. The assay was based on quantitative real-time PCR (RT-PCR) in an iCycler thermocycler (Bio-Rad, Hercules, CA) using SYBR Green (Invitrogen, Buenos Aires, Argentina) as a fluorescent dye.

For the methylation-specific PCR experiment, two pairs of primers were used: one pair was specific for bisulfite-modified methylated DNA (M primers) and the other pair was specific for bisulfite-modified unmethylated DNA (U primers). Thus, for each sample studied, two PCRs were performed simultaneously using the M primer and U primer pairs. Successful amplification from the M primers and U primers indicated methylation and unmethylation, respectively. For primer design, a sequence starting 2,000 bp upstream from the transcriptional start site (TSS) of PPARGC1A and Tfam was used in the MethPrimer program (http://www.urogene.org//methprimer/index1.html)15 to search for regions with potentially methylated CpG sites. The sequence was retrieved from the Database of Transcriptional Start Sites (http://dbtss.hgc.jp/) with the following ID numbers: NM_013261, chromosome: NCBI36: 4: 23402742..23500798, TSS: 23500789 for PPARGC1A, and NM_003201, chromosome: NCBI36: 10: 59815182..59825903, TSS: 59815182 for TFAM.

For maximal discrimination between methylated and unmethylated alleles, M and U primers were designed to contain at least one CpG site at the 3′ end.15 Primer sequences are shown in Table 2.

Table 2. Primers Used for Methylation-Specific PCR and Quantification of PPARGC1A mRNA Abundance and mtDNA
GeneForward Primer (5′[RIGHTWARDS ARROW]3′)Reverse Primer (5′[RIGHTWARDS ARROW]3′)Size (bp)
  1. Abbreviations: M, methylated-specific; PPIA, cyclophilin A; U, unmethylated-specific.

Methylation-specific PCR
 PPARGC1A-MATTTTTTATTGTTATGGGGGTAGTCAAAAATATTTAAAAACGCAAACGAA143
 PPARGC1A-UTTTTATTGTTATGGGGGTAGTTGAAAAAAATATTTAAAAACACAAACAAA141
 TFAM-MTTAATGGGTTTTATATAGATATACGGAAAAATAATAACGAAAAAAACGAA102
 TFAM-UTAATGGGTTTTATATAGATATATGGCAAAAATAATAACAAAAAAAACAAA102
mRNA gene expression
 PPARGC1ACCTGCATGAGTCTGTGCTCTGCAAAGAGGCTGGTCTTCAC164
 PPIATTCCAGGGTTTATGTGTCACTCATCTTCAAATTTCTCC89
mtDNA and nDNA amplification
 Mt F3212CACCCAAGAACAGGGTTTGTTGGCCATGGGTATGTTGTTAA108
 18S rRNATAGAGGGACAAGTGGCGTTC3CGCTGAGCCAGTCAGTGT105

The level of methylated DNA is expressed as the ratio of the estimated amount for methylated DNA to the unmethylated DNA levels, calculated for each sample using the fluorescence threshold cycle values for a previously estimated efficiency.16 We estimated the efficiency for each single sample tube using the slope of the exponential phase as described by Ramakers et al.17 Furthermore, as expected, controls for unmethylated DNA (a purified native amplicon) and fully methylated DNA (the same amplicon treated with the DNA methylase M.SssI (New England Biolabs, Ipswich, MA) yielded 0% and 100% DNA methylation patterns.

All the experiments were performed in triplicate. The CV% was observed to be less than 5%. The specificity of amplification and the absence of primer dimers were confirmed by way of melting curve analysis at the end of each run and agarose electrophoresis.

The Web-based AliBaba2 program (http://www.gene-regulation.com/pub/programs/alibaba2) was used for insilico prediction of transcription factor–binding sites in the studied DNA sequences,18a posteriori of the sequence and primers selection. To ensure the specificity of the method and to avoid variability in the results because of the presence of two CpG dinucleotides in the reverse primer, we designed a degenerated reverse primer that introduced a mismatch in the second CG site, and observed the absence of amplification, regardless of the target DNA (data not shown), indicating that primers recognize the status of both CpG dinucleotides simultaneously.

RNA Preparation and RT-PCR for Quantitative Assessment of Messenger RNA Expression.

