Depression, alcohol abuse or dependence (AUD), as well as other affective disorders and substance use disorders (SUD), co-occur in the general population more frequently than expected by chance 12. Approximately 80% of individuals with AUD report symptoms of depression and 25-40% of the people suffering from depression also report drinking problems 3. The comorbidity of depression and AUD complicates the treatment and can alter the prognosis 34. Furthermore, both depression and AUD increase the risk of suicide. Hence, having both depression and AUD is more severe than having just one of the disorders and it often leads to greater impairment 56.

A number of hypotheses have been proposed to explain the comorbidity between depression and AUD and answer to whether we are dealing with one or two independent and overlapping disorders. The comorbidity could be due to shared risk factors or highly correlated risk factors 1. Also, some symptoms of AUD overlap with some common symptoms in depression, such as sadness and sleep disturbances 3. It has also been discussed whether the co-occurrence could be the result of one of the disorders increasing the risk for or even aggravating the other disorder 13. Alcohol dependent individuals are possible at higher risk of developing depression, as a consequence of the associated interpersonal and social problems often caused by alcohol dependence 1. On the contrary, the substance induced mood disorder theory advocates that depressed persons are more vulnerable to develop an addiction/abuse. Related to this latter assumption is the self medication theory, where the depressed individual tries to self-medicate with alcohol 123.

Depression and AUD are both complex disorders meaning that both genetic and environmental risk factors have an influential role, with the interplay between genes of modest effect with several environmental risk factors contributing to disease susceptibility. Results from several studies indicate that both environmental and genetic risk factors partly overlap between depression and AUD suggesting a common etiology 78. An epidemiological study by Prescott and colleagues on depression and alcoholism conclude that the causes overlap between the disorders, though without having the same origin and they estimated that the shared overlap of genetic and environmental factors influencing depression and AUD was only 9-14% 4. Nevertheless, several genes have been proposed to be involved in the etiology of both depression and AUD, exemplified by brain-derived neurotrophic factor (BDNF), neuropeptide Y (NPY), dopamine receptor D2 (DRD2), catechol-O-methyltransferase (COMT), monoamine oxidase A (MAOA), period homolog 2 (PER2) and several subtypes of the serotonin receptor 89101112. Only few studies have investigated the genetic component of the co-occurrence between depression and AUD. McEachin et al. investigated the comorbidity between depression and AUD in silico, by modeling gene-by-environment interactions using bioinformatics and identified tumor necrosis factor (TNF) and methylenetetrahydrofolate reductase (MTHFR) as candidate genes. TNF is involved in the pathway activating MTHFR expression and excessive alcohol intake leads to reduced TNF signaling. MTHFR is a key component of the folate metabolism and previously folate levels have been associated to both depression and AUD 7. In addition, linkage to chromosome 1p13-35 locus and alcoholism or depression has been found, the region containing several genes includes two genes coding for potassium channel-related proteins 13.

Alcohol is known to influence and alter the circadian rhythm and it may even act on the central pacemaker located in the suprachiasmatic nucleus (SCN) 14. Studies also show that alcohol preference and sensitivity vary along with the circadian oscillation 1516. Studies show that rats and other rodents have a preference for alcohol during their active phase (dark-phase) 15. Drug-induced changes of gene expression have been reported for several clock genes and the CLOCK:ARNTL transcription activity was increased in in vivo experiments when stimulating the dopamine D2 receptor 16. The Period (Per) genes in rats have an decreased circadian expression pattern in SCN and various other brain areas after alcohol intake 17. Spanagel and colleagues found that a haplotype in the PER2 gene associated with high (>300 g/day) versus low (<300 g/day) alcohol intake, though it was not associated with alcohol dependence 18. Also, both non-seasonal and seasonal unipolar depressive and bipolar disorders and certain sleep disorders are associated with an abnormal circadian rhythm and display symptoms like disturbed sleep-wake cycle and appetite, as well as abnormal physical functions including changes in temperature and various hormonal levels 192021.

Our aim here was to investigate whether certain genetic variations in the circadian clock system are associated with comorbidity between depression and AUD.

Allele and genotype frequencies of SNPs in circadian genes (Table 4) were tested for association with depression and AUD comorbidity as compared to healthy controls (Sample set 1, Table 2). For SNPs with nominal allele or genotype association (P < 0.05), genetic association analyses with AUD versus healthy controls (Sample set 2) and comorbidity versus AUD and controls (Sample set 3) were performed to determine the nature of the association to the comorbidity. The results are presented in Table 5.

In the CLOCK gene the minor allele G of rs11240 was suggestively associated with depression and AUD comorbidity and showed an OR of 1.65 (P = 0.0077). This was further strengthened by the results from the recessive model (GG vs GC and CC, P = 0.013) and trend test (P = 0.0077). Rs11240 showed no associations to AUD in Sample set 2. Accordingly, in Sample set 3 the allele G was suggestively associated with risk for depression (OR = 1.59, P = 0.0084) and the Cochran-Armitage test suggested a trend (P = 0.0082).

