A concatenation of findings from preclinical and clinical studies support a preeminent function for the corticotropin-releasing factor (CRF) system in mediating the physiological response to external stressors and in the pathophysiology of anxiety and depression. Recently, human genetic studies have provided considerable support to several long-standing hypotheses of mood and anxiety disorders, including the CRF hypothesis. These data, reviewed in this report, are congruent with the hypothesis that this system is of paramount importance in mediating stress-related psychopathology. More specifically, variants in the gene encoding the CRF1 receptor interact with adverse environmental factors to predict risk for stress-related psychiatric disorders. In-depth characterization of these variants will likely be important in furthering our understanding of the long-term consequences of adverse experience.
Soon after the isolation and chemical characterization of corticotrophin-releasing factor (CRF) from ovine hypothalamus >25 years ago, the 41 amino acid-containing peptide was unequivocally established as the major physiological regulator of hypothalamic–pituitary–adrenal (HPA)-axis activity.1 Its preeminent function in mediating stress-related behaviors in rodents and primates was quickly demonstrated and hypothesized to be important in the pathogenesis of mood and anxiety disorders.2, 3, 4, 5, 6, 7, 8 These early findings were followed by a series of clinical studies and additional laboratory animal experiments that support the hypothesis that alterations in central nervous system (CNS) CRF-containing circuits have an important function in the pathophysiology of anxiety and depression (see Refs 9–11 for review).
As the CRF hypothesis of depression has earlier been comprehensively reviewed by our group and others, this monograph will only briefly summarize the CRF system, which contains four cloned ligands and two receptors, as well as major findings derived from animal and human studies that relate this system to stress, anxiety and depression. Our main focus, hitherto not comprehensively reviewed, is on recent developments in the human molecular genetics of the CRF system, how these findings further support the CRF hypothesis of depression and anxiety, and a discussion of future research directions.
The CRF system
Since the initial isolation of ovine CRF in 1981, three related ligands and two receptors have been identified (see Ref. 12 for review).
As noted above, CRF is a 41 amino-acid neuropeptide, expressed both peripherally (blood vessels, skin, lung, testes, ovaries and placenta) as well as in the CNS, with highest expression in the hypothalamus, amygdala, cerebrocortical areas and the septum.13, 14 Urocortin 1, urocortin 2 (stresscopin-related peptide) and urocortin 3 (stresscopin) are CRF-like neuropeptides.15, 16, 17, 18 Human urocortin 1 shares ∼45% homology to human CRF. In the CNS, urocortin 1 is expressed most strongly in the Edinger–Westphal nucleus and the hypothalamus,19, 20 but overall, its expression is more widespread peripherally, including in the pituitary, gastrointestinal tract, testes, cardiac myocytes, thymus, spleen and kidney.13, 20, 21 Urocortins 2 and 3 are also expressed both in the CNS and the periphery. Urocortin 2 is expressed in the magnocellular portion of the hypothalamic paraventricular and arcuate nuclei, the locus coeruleus, brainstem and spinal motor neurons; its expression pattern partially overlaps with that of CRF.17 Urocortin 3 is concentrated in the median preoptic area, the rostral perifornical area, the bed nucleus of the stria terminalis and the medial nucleus of the amygdala.16 Its expression pattern shows little overlap with CRF expression. In the periphery, urocortin 2 has been found in the heart, adrenal gland and peripheral blood cells, whereas urocortin 3 has been localized in the gastrointestinal tract, muscle, adrenal gland and skin.13, 22
The available evidence suggests that these four peptide neurotransmitters bind to two identified receptors, CRF1 and CRF2 (see Ref. 12 for review). CRF and urocortin 1 both bind with high affinity to the CRF1 receptor, whereas urocortins 1, 2 and 3, but not CRF, bind with high affinity to the CRF2 receptor. Both CRF receptors are 7-transmembrane G-protein-coupled receptors and share 70% sequence homology. Both receptors are present in the CNS. CRF1 is expressed in high density in the cerebral cortex, cerebellum, hippocampus, amygdala and pituitary; its peripheral expression is less robust and concentrated to skin, ovaries and testes and the adrenal gland. The CRF2 receptor is also expressed in the CNS, but largely restricted to subcortical areas, including the hypothalamus, amygdala, bed nucleus of the stria terminalis and raphe nuclei, and in peripheral tissues, such as the pituitary, heart, lungs, ovaries, testes and adrenal gland.14, 23, 24, 25, 26
