Molecular Psychiatry (2008) 13, 813–820 & 2008 Nature Publishing Group All rights reserved 1359-4184/08 $30.00 www.nature.com/mp ORIGINAL ARTICLE Genetic architecture of the human 2 : existence of neural isoforms and relevance for major depression F Haghighi1,2, H Bach-Mizrachi2, YY Huang2, V Arango2, S Shi1, AJ Dwork2,3, G Rosoklija2,4, HT Sheng1, I Morozova1,JJu1,5, JJ Russo1 and JJ Mann2 1Columbia Genome Center, Columbia University, New York, NY, USA; 2Department of Psychiatry, Division of Molecular Imaging and Neuropathology, Columbia University, New York, NY, USA; 3Department of Pathology, Columbia University, New York, NY, USA; 4School of Medicine, Ss Cyril and Methodius University, Skopje, Republic of Macedonia and 5Department of Chemical Engineering, Columbia University, New York, NY, USA

Impaired brain neurotransmission is a potential component of the diathesis of major depression. Tryptophan hydroxylase-2 (TPH2), is the rate limiting biosynthetic isoenzyme for serotonin that is preferentially expressed in the brain and a cause of impaired serotonin transmission. Here, we identify a novel TPH2 short isoform with truncated catalytic domain expressed in human brainstem, prefrontal cortex, hippocampus and amygdala. An exploratory study of 166 Caucasian subjects revealed association with major depression or suicide of a novel single nucleotide polymorphism (SNP) g.22879A > G located in exon 6 of this short isoform. This SNP and additional SNPs were discovered through a systematic characterization of the genetic architecture of the TPH2 gene for further genetic and functional investigations of its relationship to major depression and other psychopathology. Molecular Psychiatry (2008) 13, 813–820; doi:10.1038/sj.mp.4002127; published online 8 January 2008 Keywords: major depression; suicide; tryptophan hydroxylase 2; TPH2; Haplotype; genetic association

Introduction isoenzyme, is elevated in the dorsal raphe nuclei of depressed suicides at both the transcript9 and protein In the brain, serotonin is synthesized in the cell levels.10,11 To reconcile these observations, we had bodies of serotonergic neurons in the brainstem raphe suggested the presence of a gene variant that may nuclei that project to most cortical and subcortical affect the catalytic site resulting in the production of a structures. These brain regions include the prefrontal lower activity . This hypothesis is consistent cortex, anterior cingulate, amygdala, hippocampus with recent reports of such a tryptophan hydroxylase-2 and thalamus, structures involved in regulating (TPH2) gene variant expressing an isoenzyme with mood, memory and behavior and manifesting altered impaired catalytic activity in the Balb/c mouse, an 1 activity in major depression. Alterations in seroto- inbred strain that naturally exhibits a depressive nergic function are implicated in the pathogenesis of behavioral phenotype and has low brain serotonin. major depression and in the diathesis for suicidal Balb/c is reported to have a novel coding variant in a 2,3 behavior. Since genetic factors can contribute homologous region of a TPH2 human ortholog with causally to both major depression and the proclivity apparent impaired catalytic activity.12 However, the 4,5 to commit suicide and since the serotonergic system report that this type of variant is present in 10% of 6,7 is partially regulated by genetic factors, candidate human subjects diagnosed with unipolar depression for both major depression and suicidal behavior has not been replicated in other datasets.13–17 could involve the serotonergic system. Although We investigated the role of TPH2 in major depres- levels of serotonin and/or its main metabolite sion and suicidal behavior using a collection of case 5-hydroxyindoleacetic acid are reported to be low in and control postmortem brain samples. Through 8 suicides, we find that neuronal tryptophan hydro- psychological autopsies using validated structured xylase-2, the rate-limiting serotonergic biosynthetic clinical interviews with family members18 we ob- tained detailed clinical information, including axis I Correspondence: Dr JJ Mann, Department of Neuroscience, New and II psychiatric diagnoses based on DSM-IV York State Psychiatric Institute, 1051 Riverside Drive, Box 42, diagnostic criteria. Recent studies report associations New