A population-based association study of candidate genes for depression and sleep disturbance^†

INTRODUCTION

Depressive disorder is a common condition, affecting 6–7% of the population annually and 16% over the lifetime [Kessler et al., 2003; Pirkola et al., 2005]. Several factors, such as age, gender, and genetic background, affect the prevalence of the disease. The disorder is more common in women than in men; in the Finnish population, the respective 12-month prevalence rates are 8.3% and 4.6% [Pirkola et al., 2005]. There is a strong genetic contribution, with an estimated heritability of 40–50% [Bierut et al., 1999; Sullivan et al., 2000; Kendler et al., 2001]. In a study of 42,161 twins, heritability of major depressive disorder (MDD) was clearly higher in women (42%) than in men (29%), and some genetic risk factors were suggested to be sex-specific [Kendler et al., 2006]. Recently, we obtained similar findings for life dissatisfaction, depressed mood, and depressive disorder in a Finnish sample of 18,631 same-sex twins, with additive heritability estimates of up to 53% in females and 39% in males [Paunio et al., 2009].

Findings in molecular genetic studies of MDD vary in different study samples and populations. The most convincing evidence is for the serotonin transporter gene (SERT or SLC6A4) and stressful life events leading to depression [Uher and McGuffin, 2008]. Findings for other genes, including those encoding for the rate-limiting enzyme for serotonin synthesis tryptophan hydroxylase 2 (TPH2) [Zill et al., 2004], the brain-derived neurotrophic factor (BDNF) [Hashimoto et al., 2004], the enzyme responsible for the synthesis of gamma-aminobutyric acid (GABA) and the glutamic acid decarboxylase (GAD1) [Hettema et al., 2006], have been more controversial.

Of the molecular mechanisms that mediate depression, the serotonergic system is one of the best characterized, and it is the major target for antidepressive treatment [Carranza Gomez-Carpintero and Ochoa Estomba, 1993]. In addition, glutamatergic neuronal transmission, controlling for plasticity and synaptic excitability in limbic pathways, has been implicated in clinical depression [Witkin et al., 2007]. Multiple lines of evidence have also implicated for the role of neuroplasticity in depression, and adaptive changes in neuroplasticity are likely to be involved in therapeutic effect of antidepressant medication [Manji et al., 2001; Castren, 2005]. Episodes of depression are often triggered by stress and traumatic life events, activating the hypothalamic–pituitary–adrenal (HPA) axis [Checkley, 1996]. An individual's sensitivity for traumatic life events depends on environmental factors and genetic background [Caspi et al., 2003]. Thus, the molecules of the stress-regulating HPA axis are expected to be among the probable mediators of depression.

The majority of depressed patients suffer from sleep difficulties and have characteristic changes in their sleep architecture [Benca et al., 1997; Tsuno et al., 2005]. Sleep problems in depression include difficulty with sleep initiation, reduced rapid eye movement (REM) latency, decreased slow wave sleep (SWS), sleep fragmentation, and early morning awakenings [Argyropoulos et al., 2003; Doghramji, 2003]. Of these, early morning awakenings are considered a specific feature of severe “melancholic” depression. Depressive states associated with medical disorders are characterized by reduced REM intensity [King et al., 1981].

Treatments for insomnia typically improve depressed mood [Casper et al., 1994; Jindal and Thase, 2004]; however, insomnia is an independent risk factor for depression and other psychiatric disorders [Riemann, 2007]. An interesting link between sleep and mood is evidenced by sleep deprivation immediately alleviating symptoms in depressed patients [Giedke and Schwarzler, 2002], while in healthy individuals restriction of sleep leads to a significant increase in depression subscale in morningness chronotypes (“morning larks”) and a significant decrease in depression subscale score in eveningness chronotypes (“evening owls”) after total sleep deprivation [Selvi et al., 2007]. So, the outcome depends on the individual properties of the circadian pacemaker system that interacts with the sleep process. According to a number of epidemiological studies, sleep impairments and depressive disorder seem to appear in a chronological order such that poor sleep precedes the onset of depression [Vollrath et al., 1989; Brabbins et al., 1993; Hochstrasser, 1993; Livingston et al., 1993; Paffenbarger et al., 1994; Breslau et al., 1996; Chang et al., 1997]. Our own study on a nationwide cohort of same-sex twins revealed a one-way temporal relationship between poor sleep and subsequent depressed mood, whereas the contrary was not observed. However, one-fifth of individuals with depressed mood showed no signs of poor sleep at any of the time points analyzed [Paunio et al., 2009]. This illustrates the presence of multiple mechanisms underlying clinically manifest depressive disorder.

Fatigue is a characteristic feature of diseases involving the HPA axis [Chaudhuri and Behan, 2004], such as depression. The experience of fatigue is also associated with poor sleep quality [Lichstein et al., 1997], although it is unclear whether sleep loss alone can explain the chronic fatigue syndrome (CFS) [Ball et al., 2004]. Fatigue, along with reduced sleep quality, also affects mood [Lobentanz et al., 2004].