One unique characteristic of this system is that the biological activity of all four peptides is regulated by a CRF-binding protein (CRFBP), which is highly conserved in mammalian as well as non-mammalian species.27, 28, 29 It is present in the circulation and the interstitial spaces as a soluble 37 kDa glycoprotein that binds CRF and all urocortins with high affinity, reduces their bioavailability and prevents their binding to CRF receptors.27 CRFBP is highly expressed in a series of tissues, including brain, heart, intestines, lungs and placenta.13, 30, 31 In the CNS, CRFBP is present in all CRF-related pathways and the pituitary. The CRFBP is partially colocalized with CRF, as well as CRF receptor expression, thereby permitting its function in modulation of CRF neurotransmission.30
The CRF system, stress, depression and anxiety
The CRF system and the stress response
The CRF system has a multitude of physiological functions that are apparently related to orchestrating the stress response at several different anatomical levels (see Figure 1). In addition to its well-documented function as a hypothalamic hypophysiotropic factor that stimulates pituitary ACTH synthesis and secretion and thereby controls the activity of the HPA axis,1 CRF neurons also innervate the locus coeruleus, thus activating the other major stress response axis, the CNS noradrenergic and sympathetic nervous systems.32 The CRF system also locally regulates adrenal steroidogenesis and catecholamine synthesis and release from the adrenal gland,33 again, dually influencing the HPA axis and norepinephrine and epinephrine secretion. Furthermore, effects of CRF in limbic brain regions have been associated with increased fear, alertness, decreased appetite and libido, all functions relevant in the fight or flight response and dysregulated in depression and anxiety disorders.9 These effects seem to be mediated mainly by the CRF1 receptor.10, 34, 35 The function of the CRF2 receptor remains more obscure and likely context and brain region dependent. Anxiolytic effects in a brain region-specific and stress level-dependent manners have been described.35, 36, 37, 38, 39, 40, 41, 42, 43, 44 Data from CRF2 receptor transgenic animals suggest an inhibitory function on adrenocortical function.38, 45 Overall, it seems that the CRF1 receptor is the principal receptor mediating the stress response, whereas the CRF2 receptor modulates the effects of CRF1 signal transduction.10, 11, 46
The CRF system and depression
Laboratory animal studies in which brain intracerebroventricular or brain region-specific microinjections of CRF has been used have revealed that CRF produces behavioral responses reminiscent of major depression in human beings, including increased anxiety, reduced slow wave sleep, psychomotor alterations, anhedonia, decreased appetite and libido.47, 48 These studies are complemented by results obtained in transgenic animals either lacking or overexpressing CRF-system ligands or receptors, as well as from studies using selective CRF antagonists. This large database indicates that these behaviors are mainly mediated by CRF1 receptor activation and modulated by CRF2 receptors.10, 49 Conditional CRF1 receptor knockout mice and CRF overexpressing mice restricted to forebrain areas have further shown that these anxiety- and depression-related phenotypes are specific to activation of the CRF1 receptor in limbic forebrain regions and independent of actions on HPA-axis activity,50, 51 though the latter endocrine effects of CRF may indeed contribute to the depressive symptoms. Although a negative effect of glucocorticoid receptor (GR) activation on CRF expression has been described for the hypothalamus, glucorticoids were shown to increase CRF expression in limbic areas, including the amygdala and the lateral septum.52, 53 In support of the critical function on limbic CRF transmission, recently increased CRF expression in the amygdala induced by use of a lentiviral vector was shown to produce most of the behavioral effects that comprise the depressive syndrome, as well as HPA-axis hyperactivity.54
In human beings, increased cerebrospinal fluid (CSF) CRF concentrations have been repeatedly observed in major depression8, 55, 56 and suicide victims.57 Moreover, treatment with either electroconvulsive therapy58 or antidepressants59, 60, 61 reduces CSF CRF concentrations. Notably, persistent elevations of CSF CRF concentration in symptomatically improved depressed patients is associated with early relapse of depression.62
A series of postmortem studies have provided compelling evidence of increased CRF neurotransmission in major depression and suicide. CRF expression (both mRNA and peptide) is increased in the hypothalamus, cortical areas, pontine nuclei and the locus coeruleus.63, 64, 65, 66 This is paralleled by a down-regulation of CRF1, but not CRF2, receptors in cortical areas of suicide victims, all pointing to an overactive CRF/CRF1 receptor system in depression.67, 68 In addition, CRFBP has been found to be down-regulated in the amygdala of patients with bipolar disorder,69 and urocortin 1 expression is up-regulated in the Edinger–Westphal nuclei of suicide victims.70