Total RNA was prepared from liver tissue using a phenol extraction step method with an additional DNAse digestion. For RT-PCR, 1-3 μg of total RNA were reverse-transcribed using random hexamers and Moloney murine leukemia virus reverse transcriptase (Promega, WI). RT-PCR was performed for quantitative assessment of messenger RNA (mRNA) expression in an iCycler thermocycler (Bio-Rad) using SYBR Green (Invitrogen). All RT-PCRs were performed in triplicate. The mRNA abundance of target genes was normalized to the amount of a housekeeping gene (peptidylprolyl isomerase A, also known as cyclophilin A) to perform comparisons between the groups. Cyclophilin was found to be the most stable reference gene for testing liver mRNA expression among other housekeeping genes (β-actin, TATA box binding protein, and glyceraldehyde-3-phosphate dehydrogenase) tested before starting the experiment. The levels of mRNA were expressed as the ratio of the estimated amount of the target gene relative to the peptidylprolyl isomerase A mRNA levels using fluorescence threshold cycle values calculated for each sample, and the estimated efficiency of the PCR for each product was expressed as the average of each sample efficiency value obtained.19 The specificity of amplification and absence of primer dimers were confirmed using the melting curve analysis at the end of each run. The primer sequences and the resulting PCR product lengths are shown in Table 2.

Quantification of mtDNA.

An assay based on quantitative RT-PCR was used for both nuclear DNA (nDNA) and mtDNA quantification using SYBR Green (Invitrogen) as described.6 Primer sequences for mtDNA and nDNA for loading normalization are shown in Table 2. Results are presented as the mtDNA/nDNA ratio.

Quantitative RT-PCR was performed with a Bio-Rad iCycler (Bio-Rad). The calculation of DNA copy number involved extrapolation from the fluorescence readings in the mode of background subtracted from the Bio-Rad iCycler according as described by Rutledge.20 Specificity of amplification and the absence of primer dimers were confirmed by way of melting curve analysis at the end of each run. The two target amplicon sequences (mtDNA and nDNA) were visualized in agarose 2% and purified with a Qiagen Qiaex II Gel Extraction Kit (Tecnolab, Buenos Aires, Argentina), and dilutions of purified amplicons were used as the standard curve. The interassay variation coefficient was <20%. Specificity of the method was evaluated as described.6

Statistical Analysis.

The quantitative data are expressed as the mean ± SD unless otherwise indicated. Because a significant variance difference was observed between the groups for most of the variables, and because the distribution was significantly skewed in most cases, we chose to be conservative and assessed the differences between the groups using the Mann-Whitney U test. For multiple regression analysis, because the methylated DNA/unmethylated DNA ratio, HOMA-IR, and mtDNA/nDNA ratio were not normally distributed, we used log transformation of these variables. Correlation between the two variables was performed with a Spearman rank correlation test or Pearson correlation test after log transformation of the variables. To perform these analyses, we used the CSS/Statistica program package, version 6.0 (StatSoft, Tulsa, OK).

Results

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Hepatic DNA Methylation of the Promoter of PPARGC1A Significantly Correlates with Peripheral Insulin Resistance.

Studies in humans and rodents have demonstrated that altered PPARGC1A signaling leads to insulin resistance, as recently reviewed.21 Moreover, PPARGC1A is highly inducible in the liver and heart in response to physiological conditions that demand increased mitochondrial energy production.21PPARGC1A interacts with TFAM, which is required for the replication and maintenance of mtDNA.22 Therefore, to determine whether the status of liver DNA methylation of these two genes was associated with peripheral insulin resistance, we measured the level of DNA methylation of three putative methylation target sites (CpG) in the promoters of PPARGC1A (located at positions relative to TSS: −513, −519, and −615) and Tfam (located at positions relative to TSS: −433, −442, and −499). For this purpose, we included liver biopsies obtained from a subgroup of 30 NAFLD patients (simple steatosis, n = 17; NASH, n = 13), and nine control livers. The selection of liver samples was based upon the availability of liver DNA to perform all the methylation-specific PCR reactions (36 PCR reactions per sample for this particular experiment).

We observed that liver PPARGC1A promoter methylated DNA/unmethylated DNA ratio was significantly correlated with plasma fasting insulin levels (Pearson correlation coefficient r = 0.51, P < 0.01) and HOMA-IR (r = 0.58, P < 0.003) (Fig. 1A). Liver TFAM promoter methylated DNA/unmethylated DNA ratio was inversely correlated with fasting insulin levels (Pearson correlation coefficient, r = −0.49, P < 0.04).

thumbnail image

Figure 1. (A) Correlation between log-transformed methylated DNA/unmethylated DNA ratio for the PPARGC1A promoter in liver tissue and log-transformed HOMA-IR index. (B) Methylated DNA/unmethylated DNA ratio for the liver PPARGC1A promoter according to liver disease status.

