Since both aggression-related traits and serotonergic activity are partially heritable and correlate inversely, variations in genes of the serotonergic system might then, to some extent, account for variations in aggression-related behavior. Tryptophan hydroxylase (TPH) is the rate limiting biosynthetic enzyme in the serotonin pathway and regulates levels of serotonin. Recently, a genetic variation in TPH has been associated with aggression and anger-related traits in volunteers. We investigated a sample of community-based healthy volunteers (n = 154) and suicide attempters (n = 86), a clinical population with a high risk for elevated impulsive aggression and related traits. The subjects were genotyped for a A218C and a A779C single nucleotide polymorphism (SNP) located in the TPH gene. All subjects were administered standard psychiatric interviews as well as self-report questionnaires for aggression, irritability and anger-related traits. For anger-related traits, a multivariate effect of the tryptophan hydroxylase genotype and an interaction effect for genotype and diagnosis was observed in healthy volunteers and suicide attempters after controlling for age and educational level. U-carriers in both groups showed higher scores for State Anger, Trait Anger and Angry Temperament. These findings support the hypothesis that the A218C and the A779C SNP in the TPH gene may be associated with anger-related traits in German samples.
Low cerebrospinal fluid (CSF) concentrations of the serotonin metabolite 5-hydroxyindoleacetic acid (5-HIAA) is regarded as a putative indicator for low serotonin turnover.1 It is a relatively enduring trait which is partially genetically controlled as suggested by studies in humans,2 and demonstrated in rhesus monkeys.3,4 In these animals, low CSF 5-HIAA is associated with aggressiveness, low social affiliation, high-risk behavior, and premature mortality.5,6,7 In humans, 5-HIAA in the CSF correlates inversely with various aggressive behaviors as demonstrated in healthy and psychiatric samples throughout the life span. Although there are some negative reports, taken together, the vast majority of findings suggest that lowered 5-HIAA is related to the vulnerability for aggressive behavior.1,8 The tryptophan hydroxylase (TPH) gene encodes the rate limiting biosynthetic enzyme in the serotonin pathway, and regulates levels of serotonin.9 Therefore, variations in the TPH gene could contribute to the predisposition to low serotonergic neurotransmission. A single nucleotide polymorphism (SNP) consisting of an A779C transversion10 has been reported for the TPH gene, resulting in two alleles which are commonly addressed to as ‘U’ (upper) and ‘L’ (lower) respectively. In healthy volunteers, lower CSF 5-HIAA levels were found in men but not in women carrying the TPH U allele.11 In contrast, in a small, behaviorally extreme, impulsive group of alcoholic violent offenders, Nielsen and colleagues12 originally observed that LL carriers had lowest CSF 5-HIAA concentrations but this finding could not be replicated in a subsequent study.13
Some findings suggest an association of variations in the TPH gene with suicide-related behavior. First studies on the A779C SNP found an excess of the 799C allele in impulsive suicide attempters,12,13 whereas the association was inversed in the non-impulsive group.13 Later studies found no association14,15 or a marginal association16 with the 799C allele. Most studies investigated a A218C SNP, which is in strong linkage disequilibrium with the A779C SNP. In Caucasians, some studies associated an excess of 218A (or a deficiency of C218) with suicide-related behavior17,18,19 while others found no such association.20,21,22,23,24,25
First attempts have also been made to evaluate the role of TPH polymorphisms in impulsive aggression and related traits. Impulsive aggression, as assessed by the Buss–Durkee Hostility Inventory,26 was associated with the LL genotype in men but not in women in a small sample of subjects with personality disorders.14 In a subsequent study in a large sample of community-derived volunteers Manuck and colleagues27 have shown that individual differences in aggressive disposition are associated with the A218C SNP. A allele carriers scored significantly higher on measures of aggression and tendency to experience unprovoked anger and they were more likely to report expressing their anger outwardly. In summary, data on the association of TPH polymorphisms with aggression-related traits as measured by self report questionnaires are at least partially incongruent.
The aim of the present study was to investigate the hypothesis that the A779C and the A218C SNPs in the TPH gene are associated with aggression, irritability and anger-related personality traits. We conducted an association study in community-derived volunteers and suicide attempters, a high-risk population with elevated aggression-related traits. Behavioral assessments included self-report questionnaires for aggression, irritability and anger-related personality traits.