The relevance of overactive limbic CRF1 receptor transmission in depression is underlined by the fact that selective CRF1 receptor antagonists exert antidepressant effects at doses that do not influence baseline or stimulated HPA-axis activation.71, 72 It is important to note that, although not all studies with CRF1 receptor antagonists are positive in major depression, lack of a CRF1 receptor positron emission tomography ligand renders choice of dose problematic.73
The CRF system and early trauma
Overactivity of the CRF/CRF1 receptor system also has been showed to be one of the long-term neurobiological sequelae of early life trauma, a major risk factor for the development of affective disorders.74, 75 In fact, both rodents and non-human primates exposed to adverse experiences in early life exhibit evidence of hyperactivity of the CRF system as adults. Bonnet macaques reared under stressful conditions exhibit higher CSF CRF concentrations as adults than monkeys reared under non-stressful condition.76 Similar neuroendocrine results have been obtained in Rhesus monkeys.77 In rodents, early life stress is associated with persistent alterations in the CRF system, including increased CRF concentrations, increased CRF mRNA expression, altered CRF receptor expression and binding in the hypothalamus and limbic brain regions.78, 79, 80 This overactivity of the CNS CRF system is paralleled by a hyper-reactive HPA-axis response to stress in these animals. These findings in laboratory animals are consistent with the findings in human beings in which early life trauma is a strong predictor of CSF CRF concentrations in adults81, 82, 83 and is associated with an enhanced stress response to standardized psychosocial stressors, such as the trier social stress test. In addition, in endocrine challenge tests, including the CRF stimulation test and the combined dexamethasone suppression/CRF stimulation test,84, 85, 86, 87 patients with early life trauma exhibit evidence of marked HPA-axis hyperactivity.
The CRF system and anxiety disorders
On account of its anxiogenic effects in laboratory animals, the CRF system has also been implicated in the pathophysiology of anxiety disorders.10, 88, 89 Increased CSF CRF concentrations have been reported in posttraumatic stress disorder,90, 91, 92 but studies in adults with panic disorder and generalized anxiety disorder have failed to show such abnormalities.93, 94, 95
There is evidence from a series of animal studies that the CRF system is also highly relevant to alcohol dependence,96 particularly during alcohol withdrawal,97 and depression and anxiety disorders exhibit high rates of comorbidity with alcoholism. Indeed, several studies have revealed that acute alcohol withdrawal is associated with marked increases in CSF CRF concentrations.98, 99 It is also worth noting that clinically effective benzodiazepine anxiolytics, such as alprazolam, reduce CRF-ergic activity.100
In summary, over three decades of research have implicated CRF circuits in the pathophysiology of depression and certain anxiety disorders, particularly posttraumatic stress disorder, as well as in mediating the long-term impact of early trauma. We refer the reader to a series of review articles10, 11, 46, 101 that comprehensively review the above topics in considerably more detail. This review focuses on the human genetics of the CRF system and how recent advances in this area inform the field on the critical function of the CRF system in depression and anxiety.
Human genetic studies
Susceptibility to depressive or anxiety disorders is now well established to be due to the combined effect of genes and the environment, with heritability estimates for these disorders ranging from about 30 to 40%.102, 103, 104, 105, 106 The CRF system, being highly responsive to the environment, has been posited to serve as a key interface between environmental stressors and the development of depression. In fact, differences in the function of the CRF system because of genetic variation may lead to differences in an individual's response to stressful events and may regulate an individual's vulnerability to stress-related psychiatric disorders.
Table 1 lists the genes encoding all members of the CRF system with their human chromosomal position.
Before the sequencing of the human genome,107, 108 studies addressing the genetic basis of affective and anxiety disorders mostly consisted of linkage studies, using large pedigrees of families with multiple affected individuals or sibpairs. Most linkage studies have used short or variable tandem repeats or microsatellite markers, which often tag larger regions in the chromosome, so that inference of linkage to a certain candidate gene from these studies usually requires extensive follow-up fine mapping. Nonetheless, if genetic variation in the CRF system is related to increased vulnerability to depressive and anxiety disorders, one might expect to find regions of linkage for these disorders in the neighborhood of the chromosomal loci encoding these genes.