Download figure to PowerPoint

We further evaluated the level of liver DNA methylation of these two genes according to fatty liver disease, and observed that PPARGC1A methylation status was significantly associated with NAFLD (Fig. 1B), showing that 47.9% of alleles were methylated in NAFLD versus 30.6 % in control liver. There were no significant differences in the ratio of Tfam promoter methylated DNA/unmethylated DNA between the control liver and NAFLD (0.07 ± 0.04, 6.8% of methylated alleles, and 0.06 ± 0.03, 5.5% of methylated alleles, respectively).

The histological disease severity was not associated either with PPARGC1A (simple steatosis, 0.81 ± 0.61 versus NASH, 0.84 ± 0.50; P = 0.54) or Tfam (simple steatosis, 0.06 ± 0.03 versus NASH, 0.05 ± 0.02; P = 0.74) promoter methylation levels. Neither liver PPARGC1A nor Tfam promoter methylated DNA/unmethylated DNA ratio correlated with NAFLD activity score (r = 0.23, P = 0.23 and r = −0.22, P = 0.32, respectively).

Because DNA methylation of the cytosine residues can interfere with the binding of transcription factors and thus prevent transcription, we examined whether the DNA sequence of PPARGC1A and TFAM that we selected for the analysis encompasses putative transcription factor–binding sites. Insilico prediction showed that the methylated CpG sites in PPARGC1A promoter were located near a putative hepatocyte nuclear factor-1 (essential for insulin gene transcription and involved in the susceptibility of type 2 diabetes) and C/EBPα-binding motif (which enhances the expression of peroxisome proliferator–activated receptor γ), and the TFAM sequence showed that one selected CpG dinucleotide site resides 100 bp from a putative hepatocyte nuclear factor-1 binding site.

Liver PPARGC1A Methylation Is Associated with Decreased Liver PPARGC1A mRNA Expression.

DNA methylation is an important epigenetic mechanism in the control of gene expression. To evaluate whether the status of PPARGC1A methylation in the liver affects its mRNA expression, we measured the relative liver mRNA level in all the samples included in the above-mentioned experiment. We observed that the abundance of liver PPARGC1A mRNA inversely correlated with the methylation levels of the PPARGC1A promoter (r = −0.602, P < 0.004), showing that DNA methylation of the promoter silenced gene expression. Consistent with our previous observation, the abundance of liver PPARGC1A mRNA inversely correlated with fasting plasma insulin levels (r = −0.70, P < 0.009) and HOMA-IR (r = −0.75, P < 0.004), suggesting that peripheral insulin resistance and the transcriptional activity of PPARGC1A in the liver are closely related processes.

Mitochondrial Biogenesis Is Reduced in the Liver of NAFLD Patients and Is Associated with Peripheral IR and PPARGC1A Promoter Methylation Status.

Mitochondrial function strongly influences the MS-related phenotypes. In addition, mitochondrial dysfunction and mitochondrial biogenesis are closely correlated, as decreased mtDNA content is associated with a decreased mitochondrial oxidative capacity.4 To address the role of liver mtDNA content in the pathogenesis of NAFLD and IR, we evaluated the liver mtDNA copy number in a sample of 63 patients with NAFLD and 11 control subjects. Interestingly, we observed that the liver mtDNA/nDNA ratio was significantly (P < 0.01) lower in the liver of NAFLD patients in comparison with that of control subjects (Fig. 2A). In addition, the mtDNA/nDNA ratio was inversely correlated with HOMA-IR (r = −0.38, P < 0.01) (Fig. 2B), serum fasting glucose (r = −0.24, P < 0.03), and plasma fasting insulin (r = −0.35, P < 0.02).

thumbnail image

Figure 2. (A) Liver mtDNA/nDNA ratio in control and NAFLD livers. (B) Correlation between log-transformed liver mtDNA/nDNA ratio and log-transformed HOMA-IR index.

Download figure to PowerPoint

Finally, log-transformed mtDNA/nDNA ratio was inversely correlated with log-transformed methylation levels of the PPARGC1A promoter (r = −0.48, P < 0.006) even after adjusting for HOMA-IR (beta ± SE: −0.60, P < 0.03), but not with methylation levels of the TFAM promoter.