One LD block was formed in CLOCK (rs3805151, rs2412648, rs11240, rs2412646), spanning a 18-kb region, using Sample set 1 data. The haplotype TTGC, including the rs11240 risk allele G, suggestively conferred a risk for comorbidity (OR = 1.65, P = 0.0077) (Table 6). This TTGC haplotype was suggestively associated also with an increased risk for depression in Sample set 3 (OR = 1.50, P = 0.0084). Three additional haplotypes were formed, however none of then nominally associated, TTCT, CGCC and CTCC with overall frequencies of 26%, 36% and 3.8% respectively. To elucidate TT alleles role in the two haplotypes, analysis was performed with only the last two SNPs in the haplotype (rs11240 and rs2412646). The p-values for the risk haplotypes did not change between the TTGC and the GC haplotypes in Sample set 1 (Table 6) indicating location of the functional variation downstream of rs2412648.

The rs2306074 in ARNTL2 showed a border-line nominal Cochran-Armitage trend in Sample set 1 (P = 0.043), and no indication for association in Sample sets 2 and 3. In ACADS, the A allele of rs1799958 showed a border-line nominal allelic association of increased risk for comorbidity (OR = 1.47, P = 0.045) and a trend test supported the finding (P = 0.045). Likewise, in Sample set 3 a very modest nominal allelic association was found for rs1799958 (P = 0.046). No haplotype could be constructed for ARNTL2 or ACADS that included the SNPs nominally associated with comorbidity.

To test for a quantitative effect of the CLOCK rs11240, ACADS rs1799958 and ARNTL2 rs2306074 on AUD depression comorbidity, these variations were tested for association to GSS, BDI, GHQ and MBI among the comorbid cases only. The A allele of rs1799958 in ACADS was nominally associated with higher GHQ score among cases (P = 0.031).

Our results herein suggest that the circadian gene CLOCK is associated with comorbid depression and AUD, but not with AUD only. The haplotype TTGC formed by SNPs rs3805151, rs2412648, rs11240, rs2412646 was suggestively associated with increased risk for the comorbidity, with the odds ratio of 1.65. The SNPs of importance for this suggestive association were rs11240 and rs2412646, indicating location of the functional variation downstream of rs2412648. No indication of association with CLOCK was found when comparing AUD with healthy controls. Accordingly, the suggestive association to CLOCK was seen when comparing comorbid cases with combined group of healthy controls and persons diagnosed with AUD. This CLOCK variation may be a vulnerability factor for depression given the alcohol exposure in AUD but not considerably increasing the risk for depression without AUD. This view is supported by the findings from other studies of the Finnish general population through the Health 2000 Study. They could not detect any CLOCK association with major depressive disorder or dysthymia 33 or anxiety disorders (Sipilä et al., submitted 2009), each using a disorder focused set of samples inclusive of all the cases. These studies analyzed rs10462028 and rs1801260 33, and rs3749474 and rs1801260 (Sipilä et al., submitted 2009) that are in high LD with rs11240 analyzed in the current study (according to HapMap public release) 34. Neither could we detect any indication that CLOCK variation was associated with AUD only. In agreement, the shared overlap of genetic and environmental factors influencing depression and AUD was estimated to only 9-14% out of which 50-60% was attributed to shared genetic factors. Also, Prescott and colleagues found no support that comorbidity arises from depression causing alcoholism or alcoholism causing depression using structural modeling in twins 4.

The CLOCK gene is one of the most central genes in the circadian system and has been studied in a wide range of areas due to its crucial role in creating and maintaining the body's internal rhythm. CLOCK protein exhibits a regulatory role as transcription factor over other circadian genes, like the CRYs and PERs, together with ARNTL or ARNTL2 protein. The CRY and PER complexes exhibit a regulatory role as repressors and inhibit the transcription of CLOCK and ARNTL and thereby themselves, when reaching a critical concentration. This transcription-translation feedback loop takes approximately 24 hours 2135.

Individuals with depression or AUD often have circadian misalignment and many physiological phenomena such as the sleep-wake cycle and hormonal profiles are disrupted 36. Sleep disturbances are also pronounced symptoms of a wide range of circadian rhythm disorders such as familial advanced sleep-phase syndrome (FASPS), delayed sleep phase syndrome (DSPS), as well as other psychiatric disorders like seasonal and non-seasonal mood disorders like, bipolar, schizophrenia as well as in drug addictions 3738. Furthermore, sleep deprivation and light therapy have an antidepressant effect synchronizing the sleep-wake cycle with the circadian rhythms, indicating the important role that the circadian system plays in many psychiatric disorders 1939.