Although none of the recent meta-analyses of linkage scans for bipolar disorder provide strong support for linkage to regions containing CRF-system genes,109, 110 some individual scans in major depressive or anxiety disorders111, 112 have reported suggestive evidence for linkage in some of these regions. Two meta-analyses of linkage scans in bipolar disorder109, 110 report linkage to the 8q region distal to the corticotrophin-releasing hormone (CRH) gene. The more recent study actually reports genome-wide significance for this locus when bipolar II patients are included in the analysis.109 However, the signal in these studies (8q23) is distal to the CRH gene, which is located in 8q13.1. Testing of a short tandem repeat marker proximal to the CRH gene did not provide any evidence for linkage with bipolar disorder in a smaller sample of 22 pedigrees.113 In contrast, two studies have reported suggestive linkage (LOD <2.0) for bipolar disorder in the chromosomal regions 8q13.2 and 8q13.3 including the CRH gene,114, 115 and a recent study has also shown linkage of a region containing 8q13.1 to deficit symptoms in a cohort of families affected with schizophrenia, schizoaffective disorder and bipolar disorder.116 Suggestive evidence for linkage to bipolar disorder has also been reported for 3p21, containing UCN2.117 Interestingly, strong linkage (LOD 3.88) has been found of the combined phenotype of anxiety disorders and early onset major depression with a large locus on chromosome 3 (3p12.3-q12.3) overlapping with the earlier locus.118 UCN3 is located centromeric to a linkage peak for obsessive–compulsive disorders on 10p15.3.119
Overall, linkage studies provide some evidence for an involvement of the CRF system in the genetics of depression and anxiety disorders, but the data are not compelling. However, such approaches have significant limitations in the study of the genetics of complex disease because they are usually underpowered to detect the smaller combined effects of multiple genes as well as gene × environment interactions expected in these disorders.120 Therefore, lack of evidence from linkage studies cannot be extrapolated to an absence of genetic effects of the CRF system in depression and anxiety.
Results from whole-genome association studies
For whole-genome association studies, the markers of choice have been single-nucleotide polymorphisms (SNPs) because they are common genetic variations (single base exchanges) and their genotyping is amenable to high-throughput methods.121, 122 Over 6.5 million of these variants have been validated in human beings so far (www.ncbi.nlm.nih.gov/projects/SNP/snp_summary.cgi). In past years, data from the HapMap project, an international consortium to catalog human genetic variation (HapMap),123 and other publicly available resources have become available online (http://www.hapmap.org/, http://genome.perlegen.com/, and others) to facilitate the selection of the most informative SNPs to adequately cover the genetic variation in different populations. Most commercially available SNP assays and whole-genome SNP arrays rely on these resources for SNP selection.
Several whole-genome studies have been published for bipolar disorder124, 125, 126, 127, 128 and two for unipolar depression129, 130 and none show reproducible association of SNPs within CRF-system loci in the samples for which association data is readily available online. In fact, data from the GWAS for unipolar depression of the GAIN study130 show that within 200 kb 5′ and 3′ of the CRF-system genes, no association with a P-value <0.01 is observed, except for one SNP about 40 kb proximal of UCN3 (P-value between 0.01 and 0.001). The results from whole-genome studies for anxiety disorders or gene × environment interactions have not been published yet. It is, however, noteworthy that the arrays used in these studies contained especially few SNPs within the CRH, UCN and UCN2 loci. The Illumina hap550 v3 array, for example, with over 550 000 SNPs across the whole genome has only two SNPs within 20 kb surrounding the CRH gene (http://genome.ucsc.edu/). Overall, the genetic coverage of the used SNP arrays is not complete and especially weak for rarer variants. Negative results in these studies may thus also be confounded by a lack of coverage of the genetic variation in some of these loci.
Candidate gene approaches
In contrast to the genome-wide approaches described above, candidate gene approaches have the advantage of directly testing association of genetic variations within a gene or locus of interest. This focused approach allows to answer specific hypotheses, including gene × gene as well as gene × environment interaction analyses and to home in on the actual functional variants. Candidate gene approaches have investigated a number of different genetic polymorphisms, including putatively functional variable tandem repeats and insertion–deletions. In the absence of a know functional variant, researchers often rely on a selection of marker variants to tag a genetic site making use of information on the linkage disequilibrium structure of the investigated chromosomal region.131
In the following section, we will describe the genetic structure of the genes encoding the various components of the CRF system, as well as evidence for functional polymorphisms within the genes of interest and results from candidate gene association studies.
The CRH locus
Although the official name of the peptide is CRF, it is also often referred to as CRH and CRH is also the official gene symbol for the locus encoding CRF.
CRH is a short gene, spanning about 2 kb with two exons located on the long arm of chromosome 8. It encodes the 196 amino-acid long CRF precursor, which is then processed to the 41 amino-acid long CRF.
In the HapMap project phase II, 12 SNPs have been genotyped within the CRH locus (±1.5 kb), of which most are either non-polymorphic or rare. Shimmin et al.132 have comprehensively re-sequenced this locus in over 200 individuals of Mexican-, European- and African-American descent and have detected 37 SNPs, one insertion–deletion polymorphism and a microsatellite marker in the same region. At the time of publication, only 23% of these variants were listed in dbSNP, the most comprehensive database cataloguing these polymorphisms. This would indicate that current SNP databases, and thus commercially available genotyping assays and arrays, vastly under-represent the genetic variation in this locus. Negative association results may, therefore, also be related to the fact that none of the potentially relevant markers have been genotyped in earlier association studies.