Thus, early morning awakenings and fatigue are two features that reflect both disturbed sleep and depression. We hypothesized that the presence or absence of these features could be used to divide depressive patients into etiologically different categories. We also hypothesized that there could be genetic differences in serotonergic and glutamatergic neurotransmission systems, neural plasticity, and HPA axis between these categories. We tested these hypotheses in the population-based sample of Health 2000 from Finland, comprising 384 depressed individuals and 1,270 controls, with detailed information on their sleep as well as on day-time activity levels.

SUBJECTS AND METHODS

RESULTS

The genetic background of depression and characteristic symptoms related to disturbances in sleep, that is, early morning awakenings and fatigue, were investigated separately in both sexes (Tables II and III). Additional data from permutation-based allelic association analysis and the LD patterns for all genotyped variations within each gene are available in Supplementary 5 and Supplementary 6, respectively.

DISCUSSION

The current findings illustrate sex-dependent and symptom-specific differences in the genetic background of depression in a representative sample of the genetically relatively homogeneous population of Finns. The data support the hypothesis that some genes underlying depressive disorders affect the core component of depressive disorder with depressive mood and anhedonia (unrelated to sleep disturbance), while others are more strongly associated with the mood disorder only in the presence of disturbed sleep. Some genes, such as TPH2 or GAD1 in females, are associated with depression accompanied by fatigue, whereas others, such as CREB1 in males, are associated with depressive disorder itself, regardless of the presence or absence of symptoms of sleep disturbance. Our results also support the hypothesis that neurobiological mechanisms of depression may differ in females and males since in the analysis of 72 genetic markers in functionally relevant candidate genes, only one variant (rs11904814 of CREB1) was associated with depression in both sexes. These findings reflect the heterogeneity of the mood disorders at both the level of symptoms and underlying mechanisms. They are also in accordance with our previous data on 18,631 same-sex twins from Finland which revealed only a weak shared genetic component for depressed mood and poor sleep quality and different heritability estimates for females and males [Paunio et al., 2009].

The most robust finding was that the genes of the serotonergic neurotransmission system were associated with depressive disorder in females but not in males. The strongest association was found for the brain-specific, rate-limiting enzyme of serotonin synthesis, TPH2, in females with depression and fatigue (P = 0.005). An intronic variant of SLC6A4 was associated with depression (P = 0.025), and the associating allele was more frequent in patients with day-time fatigue. The high-activity alleles of the MAOA gene promoter region (MAOA-uVNTR) also showed a tendency of association with depression in females (P = 0.055), in accordance with earlier reports describing a marginal association with seasonal pattern and the presence of psychotic symptoms in females with major depression [Gutierrez et al., 2004]. A similar tendency was observed between depression in women and the functional SNP rs4680 of COMT (P = 0.068 in females with depression and fatigue, associating allele Met^108/158).

The serotonin system is recognized as sexually dimorphic. During stressful events there is a lower rate of serotonin synthesis in females, making women more susceptible to depressive disorder than men [Nishizawa et al., 1997]. Interestingly, animal studies suggest that gender differences exist in the expression of serotonin transporter gene due to transcriptional modification by sex hormones [Gubbels Bupp et al., 2008], and in humans, estrogens can downregulate COMT gene transcription [Xie et al., 1999].

Previous studies have reported that the TPH2 gene is associated with major depression [Zill et al., 2004], suicide in MDD [Ke et al., 2006; Lopez de Lara et al., 2007], bipolar disorder [Van Den Bogaert et al., 2006; Harvey et al., 2007; Lopez et al., 2007; Cichon et al., 2008], and schizoaffective disorder [Zhou et al., 2005]. The variant of TPH2 that gave the strongest signal in our study, rs12229394, is located on the 8th intron of the gene. Interestingly, the same allele “A” of rs12229394 tentatively associated to fatigue also in our control individuals (P = 0.049; OR = 1.34) (data not shown), suggesting that there could be a relationship between this allelic form of TPH2 and fatigue without manifestation of depressive disorder. However, we may also further hypothesize that these control individuals who do not have depressive disorder but show some symptoms of fatigue are prone to develop depressive disorder later in their lives as we know by earlier studies that signs of poor sleep are an important risk for depression [Paunio et al., 2009]. Variants from the same part of the gene were found to be associated with suicide in MDD [Ke et al., 2006; Lopez de Lara et al., 2007], schizoaffective disorder [Zhou et al., 2005], and a three-marker haplotype spanning intron 8, which is associated with bipolar disorder [Lopez et al., 2007]. These variants are in moderate LD with each other or with rs12229394 (D′ = 1, r² = 0.576), suggesting that part of TPH2 is important in regulation of mood, although the exact functional polymorphism remains to be clarified.

