TPH2 haplotype blocks derived from HapMap Data Release 16, Phase I, June 2005. TPH2 exons (black boxes) and untranslated regions (open boxes) are drawn with indication of the 7 haplotype-tagging single nucleotide polymorphisms (htSNPs) respective to their linkage disequilibrium (LD) blocks (hapblocks). Only haplotypes with a frequency higher than 5% are represented. The haplotype block definition is based on confidence interval minima for strong LD (upper, 0.9 and lower, 0.65); upper confidence interval maximum for strong recombination (0.9); and fraction of strong LD in informative comparisons (must be at least 0.9); markers lower than 0.01 minor allele frequency are excluded. rs1487275 is described in the opposite strand in HapMap. htSNP rs11178997 is colored red because it is the only differing htSNP in the haplotypes associated with unipolar (AGTT) and bipolar disorder (TGTT).
TPH2 haplotype blocks derived from HapMap Data Release 20, Phase II, January 2006. TPH2 exons (black boxes) and untranslated regions (open boxes) are drawn with indication of the 7 haplotype-tagging single nucleotide polymorphisms (htSNPs) respective to their linkage disequilibrium (LD) blocks (hapblocks). Only haplotypes with a frequency higher than 5% are represented. The haplotype block definition is based on confidence interval minima for strong LD (upper, 0.9 and lower, 0.65); upper confidence interval maximum for strong recombination (0.9); and fraction of strong LD in informative comparisons (must be at least 0.9); markers lower than 0.01 minor allele frequency are excluded. rs1487275 is described in the opposite strand in HapMap. The associated haplotypes in these studies are colored red.
Van Den Bogaert A, Sleegers K, De Zutter S, Heyrman L, Norrback K, Adolfsson R, Van Broeckhoven C, Del-Favero J. Association of Brain-Specific Tryptophan Hydroxylase, TPH2, With Unipolar and Bipolar Disorder in a Northern Swedish, Isolated Population. Arch Gen Psychiatry. 2006;63(10):1103-1110. doi:10.1001/archpsyc.63.10.1103
Tryptophan hydroxylase is the rate-limiting enzyme in the serotonin (5-HT) biosynthetic pathway responsible for the regulation of serotonin levels. Tryptophan hydroxylase 2 (TPH2) was found to be solely expressed in the brain and therefore considered an important susceptibility gene in psychiatric disorders.
To determine the role of the brain-specific TPH2 gene in unipolar (UP) disorder and bipolar (BP) disorder in a northern Swedish, isolated population.
HapMap-based haplotype-tagging single nucleotide polymorphism (htSNP) patient-control association study.
A northern Swedish, isolated population.
One hundred thirty-five unrelated patients with UP disorder, 182 unrelated patients with BP disorder, and 364 unrelated control individuals.
Significant allelic association was identified in our UP disorder association sample for an htSNP located in the 5′ promoter region (rs11178997; P = .001). Haplotype analysis supported this significant result by the presence of a protective factor on hapblock 2 (Pspecific = .002). In the BP disorder association sample, single-marker association identified a significant htSNP in the upstream regulatory region (rs4131348; P = .004). Moreover, haplotype analysis in the BP disorder sample showed that the same htSNPs from hapblock 2 associated with UP disorder were also significantly associated with BP disorder (Pspecific = .002).
Haplotype-based analysis of TPH2 in patients with UP and BP disorder and controls from northern Swedish descent provides preliminary evidence for protective association in both disorders and thus supports a central role for TPH2 in the pathogenesis of affective disorders.