A set of four polymorphisms (SNPs) in the 5′ regulatory region of the CRH locus up to 4000 bps upstream of the transcription initiation site have been described and seem to be associated with differential transcription of the gene product using reporter gene assays.133, 134, 135 One of these polymorphisms, rs5030876 seems to alter binding of activating transcription factor 6 to the putative promoter region,135, 136 thus providing a potential molecular mechanism for differences in gene transcription. Rosmond et al.137 tested one of these putatively functional promoter variants (rs503875) for its effects on baseline and dexamethasone-suppressed saliva cortisol measures. This T/G SNP is rare in European populations (8–10% carrying the G-allele), but more frequent in Africans (close to 40% carrying at least one G-allele) (Hapmap and Baerwald et al.133). In a Swedish sample of 284 men, no effects of the SNP on diurnal saliva cortisol variation or dexamethasone suppression (0.5 mg) were observed. However, its rare allele showed a significant interaction with a functional SNP (ThtIII1) in the GR gene. Only the combination of the rarer CRH promoter variant and the allele of the GR polymorphism earlier associated with higher diurnal cortisol secretion was also associated with higher salivary cortisol levels in this study.137 The sample size (N=12) in the combined genotype group that carried the effect was, however, very small.
With the exception of another small study that linked the CRH locus to a congenital-isolated ACTH deficiency in one family with two affected and two non-affected siblings,138 we are not aware of any other studies relating genetic variation in the CRH locus to endocrine measures. Recently, a functional variant in the promoter of the CRH gene of Rhesus monkeys has been reported that disrupts a glucocorticoid response element. This functional variant was associated with lower CSF CRH concentrations as well as more exploratory and bold behavior as infants and adolescents.139
A handful of studies have investigated the association of the CRH locus to anxiety-related phenotypes. Smoller et al.141, 142 have investigated the phenotype of behavioral inhibition in children of parents with anxiety disorders and found association with a repeat polymorphisms about 20 kb 3′ of the gene140 and three SNPs located just 3′ as well as 5′ of CRH. Two of these SNPs were also tested for association with panic disorder in a case–control study in adults, but no significant associations were observed.143 Finally, the synonymous coding SNP rs6159 (amino-acid position 96 Gly/Gly), associated with behavioral inhibition142 as well as two SNPs 5′ of the CRH genes were neither associated with major depression nor with response to antidepressant treatment in a German sample.144 Wasserman et al.145 also reported negative results for an association of two CRH SNPs with suicide attempt.
Even though some interesting functional SNPs regulating CRH expression have been identified, so far no definitive conclusions can be reached on the involvement of genetic variation within the CRH locus in anxiety- and depression-related phenotypes because of the limited sample sizes in these studies and the small number of studies overall. In addition, putatively important gene × environment interactions have not yet been investigated for this gene.
Urocortins and CRHBP
Genetic architecture and functional variants
The genes encoding the preproproteins for the three urocortins are located on different chromosomes and each has two exons. The CRHBP locus, encoding CRFBP, is located on chromosome 5 and contains seven exons. Public databases also indicate two additional exons 3′ of the gene and an alternate isoform skipping exon 7 and containing these two 3′ exons.
We are currently not aware of any validated functional variants in these genes, but putative functional variants such as non-synonymous coding SNPs and other variants potentially influencing transcription or splicing regulation are predicted in silico.
To our knowledge, except for negative associations with response to citalopram treatment in major depression146 no psychiatric genetic association studies have specifically investigated the urocortin gene variants. An initial positive study on the association of CRHBP SNPs with recurrent unipolar depression147 could not be verified in an enlarged sample of the same cohort and did not replicate in a second independent sample.148 The same group has also reported a lack of association of these variants with bipolar disorder.149 Enoch et al.,150 report the association of variants within a CRHBP haplotype block with anxiety disorders in Plains Indians and alcohol use disorders in Caucasians. The same haplotypes were also associated with α-power as measured by resting EEG in these individuals.150 These results may suggest that CRHBP variants could have a function in anxiety and addiction by moderating arousal states. A recent study from our group showed an ethnicity-specific association of a CRHBP SNP with response to citalopram treatment in the STAR*D cohort.146 This study investigated the association of polymorphisms within all CRF-system genes discussed in this article and response to antidepressant treatment. The only association withstanding correction for multiple testing was with an SNP located 3′ of the CRHBP gene, rs10473984. In the African American and Hispanic subset of STAR*D patients, the T-allele of rs10473984 was associated with poorer treatment outcome. This allele was also associated overall higher plasma ACTH levels, suggesting functional effects of this variant on CRHBP levels and thereby CRF signaling at the pituitary level. Finally, the association with poorer treatment outcome was most pronounced in individuals with anxious depression, supporting the function of this system in anxiety-related behaviors.146
The CRF receptor genes
In the following, we will present data for the two CRF receptor genes, and begin with the gene encoding the CRF2 receptor (CRHR2), for which fewer studies are available, followed by the more intensely studied CRF1 receptor gene (CRHR1).