The distributions of genotype and allele frequencies of the A218C and A779C SNPs in the TPH gene in healthy controls and suicide attempters are shown in Table 1. These SNPs are in strong linkage disequilibrium and only nine out of the 240 genotyped subjects (3.7%) did not have identical genotypes. The genotype distribution was in Hardy–Weinberg equilibrium in both groups (Table 1). Genotype distribution was similar in healthy volunteers and suicide attempters (A779C: χ2 = 0.621, df = 2, P = 0.733; A218C: χ2 = 0.613, df = 2, P = 0.736). Furthermore, there were no differences in allele frequency between healthy volunteers and suicide attempters (A779C: χ2 = 0.371, df = 1, P = 0.578; A218C: χ2 = 0.164, df = 1, P = 0.696).
A MANOVA in healthy volunteers and suicide attempters was computed by integrating the seven subscales of the State-Trait Angry Expression Inventory28 and the two subscales Irritability and Assault of the Buss–Durkee Hostility Inventory26 as well as the factors: genotype (UU+UL vs LL), gender (males, females) and diagnosis (healthy volunteers, suicide attempters), controlled for age and educational level. Effects were seen for each of the three factors: genotype (A779C: F = 2.275, df = 8/197, P = 0.02; A218C: F = 2.255, df = 8/194, P = 0.02), diagnosis (A779C: F = 8.098, df = 8/197, P < 0.0001; A218C: F = 8.015, df = 8/194, P < 0.0001) and gender (A779C: F = 2.658, df = 8/197, P = 0.01; A218C: F = 2.445, df = 8/194, P = 0.01). Furthermore, an interaction effect was seen between genotype and diagnosis (A779C: F = 2.406, df = 8/197, P = 0.02; A218C: F = 2.535, df = 8/194, P = 0.01) (see Tables 2 and 3). Further detailed univariate analysis followed for genotype and interaction effects (see Tables 4 and 5). U-carriers (UU + UL) scored higher on the subscales State Anger (A779C: F = 5.112, df = 1/204, P = 0.03; A218C: F = 4.850, df = 1/201, P = 0.03), Trait Anger (A779C: F = 4.552, df = 1/204, P = 0.03; A218C: F = 3.857, df = 1/201, P = 0.05) and Angry Temperament (A779C: F = 4.368, df = 1/204, P = 0.04; A218C: F = 4.018, df = 1/201, P = 0.04) (see Tables 4 and 5). No sociodemographic differences between these two genotype groups were present for gender, age and educational level (data not shown). Looking only at the subgroup of suicide attempters, U-carriers (UU+UL) scored higher on the subscales State Anger (A779C: F = 3.802, df = 1/204, P = 0.05; A218C: F = 4.862, df = 1/201, P = 0.03), Trait Anger (A779C: F = 8.997, df = 1/204, P = 0.003; A218C: F = 8.611, df = 1/201, P = 0.004), Angry Temperament (A779C: F = 8.859, df = 1/204, P = 0.003; A218C: F = 9.614, df = 1/201, P = 0.002), Angry Reaction (A779C: F = 5.704, df = 1/204, P = 0.018; A218C: F = 4.790, df = 1/201, P = 0.03) and Anger-In (A779C: F = 4.330, df = 1/204, P = 0.04; A218C: F = 4.073, df = 1/201, P = 0.04). There were no sociodemographic differences with regard to gender, age and educational level between these two genotype subgroups in suicide attempters (data not shown).
It can be argued that, since neither the A779C nor the A218C SNPs are believed to have functional consequences, another SNP in the TPH gene might prove more useful for such a study. Although recently successful attempts were undertaken in order to identify other SNPs,29,30 their functional relevance still remains to be demonstrated.
The issue of association of TPH SNPs with suicide-related behavior is a complex and controversial one. This prompted us to conduct a meta-analysis which showed that U allele carriers had an enhanced risk for suicide-related behavior with a modest but clear effect in Caucasians (manuscript submitted). Nevertheless, we failed to find an association of suicidal behavior with SNPs in the TPH gene in the present study. This is not surprising since the sample size is far too small to detect the expected effect, although the genotype distribution points towards the same direction, without being significantly different. Thus, this apparently negative finding should be interpreted with extreme caution since much larger samples are needed for case control association studies of suicide-related behavior with the SNPs under investigation.
