Alcohol dependence (AD) is a complex behavioral disorder characterized by impaired control over drinking and a syndrome of serious problems related to alcohol abuse.1 It is determined by both genetic and environmental factors with an estimated heritability of 40%–60%.2 Although the prevalence of AD is moderate,3 it is not easy to choose the best AD phenotype for genetic research because of the difficulties in psychiatric diagnosis of AD through personal interviews.4 A number of alcohol-related phenotypes have been proposed and used in AD genetic studies,5 such as broader definition of AD, withdrawal symptoms after cessation of heavy alcohol intake, history of alcohol-induced blackouts, age of onset of regular drinking, age at first drunkenness, alcohol drinking, maximum number of drinks,6 daily alcohol consumption,7 heavy consumption8 and tolerance to alcohol’s effects.9
Recent genetic research on AD employing both genome-wide linkage scans (GWLS)6,8,10-18 and genome-wide association studies (GWAS)19-23 have provided preliminary data of the genetic determination of AD. For example, GWLS results have implicated the importance of the genomic region-4q to AD risk;6,8,10-13 and a GWAS using 100K single-nucleotide polymorphisms (SNPs) microarray in pooled DNA samples found 188 SNPs that were potentially associated with AD.20 In particular, the state-of-the-art GWAS is a powerful tool for identifying susceptibility alleles of complex diseases.24 However, currently a few GWAS or high-throughput association studies have been conducted for AD,19,20,22,23 leaving the majority of the genetic factors underlying AD remain unknown.
In this study, we conducted a GWAS in a sample of 904 unrelated Caucasian subjects using Affymetrix 500K SNP microarrays, for one commonly used AD phenotype- alcohol drinking
. Alcohol drinking information is obtained by self-reporting of the subjects. In Diagnostic Interview for Genetic Studies (DIGS),25
question 5 in section of ALCOHOL ABUSE AND DEPENDENCE is “Did you ever drink regularly—that is, at least once a week, for six months or more?” (http://zork.wustl.edu/nimh/home/m_DIGS_2.0.html/
Obviously, alcohol drinking is a necessary criterion for diagnosis of AD. For example, if a person does not drink any alcohol every week, he/she certainly does not have AD problems. Alcohol drinking is genetically determined because of its high heritability of 40% as calculated by our study of 6006 Caucasians from 1131 pedigrees (data not shown). It was also reported that alcohol drinking behavior and AD may have common genetic causes.26 The present GWAS results suggested the importance of the ankyrin repeat domain 7 (ANKRD7) and cytokine-like1 (CYTL1) genes in the determination of alcohol drinking; and this implicated their potential importance for AD risk.
Nine hundred and four unrelated Caucasian subjects were recruited from Omaha and its surrounding areas in Nebraska for initial GWAS. Only healthy people (defined by the exclusion criteria that are detailed below) were included in the analysis. The sample of 904 subjects for this alcohol drinking study was part of our initial GWAS sample comprising 1000 unrelated Caucasians for osteoporosis studies,27,28 of which 96 subjects were excluded because we do not have the alcohol drinking or age information for them. The other exclusion criteria were: subjects with certain conditions were excluded, including chronic disorders involving vital organs (heart, lung, liver, kidney, brain), serious metabolic diseases (diabetes, hypo- and hyper-parathyroidism, hyperthyroidism, etc.), skeletal diseases (Paget disease, osteogenesis imperfecta, rheumatoid arthritis, etc.), chronic use of drugs affecting bone metabolism (hormone replacement therapy, corticosteroid therapy, anti-convulsant drugs), malnutrition conditions (such as chronic diarrhea, chronic ulcerative colitis, etc.), and so forth. The study was approved by the Institutional Review Board at Creighton University. Informed-consent documents were obtained from all study participants before they entered the study.