York, NY 10032, USA. E-mail: [email protected] of TPH2 with major depression and with suicide, Received 2 May 2006; revised 31 October 2007; accepted 4 involving single nucleotide polymorphisms (SNPs) in November 2007; published online 8 January 2008 intron 5 and exon 11.12,13,19,20 Thus, while we Novel TPH2 gene variant associated with major depression F Haghighi et al 814 explored the coding portion of the TPH2 gene and tained 50 ng genomic DNA extracted from brain sections of each intron, we focused our analyses on tissue, 1X PCR buffer, 0.25 U Taq polymerase (Accu- the SNPs that lie in these two regions in an effort to Taq, SIGMA, Sigma-Aldrich, St Louis, MO, USA), replicate the previously reported positive findings 0.1 mM dNTPs and 5 pmol forward and reverse in our dataset. In addition, we report on the first primers. After PCR, 3 ml of each PCR product was systematic examination of the TPH2 gene variants run on 2% agarose gel with 1X TAE buffer to visualize covering both the coding and non-coding regions and the PCR product quality. PCR products were purified transcript variants in human brain. using shrimp alkaline phosphatase and exonuclease 1 to degrade the excess dNTPs and primers. Bi-direc- tional DNA sequencing reactions were conducted Materials and methods with the ABI PRISM Big Dye Terminator Ready Samples and subjects Reaction kit (Applied Biosystems, Foster City, CA, Brain tissue was obtained from autopsies conducted USA) on 1 ml of purified PCR products with the same for medical or forensic reasons unrelated to the study. primers used for PCR. Sequences were analyzed on Next of kin gave informed consent to the use of brain an ABI 3730 Â l sequencer. Raw sequences of both tissue for research and to undergo a psychological DNA strands were assembled and SNPs were checked autopsy interview.18 Clinical axis I and axis II diag- against the ENSEMBL reference sequences using noses, including a major depressive disorder were Seqman (DNASTAR, Madison, WI, USA). based on DSM-IV criteria established from structured clinical interviews, SCID-I and SCID-II and when Quantitative real-time RT-PCR available, from review of medical records with the Quantitative real time PCR was carried out on the MJ modified Diagnostic Evaluation After Death.21 These Research DNA Engine Opticon System. Each 20 ml psychiatric diagnoses (or the absence of any psychia- reaction contained 3 mM MgCl2, 1X PCR Buffer, tric diagnosis), based on all available data, were 1 pmol of each primer, 0.5 Units of Taq DNA confirmed as meeting published criteria,22–24 at a polymerase (Invitrogen, Carlsbad, CA, USA), 2 mls of consensus conference by the authors (JJM, AJD and cDNA (100 ng) and 0.3 Â Syber Green (Molecular GR), which also reviewed the medical examiner’s Probes, Eugene, OR, USA) diluted from a 10 000 Â determination of suicide or other manner of death. stock solution, according to the manufacturer. PCR The samples in the study included: 72 suicides parameters were an initial denaturation at 95 1C for with major depression, four suicides without major 30 s followed by seven cycles with an annealing depression; 10 non-suicides with major depression, temperature ramping down to 53 1C. This was 12 non-suicides with other psychiatric diagnoses and followed by 40 cycles of 95 1C for 15 s, 53 1C for 20 s 68 non-suicides with no psychiatric diagnosis. With and 72 1C for 20 s. Each sample was run in triplicate such a sample we carried out the following compa- using the primers described above and primers risons: (1) 82 major depression; or (2) 86 depressed or specific for b-actin and GAPDH mRNAs. The TPH2 suicide were contrasted with a common control group results were normalized to curves obtained for of 80 non-major depression and non-suicide. The amplification of actin cDNA. The fluorescence of the major depression group was 49.8±18.7 years at time accumulated PCR products was acquired after each of death, 61% were male and 87.8% died by suicide. cycle. The lowest number of cycles yielding above In the non-major depression group there were 12 background was compared between the long and subjects with a psychiatric disorder, one subject had short forms. These measurements were combined to schizophrenia spectrum disorder and the remaining a mean value. 