The interaction between TPH2 promoter haplotypes and clinical–demographic risk factors in suicide victims with major psychoses


Dr A. H. C. Wong, Centre for Addiction and Mental Health, 250 College Street, Room 711, Toronto, Ontario, Canada M5T 1R8. E-mail:


Tryptophan hydroxylase isoform 2 (TPH2) is a rate-limiting enzyme in the biosynthesis of serotonin (5-HT) and is predominantly localized in the brain. Previous studies have suggested that there is an association between serotonergic dysfunction in the brain and suicidality. This study was designed to examine whether the −473T > A and −8396G > C polymorphisms of the TPH2 gene may be associated with completed suicide in subjects with major psychoses from the Stanley Foundation Brain Bank sample. TPH2 genotypes were determined in 69 subjects with a diagnosis of schizophrenia or bipolar disorder, among which 22 died by suicide. Genomic DNA was amplified by polymerase chain reaction and typed by automated methods. Both markers were found to be in Hardy–Weinberg equilibrium and in strong linkage disequilibrium. No association with history of suicide was found for either polymorphism. Haplotype analysis with ehap showed no association between completed suicide and haplotype distribution (χ2 = 1.877; 3 df; P = 0.598). Nor was there any association between suicide and these genetic markers even when clinical–demographic factors were considered as covariates in the haplotype analysis. These findings suggest that these 5′ marker haplotypes in the TPH2 gene do not influence suicidal behaviour.

Suicide is a major contributor to morbidity and mortality in schizophrenia (SCZ) and bipolar disorder (BD). The lifetime suicide risk is between 4 and 13% in SCZ (Meltzer 2005) and 19% in BD (Goodwin & Jaminson 1990). Between 25 and 60% of BD patients make at least one suicide attempt during the course of their illness (Chen & Dilsaver 1996). Demographic factors such as sex, age, geography and religious affiliation are known to influence suicide risk (Stack 2000). In our own sample of BD patients, suicide attempts were associated with comorbid substance abuse disorders (Dalton et al. 2003).

Genetic factors also affect suicide risk, but the mechanism and magnitude of the genetic contribution are unknown (Roy 1993). The serotonin system is thought to be associated with disturbances in the regulation of anxiety, impulsivity and aggression (Mann 1995). Furthermore, several studies have reported altered serotonergic function in people who die by suicide. For example, low concentrations of the serotonin metabolite 5-hydroxyl indole acetic acid (5-HIAA) in the cerebrospinal fluid (CSF) have been associated with suicide attempts and completed suicide (Asberg et al. 1976). Low levels of 5-HT or 5-HIAA were also found in CSF of suicide victims (Nordstrom et al. 1994).

5-HT is synthesized in two steps, with tryptophan hydroxylase (TPH) as the rate-limiting enzyme (Fitzpatrick 1999). Thus far, two TPH genes have been identified. The TPH1 gene is located on chromosome 11p15 and has been the focus of a number of genetic studies on suicide (Nielsen et al. 1998). A recent meta-analysis concluded that a polymorphism in intron 7 of the TPH1 gene (A218) is associated with suicide-related behaviours (Rujescu et al. 2003); however, another meta-analysis (Lalovic & Turecki 2002) showed no association. A new TPH gene isoform (TPH2), located on chromosome 12q15, has recently been identified. In human, TPH1 and TPH2 proteins are highly homologous exhibiting 71% of amino acid identity (Walther & Bader 2003). In mouse brain stem, the expression of TPH1 appears to be 150 times lower than that of TPH2, suggesting that TPH2 may be much more important for 5-HT synthesis in the brain than TPH1 (Walther et al. 2003). Zill et al. (2004a) have demonstrated that TPH2 is expressed at high levels in human brain, while transcripts are present only at very low levels in other human tissues.

TPH2 has two known single-nucleotide polymorphisms (SNPs: −8396G > C and −473T > A) in the putative promoter region (Kennedy et al. 2003) and many highly polymorphic variants in the introns, spanning most of this 90 kb gene. Polymorphisms in the regulatory regions do not alter protein structure but may alter gene expression and consequently may affect behaviour that is regulated by 5-HT levels (Weiss & Terwilliger 2000). In the light of these findings, we tested for genetic association between the TPH2 −8396G > C and −473T > A haplotypes and suicide in patients with either SCZ or BD. We also incorporated clinical–demographic risk factors into this haplotype analysis.