In this study, “alcohol drinking” indicates whether a study subject reported regularly drinking at least once every week and has two values: 1 for yes and 0 for no. “Alcohol consumption” is graded as 0, 1, 2, 3 and 4, representing the number of alcohol drinking as 0, 1–2, 3–6, 7–10 and >10 times per week, respectively. Regular alcohol drinking in early ages was associated with later AD.29 Relevant characteristics of the study subjects were summarized in Table 1. The mean percentages of drinkers were 83.2% for males and 74.0% for females.
|view in a new window |
Table 1. Characteristics of the subjects of the initial Caucasian GWAS sample
Genomic DNA was extracted from human peripheral blood using a commercial isolation kit (Gentra systems, Minneapolis, MN, USA). Genotyping with the Affymetrix Mapping 250K Nsp1 and Affymetrix Mapping 250K Sty1 arrays was performed at the Vanderbilt Microarray Shared Resource at Vanderbilt University Medical Center, using the protocol recommended by the manufacturer. Briefly, genomic DNA was first digested with either Nsp1 or Sty1 and then ligated to adapters. One primer amplification was performed to amplify the ligation products in a polymerase chain reaction (PCR). The amplified DNA was purified and fragmented. Samples with fragment distributions were hybridized to the Affymetrix array. Arrays were stained, washed and scanned with the Affymetrix array scanner 30007G. Data management and analyses were performed using the Affymetrix GeneChip Operating System. Genotyping calls were determined from the fluorescent intensities using the DM algorithm with a 0.33 confidence score setting and the BRLMM algorithm. According to Affymetrix guidelines, DM calls were used for quality control. Specifically, subjects with DM call rates <93% were subject to re-genotyping. All 904 subjects passed this quality control criterion. The BRLMM algorithm was used for the association analyses due to its improved genotype calling. BRLMM clustering was performed with 94 samples per cluster. The final average BRLMM call rate across the entire sample was 99.14%. However, of the initial full-set of 500 568 SNPs, we discarded 32 961 SNPs whose sample call rates were less than 95%, 36 965 SNPs with allele frequencies deviating from Hardy-Weinberg equilibrium (HWE) (P <0.001) and 51 323 SNPs with minor allele frequencies (MAFs) <1%. Consequently, association analyses were performed using the final SNP set containing 379 319 SNPs.
To validate our GWAS results, we performed replication studies using three independent samples: (1) 1972 Caucasians in 593 nuclear families; (2) 761 unrelated Caucasian subjects; (3) 2955 unrelated Chinese. The selection criteria for replication include statistical significance, haplotype block structure and minor allele frequency. Specifically, ten SNPs of ANKRD7 and one SNP of CYTL1 showed most significant associations in initial GWAS. Together with the linkage disequilibrium (LD) situations of ANKRD7 SNPs, we chose four SNPs in ANKRD7 and one SNP in CYTL1 for the replication analyses.
Caucasian nuclear Families
A total of 1972 Caucasian subjects from 593 nuclear families living in Omaha, Nebraska were included in the analyses. The sample is composed of 831 parents (age range: 38.3–79.7 years), 246 male children (age range: 19.6–59.2 years) and 895 female children (age range: 19.0–61.1 years). The 40.3%, 35.9%, 18.5% and 5.3% of these nuclear families have 1, 2, 3 and more than 3 children, respectively, resulting in 824 sibling pairs and 1470 parent-offspring pairs. The overall alcohol drinking rates were 75.8% for females (age: (47.8±15.2) years) and 83.5% for males (age: (52.8±16.2) years), respectively. The genotyping concordance rate for duplicate samples was 99.7% through competitive allele specific PCR.30 Association tests were carried out by the use of family-based association test (FBAT) software.
Unrelated Caucasian subjects
A total of 761 unrelated Caucasian subjects previously explored for genetic determination of nicotine dependence,31 were used as the second replication sample. The mean age of these subjects at recruitment were (37.8±10.2) years, and the AD rate was 42.5%. The alcohol-related phenotype-maximum drinking was used for replication. Maximum number of drinks as an alcoholism phenotype has been used in the genetic dissection of AD.6 All of the SNPs were genotyped using TaqMan assay and FBAT was conducted to test for association.