11 subjects had other psychiatric disorders (not related to MDD). The group with MDD had an average Statistical analysis of 2.4±2.7 prior episodes of major depression and All SNPs examined were tested for deviation from 48.8% (40 subjects) had other comorbid axis I or II Hardy–Weinberg proportions, where no deviation disorders. Brain tissue was either rapidly frozen after was observed. Single-locus analyses were performed removal at autopsy and stored at À80 1C or fixed and using both allelic and genotypic tests for association, stored in phosphate-buffered 10% formalin. Brain pH comparing allele and genotype frequencies between was measured from cerebellar tissue and was found to cases and controls. Statistical evaluations were be in the physiological range (ranging from 6.0 to 7.0). performed based on maximum likelihood methods, Postmortem interval for all samples was under 24 h. which are analogous to contingency table w2 approaches, All subjects for the quantitative studies were drug- but allow for more straightforward nesting and combi- free at the time of death as verified by peripheral and nation of hypotheses. Statistical significance was deter- brain toxicology. mined empirically by randomization, where we held the multi-locus genotypes for the set of loci fixed, and PCR and sequencing procedures randomized the assignment of individuals to case– PCR primers for the human TPH2 gene were selected control groups 10 000 times and re-evaluated each of based on the published genomic sequence and SNP the test statistics.25 The P-value for each statistic was information from NCBI; primer sequences are pro- defined as the proportion of replicates in which a given vided as supplementary data. The 15 ml PCRs con- statistic exceeded the value observed in the actual data.

Molecular Psychiatry Novel TPH2 gene variant associated with major depression F Haghighi et al 815 As multiple hypotheses were tested for multiple loci, leading to a multiple testing problem that is not easily dealt with by the conservative Bonferroni approach, the 10 000 replicates were used to evaluate the effect of such a multiple testing as well. For each test in each of the 10 000 replicates, the ‘P-value’ was Figure 1 The short and long isoforms of TPH2 in the computed, and for each replicate, the minimum brainstem of four non-psychiatric control subjects. RT-PCR P-value over the tests evaluated in that replicate. was used to identify the long (768 bp) and short (376 bp) The proportion of replicates with minimum P-value isoforms. The sequence of primers specific for the long lower than that found in the real data over all tests isoform (lanes 1–4) is forward 50-CTCTCCAAACTCTATCCC evaluated at a given locus was used as the corrected ACTC-30 and reverse 50-AGGCATCAAATCCCCAGATA-30. The short isoform (lanes 5–8) primer sequences are: P-value. These P-values were minimized over all 0 0 loci in the real data, and likewise in each replicate, to forward—5 -AGTGGGAAGCCCTATCCTGT-3 and reverse 50-CTTGACGGGGACTCATCTGT-30. MW is a 100 bp mole- evaluate P-values corrected for multiple loci consid- cular weight marker. ered. The global P-values were computed in a similar fashion, modelling all SNPs and hypotheses tested in each replicate and comparing to the observed amygdala from one representative case. Expression of data. Multilocus haplotype analysis was done using the short transcript was also seen in these regions PHASE case–control option,26 and genetic power (data not shown). Quantitative RT-PCR of the short analyses were performed using the Genetic Power and long forms in four non-psychiatric subjects Calculator web resource.27 revealed that compared with the long transcript there is relatively lower abundance of the short transcript Results (Figure 2) that seemed unrelated to PMI or brain pH. Since the major depression case sample population We conducted an exploratory study using 166 indivi- was drawn from a postmortem brain collection, duals who were genotyped for SNP loci spanning the majority of the subjects with major depressive both intron 5,19,20 and exon 11.12,13 Exon 11 is of disorder had committed suicide. Thus, we assessed interest because it encodes the catalytic site of the the potential associations between these gene variants enzyme and may reveal hypofunctional variants. and major depression separately and together with Therefore, we sequenced that region in