Materials and methods

DNA genotypes were determined in samples (69) donated by the Stanley Medical Research Institute. This sample is part of Stanley Array Collection and consists of genomic DNA from 34 subjects with diagnosis of BD and 35 individuals with diagnosis of SCZ. Among these, 22 died by suicide and all the others died by other causes. The sample demographics and clinical information are summarized in Table 1.

Table 1.  Demographic characteristics of the subjects
 Suicide subjectsControl subjects
  1. * Lifetime alcohol and drug abuse were assessed on a semiquantitative scale, with the following order: 0 = little or none, 1 = social, 2 = moderate in the past, 3 = moderate at time of death, 4 = heavy in the past, 5 = heavy at time of death.

Sex (male : female)11:1132:15
Principal Axis I diagnosis [number (%)]
 Schizophrenia6 (27%)29 (62%)
 Bipolar disorder16 (73%)18 (38%)
Mean ± SD
 Age of death (years)42.73 ± 9.9044.98 ± 9.37
 Lifetime alcohol abuse*2.45 ± 1.952.33 ± 1.93
 Lifetime drug abuse*1.82 ± 1.791.93 ± 1.99

These specimens were collected, with informed consent from next of kin, by participating medical examiners between January 1995 and June 2002. The specimens were all collected, processed and stored in a standardized way. Exclusion criteria for all specimens included the following: (a) significant structural brain pathology on post-mortem examination by a qualified neuropathologist or by pre-mortem imaging; (b) a history of significant focal neurological signs pre-mortem; (c) a history of central nervous system (CNS) disease and (d) documented IQ <70. Diagnoses were made by two senior psychiatrists, using DSM-IV criteria (American Psychiatric Association 1994), based on medical records and, when necessary, telephone interviews with family members. The information about lifetime alcohol and drug abuse was recorded using a non-parametric scale ranging from 0 = absent to 5 = heavy present (Table 1).

The −473T/A and −8396G/C flanking sequences were analysed for the presence of transcription factor-binding site using the Transcription Element Search System (

The genotype of the TPH2 −473T/A (rs11178997) polymorphism was determined by polymerase chain reaction (PCR) amplification and sequence-specific fluorescent probes using Taqman assays on the ABI PRISM 7000 (Applied Biosystem, Foster City, CA).

The genotype of the −8396G/C SNP was determined by restriction fragment length polymorphism, because this SNP (rs4131347) is within an Alu-repeat sequence. Thus, we designed specific PCR amplification primers flanking the Alu sequence: 5′-CAG TCT TCG TTG TTA AGC TTG C-3′ and 5′-AGA GAA CAA GAA TAA AAA TAA GGC ATA-3′. The −8396C/T polymorphism was detected by SmaI restriction enzyme digestion, followed by electrophoresis. The digest generated DNA fragments of the following sizes: −8396G = 379 bp, −8396C = 241 bp + 138 bp). Continuous data were presented as means and standard deviations and analysed by the t-test (Mann–Whitney when non-parametric). Dichotomous variables were analysed using the χ2-test and odds ratios (ORs).

Haplotype analysis was performed using ehap v2.0 software (, which allows for the incorporation of covariates. ehap determines whether there is an association between haplotypes and phenotypes based on the evolutionary relationships among the haplotypes (Seltman et al. 2003).


The transcription factor analysis showed that −8396G/C does not affect any transcription factor-binding sites; however, the −473A allele eliminates the DNA-binding site for the transcription factor Brain-2 (Brn-2).

We performed a preliminary analysis on the clinical and demographic data to determine which factors could be predictors for suicide. The mean age of death in the suicide group and non-suicide group was 42.73 ± 9.90 and 44.98 ± 9.37, respectively, and there were no significant differences between the two groups (t = −0.914; 67 df; P = 0.364). When diagnosis was examined as a potential predictor for suicide, we found that proportionally more bipolar patients than schizophrenics committed suicide in this sample (χ2= 7.107; 1 df; P = 0.008). The estimated increase in suicide risk in BD vs. SCZ patients was twofold (OR: 2.262; CI 95%: 1.103–4.641). In this sample, suicide was more frequent in females than in males (42% vs. 26%), but this difference was not statistically significant (χ2 = 2.087; 1 df; P = 0.149). There are more females proportionally in the bipolar group (χ2 = 4.332; 1 df; P = 0.037); however, a separate analysis in female and male patients was not possible for the small sample size. Lifetime alcohol use or abuse was evaluated using a non-parametric scale, and the Mann–Whitney test did not reveal a significant association between suicide and higher alcohol consumption (z = −0.341; P = 0.733). Nor was suicide associated with drug abuse (z = −0.28; P = 0.978) or the presence of psychotic symptoms (χ2 = 1.953; 1 df; P = 0.162). However, suicide was associated with late age of onset (t = 3.528; 66 df; P = 0.001) and shorter time in the hospital (t = −2.033; 66 df; P = 0.049).