The CRHR2 locus
The CRHR2 locus contains 15 exons with three different promoter sites and three different first exons151 encoding the CRF2α, CRF2β and CRF2γ isoforms.152, 153, 154 These three isoforms differ in the initial part of the N-terminal ligand-binding domain, but otherwise the receptor structure is the same, starting with the common second exon and they do not seem to differ in their ligand-binding profile.152, 153 The CRF2α is the predominant isoform in human brain and its expression is responsive to glucocorticoids.155
For variants within the CRHR2 locus, positive associations have been reported. Tochigi et al.156 reported an association of an intronic SNP in CRHR2 with the personality trait openness on the NEO personality inventory in Japanese individuals. De Luca et al.157 identified a three microsatellite marker haplotype that was associated with the severity of suicidal behavior, but not suicide attempts per se in families with bipolar disorder. Another intronic SNP (not in the same haplotype block as the one reported by Tochigi et al.156 was also associated with response to citalopram treatment in a Spanish cohort.158 Negative associations of CRHR2 variants with unipolar, bipolar and panic disorder as well as antidepressant treatment response have also been published,143, 146, 159, 160 painting a controversial picture of the association of variants in this gene with depression and anxiety-related phenotypes.
The CRHR1 locus
The CRHR1 gene spans 51 kb on the long arm of chromosome 17 (see Figure 2). Its 13 exons encode the 415 amino-acid CRF1 receptor for which eight splice variants have been reported.161, 162, 163 CRF1α is the main isoform, containing all 13 exons. CRF1β has 29 amino acids inserted into the first intracellular loop (encoded by an additional exon without any homologous sequence in mouse). CRF1c has 40 amino acids deleted from the N-terminus (lacking exon 3) and CRF1d results from a skipping of exon 12 and thus a 14 amino-acid deletion from the 7th transmembrane domain. CRF1e lacks exons 3 and 4—coding for the N-terminus, CRF1f lacks exon 11 and CRF1g lacks exon 10 and part of exons 9 and 11. CRF1h has a cryptic exon inserted between exons 4 and 5. All receptor variants result in either altered/impaired binding affinity or signal transduction. In addition, CRF1h, in which the cryptic exon generates a translation terminator in exon 4, only has a functional CRF-binding domain, but no transmembrane domains. CRF1e and CRF1h are thus soluble isoforms and might serve in an analogous function to the CRFBP. In fact in in vitro systems, the presence of CRF1e decreases and CRF1h increases CRF1α activation.162 CRF receptors isoform CRF1α is expressed in both the CNS and the periphery. CRF1e through CRF1h have been discovered in the skin and CRF1β-1h in the myometrium.161, 164, 165 The lack of isoform-specific antibodies has made it difficult to test in which tissues the receptor isoform proteins are actually expressed. So far, no report has described how genetic variation might influence this alternative splicing, but polymorphisms altering the ratio of the different isoforms are likely to have a strong impact on CRF1 signaling.
An interesting feature of the chromosomal region containing the CRHR1 locus on chromosome 17q21.31 is the presence of a 900 kb inversion polymorphism.166 This polymorphism is rare in African and Asian populations, but more frequent in populations of European descent. It is associated with a higher number of off-springs in female carriers and thus under positive selection. Owing to an increased potential for misalignment during meiosis in heterozygous carriers of this polymorphism, there is an increased risk for deletion and duplications of this locus. Indeed, this inversion polymorphism has been associated with the occurrence of a microdeletion syndrome of this region encompassing the CRHR1 and microtubule-associated protein tau (MAPT) genes.167, 168, 169, 170, 171 The clinical features of this syndrome include a marked hypotonia, dysmorphic features and moderate mental retardation. One report describes friendly and amiable behavior in these patients.167 A duplication of this region has been reported in one female patient,172 in whom hirsutism and sleep disturbances were noted besides mental retardation and dysmorphic features. No reports about the effects of this polymorphism on endocrine regulation or a more in-depth psychiatric characterization have been published to our knowledge.