A total of 2955 healthy, unrelated Chinese Han adults from Xi’an and Changsha cities in China were used as the third replication sample. The study was approved by the Institutional Review Boards of Xi’an Jiaotong University and Hunan Normal University, respectively, and all study participants provided written informed consent before being enrolled into the study. The case sample of alcohol drinking comprised 536 subjects, including 431 males aged (29.6±10.6) years and 105 females aged (32.3±14.2) years. The control sample consisted of 2419 subjects, including 1006 males aged (30.9±13.3) years and 1413 females aged (35.7±15.8) years. The mean percentages of drinkers were 30.0% for males and 7.0% for females. Exclusion and inclusion criteria were the same as those adopted in the initial Caucasian GWAS sample. Genotyping was performed using the MassARRAY system with MALDI-TOF mass spectrometry (Sequenom, Inc., San Diego, CA, USA).32 The concordance rate for duplicate samples was 99.8%. Allelic and genotypic association tests were conducted using HelixTree.
Allelic, genotypic and haplotypic association tests were performed using HelixTree 5.3.1 (Golden Helix, Bozeman, MT, USA). The haplotype trend regression (HTR) approach was used to test haplotype associations.33 The GWAS level significance threshold was set at 4.2×10-7 based on the criterion adopted by the previous GWAS researches.34 False discovery rate (FDR) was also applied to address the multiple testing problems. Haploview (http://www.broad.mit.edu/mpg/haploview/) was used to analyze and plot LD patterns.35 MAPPER (http://mapper.chip.org) was used to search for transcript factor binding sites.36 In addition, EIGENSTRAT was used to minimize spurious associations due to population stratification,37 though no evidence of population stratification was found.
The z transformation method was applied to combine P values from different samples for meta-analysis.38 Each P value is transformed into a z score, a normal deviate via the normal probability transformation. The sum has a normal distribution with N(0, n), thus, has a normal N(0, 1) distribution. The combined P value is computed as the normal distribution function associated with .
Initial GWAS results
In the initial GWAS, ten SNPs in the ANKRD7 gene were most significantly associated with alcohol drinking (FDR q values <0.05). These SNPs still showed highly suggestive associations with alcohol drinking adjusted by the significant covariates of sex and age (Table 2). Consistent results were obtained in HTR (haplotype trend regression) analyses for the 5-SNP haplotype windows consisting of ANKRD7 SNPs rs7812233-rs4727889- rs6466686-rs4295599-rs12531086 (P=2.47×10–7, FDR q=0.02) and rs4320489-rs12534529-rs10230384- rs10271557-rs11487118 (P=4.76×10–7, FDR q=0.02) (Figure 1). In addition, the haplotype block rs6466686-rs4295599-rs12531086 within ANKRD7 (Block 9 in Figure 1) was highly significantly associated with alcohol drinking (P=6.51×10–8).
|view in a new window |
Table 2. Associations between ANKRD7 SNPs and AD phenotypes
|view in a new window |
. Associations of ANKRD7
gene with alcohol drinking in the total initial Caucasian sample. A:
Assocoation results. The y
-axis showed the -log10 P
values and x
-axis showed the identity of 31 SNPs in three haplotype blocks which contained significant SNP association with alcohol drinking. Significant haplotypes were plotted with the solid lines. B:
Linkage disequilibrium patter
n of ANKRD7
gene in the three haplotype blocks. The linkage disequilibrium extent was denoted using the pairwise r2
Possible functional significance of the most promising SNPs in ANKRD7 was predicted through the analysis of the transcript factor binding sites. The results suggested that the change of alleles at certain SNP loci can alter the binding affinity of some transcriptional factors such as REVERB-alpha and SRY (Table 3). This may lead to the altered expression of ANKRD7 gene that could be associated with the variance in alcohol drinking behavior.
|view in a new window |
Table 3. Analysis of the transcript factor binding sites (52 bp) around different alleles at the interested SNP loci of ANKRD7
For CYTL1, the 5-SNP haplotype window consisting of rs16836480-rs11728854-rs16836497-rs11732156- rs11723786 showed the most significant association with alcohol drinking (P=4.42×10–7, FDR q=0.02) (Figure 2). In allelic and genotypic association tests, rs16836497 in CYTL1 demonstrated significant associations with alcohol drinking (P = 1.08E-06 and 3.36E-06, respectively). In our genotyped SNP set, rs16836497 was an independent SNP that was not in strong LD with other CYTL1 SNPs (Figure 2).
|view in a new window |
Figure 2. Associations of CYTL1 with alcohol drinking in the initial Caucasian sample. A: Assocoation results. The y-axis showed the -log10 P values and x-axis showed the identity of SNPs. B: Linkage disequilibrium pattern of the CYTL1 gene. The linkage disequilibrium extent was denoted using the pairwise r2 values.
