Alcohol abuse and cigarette smoking are associated with global DNA hypermethylation: results from the German Investigation on Neurobiology in Alcoholism (GINA).

Recent studies have shown that smoking and alcoholism may be associated with altered DNA methylation and that alcohol consumption might induce changes in DNA methylation by altering homocysteine metabolism. In this monocenter study, we included 363 consecutive patients referred for hospitalization for alcohol detoxification treatment. Blood samples were obtained on treatment days 1, 3, and 7 for measurement of global DNA methylation in leukocytes by liquid chromatography tandem mass spectrometry. Genomic DNA was used for genotyping the following seven genetic variants of homocysteine metabolism: cystathionine beta-synthase (CBS) c.844_855ins68, dihydrofolate-reductase (DHFR) c.594 + 59del19bp, methylenetetrahydrofolate-reductase (MTHFR) c.677C > T and c.1298A > C, methyltetrahydrofolate-transferase (MTR) c.2756A > G, reduced folate carrier 1 (RFC1) c.80G > A, and transcobalamin 2 c.776C > G. Multivariate linear regression showed a positive correlation of global DNA methylation with alcohol consumption and smoking on day 1 of hospitalization. DNA methylation was not correlated with homocysteine or vitamin plasma levels, nor with the tested genetic variants of homocysteine metabolism. This suggests a direct effect of alcohol consumption and smoking on DNA methylation, which is not mediated by effects of alcohol on homocysteine metabolism.


Introduction
Tobacco and alcohol are the most commonly used noxious compounds worldwide.
Cigarette smoking and alcoholism significantly increase the risk for a variety of medical and psychiatric conditions and different forms of cancer (Thun et al., 1997;Zaridze et al., 2009). Because of the high prevalence of alcoholism and tobacco abuse and their negative health consequences, it is important to understand the mechanisms involved in alcohol and tobacco dependence and toxicity. A growing number of studies have shown that alcoholism and chronic alcohol intake in non-addicted subjects may be associated with altered DNA methylation (Bonsch, Lenz, Reulbach, Kornhuber, & Bleich, 2004;Harlaar & Hutchison, 2013;Starkman, Sakharkar, & Pandey, 2012).
DNA methylation depends on S-adenosylmethionine (SAM) as a methyl group donor.
The demethylated residue of SAM is S-adenosylhomocysteine (SAH), which is reversibly hydrolyzed to homocysteine. In a vitamin B6-dependent pathway, homocysteine can be transsulfurated to cystathionine and cysteine ( Figure 1).
Alternatively, homocysteine can be remethylated, depending on the essential co-factors folate, vitamin B2 and vitamin B12. As vitamin deficiency commonly occurs in alcoholdependent patients, alcohol-induced changes in DNA methylation and homocysteine might be explained by vitamin deficiencies (Cravo & Camilo, 2000;Heese et al., 2012).
Not only the vitamin status, but also genetic variants may modify folate, vitamin B12 and homocysteine metabolism (Stover, 2011). Alcohol-induced changes of DNA methylation are possibly influenced by these genetic variants which are common in the general population.
In a cohort of 363 patients with alcohol dependency, we analyzed changes of DNA methylation and attempted to identify parameters related to homocysteine metabolism which may mediate or modify the association of alcohol and DNA methylation, i.e. plasma levels of homocysteine and vitamins involved in homocysteine metabolism as well as genetic variants of homocysteine metabolism.

Patients
The present study is part of the German Investigation on Neurobiology in Alcoholism (GINA) (Heese et al., 2012). Consecutive patients were recruited from the Department of Addiction and Psychotherapy of the LVR-Clinic in Bonn, Germany (Heese et al., 2012). All participants were diagnosed with alcohol dependency according to ICD-10 and were included in the study on admission for alcohol detoxification. Patients were mainly detoxified with clomethiazole following a symptom-triggered regime using the Banger-Score (Banger, Philipp, Herth, Hebenstreit, & Aldenhoff, 1992). If, for clinical reasons, clomethiazole could not be used, benzodiazepines were administered.
Patients diagnosed with dependence from other substances were excluded. Daily alcohol consumption was calculated per day according to patients' self-reported alcohol consumption within the last week before admission to the hospital. Fasting blood samples were obtained on days 1 (admission), 3 and 7 of the detoxification treatment.
Blood samples were centrifuged and consecutive serum and lithium heparin plasma samples were stored at −80°C directly after collection. Homocysteine and global DNA methylation were assessed at all three time points, while vitamin serum levels were obtained at admission. This study was approved by the local ethics committee. All patients gave their informed written consent.