No data on polymorphisms that change CRHR1 gene function have yet been published. In silico analysis of potentially functional SNPs indicates that several SNPs within this locus might have an impact on gene function (http://fastsnp.ibms.sinica.edu.tw). It is noteworthy that for rs12936511, a synonymous coding SNP in exon 2 (Pro/Pro at amino-acid position 20) and the presence of the rare allele (minor allele frequency <10%) result in the loss of an exonic splice enhancer element and could lead to altered splicing regulation. Two missense mutations in exons 3 and 4 (rs41280114 and rs16940655) result in conservative amino-acid exchanges (amino-acid position 60—Val/Ala and amino-acid position 96—His/Arg, respectively) and likely do not have a major impact on protein function.
Several association studies with CRHR1 polymorphisms have been published and center around anxiety, depression, response to antidepressant drugs and gene × environment interactions.
Keck et al.143 have reported an association of CRHR1 polymorphisms with panic disorder. In this study, the authors report nominally significant associations of CRHR1 SNPs in two independent German panic disorder samples. The strongest results were, however, the combined effects of rs878886 in CRHR1 and rs28632197 in AVPR1B, encoding the vasopressin 1B receptor. A two-SNP model showed significant associations with panic disorder in both samples separately and the combined sample of 359 cases and 794 controls. Both SNPs are of potential functional relevance because rs878886 is located in the 3′ untranslated region of the CRHR1 gene and rs28632197 leads to an arginine to histidine amino-acid exchange at position 364 of AVPR1B, which is located in the intracellular C-terminal domain of the receptor, likely involved in G-protein coupling. These genetic data support the large body of evidence showing interactions of the vasopressin and CRF systems in anxiety.48 Another family-based study failed to find association of four polymorphisms in the CRHR1 locus with panic disorder. However, fewer CRHR1 and no AVPR1B polymorphisms were tested in this study.173
Four studies in ethnically different samples have investigated associations of CRHR1 SNPs with response to antidepressant treatment in major depression.146, 158, 174, 175 Licinio et al.174 initially reported an association of a three SNP haplotype within CRHR1 with response to fluoxetine or desipramine in Mexican-Americans and this has been replicated in a Chinese population.175 Interestingly, both studies reported that the effects of this CRHR1 haplotype were most pronounced in patients with anxious depression. Nominally significant associations with response to citalopram of an SNP within the putative CRHR1 promoter region were also observed in the STAR*D sample.146 Papiol et al.158 found that another intronic SNP (rs110402) was not associated with response to citalopram treatment in 159 Spanish patients, but was correlated with an earlier age at onset for major depression. A significant case–control association of CRHR1 SNPs with major depression was only reported by Liu et al.176
Another set of studies have examined gene × environment interactions of CRHR1 SNPs and stressful events on different psychiatric phenotypes. Wasserman et al.145 have investigated over 500 family trios of male suicide attempters and found that the rarer T-allele of the intronic CRHR1 SNP rs4792887 is overtransmitted in these families. This overtransmission was restricted to individuals with less stressful life events. As the majority of suicide attempters were also depressed, these data suggest an interaction of CRHR1 polymorphisms with stressful life events on suicide attempts in depression. Our group177 showed a strong interaction of CRHR1 SNPs with child abuse predicting adult depressive symptoms in an inner city African-American sample of over 400 individuals. We reported protective effects of two three-SNP CRHR1 haplotypes, whereby individuals with high levels of child abuse and individuals homozygous for the protective haplotypes had significantly less current depressive symptoms than abused individuals carrying other haplotype combinations. In fact the latter groups' scores on the Beck Depression Inventory were similar to the non-abused group, in which the SNPs had no effects on depressive symptom severity. This result was supported by similar findings in a sample of mostly Caucasian women described in the same report. All three SNPs were intronic and one of the two associated haplotypes contained rs479887 for which interaction with life events and suicide attempts had earlier been reported.178 These early trauma × CRHR1 interactions on depression have been replicated and extended in independent studies. Polanczyk et al.179 report a replication of the interaction on past year and recurrent major depression using the same three-SNP haplotype in a female cohort retrospectively assessed for childhood maltreatment. The interaction was, however, not observed in the prospective Dunedin cohort and the authors argue that this might be due to differences in the assessment of childhood trauma. Furthermore, two reports corroborate the putative functional role of these same variants using data from an endocrine challenge test, the combined dexamethasone/CRF test. Here, the protective alleles are associated with a lower cortisol response in this test, either in interaction with early life events180 or as a main genetic effect.181 The CRHR1 × environment interaction on depression and HPA-axis regulation are further strengthened by the fact that the effects seem to replicate across different ethnicities. As mentioned above, gene × gene interactions are likely important in the genetics of depression and anxiety. Ressler et al.182 have reported a significant gene × gene × environment interaction, with these CRHR1 polymorphisms interacting with the serotonin transporter-linked polymorphic region (5-HTTLPR) and child abuse to predict adult depression.