Replication studies were conducted in the three independent samples. The threshold for judging significant replications was set at P ≤0.05. Interestingly, the ANKRD7 SNP rs4295599 and the CYTL1 SNP rs16836497 showed the most significant associations with alcohol consumption in the replication samples. The combined P values from meta-analyses reached 5.79×10–5 and 7.46×10–5 for those two SNPs, respectively (Table 4). However, replication studies did not reveal the significance for alcohol drinking trait. Figure 3 shows the alcohol drinking rates stratified by the genotypes of the replicated SNPs in the ANKRD7 and CYTL1 genes, which showed significant differences (P <0.05).
|view in a new window |
Table 4. Replication results of five interested SNPs in ANKRD7 and CYTL1 for alcohol consumption trait
|view in a new window |
Figure 3. Alcohol drinking rates stratified by the genotypes of the significant SNPs in the ANKRD7 and CYTL1 genes in the total initial Caucasian sample.
GWAS is beginning to reveal genes contributing to the risk of AD.19-23 In our GWAS using Affymetrix 500K SNP chips, we identified two genes – ANKRD7 and CYTL1 that were significantly associated with an important phenotype for AD: alcohol drinking behavior, which was reported to share the common genetic components with AD.26 In particular, the significant associations of ANKRD7 and CYTL1 genes with alcohol drinking were successfully replicated in the other independent Caucasian and Chinese samples for alcohol consumption, suggesting the validity of our initial GWAS findings. It should be noticed that Caucasians have much higher alcohol drinking rate than Chinese, partially due to the different polymorphisms of the alcohol dehydrogenase (ADH) and acetaldehyde dehydrogenase (ALDH) genes in different races.
ANKRD7 gene is located on chromosome 7q31. In the Collaborative Study on the Genetics of Alcoholism (COGA) sample, the 7q31 region containing ANKRD7 was significantly linked to AD and the other AD related phenotypes.39 In another genome-wide scan from the Framingham Heart Study (FHS), sib-pair regression analysis provided nominal evidence for linkage of alcohol consumption to the regions of 7q21, 7q32 and 7q36 (D7S2204, P=0.0027; D7S1804, P=0.0049; and D7S559, P=0.001, respectively)40 that are very close to 7q31. These lend support to our detected significant associations between ANKRD7 and alcohol drinking.
Functional studies of ANKRD7 provide additional evidence supporting its importance to AD. First, ANKRD7 was differentially expressed in cerebella between the inbred long-sleep (ILS) mice and the inbred short-sleep (ISS) mice.41 In general, ILS mice are more sensitive to ethanol than ISS mice, and the expression level of ANKRD7 in the cerebella of ISS mice was much higher than that of ILS mice (P ≤0.05) (http://www.ncbi.nlm.nih.gov/projects/geo/). Because of the high conservation in the sequence of the ANKRD7 gene between humans and mice, the differential expression of ANKRD7 in mice with discordant alcohol-related traits may also support the importance of this gene to alcohol-related traits in humans. Furthermore, it was reported that the ANKK1 (ankyrin repeat and kinase domain containing 1) gene was significantly associated with AD.42,43 Since ANKK1 protein is similar to ANKRD7 protein in structure (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi/), it was logical to deduce that ANKRD7 was involved in the etiology of AD in a similar way.
We also explored the potential functional changes that may be caused by different alleles of ANKRD7 SNPs (Table 3). For example, one REVERB-alpha binding site around rs7812233 in ANKRD7 will be removed for the “A” allele. REVERB-alpha is encoded by nuclear receptor subfamily 1 group D member 1 (NR1D1) gene, which is related to the function of vitamin D3, thyroid and sex hormones that is important to alcohol metabolism. The deletion of the binding site of REVERB-alpha may affect the expression of the ANKRD7 gene at the transcriptional level, which could result in the variance in alcohol drinking and AD risk.