Biochemical measurements
Serum alcohol concentrations were measured by an enzymatic test (alcohol dehydrogenase method) with a Dimension Vista™ system (Siemens Healthcare Diagnostics, Eschborn, Germany).
Serum alanine aminotransferase (ALT) activity, aspartate aminotransferase (AST) activity and gamma glutamyl transferase (GGT) activity were measured by means of an enzymatic test (ALTI method, AST method, GGT method) with a Dimension Vista™ system (Siemens Healthcare Diagnostics). Reference intervals for ALT ranged up to 45 U/l for men and 34 U/l for women, reference intervals for AST ranged from up to 35 U/l for men and 31 U/l for women, and reference intervals for GGT ranged up to 55 U/l for men and 38 U/l for women.
Serum carbohydrate-deficient transferrin (CDT) and serum transferrin were measured by means of particle-enhanced immunonephelometry using a BN Prospec™ System Homocysteine was determined by fully automated particle-enhanced immunonephelometry with a BN II System (Siemens Healthcare Diagnostics, Eschborn, Germany) by enzymatic conversion to S-adenosyl-homocysteine (SAH). The reference range for homocysteine is 5.8 -11.9 µmol/L. The intra-assay coefficient of variation of the homocysteine assay was 3.4% (mean: 11 µmol/L, n=20), the inter-assay coefficient was 5.6% (mean: 11 µmol/L l, n=20).
Plasma concentrations of vitamin B12 and folate were measured by means of a competitive chemiluminescence immunoassay with an Access™ Immunoassay System (Beckman Coulter, Krefeld, Germany) according to the manufacturer's instructions. The reference range of vitamin B12 is 130-680 pmol/L and the reference range of folate is 6.8-30 nmol/L. The intra-assay coefficient of variation of the vitamin B12 assay was 3.8% (mean: 487 pmol/L; n=20), the inter-assay coefficient was 4.2% (mean: 492 pmol/L; n=20). The intra-assay coefficient of variation of the folate assay was 3.1% (mean: 14.1 nmol/L; n=20), while the inter-assay coefficient of variation was 3.6% (mean: 14.3 nmol/L; n=20).

Determination of global DNA methylation
Global DNA methylation was measured by ultra-high performance liquid chromatography (UHPLC) tandem mass spectrometry with a method adapted from the work done by Kok and coworkers (Kok et al., 2007). In short, 2 µg of DNA were hydrolyzed using formic acid. Cytosine (cyt) and 5-methylcytosine (mcyt) were separated using an Ultimate 3000 LC system (Dionex, Sunnyval, CA, USA) and an Acquity BEH Amide column (1.7 µm, 2.1x100 mm, Water Corporation, Milford, MA, USA). The mobile phase was consisting of (A) MeCN/H 2 O 1:1 and (B) MeCN/H 2 O 95:5, both buffered with 10 mM NH 4 HCO 2 and 0.125% HCOOH (v/v). The compounds were eluted with a linear gradient from 95 to 85% in 4 min at 0.5 mL•min -1 flow rate. The column was then washed with 50% of solvent B during 2 min and reconditioned during 2 min until the next injection. The UHPLC system was coupled to a 3200 QTRAP The intra-and inter-assay coefficient of variation (CV) for the mcyt/tcyt was 1.7% (n = 9) and 3.5% (n = 9) for calf thymus DNA (mean mCyt/tCyt ratio 6.5%), respectively.

Statistics
Deviations from Hardy-Weinberg equilibrium were separately analyzed using the χ 2 goodness of fit test, comparing observed and expected numbers for each genetic variant (α = 0.05). We used multivariate linear regression with the global leukocyte DNA methylation on the day of admission as dependent variable and age, gender, body mass index, cigarette consumption per day, daily alcohol consumption, vitamin plasma levels and the seven included single nucleotide polymorphisms as independent variables. α = 0.05 was defined as significance level.

Results
We included 363 serial patients (250 men, 113 women). Demographic and laboratory data determined at admission are shown in Table 1.
From the day of admission (day 1) 344 samples were available. A significant number of patients left the hospital before alcohol detoxification was completed or refused participation on some days of the study. Therefore, on day 3, n=59 samples and on day 7, n=75 samples were available. Self-reported alcohol consumption before admission was 218 g/d (± 119), Figure 2. 23% (n=83) of the study population were non-smokers.
The rest of the study population reported tobacco consumption with a mean consumption of 18.4 (± 13.6 1SD) cigarettes per day.
Multivariate linear regression showed that on the day of admission (day 1), global leucocyte DNA methylation correlated with alcohol consumption per day and cigarette consumption per day. There was no correlation of global leukocyte DNA methylation with age, gender, vitamin plasma levels or the seven single nucleotide polymorphisms (table 3).
Samples drawn on day 3 of the study showed a positive correlation of global DNA methylation with cigarette consumption, but not with alcohol consumption. On day 7, neither alcohol consumption nor cigarette consumption were significantly associated with global DNA methylation in leucocytes. When analysis was confined to patients who remained in hospital until days 3 or 7, no significant change in DNA methylation was detectable over time.
When the small group of non-smokers was analyzed separately, there was no significant association between the global DNA methylation and daily alcohol consumption (Beta 0.162; p=0.27). When only smokers were included analysis showed an association between global DNA methylation and daily alcohol consumption on the day of admission (Beta 0.154; p=0.04).