Serotonin (5-HT) has been linked to mood control and a wide variety of functions, including regulation of sleep, pain perception, body temperature, blood pressure and hormonal activity, cognition, perception and attention, aggression, sexual drive, appetite, and energy level. Many of these functions are relevant to the etiology of unipolar (UP) and bipolar (BP) disorder, and therefore, it has been hypothesized that disturbances of 5-HT activity are crucial in the etiology of these conditions. Tryptophan hydroxylase (TPH) is the rate-limiting enzyme in the 5-HT biosynthetic pathway responsible for the regulation of 5-HT levels. Numerous genetic and functional studies have investigated the role of the TPH isoform 1 gene (TPH1) in UP disorder and BP disorder, schizophrenia, alcoholism, drug abuse, aggression, and suicide, but none led to persuasive evidence for the involvement of TPH1 in disease pathogenesis.1- 4 In 2003, Walther and Bader5 demonstrated that the tph1 homozygous knockout mice continued to produce 5-HT in the brain, while expression in other tissues was absent or reduced significantly, through a novel tph isoform, tph2. The human homologue TPH2 is located on chromosome 12q21, a positional candidate region for UP disorder and BP disorder.6- 9 In the human brain, it has now been shown that both TPH isoforms are expressed in different human brain regions, showing the highest expression of TPH2 messenger RNA in the raphe nuclei, a region where the serotonergic neurons are the main neuronal components.10
Since its discovery in 2003, numerous genetic studies have investigated TPH2 for its involvement in affective disorders. Zhang et al11 found a tph2 Pro447Arg missense variant in several mouse strains, altering brain 5-HT levels. We performed a sequence analysis of the corresponding exon in humans but were unable to detect this variant in 2 populations comprising 317 patients with UP disorder, 364 patients with BP disorder, and 728 healthy control individuals (A.V.D.B., K.S., S.D.Z., L.H., K.F.N., R.A., C.V.B., and J.D., unpublished data, 2005). Zhang et al12 also reported a functional Arg441His human missense mutation linked to major depression. However, this variant is very rare since our group and several other groups could not detect it by resequencing the relevant region in a total of 3665 patients with UP disorder, 396 patients with BP disorder, and 1266 control individuals.13- 16
Furthermore, 4 genetic association studies reported significant associations between TPH2 and UP disorder, suicide, and BP disorder, supporting the involvement of TPH2 in the pathogenesis of both diseases.12,17- 20 These association studies were based on resequencing parts of the gene, mostly codon regions, and genotyping and analyzing relevant single nucleotide polymorphisms (SNPs) and haplotypes.17- 20
In the present study, we used the information provided by the HapMap Project and selected a panel of 7 haplotype-tagging SNPs (htSNPs) for genetic association studies, representing all TPH2 haplotypes with a frequency higher than 5%. Herein, we analyzed the 7-htSNP panel in both patients with UP and BP disorder and controls from a geographically isolated, northern Swedish population.
The northern Swedish patient sample comprised 135 unrelated patients with UP disorder (87 women, 48 men) and 182 unrelated patients with BP disorder (96 women, 86 men). The mean ± SD age at inclusion for the patients with UP disorder was 61.6 ± 13.4 years and for the patients with BP, 55.8 ± 14.25 years. The individuals were thereafter examined at multiple points, the majority up to 2005, by research psychiatrists and nurses. Relevant parts of a semistructured interview instrument (ie, the Mini International Neuropsychiatric Interview,21 the Diagnostic Interview for Genetic Studies,22 the Family Interview for Genetic Studies,23 and the Schedules for Clinical Assessment in Neuropsychiatry24) were used when deemed necessary. The diagnoses were made according to DSM-IV criteria.25 The mean ± SD age at disease onset for patients with UP disorder was 37.8 ± 16.7 years and for patients with BP disorder, 27.8 ± 11.5 years. The following disease characteristics were gathered for both patient groups: age at disease onset, first-episode symptoms (depression, hypomania/mania), heredity for affective disorders, and presence of phase-related psychotic symptoms, suicide attempts, melancholia, and atypical features, and mean follow-up age (Table 1).
The control individuals were selected from the Betula project, a large, population-based prospective study that is described in detail elsewhere.26,27 The random, unrelated control population (n = 364; 202 women, 162 men; mean ± SD age at inclusion, 58.0 ± 13.0 years) was recruited from the same geographical region of northern Sweden as the patients.
All control individuals and patients were white and their mother tongue was Swedish. Participants were only included in the study after signing an informed consent, and the study was approved by the medical ethical committees of the universities of Umeå and Antwerp.
We controlled all our association samples for population stratification by genotyping 37 microsatellite markers using standard genotyping and scoring methods. Statistical tests for population stratification were as described by Pritchard and Rosenberg.28 No population substructure was observed in the association samples (data not shown).