Discussion

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

In this study, we evaluated whether epigenetic factors occurring in the liver, like altered DNA methylation of PPARGC1A and TFAM promoter associated CpG dinucleotides, are associated with peripheral insulin resistance. We elected to test this hypothesis in patients with NAFLD because IR is a hallmark feature in the pathophysiology of NAFLD, and NAFLD represents the hepatic component of MS. We observed that methylation levels of PPARGC1A promoter CpGs were correlated with HOMA-IR and plasma fasting insulin levels, and those of the TFAM promoter CpGs were inversely correlated with fasting insulin. Bisulfite treatment of liver DNA reveled that a moderate proportion (47.9%) of the promoter of PPARGC1A was methylated in NAFLD livers, which is significantly higher compared with that found in the control livers. The technique does not allow us to discriminate whether this proportion corresponds to a variable methylation within or between hepatic cells. Because DNA methylation located within or close to the 5′ region of genes is associated with regulation of gene expression, we examined whether liver PPARGC1A mRNA expression was altered. We observed that the liver abundance of PPARGC1A mRNA was inversely correlated with the methylation levels of PPARGC1A promoter CpGs, and also with the status of peripheral IR, suggesting that methylation of the three CpG sites in the gene promoter efficiently repressed the transcriptional activity of its mRNA.

The liver DNA methylation status of PPARGC1A promoter did not correlate with the histological disease severity or NAFLD activity score. Although our sample included patients covering the histological disease spectrum from steatosis alone to NASH, HOMA-IR was not different between the two histological groups (3.3 ± 2.10 and 3.8 ± 2.34, P = 0.31). This observation may explain our findings.

The epigenetic events observed in our study, regardless of the presence of hepatocyte injury, strengthens the concept that the NAFLD phenotype should be targeted for close observation and that early intervention is likely to play a pivotal role in early stages of the disease. Nevertheless, increasing the sample size of the control group may provide a better understanding of the observed results.

We next looked for quantitative changes in the mtDNA content of the liver of NAFLD patients and control subjects. Interestingly, we provided evidence that NAFLD was associated with reduced mtDNA copy number in the liver and linked quantitative changes in the liver mtDNA content with the peripheral status of IR, because we observed that mtDNA abundance was inversely correlated with HOMA-IR, plasma fasting insulin and glucose, and PPARGC1A promoter methylation status.

Overall, the results summarized above highlight the potential physiological role of the liver in MS-associated phenotypes, and suggest that MS is more than a clustering of related diseases and should be regarded as a multiple-organ incapacity to control the metabolic homeostasis. Our results also suggest that the abnormal hepatic triglyceride accumulation is not a bystander comorbidity but adversely affects the peripheral insulin sensitivity. This observation is supported by previous evidences that showed the impact of the progressive increase in the hepatic triglyceride content on the progressive impairment of insulin action in the liver, skeletal muscle, and adipose tissue.2, 23 Therefore, we may speculate that hepatic steatosis plays an important role in the metabolic events associated with MS, suggesting that the “pancreatic β cells–centric” approach of the pathophysiology of IR should be revised, at least during the prediabetic state. A recently published human study may confirm this hypothesis, as it revealed that the amount of the intrahepatic triglyceride content is an independent indicator of multiorgan insulin resistance.24

The close relationship we observed between the status of liver DNA methylation of the promoter of PPARGC1A and peripheral IR has not been demonstrated in humans. Indeed, it has been shown that high-fat diets down-regulate PPARGC1A as well as genes coding for proteins of the electron transport chain in human skeletal muscle, but these observations were never linked to an epigenetic mechanism.25 Interestingly, PPARGC1A is a coactivator wherein biological and physiological functions are particularly centered on striated muscle, liver, and heart, a triad of organs particularly relevant to the pathophysiology of IR and its cardiovascular complications.