Discussion
This study shows that global DNA methylation in leukocytes correlates with the amount of daily alcohol consumption and cigarette smoking in alcohol dependent patients suggesting that both alcohol and tobacco provide an increase in global DNA methylation. We found no positive correlation with alcohol consumption on day 3, and no correlation with alcohol consumption or smoking on day 7. However, a large number of patients left the hospital before detoxification was completed on day 7. When analysis was confined to patients who remained in hospital until days 3 or 7, no significant change in DNA methylation was detectable over time. Thus, the lack of a correlation between alcohol consumption and DNA methylation on days 3 and 7 might be explained by drop-outs rather than by a change of methylation during detoxification.
When smokers and non-smokers were analyzed separately, the association of global DNA methylation and daily alcohol consumption could only be found in smokers. The data of this study does not allow to decide whether this finding is due to separate analysis of a small subgroup (non-smokers), or if any impact of alcohol on DNA methylation only becomes effective in smokers.
No association of DNA methylation with vitamin or homocysteine plasma levels was found. Therefore, the increase in global DNA methylation cannot be explained by vitamin deficiency or metabolic consequences of chronic alcohol consumption, such as liver dysfunction. This is surprising as dietary and metabolic factors are known to influence DNA methylation (Davis & Uthus, 2004;Ulrey, Liu, Andrews, & Tollefsbol, 2005), and malnutrition and liver dysfunction are common in alcohol-dependent patients. Possibly, DNA hypermethylation in alcohol-dependent patients is a direct, dose-related consequence of alcohol consumption rather than a dietary and metabolic consequence.
Our findings also show that genetic variants of key enzymes of methylation metabolism are not associated with DNA methylation in alcohol dependent patients, suggesting that the tested variants do not contribute to hypermethylation in alcohol dependent patients.
According to our findings, changes in global DNA methylation in alcohol dependent patients cannot be explained by changes in homocysteine metabolism. Direct inhibition of DNA methyltransferases by alcohol has previously been described as a potential mechanism of DNA hypermethylation in alcohol dependence, which could also be the case in our patient sample (Bonsch et al., 2006;Varela-Rey et al., 2013).
Cigarette smoking and alcoholism significantly contribute to a great variety of medical conditions, especially to different forms of cancer (Varela-Rey et al., 2013;Zaridze et al., 2009). Several factors potentially promote alcohol-induced cancer development, such as toxic effects of ethanol metabolites and oxidative stress. This study provides additional evidence that aberrant patterns of DNA methylation could also contribute to alcohol-tobacco induced disease development.
Other authors have described different results. Thapar et al. found decreased DNA methylation when alcohol dependence was combined with smoking (Thapar et al., 2012). The difference between this these findings and the present study may be explained by a smaller sample size and a lower amount of alcohol consumption in the alcohol dependent subjects included in the study of Thapar et al. (n=25 alcoholics). It has also been reported that occasional alcohol consumption in healthy subjects leads to DNA hypomethylation (Zhu et al., 2012). However, studies including healthy subjects with occasional alcohol consumption and studies including subjects with alcohol dependence and heavy consumption may not be comparable. The same argument holds true for studies with negative results of alcohol consumption and DNA methylation in healthy subjects (Ono et al., 2012;Zhang et al., 2011).
The strengths of the present study are its sample size, the characteristics of the study population including only alcohol dependent patients with heavy alcohol consumption, and the inclusion of important dietary and genetic factors, such as homocysteine and vitamin plasma levels and the polymorphisms of seven important genetic variations of methylation metabolism. One important limitation is that a significant number of study participants dropped out of the study after day 1, considerably limiting the number of available samples from day 3 and day 7 after alcohol cessation. Unfortunately, this study does not contain data which relate the differences in DNA methylation with a functional effect. Neither can it provide data on the molecular mechanisms underlying the effects of alcohol and smoking on global DNA methylation. Also, DNA methylation patterns are frequently tissue-and cell-specific, reflecting tissue-specific regulation of DNA methylation (Khavari, Sen, & Rinn, 2010). Therefore, methylation patterns should ideally be examined in the tissue that is directly related to the outcome of interest. Thus, further tissues of interest, such as brain and liver, should be tested in animal studies. The sulfur-containing amino acid methionine is activated to S-adenosylmethionine (SAM), which is a ubiquitous methyl group donor. The degradation product of SAM is Sadenosylhomocysteine (SAH), which is hydrolyzed to homocysteine. Homocysteine can be remethylated to methionine and SAM via methionine synthase (MTR), which depends on derivatives of folate and vitamin B12 as cofactors. Lack of these vitamins is a common cause of hyperhomocysteinemia (Mudd et al., 2001). The folate derivative is synthesized by methylenetetrahydrofolate reductase (MTHFR) and dihydrofolate reductase (DHFR), and the derivative of vitamin B12 is transported by transcobalamin 2 (Tc2). Alternatively, homocysteine can be transsulfurated by vitamin B6 dependent cystathionine β-synthase (CBS) and cystathionine gamma-lyase (CGL) to cysteine as a component of glutathione. Due to the existence of several functional variants in the genes involved in homocysteine metabolism, and to differences in dietary vitamin and amino acid uptake, disorders of homocysteine metabolism exhibit marked inter-individual differences.