Genotyping was performed by pyrosequencing on a PSQ HS96 pyrosequencer (Biotage, Uppsala, Sweden, http://www.biotage.com/). Biotinylated polymerase chain reaction (PCR) products were immobilized onto streptavidin-coated sepharose beads (Amersham Biosciences, Uppsala). Biotinylated single-stranded DNA was obtained by incubating the immobilized PCR products in 0.5M sodium hydroxide, followed by 2 sequential washes in 10mM Tris acetate, pH 7.6. Primer annealing was performed by incubation at 80°C for 2 minutes and then at room temperature for 5 minutes.
Pyrosequencing assays were developed for all htSNPs except for rs2129575, which was genotyped by direct PCR sequencing. Hereto, 10 ng of genomic DNA and 10 pmol of each primer were used in a standard PCR reaction, followed by ExoSAPit treatment (Amersham Biosciences) and subsequent sequencing using the Big Dye terminator cycle sequencing kit (Applied Biosystems, Foster City, Calif) version 3.1 according to the manufacturer's instructions. Sequencing reactions were run on an ABI 3730 automated sequencer (Applied Biosystems). The resulting trace files were analyzed and scored using novoSNP.29
The linkage disequilibrium (LD) measures D′ and r2 were calculated using the Haploview program30 (http://www.broad.mit.edu/mpg/haploview/). Haplotype-tagging SNPs were chosen as predicted by Haploview (the haplotype block defintion is based on confidence interval minima for strong LD [upper, 0.9 and lower, 0.65]; upper confidence interval maximum for strong recombination [0.9]; and fraction of strong LD in informative comparisons [must be at least 0.9]; markers lower than 0.01 minor allele frequency are excluded). Only haplotypes with an estimated overall frequency of 5% or greater were considered in the analyses.
We calculated Hardy-Weinberg equilibrium in both patients and control groups for the htSNPs using GENEPOP.31 Allelic SNP and haplotype associations were calculated using 5000 permutations with Whap32 (http://pngu.mgh.harvard.edu/purcell//whap/). Haplotype frequencies for marker combinations in the patient-control samples were estimated using the program Haplotyper.33 The level of significance for all statistical tests was defined as P<.05.
In HapMap (HapMap Data Release 16, Phase I, June 2005), 37 SNPs across TPH2 and the upstream regulatory region were successfully genotyped in the 30 HapMap Centre d’Etude du Polymorphisme Humain (CEPH) trios.34 Furthermore, we genotyped a known SNP (rs11178997) located in the promoter region of TPH235 in the 30 HapMap CEPH trios because of its potential effect on TPH2 expression. Haploview analysis of the 38 TPH2 SNPs resulted in 3 distinct haplotype blocks, hapblocks 1, 2, and 3, across the TPH2 genomic region (Figure 1). We selected a panel of 7 htSNPs for subsequent genetic association analyses, representing all TPH2 haplotypes with a frequency higher than 5%. This htSNP panel comprises the SNPs rs4131348, rs11178997, rs10748185, rs2129575, rs1843809, rs1487275, and rs4474484 (Table 2).
We performed single-marker allelic and genotypic association analysis for completeness (Table 3 and Table 4) to test whether any of the 7 htSNPs are functionally relevant, although this study is basically a haplotype-based study to reduce testing and to maximize genetic variation coverage.
The 7 htSNPs were genotyped in both the UP and BP disorder association sample ascertained from an isolated population from northern Sweden. All SNPs were in Hardy-Weinberg equilibrium in both the control as well as the patient populations.
In the sample of patients with UP disorder and controls, significant association was observed for the A allele of rs11178997 (P = .001). This association is the result of a 4% decrease in the A allele frequency of patients with UP disorder (1%) compared with control individuals (5%) (Table 4).
In the sample of patients with BP disorder and controls, significant association was observed because of a decrease of the C allele of rs4131348 in the patients with BP disorder, from 21.9% to 14.6% (P = .004).
We performed a haplotype analysis based on the HapMap-defined haplotype blocks (Figure 1). We identified a significant difference in haplotype distribution between patients with UP disorder and control individuals for hapblock 2 (Pglobal = .04), which is the result of a specific decrease in the frequency of haplotype AGTT, from 5% in control individuals to 0.4% in patients with UP disorder (Pspecific = .002). No associations were observed between haplotypes in hapblocks 1 or 3 and UP disorder (Table 5).