Blomeyer et al.183 described the interaction of another CRHR1 tagging polymorphism (rs1876831) and negative life events on adolescent alcohol consumption. In the presence of a large number of negative life events, individuals homozygous for the C-allele of this SNP reported about twice as much life-time heavy drinking as well as drinks per occasion than carriers of the T allele. This difference was not seen in the presence of a low number of life events. These findings have been extended to young adults, in which this tagging SNP and stressful life events interact to predict progression of heavy alcohol use in young adulthood.184 The interaction of genetic variation of CRF-system genes and stressful life events on alcohol consumption is further supported by data from rhesus macaques. A functional promoter polymorphisms in the CRH gene interacts with prior stress exposure to predict voluntary alcohol consumption in these animals.185
In human beings, further fine mapping and re-sequencing will be needed to identify the putative functional variants responsible for these gene × environment interactions and whether the SNPs interacting with life events to predict alcoholism, suicide attempts or depression in fact represent distinct functional loci. Overall, these recent human genetic data strongly support a function for the CRF1 receptor in depression and anxiety, most likely by moderating the effects of early trauma or life events in general.
Conclusions and future directions
Even though the majority of human genetic studies investigating genes encoding the CRF system are limited by small sample sizes or insufficiently dense coverage, the sum of these findings, especially those related to variants within CRHR1, supports the concatenation of preclinical and clinical data implicating the CRF system in the pathophysiology of depression and anxiety. Together with the serotonergic and neurotrophic system,111 the CRF system thus represents an example of how human genetic studies can contribute to specific pathophysiologic hypotheses in depression and anxiety.
It has to be noted that the presented genetic association studies, including data in GWAS, are only powered to detect associations with common genetic variants. Rare, functional variants in these genes may, however, also have a relevant impact for these disorders. Next generation sequencing methods will allow to catalog and compare all sequence variation in these loci and to identify, rare, but functionally relevant, polymorphisms, including copy number variations.
In addition to expanding genetic association studies of this system in larger samples with sufficiently dense genetic markers as well as re-sequencing, the next generation of CRF-related genetic studies should focus on system genetics, analyzing the impact of the whole system including ligands, receptors as well as other genes upstream and downstream of this signaling pathway. There is already evidence for the association of genetic variants in genes intricately tied to the CRF system with depression and anxiety, such as the GR gene186 and the gene encoding the GR co-chaperone FKBP5187 for example.
On account of their critical function in stress regulation, it is unlikely that detrimental variants in single genes within the CRF system will be common in the population. More likely, the sum of genetic variations with discrete functional effects in a number of genes within this system will impact stress-axis regulation and susceptibility to stress-related psychiatric disorders. Similar to the number of studies now investigating the 5-HTTLPR and the BDNF Val66Met polymorphism,111 gene × gene × environment interactions of CRHR1 variants with this polymorphism have been reported.182 It will now be important to identify and characterize the putative functional variants in CRHR1 and their interaction with the other genes. From the above presented data, it seems that loci within this gene moderate the impact of life stress on the development of psychiatric disorders. Stratifying individuals by the protective vs susceptibility variants of these loci in neuroimaging and endocrine studies, for example, might yield a better understanding of the long-term impact of trauma on neuronal circuitry and hormone systems, and more specifically, which compensatory mechanisms are involved in the development of stress-related psychiatric disorders. The characterization of these variants may also be a tool to dissect possible epigenetic impacts of trauma on the CRF system. As described above, early life adversity has been shown to result in persistent effects on CRF neurotransmission and HPA-axis regulation (in an animal model of early prenatal stress, methylation of the CRH gene locus correlates with long-term changes in gene expression188). Genetic variations within the CRHR1 locus seem to moderate this impact, possibly by enhancing or preventing epigenetic changes within this gene. In fact, early adversity and thus stress have been shown to lead to epigenetic changes, including difference in the methylation status of HPA-axis genes.189 One possibility would be that these putative variants could alter the epigenetic impact of stress, for example the methylation of CpG islands by changing a CG sequence. Indeed, an SNP abolishing one CpG can be associated with global differences in methylation of the whole surrounding CpG island.190
Despite some controversial results, the initial series of human genetics studies support the importance of this system in mediating stress-related psychopathology. In-depth characterization of these variants could represent an important tool in elucidating the long-term consequences of early life adverse experience.
Support was received from NIH grants MH-42088, MH-69056, MH-58922 and RR-25008. Dr Binder is supported by a Doris Duke Clinical Scientist Development Award.