Recently, Manuck and colleagues27 examined an ethnically heterogeneous community sample and reported that U allele carriers scored significantly higher on measures of aggression and tendency to experience unprovoked anger and that they were more likely to report expressing their anger outwardly. Although this well conducted study shows a clear association, these findings warrant further replication in order to avoid any spurious associations, due to eg ethnic stratification effects and sampling errors. The aim of this study was to test the hypothesis that two SNPs (A218C and A779C) in the TPH gene, which are in strong linkage disequilibrium, are associated with aspects of aggression-related traits as measured with self-report questionnaires. We conducted genetic association studies in two well characterized independent populations of German origin. First, we included a group of community-derived volunteers who were rigorously screened for absence of mental disorders. Next, we hypothesized that the TPH SNP might influence anger-related traits in behaviorally extreme populations and included an independent high risk clinical sample, suicide attempters. In order to stay compatible with the report by Manuck and colleagues,27 we chose similar anger assessments. In line with the above mentioned report, we found an association of the A779C (and A218C) SNP with anger-related phenotypes in our samples of German descent and provide further evidence for the involvement of the TPH gene or a gene nearby in the regulation of this complex behavior. It is remarkable that our findings replicate the earlier study so closely. In particular, univariate analyses reveal that subscales associated with the A218C TPH polymorphism in the study by Manuck and colleagues27 are similarly associated with this polymorphism in the present report, and subscales not found to be associated previously similarly showed no univariate association.
Interestingly, an interaction effect of genotype and diagnosis was present, with the U-allele having a greater effect on anger in the group of suicide attempters. This suggests that other genetic and/or environmental factors are probably operating in patients and are less pronounced in the controls. Having only a patient group consisting of suicide attempters with various diagnoses makes it hard to differentiate between the relationship to suicide attempt status compared with the relationship to psychiatric disorders. This point is important because the more aggressive patients with a variety of diagnoses are more likely to be suicide attempters.31 In order to address this highly interesting question, further studies should include a parallel group of patients with psychiatric disorders but no suicide attempts.
In summary, our findings suggest that the A779C and the A218C SNPs in the TPH gene are markers for the genetic susceptibility to anger-related traits in Germans.
Materials and methods
Subjects and psychiatric assessment
The healthy volunteers were aged from 19–79 years (48 ± 16 years) and of German descent. They were randomly selected from the general population of Munich, Germany, and contacted by mail. In order to exclude neuropsychiatric disorders in the volunteers as well as in their first degree relatives, we conducted further screenings before the volunteers entered the study. First, subjects who responded were initially screened by phone. Subsequently, they were assessed with detailed medical and psychiatric history forms, both for themselves and their first-degree relatives. Third, volunteers were invited to an interview and psychiatric disorders in any of the subjects were excluded using the Structured Clinical Interview for DSM-IV (SCID I and SCID II).32,33 Psychiatric diagnosis in first-degree relatives was excluded using the Family History Assessment Module.34 One hundred and fifty-four healthy control subjects (66 males and 88 females) without relevant somatic, and with no psychiatric disorders, were included in this study.
The patient group of suicide attempters consisted of inpatients of German descent aged from 18–74 years (39 ± 15 years). They were consecutively recruited from general psychiatric wards of the Department of Psychiatry, University of Munich. The patients had a current or past history of suicide attempts and various psychiatric disorders (n = 86, 27 males, 59 females, detailed list of diagnoses is given in Table 1). The diagnosis was established by applying the SCID I and SCID II interviews.32,33 The patients were investigated close to discharge. Exclusion criteria included any serious medical condition, psychiatric disorders due to a general medical condition, dementia, and also schizophrenia and related psychotic disorders.
Written informed consent was obtained from all subjects after a detailed and extensive description of the study. The study was approved by the local ethics committee and performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki.
Both groups were characterized with the Buss–Durkee Hostility Inventory (Subscale Assault and Subscale Irritability). A total score was computed separately for both subscales.26 We also applied the State-Trait Anger Expression Inventory (STAXI).31,35 This instrument assesses actual anger as well as anger disposition, and distinguishes between inwardly and outwardly directed facets of its expression. The instrument consists of seven subscales: State Anger, Trait Anger, Angry Temperament, Angry Reaction, Anger-In, Anger-Out and Anger Control. A total score was computed for each subscale.
An intercorrleation analysis was computed for all seven subscales of the State-Trait Angry Expression Inventory and the two subscales Irritability and Assault of the Buss–Durkee Hostility Inventory. The subscales were highly intercorrelated (data not shown).