CYTL1 in 4p16 is a cartilage-related gene.44 Both our initial GWAS and the follow-up replication results supported the association of CYTL1 with alcohol consumption. Previously, it was also reported that 4p15.3 was linked to the maximum alcohol consumption.40 The linkage of 4p16 to psychiatric disorders has also been reported, such as its significant linkages to bipolar disorder (BPD)45 and schizophrenia.46 Since AD is correlated with the psychiatric traits,47 the above linkages indirectly support the existence of some genetic factors in 4p16 that may contribute to AD phenotypes. This is consistent with our data showing that CYTL1 should be among the list of AD candidate genes in the 4p16 region.
In summary, our GWAS and replication results provide the evidences supporting that ANKRD7 and CYTL1 are significantly associated with alcohol drinking behavior, which is an important phenotype of AD. Replication by other investigators and the elucidation of the underlying molecular mechanisms are needed to verify our findings.
Institute contributions: Hunan Normal University: Research design, organization, manuscript drafting and recruitment of Chinese samples in Changsha; Xi’an Jiaotong University: Recruitment of Chinese samples in Xi’an; University of Missouri-Kansas City: Genotyping for Caucasian nuclear families samples, statistical analyses and critical revising of manuscript; Vanderbilt University Medical Center: Genotyping of initial Caucasian samples; University of Connecticut Health Center: Direction of alcohol dependence study and manuscript revising; Boston University School of Medicine: Genotyping for replication in unrelated Caucasians; Yale University School of Medicine: Replication research in unrelated Caucasians; Creighton University: Recruitment of initial unrelated Caucasian and nuclear family samples; University of Shanghai for Science and Technology: Genotyping of Chinese population.
Hasin D, Grant B. 1994 draft DSM-IV criteria for alcohol use disorders: comparison to DSM-III-R and implications. Alcohol Clin Exp Res 1994; 18: 1348-1353.
Prescott CA, Kendler KS. Genetic and environmental contributions to alcohol abuse and dependence in a population-based sample of male twins. Am J Psychiatry 1999; 156: 34-40.
3. Grant BF,
Dawson DA, Stinson FS, Chou SP, Dufour MC, Pickering RP. The 12-month prevalence and trends in DSM-IV alcohol abuse and dependence: United States, 1991–1992 and 2001–2002. Drug Alcohol Depend 2004; 74: 223-234.
Kendler KS, Prescott CA, Jacobson K, Myers J, Neale MC. The joint analysis of personal interview and family history diagnoses: evidence for validity of diagnosis and increased heritability estimates. Psychol Med 2002; 32: 829-842.
5. Dick DM,
Plunkett J, Wetherill LF, Xuei X, Goate A, Hesselbrock V, et al. Association between GABRA1 and drinking behaviors in the collaborative study on the genetics of alcoholism sample. Alcohol Clin Exp Res 2006; 30: 1101-1110.
6. Saccone NL, Kwon JM,
Corbett J, Goate A, Rochberg N, Edenberg HJ, et al. A genome screen of maximum number of drinks as an alcoholism phenotype. Am J Med Genet 2000; 96: 632-637.
9. Wilhelmsen KC, Schuckit M, Smith TL,
Lee JV, Segall SK, Feiler HS, et al. The search for genes related to a low-level response to alcohol determined by alcohol challenges. Alcohol Clin Exp Res 2003; 27: 1041-1047.
Prescott CA, Sullivan PF, Kuo PH, Webb BT, Vittum J, Patterson DG, et al. Genomewide linkage study in the Irish affected sib pair study of alcohol dependence: evidence for a susceptibility region for symptoms of alcohol dependence on chromosome 4. Mol Psychiatry 2006; 11: 603-611.
Williams AH, Brown WM, Langefeld CD. Multilocus and interaction-based genome scan for alcoholism risk factors in Caucasian Americans: the COGA study. BMC Genet 2005; 6 Suppl 1: S37.
