The involvement of the TPH1 gene in the pathogenesis of affective disorder is supported by several lines of evidence. Previous studies have reported a significant association between TPH1 intron 7 A218C polymorphism and BPD in the French population . In addition, two current meta-analyses confirmed the significant association of suicide-related behavior with TPH1 A218 polymorphism [42, 43] and suggested that the A allele has a dose-dependent effect on the risk of suicidal behavior . Since these results suggested that TPH1 could be a strong candidate for involvement in BPD, we have systematically searched all the exons and promoter region of the gene for SNPs. Five polymorphisms were identified, one of which, 3'UTRSNP2, is a novel polymorphism not previously reported. The only sequence variant found in all TPH1 exons was in the exon 1c/intron1 region (T3804A, dbSNP ID: rs623580), but the polymorphism is within the 5'-UTR and therefore does not result in an amino acid substitution. This finding is consistent with a previous report in which no coding sequence variant of TPH1 was detected in Americans (Indians or Caucasians), Italians, and Finns . These results demonstrate the highly conserved nature of the human TPH1 gene. The exon 1c/intron 1 T3804A polymorphism has been previously reported and has an estimated minor allele frequency of 0.001 in the Swiss population [34, 45].
The five identified SNPs, together with five additional SNPs taken from the literature and a public database, were tested for an association with BPD in Taiwanese patients. In general, the alleles in the TPH1 gene were common, with minor allele frequencies ranging from 21% to 49% and from 22% to 49% in cases and controls, respectively. The genotypic frequencies of all markers, except the G-347T (5'flankingSNP3) marker in the promoter region, showed Hardy-Weinberg equilibrium (HWE) in both populations. The departure from HWE of 5'flankingSNP3 was observed in both controls (p = 0.024) and patients (p = 0.024). A separate study in our lab indicated that the location of the G-347T polymorphism is on the transcriptional repressor GATA1 binding site  and alteration of alleles indeed change promoter activity in a luciferase reporter gene system . The departure from HWE could be explained by the negative selection of the homozygous TT individuals. And the presence of very low frequency of TT genotype in the population may be the reason for the impossibility of detecting any association. Despite the strong functional role of the TPH1 G-347T polymorphism, no association was detected between any of the polymorphisms and BPD in Taiwanese.
The extent and distribution of LD in humans has been a hot topic, especially for gene mapping of complex diseases. In this study, significant LD, as measured by the D' and P values using the SNP Alyze® program, could be detected between T-1721G and C27224T polymorphisms of the TPH1 gene, which are separated by roughly 29 kb (Table 3). Because LD-induced association between multiple loci that harbor disease-predisposing alleles can be identified by haplotype-based analyses , haplotypes were constructed and their frequencies were compared between cases and controls. Haplotype distributions among 10 TPH1 SNPs were estimated and twelve haplotypes with frequency larger than 2 % in at least one group were listed (Table 4). The three most common haplotypes were identical in both groups and were found in 58 % and 56.7 % of the patients and controls, respectively. Although significant difference in haplotype distribution was seen in one comparison, the association was weak and was lost after the Bonferroni correction for multiple tests. Additional permutation tests for haplotype distributions in case and control groups were performed and result indicated no differences in haplotype profile between two groups (p value = 0.602). The data suggest that both BPD patients and controls are actually from the same population thus it implies the TPH1 gene may not be related to BPD etiologies.
Positive association between TPH1 intron 7 polymorphism and BPD was identified in our previous study , but the replication with extended samples has failed to confirm this association in both single-locus and haplotype analyses. One possible reason for this discrepancy could be at the sampling bias as the age of controls in the extended population is much younger than the cases. Since mood disorders may occur late in life, the sampling bias represents a major limitation of this study. Another possible explanation may be that the TPH1 gene has only a minor effect on BPD etiology, this effect being missed when heterogeneous samples are used. Alternatively, the positive association we obtained initially could be a false positive outcome from a very small sample size being used. Recently, studies of Tph1 (the original Tph) knockout mice were found to express normal amounts of serotonin in brain, but not in the periphery , and resulted in a cardiac dysfunction phenotype . Follow up studies found that the second tryptophan hydroxylase (TPH2, also known as nTPH) gene is predominantly expressed in the brain stem, while the classical TPH1 is expressed in the pineal gland  and peripheral tissues (duodenum, kidney or lung;. The amount of TPH2 mRNA expression in individual raphe cells was estimated to be approximately 2.5-fold greater than the level of TPH1 expression in pinealocytes . These findings have changed the consideration of linking polymorphism of TPH1 gene with various psychiatric diseases  and perhaps, could explain the lack of association between the TPH1 gene and BPD obtained in this study. Whether these two paralog proteins are regulated independently or if they have distinct functions in the brain are still under investigation; studies to establish connections between TPH2 gene and various psychiatric diseases including BPD are on going and should provide more insights regarding the TPH2 function in the brain.
Studies on sequence variations in the human genome have revealed that the human genome can be parsed objectively into haplotype blocks, with limited diversity within each block [50, 51]. Johnson et al.  determined the extended haplotype at any given locus in a population to identify the SNPs in a gene or a LD region, information that is essential for association studies. The so-called "haplotype tag SNPs (htSNPs)" capture the majority of haplotype diversity within a region and thus represent the minimal number of markers that need to be typed to define the common haplotypes (higher than 5% frequency in the population). In this study, we used a publicly program to define htSNPs that represent all haplotypes of the TPH1 locus with a frequency > 2% . Three htSNPs, corresponding to all common haplotypes were generated using SNPtagger software  (Table 5). Although further analysis using haplotypes constructed with htSNPs revealed no differences in frequency distributions between cases and controls (Table 5), the htSNPs selected for human TPH1 gene significantly reduce the number of markers required for genotype analysis (in this case from 10 to 3 SNPs), which may be useful in other studies on the Taiwanese Han population.