Our finding of decreased mtDNA content in the livers of NAFLD patients may additionally explain the mitochondrial dysfunction associated with hepatic steatosis, and the finding of an inverse correlation between mtDNA content and the methylation levels of the PPARGC1A promoter may provide an explanation for this phenomenon. Supporting this observation, it was shown in patients with type 2 diabetes that low mtDNA content is responsible for alterations in oxidative phosphorylation and electron transport capacity.26 Moreover, severe mtDNA depletion can decrease the synthesis of the mtDNA-encoded polypeptides and impair the ability to oxidize fatty acids.27 From the present study, we can also propose that the decrease in the mtDNA copy number is closely linked to IR, indicating that the mitochondrial system is unable to maintain its adaptive response in the presence of hyperinsulinemia and hyperglycemia. This observation is in line with our previous report in an animal model of diet-induced fatty liver about a compensatory increase in the liver mtDNA content in the absence of IR to accommodate the metabolic load.28 Similar results have been shown to occur in skeletal muscle until the development of IR.29 However, when IR is established, the decrease in mtDNA content is a common finding, even in leukocytes.6

Finally, despite an inability to establish a cause and effect relationship between liver DNA methylation of PPARGC1A and IR, this study has novel findings with potentially important implications. We demonstrate for the first time that epigenetic changes occurring in the livers of patients with NAFLD may play a critical role in the modulation of peripheral IR and may be related to liver mtDNA content. These findings may have a direct clinical translation, as epigenetic changes are potentially reversible with any intervention, including pharmacological manipulation.

Moreover, given the importance of PPARGC1A as a key regulator of the transcriptional activity of several transcription factors involved in critical metabolic pathways, for instance, mitochondrial fatty acid oxidation, gluconeogenesis, and lipogenesis, it is an attracting candidate gene for pharmacological intervention. Likewise, any therapeutic intervention that stimulates mitochondrial biogenesis may also have a direct impact on reverting not only NAFLD but also the associated IR. This may explain the beneficial effects of pioglitazone in improving IR in NAFLD patients30 or angiotensin II type 1 receptor (AT1R) antagonists in animal models of NAFLD,31 because it was shown that these agents may increase mitochondrial biogenesis or protect mitochondria from oxidative stress.