Individuals addicted to alcohol show circadian alterations, for example sleep disturbances 36. As previously mentioned, alcohol has the ability to induce clock gene expression in different brain areas 18. Ruby et al. showed evidence that ethanol significantly affects photic and non-photic phase-resetting responses in hamsters, critical for circadian regulation, by blocking the phase-resetting action of glutamate and increase the non-photic phase-resetting action of serotonin. This signal inhibition from ethanol was manifested through direct action in the core clock in SCN 40. The preference and sensitivity to alcohol also seems to vary with the time of day 15.

Clinical effect of variations in CLOCK has been quite extensively studied. The SNP rs1801260 in CLOCK was shown to influence diurnal preference in healthy individuals, where C allele carriers had a higher evening preference. The same allele was also associated with the delay in sleep phase and insomnia in major depression and bipolar affective disorder patients 4142. Later, Serretti and colleagues also showed that the C/C genotype of rs1801260 was associate with the severity of insomnia in depressed and bipolar patients during SSRIs treatment 43. SSRIs have earlier been reported to have circadian properties, inducing phase shift in neuronal firing in the SCN in rats 44. The SNPs rs1801260 and rs11240 analyzed in the current study are probably reflecting association to the same functional polymorphism, as they are in LD with each other and display the same minor allele frequency (according to HapMap public release and NCBI Entrez dbSNP) 3445. Alcohol and other drugs of abuse modulate the dopamine neurotransmission, and McClung and colleagues showed that the Clock gene seems to increase the excitability of dopamine neurons and also the cocaine reward in mice with a dominant negative CLOCK protein that cannot activate transcription 46. These Clock-mutant mice also display alcohol preference 47. The CLOCK gene has also been proposed to be involved in the metabolic syndrome, which involves obesity and increased the risk for diabetes and heart disease 48.

Weak suggestive association to comorbidity was found for rs2306074 in ARNTL2 and rs1799958 in ACADS. A family study by Shi et al. found weak association between rapid cycling and a diurnal mood, with worse symptoms in the afternoon or evening, in bipolar subjects and SNPs in ARNTL2 49. There is also some association of ARNTL2 with social phobia (Sipilä et al., submitted 2009). Interestingly, there is a functional partnership between ARNTL2 and PER2 50 that might bridge social phobia and alcohol use 51 to end in high alcohol intake. Support for involvement of ARNTL2 in seasonal affective disorder (SAD), a subtype of mood disorder that is closely related to AUD 52, has been reported by our group, where a SNP association was seen in both Swedish and Finnish materials (Sjöholm et al., submitted 2009). In addition, ARNTL has been reported by our group to show associations with depression in a Swedish population-based and case-control material 1153. Moreover, NPAS2 is indicated to associate with SAD 5455 and NPAS2 and ARNTL or ARNTL2 heterodimerize and possess a transcriptional modulation function as the CLOCK/ARNTL complex 35.

ACADS, acyl-Coenzyme A dehydrogenase, C-2 to C-3 short chain, is an enzyme participating in the fatty acid β-oxidation and has not previously been reported to associate with mood disorder or AUD. Deficiency in ACADS leads to changes in theta oscillations during rapid eye movement (REM) sleep in mice 56, and in the majority of depressed patients disturbed sleep architecture is characterized by abnormal timing and distribution of REM and non-REM sleep stages give feedback to the SCN 57. Furthermore, there are abnormal long-range temporal correlations in theta oscillations during wakefulness 58 and profound REM sleep abnormalities in patients with non-seasonal depressive disorder that have melancholic depressive symptoms 1439. On the other hand, non-REM sleep abnormalities, such as abnormal cross-correlations between facial temperatures and delta and theta frequencies, are found in patients with SAD that have atypical depressive symptoms 59. The rs1799958 SNP (G>A) in ACADS results in the conversion of glycine to serine and associates with the short chain acyl-CoA dehydrogenase deficiency 60 that is characterized by lipid storage myopathy and muscle weakness.

An advantage of this study is that the individuals are derived from a big Finnish population based study of an ethnically homogenous population that is nationwide and representative of the general population aged 30 or over. We were able to investigate whether our results found with the comorbid versus control (without psychiatric symptoms), Sample set 1, reflected genetic vulnerability to AUD. A limitation in our study is the small size of the comorbid sample and the lack of a group of patients having depressive disorder only. For now, replication of the findings in independent study samples is needed as the most practical way to increase the probability of a true association.

The comorbid condition of alcohol use and depressive disorders in the Finnish population was associated with CLOCK genetic variations and there was no indication for CLOCK gene association with AUD only. This finding together with previous reports indicates that the CLOCK variations we found here may be a vulnerability factor for depression given the exposure to alcohol in individuals having AUD.

The authors declare that they have no competing interests.

All the authors designed the study and the analysis. LK performed the genotyping. LS performed the statistical analysis. SS, MS, CL and TP as seniors guided the work. LS wrote the first draft of the manuscript, and the remaining authors reviewed the manuscript. Thus, all the authors contributed to and have approved the final manuscript.