Narita et al. 2003 reported an association of 5-HTTLPR with CFS, and proposed that an attenuated concentration of extracellular serotonin may cause higher susceptibility to CFS [Narita et al., 2003]. We did not find robust association with the widely investigated 5-HTTLPR polymorphism. This is in line with the results from a meta-analysis demonstrating that the 5-HTTLPR polymorphism did not associate with recurrent MDD with a seasonal pattern [Johansson et al., 2003]. It is noteworthy that in most previous studies, the association signal was detected only under induced environmental stress [Uher and McGuffin, 2008], which we did not examine in our study.

We found that an allelic variant of the enzyme responsible for synthesizing GABA, GAD1, was associated with depression (P = 0.009 with SNP rs12185692 on promoter region), and intronic SNP rs769407 was suggestively associated with depression accompanied by early morning awakenings or fatigue. Earlier studies have revealed associations of other variants in the GAD1 promoter with bipolar affective disorder [Lundorf et al., 2005] and with schizophrenia [Addington et al., 2005; Straub et al., 2007], as well as associations of variants from other parts of GAD1 with anxiety and depressive disorders [Hettema et al., 2006], with childhood onset bipolar disorder [Geller et al., 2008], and with alcoholism [Loh el et al., 2006]. Previous studies have suggested a role for GABA (A) receptor in the regulation of sleep, anxiety, and mood [Gottesmann, 2002; Harrison, 2007]. The current results support the involvement of the GABAergic system in the genetic background of depression and disturbed sleep.

We also found an association of the X-chromosomal gene encoding for the inotropic glutamate receptor, GRIA3, with depression, depression accompanied by early morning awakenings, and depression accompanied by fatigue (P = 0.002, 0.008, and 0.002, respectively, with rs3848874) in females but not in males. A study in an Italian population reported that another intronic variant of GRIA 3 was related to schizophrenia in females [Magri et al., 2008], but the associating variants were not in LD to rs3848874.

The strongest signal was found for an intronic variant of CREB1 that was associated with depression in males (P = 0.0008 with rs11904814). The same allele “T” was associated with depression accompanied by fatigue in females also, but only with modest statistical significance. CREB1 is a member of the basic leucine zipper family of transcription factors consisting of 341 amino acids. Promoter variants of this gene have earlier been observed to be associated with a familial mood disorder in females [Zubenko et al., 2003] and with suicidality with a variant (rs7569963) [Perlis et al., 2007] in strong LD to rs11904814 (D′ = 1, r² = 0.892). Furthermore, the G to A transition at position −656, the promoter region of CREB1, has been found to change the activity of the promoter in the presence of female gonadal steroids [Zubenko and Hughes, 2009]. Taken together, these findings make CREB1 a highly appealing candidate gene for mood disorders that can exert its effect in different ways in the two genders.

Finally, we observed a suggestive association of a functional variation of BDNF, Val66, with depression accompanied by fatigue in females (P = 0.027). BDNF is involved in plasticity of the brain and is one of the key components activated in successful treatment of depression. It has been widely studied in psychiatric diseases, and the polymorphism Val66Met is associated with bipolar disorder, major depression, and schizophrenia. The associating allele varies between studies. Val66 has been found to be associated with bipolar disorder [Neves-Pereira et al., 2002; Sklar et al., 2002; Rybakowski et al., 2003; Geller et al., 2004; Lohoff et al., 2005; Green et al., 2006; Muller et al., 2006] and MDD and related personality traits [Sen et al., 2003; Lang et al., 2005]. However, some studies have detected no such associations [Tsai et al., 2003; Schumacher et al., 2005; Iga et al., 2007]. The suggestive association of CRHR1 with depression accompanied by early morning awakenings in females is of interest considering that dysregulation of the HPA axis is one of the major neuroendocrine abnormalities in depression. CRHR1 has been associated with major depression also in previous studies [Liu et al., 2006; Bradley et al., 2008], and interestingly, the same allele “A” of SNP rs173365 showing a suggestive association in our study was also associated with adult depression arising from childhood abuse [Bradley et al., 2008].

Given the expected low odd ratios of the genetic risk variants for complex disorders, such as depression, at the population level, our sample of 384 affected individuals has relatively limited power for detecting any associations. This is reflected by the finding of associations that are only suggestively significant when taking into account the number of genes and markers tested; we have noted that the findings will need to be replicated in an independent sample. On the other hand, our sample was carefully ascertained and is representative of the Finnish population. This diminishes the risk for obtaining spurious associations due to poor sample ascertainment and population structure, rather than arising from the genetic background of the examined trait. Furthermore, all genes included in the study are well-established functional candidate genes for mood disorders, with previous positive associations in various populations.

The main findings of our study, the association between serotonergic genes and depression and fatigue in women, the association between CREB1 and depression alone in men, and the association between GAD1, GRIA3, BDNF, and CRHR1 and depression with disturbed sleep in women—are of particular interest and add more pieces to the complex puzzle of mood disorders and their underlying mechanisms.