References

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  • 1
    Reaven GM. Banting lecture 1988. Role of insulin resistance in human disease. Diabetes 1988; 37: 1595-1607.
  • 2
    Kotronen A, Yki-Jarvinen H. Fatty liver: a novel component of the metabolic syndrome. Arterioscler Thromb Vasc Biol 2008; 28: 27-38.
  • 3
    Cheung O, Sanyal AJ. Recent advances in nonalcoholic fatty liver disease. Curr Opin Gastroenterol 2010; 26: 202-208.
  • 4
    Kim JA, Wei Y, Sowers JR. Role of mitochondrial dysfunction in insulin resistance. Circ Res 2008; 102: 401-414.
  • 5
    Kelley DE, He J, Menshikova EV, Ritov VB. Dysfunction of mitochondria in human skeletal muscle in type 2 diabetes. Diabetes 2002; 51: 2944-2950.
  • 6
    Gianotti TF, Sookoian S, Dieuzeide G, Garcia SI, Gemma C, Gonzalez CD, et al. A decreased mitochondrial DNA content is related to insulin resistance in adolescents. Obesity (Silver Spring) 2008; 16: 1591-1595.
  • 7
    Song J, Oh JY, Sung YA, Pak YK, Park KS, Lee HK. Peripheral blood mitochondrial DNA content is related to insulin sensitivity in offspring of type 2 diabetic patients. Diabetes Care 2001; 24: 865-869.
  • 8
    Gemma C, Sookoian S, Alvarinas J, Garcia SI, Quintana L, Kanevsky D, et al. Mitochondrial DNA depletion in small- and large-for-gestational-age newborns. Obesity (Silver Spring) 2006; 14: 2193-2199.
  • 9
    Sanyal AJ, Campbell-Sargent C, Mirshahi F, Rizzo WB, Contos MJ, Sterling RK, et al. Nonalcoholic steatohepatitis: association of insulin resistance and mitochondrial abnormalities. Gastroenterology 2001; 120: 1183-1192.
  • 10
    Kleiner DE, Brunt EM, Van NM, Behling C, Contos MJ, Cummings OW, et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. HEPATOLOGY 2005; 41: 1313-1321.
  • 11
    Brunt EM, Janney CG, Di Bisceglie AM, Neuschwander-Tetri BA, Bacon BR. Nonalcoholic steatohepatitis: a proposal for grading and staging the histological lesions. Am J Gastroenterol 1999; 94: 2467-2474.
    Direct Link:
  • 12
    Neuschwander-Tetri BA, Caldwell SH. Nonalcoholic steatohepatitis: summary of an AASLD Single Topic Conference. Hepatology 2003; 37: 1202-1219.
  • 13
    Kawasaki ES. Sample preparation from blood, cells, and other fluids. In: Innis MA, Gelfand DH, Sninsky JJ, White TJ, eds. PCR Protocols. A Guide to Methods and Applications. Sandiego: Academic Press; 1990: 146-152.
  • 14
    Gemma C, Sookoian S, Dieuzeide G, Garcia SI, Gianotti TF, Gonzalez CD, et al. Methylation of TFAM gene promoter in peripheral white blood cells is associated with insulin resistance in adolescents. Mol Genet Metab 2010; 100: 83-87.
  • 15
    Li LC, Dahiya R. MethPrimer: designing primers for methylation PCRs. Bioinformatics 2002; 18: 1427-1431.
  • 16
    Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 2001; 25: 402-408.
  • 17
    Ramakers C, Ruijter JM, Deprez RH, Moorman AF. Assumption-free analysis of quantitative real-time polymerase chain reaction (PCR) data. Neurosci Lett 2003; 339: 62-66.
  • 18
    Grabe N. AliBaba2: context specific identification of transcription factor binding sites. In Silico Biol 2002; 2: S1-S15.
  • 19
    Ruijter JM, Ramakers C, Hoogaars WM, Karlen Y, Bakker O, van den Hoff MJ, et al. Amplification efficiency: linking baseline and bias in the analysis of quantitative PCR data. Nucleic Acids Res 2009; 37: e45.
  • 20
    Rutledge RG. Sigmoidal curve-fitting redefines quantitative real-time PCR with the prospective of developing automated high-throughput applications. Nucleic Acids Res 2004; 32: e178.
  • 21
    Finck BN, Kelly DP. PGC-1 coactivators: inducible regulators of energy metabolism in health and disease. J Clin Invest 2006; 116: 615-622.
  • 22
    Kelly DP, Scarpulla RC. Transcriptional regulatory circuits controlling mitochondrial biogenesis and function. Genes Dev 2004; 18: 357-368.
  • 23
    Hwang JH, Stein DT, Barzilai N, Cui MH, Tonelli J, Kishore P, et al. Increased intrahepatic triglyceride is associated with peripheral insulin resistance: in vivo MR imaging and spectroscopy studies. Am J Physiol Endocrinol Metab 2007; 293: E1663-E1669.
  • 24
    Fabbrini E, Magkos F, Mohammed BS, Pietka T, Abumrad NA, Patterson BW, et al. Intrahepatic fat, not visceral fat, is linked with metabolic complications of obesity. Proc Natl Acad Sci USA 2009; 106: 15430-15435.
  • 25
    Sparks LM, Xie H, Koza RA, Mynatt R, Hulver MW, Bray GA, et al. A high-fat diet coordinately downregulates genes required for mitochondrial oxidative phosphorylation in skeletal muscle. Diabetes 2005; 54: 1926-1933.
  • 26
    Boushel R, Gnaiger E, Schjerling P, Skovbro M, Kraunsoe R, Dela F. Patients with type 2 diabetes have normal mitochondrial function in skeletal muscle. Diabetologia 2007; 50: 790-796.
  • 27
    Fromenty B, Robin MA, Igoudjil A, Mansouri A, Pessayre D. The ins and outs of mitochondrial dysfunction in NASH. Diabetes Metab 2004; 30: 121-138.
  • 28
    Carabelli J, Burgueno AL, Rosselli MS, Gianotti TF, Lago NR, Pirola CJ, et al. High fat diet-induced liver steatosis promotes an increase in liver mitochondrial biogenesis in response to hypoxia. J Cell Mol Med 2010; doi:10.1111/j.1582-4934.2010.01128.x.
  • 29
    Lenaers E, De Feyter HM, Hoeks J, Schrauwen P, Schaart G, Nabben M, et al. Adaptations in mitochondrial function parallel, but fail to rescue, the transition to severe hyperglycemia and hyperinsulinemia: a study in Zucker diabetic fatty rats. Obesity (Silver Spring) 2010; 18: 1100-1107.
  • 30
    Sanyal AJ, Chalasani N, Kowdley KV, McCullough A, Diehl AM, Bass NM, et al. Pioglitazone, vitamin E, or placebo for nonalcoholic steatohepatitis. N Engl J Med 2010; 362: 1675-1685.
  • 31
    Rosselli MS, Burgueno AL, Carabelli J, Schuman M, Pirola CJ, Sookoian S. Losartan reduces liver expression of plasminogen activator inhibitor-1 (PAI-1) in a high fat-induced rat nonalcoholic fatty liver disease model. Atherosclerosis 2009; 206: 119-126.