We also observed differences in haplotype distribution between patients with BP disorder and control individuals in hapblock 2 (Pglobal = .02). This significant finding is the result of a decrease in the frequency of haplotype TGTT, from 25% in control individuals to 15% in patients with BP disorder (Pspecific = .002). Moreover, the proportion of homozygotes was significantly lower in patients with BP disorder compared with healthy individuals (n = 2 vs n = 29), resulting in a relative risk of 0.13 (95% confidence interval, 0.03-0.54) (P = .001) for homozygotes of the TGTT haplotype. In addition, a protective BP disorder association was also observed for hapblock 1 because of a decrease of 7.3% of the C allele of rs4131348 (Pallelic = .004). For hapblock 3, no significant association was identified with BP disorder (Table 5).
We used the open resource HapMap database34 to select a panel of 7 htSNPs representing all TPH2-spanning haplotypes, including the 5′ regulatory region, with a frequency higher than 5% (Figure 1) (Table 2). Next, this htSNP panel was used in a genetic association analysis comprising patients with UP and BP disorder and controls originating from a geographically isolated, northern Swedish population. Subsequently, single SNP–based association analysis of the UP disorder sample identified a significant protective association with htSNP rs11178997 located in the promoter region (P = .001). Also, a significant protective haplotype-based association was found with a rare 4-marker haplotype (AGTT) on hapblock 2 (P = .002) (Table 4 and Table 5). The decrease in the AGTT haplotype frequency comprising rs11178997-rs10748185-rs2129575-rs1843809 observed in the patients with UP disorder suggests the presence of a protective genetic factor localized on this haplotype. This finding is in accordance with previous TPH2 studies on major depression that showed a protective effect of TPH2 on major depression and suicide attempt.17,18,20 Direct haplotype comparison between this study and previous studies is problematic since different SNPs were used. However, the 10 SNPs genotyped by Zill et al,17,18 although different, are all located in hapblock 2 between exons 5 and 7. The reported haplotype analysis of these 10 SNPs showed the involvement of protective TPH2 haplotypes in the etiology of major depression and suicidality. Also, a recent study by Zhou and colleagues20 showed preliminary haplotype association of TPH2 to suicide attempt and major depression. The haplotype structure was determined by identifying and genotyping 15 SNPs in 4 populations across the TPH2 gene and flanking sequence. They found evidence for a protective association with US white, Finnish white, and African American individuals within a haplotype block at least 52 kilobases. The SNPs in this associated haplotype block, ranging from intron 5 to intron 8, were again located on hapblock 2, in which we observed significant association with UP disorder.
In the BP disorder sample, we identified a significant association for htSNP rs4131348 located in the upstream regulatory region of TPH2, representing a protective factor in hapblock 1 (P = .004) (Table 4 and Table 5). This finding was further supported by a significant, common 4-marker haplotype (TGTT) from hapblock 2 (P = .002) (Table 5). The significant decrease of the TGTT haplotype frequency comprising rs11178997-rs10748185-rs2129575-rs1843809 observed in patients with BP disorder suggests the presence of a protective genetic factor localized on the TGTT haplotype. To our knowledge, Harvey et al19 reported the only positive BP disorder association study on TPH2 so far, in which they identified and genotyped 5 SNPs between exons 7 and 9. Haplotype-based association analysis with the 5-marker haplotype showed a global, significant P value of .03. Based on the frequencies of the individual haplotypes, it can be seen that the largest difference in frequency is observed for protective effects. Although this observation is indirect, it again suggests a protective locus within TPH2 in the etiology of BP disorder.