13. Long JC, Knowler WC,
Hanson RL, Robin RW, Urbanek M, Moore E, et al. Evidence for genetic linkage to alcohol dependence on chromosomes 4 and 11 from an autosome-wide scan in an American Indian population. Am J Med Genet 1998; 81: 216-221.
15. Dick DM, Nurnberger J Jr,
Edenberg HJ, Goate A, Crowe R, Rice J, et al. Suggestive linkage on chromosome 1 for a quantitative alcohol-related phenotype. Alcohol Clin Exp Res 2002; 26: 1453-1460.
D’Alfonso S. Nonparametric linkage analysis of alcohol dependence with chromosome 1 and 7 markers. Genet Epidemiol 1999; 17 Suppl 1: S127-S131.
Foroud T, Bucholz KK, Edenberg HJ, Goate A, Neuman RJ, Porjesz B, et al. Linkage of an alcoholism-related severity phenotype to chromosome 16. Alcohol Clin Exp Res 1998; 22: 2035-2042.
Gelernter J, Kranzler HR, Panhuysen C, Weiss RD, Brady K, Poling J, et al. Dense genomewide linkage scan for alcohol dependence in African Americans: significant linkage on chromosome 10. Biol Psychiatry 2009; 65: 111-115.
19. Liu QR,
Drgon T, Johnson C, Walther D, Hess J, Uhl GR. Addiction molecular genetics: 639 401 SNP whole genome association identifies many “cell adhesion” genes. Am J Med Genet B Neuropsychiatr Genet 2006; 141B: 918-925.
20. Johnson C,
Drgon T, Liu QR, Walther D, Edenberg H, Rice J, et al. Pooled association genome scanning for alcohol dependence using 104 268 SNPs: validation and use to identify alcoholism vulnerability loci in unrelated individuals from the collaborative study on the genetics of alcoholism. Am J Med Genet B Neuropsychiatr Genet 2006; 141B: 844-853.
Zeggini E, Weedon MN, Lindgren CM, Frayling TM, Elliott KS, Lango H, et al. Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes. Science 2007; 316: 1336-1341.
25. Berney A,
Preisig M, Matthey ML, Ferrero F, Fenton BT. Diagnostic interview for genetic studies (DIGS): inter-rater and test-retest reliability of alcohol and drug diagnoses. Drug Alcohol Depend 2002; 65: 149-158.
28. Xiong DH,
Liu XG, Guo YF, Tan LJ, Wang L, Sha BY, et al. Genome-wide association and follow-up replication studies identified ADAMTS18 and TGFBR3 as bone mass candidate genes in different ethnic groups. Am J Hum Genet 2009; 84: 388-398.
McClay JL, Sugden K, Koch HG, Higuchi S, Craig IW. High-throughput single-nucleotide polymorphism genotyping by fluorescent competitive allele-specific polymerase chain reaction (SNiPTag). Anal Biochem 2002; 301: 200-206.
Braun A, Roth R, McGinniss MJ. Technology challenges in screening single gene disorders. Eur J Pediatr 2003; 162 Suppl 1: S13-S16.
Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005; 21: 263-265.
37. Price AL,
Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet 2006; 38: 904-909.
39. Reich T,
Edenberg HJ, Goate A, Williams JT, Rice JP, Van Eerdewegh P, et al. Genome-wide search for genes affecting the risk for alcohol dependence. Am J Med Genet 1998; 81: 207-215.
Bergen AW, Yang XR, Bai Y, Beerman MB, Goldstein AM, Goldin LR. Genomic regions linked to alcohol consumption in the Framingham Heart Study. BMC Genet 2003; 4 Suppl 1: S101.
Kim JS, Ryoo ZY, Chun JS. Cytokine-like 1 (Cytl1) regulates the chondrogenesis of mesenchymal cells. J Biol Chem 2007; 282: 29359-29367.
Ewald H, Flint T, Kruse TA, Mors O. A genome-wide scan shows significant linkage between bipolar disorder and chromosome 12q24.3 and suggestive linkage to chromosomes 1p22-21, 4p16, 6q14-22, 10q26 and 16p13.3. Mol Psychiatry 2002; 7: 734-744.