Genomic DNA was extracted from leukocytes or whole blood according to standard procedures. Genotyping of the A779C TPH SNP was performed by fluorescence melting curve detection using the Light Cycler System (Roche Diagnostics, Mannheim, Germany) with the following conditions: forward primer: 5′-CTT ATA TGT GTG AGT CTG AGT GG-3′; reverse primer: 5′-GGA CAT GAC CTA AGA GTT CAT GGC A-3′; acceptor hybridisation probe: 5′-LCRed640-CAC GTC GCA GTG CTT AAC ATA CGT TTA TAA-phosphate-3′; donor hybridisation probe: 5′-CTG AAA GAG AGG TAC AAG TT-fluorescein-3′. PCR was performed with 50 ng DNA according to the manufacturer's instructions for 45 cycles of denaturation (95°C, 0 s, ramp rate 20°C s−1), annealing (60°C, 10 s, ramp rate 20°C s−1) and extension (72°C, 10 s, ramp rate 20°C s−1). After amplification a melting curve was generated by holding the reaction at 40°C for 2 min and then heating slowly to 95°C with a ramp rate of 0.2°C s−1. The fluorescence signal was plotted against temperature to give melting curves at 53°C (A-allele) and at 48°C (C-allele). Primers 5′-TTC AGA TCC CTT CTA TAC CCC AGA-3′ and 5′-GGA CAT GAC CTA AGA GTT CAT GGC A-3′ were used in order to assess allele and genotype frequency of the A218C TPH SNP. PCR was performed as previously described22 with minor modifications. The 50 μl reactions contained 50 ng genomic DNA, 0.4 mM dNTPs, 15 mM ammonium sulfate, 60 mM Tris-HCl (pH 8.4), 2 mM MgCl2, 0.3 μM of each primer, and 1 U Taq polymerase (Life Technologies). Following an initial denaturation step at 95°C for 3 min, DNA was amplified in 35 cycles of PCR (92°C for 60 s; 56°C for 30 s; 72°C for 1 min). The final extension step was 72°C for 5 min. Twenty microlitres of the PCR product were digested with 8 U of BfaI (New England Biolabs), analyzed by gel electrophoresis in a 1.7% agarose gel containing ethidium bromide, and visualized under UV light. When cytosine is present at position 218, digestion results in fragments of 615 bp and 245 bp, whereas the absence of the BfaI recognition site leaves the 860 bp PCR product intact. A fragment of 58 bp is constantly visible due to an additional BfaI recognition site.
Statistics were performed using the SPSS 10.0 Software (Statistical Package for Social Sciences, SPSS Inc, Chicago, 1999). Demographic characteristics like gender, age, educational level as well as axis I and II diagnoses are presented in Table 1. t-Tests or Chi-square tests were performed, as appropriate for the measurement, to test for differences concerning sociodemographic variables between the diagnostic subgroups and between the genotype subgroups. Since previous studies on aggression-related traits15,16 investigated U carriers vs L homozygotes, we used the same groups in an attempt to be comparable with the previous findings. Furthermore, New and colleagues15 genotyped the A779C and Manuck and colleagues16 the A218C SNP, which prompted us to genotype both polymorphisms. For each SNP, a three-factor multivariate analysis of variance (MANOVA) was computed for healthy volunteers and suicide attempters integrating the seven subscales of the State-Trait Angry Expression Inventory (State Anger, Trait Anger, Angry Temperament, Angry Reaction, Anger-In, Anger-Out, Anger Control), the two subscales Irritability and Assault of the Buss–Durkee Hostility Inventory, as well as the three factors: tryptophan hydroxylase genotype (UU + UL vs LL), gender (males, females) and diagnosis (healthy volunteers, suicide attempters) controlled for the covariates age and educational level (unfinished school, low, middle, high). Gender was included because of the higher number of females in the suicide group (χ2 = 3.836, df = 1, P = 0.04). Furthermore, the covariate's age and educational level were added since the suicide group was younger than the controls (T = 3.105, df = 212, P = 0.002) and had a lower educational level (χ2 = 8.433, df = 3, P = 0.04). MANOVA was followed by univariate analysis. Analyses used two-tailed estimation of significance, an α-significance level of P < 0.05 was defined to be statistically significant.
We thank Christopher Murgatroyd for reviewing the manuscript and Sylvia de Jounge for technical assistance. The project is supported by the German Federal Ministry with the promotional emphasis ‘Competence Nets in Medicine’. Part of the material presented here is taken from the PhD thesis of IG.
About this article
TPH1 is associated with major depressive disorder but not with SSRI/SNRI response in Taiwanese patients