The genotype counts for −8396G/C were GG = 42, GC = 24 and CC = 1 and the genotype counts for −473T/A were TT = 50, TA = 17 and AA = 2. The frequency of the (−473T/A) T allele was 12%. The AA genotype (−473T/A) was present only in the suicide group, and the Fisher's exact test showed a slight trend (P = 0.098). The −8396G/C SNP genotype CC was present only in the non-suicide group, but this was due to the larger size of this group (P = 1.000).

In the ehap analysis, only the statistically significant individual demographic risk factors for suicide were included: diagnosis, age of onset and time in hospital. The frequencies of the haplotypes −8396C/−473A, −8396G/−473A,−8396G/−473T and −8396C/−473T were 6.6, 8.6, 74.2 and 10.6%, respectively. The omnibus test for association yielded a P-value that was not significant (χ2 = 1.877; 3 df; P = 0.598). The analysis based on the evolutionary relationships among haplotypes showed that −8396G/−473A has the same impact on suicide as –8396G/−473T (P = 0.717) and that −8396C/−473A has the same impact on suicide as −8396C/−473T (P = 0.957).


To our knowledge, this is the first genetic association study to examine the relationship between TPH2 promoter polymorphisms and suicide in major psychoses.

The SNPs chosen in this study might have a putative role in the expression of TPH2 because the −8396G > C is located within an Alu sequence and these repetitive DNA elements are suggested to promote DNA strand exchange and genomic rearrangement; in addition, the −473A allele generates a new DNA-binding site for the transcription factor Brn-2.

Our results indicate that it is unlikely that the (−8396G > C/−473T > A) haplotypes have an influence on suicidal behaviour in our sample. For the genotype AA at locus −473, we found an interesting trend; however, with this sample, we are able to detect effect size as low as 2 with a power >80% for alpha 0.05 (Sample Power release 2.0; SPSS Inc., Chicago, IL). In the analysis of clinical–demographic variables, we find that the rate of suicide in the BD group is higher, which is expected. In contrast, late onset and shorter time spent in the hospital are associated with a higher rate of suicide, a rather unexpected result.

The lack of an association with suicide could be confounded by an association between TPH2 and mood disorders (Zill et al. 2004b); however, even in the psychiatric populations at high risk of suicide, most patients never attempt suicide, indicating the importance of predisposition to suicidal behaviour that is independent of the main psychiatric diagnosis. Furthermore, Zill et al. (2004b) did not demonstrate any association between the SNPs in this study and diagnosis of mood disorder.

The main strengths of this study are the haplotype analysis combined with an analysis of clinical–demographic risk factors as covariates and the investigation of the new TPH2 isoform that is specific to the CNS. One possible limitation is that the number of suicide victims was small. However, the promoter haplotypes that we have investigated may have more functional meaning than the many intronic SNPs investigated previously (Zill et al. 2004c), because the probability of finding false positives by chance among the many intronic variants of large genes is quite high. Another limitation is the paucity of clinical information available; variables such as bipolar type or SCZ subtype would enhance the quality of future studies.

In conclusion, we found no association between suicide and the promoter variants −473T/A and −8396G > C in the TPH2 gene. However the presence of the rare −473 genotype AA in two suicide victims and its absence in the control group should be investigated further. A larger sample size for future investigations of the TPH2 promoter in suicide would address many of the limitations of this study. Because sex, age of onset and diagnosis may affect suicide risk, these demographic factors should be incorporated in future genetic studies of suicide.


This work was made possible through the support of the Canadian Institutes for Health Research (CIHR), and the National Alliance for Research in Schizophrenia and Depression (NARSAD). AHCW is a CIHR Clinician-Scientist Fellow and NARSAD Young Investigator, and HHMVT holds a Canada Research Chair in Neurobiology. VDL is supported by a Postdoctoral Research Fellowship and a Pilot Grant from the American Foundation for Suicide Prevention. DNA was donated by The Stanley Medical Rescarch Institute's Array Collection courtesy of Drs. Michael B. Knable, E. Fuller Torrey, Maree J. Webster, Serge Weis, and Robert H. Yolken.