The presence of a protective TPH2 haplotype for both UP disorder and BP disorder in our association study, as well as in at least 3 other independent studies, provides evidence for the involvement of TPH2 in the etiology of affective disorders. Furthermore, these data provide preliminary evidence for a shared etiology between patients with UP disorder and patients with BP disorder. In fact, based on our data, it appears that both protective haplotypes are closely related to each other. The only difference between the complete hapblock 2 haplotypes (comprising 30 SNPs), represented by htSNP-based haplotypes AGTT for UP disorder and TGTT for BP disorder, is the allele of only 1 SNP: rs11178997 (relevant alleles are indicated in red in Figure 1). This observation can be explained by 2 mechanisms. A first mechanism is that 1 of the 2 haplotypes originated from the other haplotype by a mutational event at htSNP rs11178997. This further indicates that the observed protective variant is located on the common part of the haplotype and that UP disorder and BP disorder share common genetic factors contributing to the susceptibility of affective disorders. This mechanism is unlikely because the frequency of the TGTT haplotype (which is associated in BP disorder) does not differ significantly between patients with UP disorder and control individuals, and this is also the case for the UP disorder–related AGTT haplotype frequencies between patients with BP disorder and control individuals (Table 5). A second mechanism that can explain this observation is that the HapMap-based haplotype structure used (HapMap Data Release 16, Phase I, June 2005) is not detailed enough. After we performed this study, a new HapMap release was made available (HapMap Data Release 20, Phase II, January 2006), showing that hapblock 2 is divided in 2 distinct hapblocks (Figure 2, hapblocks 2A and 2B) clearly separating the 5′ part of the former hapblock 2. The direct consequence of this is that htSNP rs11178997 is located on hapblock 2A and htSNPs rs10748185, rs2129575, and rs1843809, on hapblock 2B. With the currently genotyped htSNPs, it is not possible to determine all actual haplotypes with a frequency higher than 5% for hapblock 2A and hapblock 2B. This indicates that either the TPH2 functional variant(s) is located on hapblock 2B and that both disorders are influenced by a variant on the same haplotype (GTT haplotype) or that the functional locus is on hapblock 2A comprising the 5′ region of TPH2, potentially affecting the regulation of transcription. For the latter, a possible hypothesis could be that the protective haplotypes enhance TPH2 expression and consequently increase serotonin levels in the brain. It has been shown that increased serotonin levels induced by administering selective serotonin reuptake inhibitors can effectively treat depression.36 There is also evidence that dietary depletion of tryptophan results in depression in patients previously treated with selective serotonin reuptake inhibitors.37,38 Therefore, an enhancement of TPH2 expression might protect against the development of UP disorder and BP disorder. Brown et al39 examined the effects of SNP rs4570625 in the upstream regulatory region of TPH2 on the reactivity of the amygdala, a neural structure critical in the generation and regulation of emotional behaviors. They found that the T allele had greater activity in the amygdala than the G allele homozygotes. This suggests that the T allele may be associated with greater promoter activity and TPH2 expression, resulting in increased 5-HT synthesis. We determined the actual haplotype block position of rs4570625 by genotyping it in the CEPH trios of the HapMap Project. We observed that rs4570625 is in strong LD with htSNP rs2129575 (D′ = 1 and r2 = 0.946), which does not show association in our study.
In this study, we presented a HapMap-based genetic association study of TPH2 and demonstrated a genetic protective involvement of TPH2 in UP disorder and BP disorder by a rare protective haplotype and a common protective haplotype, respectively. First, we confirmed the studies of Zill and colleagues17 and Zhou et al20 showing that TPH2 is involved in the development of major depression. Second, the reported association of Harvey and colleagues19 of TPH2 and bipolar disorder is strongly replicated in our patient-control study. To our knowledge, this is the first association study using a HapMap-based htSNP analysis of TPH2. However, since our current study was based on a previous release of HapMap, genotyping extra htSNPs is required for detailed analysis of the contemporary TPH2 haplotype structure, which will guide us to the identification of the underlying genetic variant(s).
Correspondence: Jurgen Del-Favero, PhD, Applied Molecular Genomics Group, Department of Molecular Genetics (VIB8), University of Antwerp–Campus Drie Eiken, Universiteitsplein 1, B-2610 Antwerpen, Belgium (firstname.lastname@example.org).
Submitted for Publication: August 8, 2005; final revision received February 16, 2006; accepted February 23, 2006.
Funding/Support: This research was funded by the Special Research Fund of the University of Antwerp, the Fund for Scientific Research Flanders, the Interuniversity Attraction Poles program P5/19 of the Belgian Science Policy Office, and ALF Västerbottens Läns Landsting.
Additional Information: Dr Van Den Bogaert holds a PhD fellowship of the Institute for Science and Technology in Flanders.
Acknowledgment: We are grateful to the patients and the control individuals for their cooperation and participation in this research study. We acknowledge the contribution of the VIB Genetic Service Facility (http://www.vibgeneticservicefacility.be/) for the genetic analyses.