Genomewide Association Study of Alcohol Dependence Identifies Risk Loci Altering Ethanol-Response Behaviors in Model Organisms

Abstract

Background: Alcohol dependence (AD) shows evidence for genetic liability, but genes influencing risk remain largely unidentified.

Methods: We conducted a genomewide association study in 706 related AD cases and 1,748 unscreened population controls from Ireland. We sought replication in 15,496 samples of European descent. We used model organisms (MOs) to assess the role of orthologous genes in ethanol (EtOH)-response behaviors. We tested 1 primate-specific gene for expression differences in case/control postmortem brain tissue.

Results: We detected significant association in COL6A3 and suggestive association in 2 previously implicated loci, KLF12 and RYR3. None of these signals are significant in replication. A suggestive signal in the long noncoding RNA LOC339975 is significant in case:control meta-analysis, but not in a population sample. Knockdown of a COL6A3 ortholog in Caenorhabditis elegans reduced EtOH sensitivity. Col6a3 expression correlated with handling-induced convulsions in mice. Loss of function of the KLF12 ortholog in C. elegans impaired development of acute functional tolerance (AFT). Klf12 expression correlated with locomotor activation following EtOH injection in mice. Loss of function of the RYR3 ortholog reduced EtOH sensitivity in C. elegans and rapid tolerance in Drosophila. The ryanodine receptor antagonist dantrolene reduced motivation to self-administer EtOH in rats. Expression of LOC339975 does not differ between cases and controls but is reduced in carriers of the associated rs11726136 allele in nucleus accumbens (NAc).

Conclusions: We detect association between AD and COL6A3, KLF12, RYR3, and LOC339975. Despite nonreplication of COL6A3, KLF12, and RYR3 signals, orthologs of these genes influence behavioral response to EtOH in MOs, suggesting potential involvement in human EtOH response and AD liability. The associated LOC339975 allele may influence gene expression in human NAc. Although the functions of long noncoding RNAs are poorly understood, there is mounting evidence implicating these genes in multiple brain functions and disorders.

Keywords: COL6A3; KLF12; LOC339975; RYR3; Alcohol Dependence.

Figure 1. Manhattan plot for case/control analysis of Alcohol Dependence (AD)

Horizontal red line indicates genome-wide significance (5 × 10⁻⁸).

Figure 2. Q-Q plot for final post-imputation dataset after all quality control (QC) and exclusions

Gray shading indicates 95% confidence interval for expected values. The post-imputation lambda (1.046) and sample size-standardized lambda1000 (1.045) indicate there is little inflation of test statistics.

Figure 3. LocusZoom plots of COL6A3, RYR3, KLF12 and LOC339975

Genomewide significant results in COL6A3 (A) and three other regions of suggestive association supported by additional data, RYR3 (B), KLF12 (C) and LOC339975 (D).

Figure 4. Mutations in C. elegans orthologs of human candidate genes cause ethanol response phenotypes

Speed of locomotion was measured at 10 and 30 minutes, and expressed as a percent of the untreated control speed. Control worms were tested simultaneously on the same plates. Worms were treated with 400 mM exogenous ethanol. The waxy cuticle of worms excludes most of the exogenous ethanol, tissue concentrations are approximately 12% of the exogenous dose (~48 mM). A, B, C: RNAi-induced gene knockdown reduced sensitivity to ethanol for C16E9.1 (a COL6A3 ortholog) but not for two other orthologous genes relative to untreated worms. D: Loss-of-function of the KLF12 ortholog, klf-3, prevented the development of acute functional tolerance between the 10- and 30-minute time points relative to wild-type N2. E, F: unc-68 mutant animals demonstrate reduced sensitivity to ethanol relative to wild-type N2. Statistical significance is shown for 2-way ANOVA followed by post-hoc comparisons across genotypes (*, p<0.05, **, p<0.01, ***, p<0.001).

Figure 5. Bioinformatic analysis of correlation of Col6a3, Klf12 or Ryr3 expression with ethanol behaviors in C57BL/6J × DBA/2J recombinant inbred (BXD) mouse lines

The GeneNetwork (GN) web-based analysis platform was used to identify correlations between between basal expression levels of Col6a3, Klf12 or Ryr3 and ethanol behaviors. Panels A-D display Spearman ranked order correlations between microarray gene expression (x-axis) and behavioral phenotypes (y-axis). Col6a3 (probeset 1424131_at) in GN113 whole brain expression dataset showed significant positive correlation with total handling induced convulsion (HIC) score (sum of baseline subtracted HIC at 4, 6 and 7 hrs) after 4 g/kg intraperitoneal (IP) ethanol in males (GN record 11382; Panel A) and negative correlation with ethanol 2-bottle choice voluntary consumption (GN record 10479; Panel B). Klf12 basal expression in prefrontal cortex (GN135, probeset 1455521_at) was significantly positively correlated with locomotor activity 0–5 minutes after 2.25 g/kg IP ethanol (GN record 11708; Panel C) and in nucleus accumbens (GN156, probeset 1439827_s_at) was positively correlated with ethanol acute functional tolerance (GN record 10348; Panel D).

Figure 6. Expression of long noncoding RNA LOC339975 in human post-mortem nucleus accumbens (NAc) and dorsolateral pre-frontal cortex (PFC) stratified by diagnostic status and by genotype

Control mean RIN was 5.6 (SD=1.8), case mean RIN was 5.8 (SD=1.5), and did not differ between cases and controls (Mann-Whitney p=0.78). Samples that did not amplify (NAc, N=13; PFC, N=24) and 4 samples missing rs11726136 genotypes were excluded. Case:control comparisons of LOC339975 expression in A. 34 case and 35 control NAc samples, and B. 28 case and 30 control PFC samples. There were no significant differences in expression between diagnostic groups. Genotypic comparisons of LOC339975 expression in C. 58 reference (T/T) homozygote and 7 T/G heterozygote NAc samples, and D. 50 T/T homozygote and 5 T/G heterozygote PFC samples. While no significant differences in expression by genotype were observed in the PFC, NAc expression was significantly reduced in carriers of the associated non-reference G allele compared to reference allele homozygotes (F=9.72, p=0.003).