all indivi- suicide, with the caveat that the major depression and duals. We observed no polymorphic sites in exon 11 suicide case samples in the two comparisons have after extensive sequencing of the 166 individuals, as substantial overlap. It is noteworthy that, this is the also observed in other studies.14–16 Thus, we did not only TPH2 mapping study, to our knowledge, where replicate the reported polymorphism in this region the control sample is carefully ascertained to select that is associated with altered levels of serotonin in subjects where a diagnosis of major depression has Balb/c mice and reported by the same group to be been ruled out and who have died from causes other present in 10% of subjects diagnosed with major than suicide. While the case sample may represent a depressive disorder.13 We did identify five previously spectrum of mood disorder and/or suicidal behavior, reported SNPs in intron 5: rs11316791, rs1843809, this control sample is an ideal complement. In this rs11386496, rs1386494 and rs1386493 and a novel exploratory aim we sought to elucidate the potential SNP, g.22879A > G. This is the region of the TPH2 relationship of such genetic variants to mood disorder gene predicted to undergo splicing into two transcript and/or suicide. We used one narrowly defined and isoforms, a short and a long form. As predicted by one more broadly defined clinical model, where affec- Ensemble annotation, these alternative transcripts, ted individuals: (1) were diagnosed with major ENST00000266669 and ENST00000333850, have depressive disorder; and (2) had major depression or 7 and 11 coding exons, respectively.28 The predicted died by suicide. The number of affected (case) and truncated transcript was based on evidence derived unaffected (control) individuals as defined by each from a study characterizing all human cDNAs in of these models is given in Table 1. We analyzed the a brain cDNA library.29 Depending on the annotation SNP data by conducting allelic tests for association considered, the variant, g.22879A > G, is in exon 6 and found two SNPs yielding statistically significant of the short transcript, or in intron 5 of the long results (Table 1). The first is locus rs1386493 that is transcript. We sought to validate the existence of the 2636 bp downstream from rs1386494, the locus predicted truncated TPH 2 transcript isoform experi- reported previously to be associated with depression mentally by RT-PCR. Primers specific to the short and and suicide.19,20 The second is the novel variant long transcripts were designed and used to amplify g.22879A > G discovered in our present study. The each isoform from cDNA prepared from human brain- strongest evidence for association was observed with stem. Both the long and short forms are present the major depression clinical model (rs1386493— in human brainstem (Figure 1). To determine whether P = 0.001 and g.22879A > G—P = 0.05), that remained the short form is present in serotonergic nerve termi- statistically significant after adjustment for multiple nals, RT-PCR was used to amplify the short and long models and markers tested (global empirical P = 0.01). forms from the prefrontal cortex, hippocampus and The broader clinical model including individuals

Molecular Psychiatry Novel TPH2 gene variant associated with major depression F Haghighi et al 816 c(T) Normalized for Actin coding region, including exons 1 and 7 through 11 of the long transcript, and the complete coding region of 25 exon 6 of the short transcript. Additionally, sequen- cing was fully completed for the 50 UTR, and was partially completed for the 30 UTR and the intronic regions throughout the gene. This effort was further 20 augmented by inclusion of existing SNPs in the public repository, including 57 known SNPs evenly distributed across the gene for maximal genomic coverage. Of these, 39 SNPs displayed a polymorph- ism in our sample. In addition, we discovered 16 15 novel SNPs by direct sequencing. In total, approxi- mately 13 000 base pairs of the TPH2 gene were sequenced for 78 individuals consisting of 48 cases with depression/suicide and 28 healthy controls. 10 Given this sample size, we can expect to detect 95% of the polymorphic sites with minor allele frequency of at least 1 and 99% of the sites with minor allele Cycle Threshold [c(T)] frequency of at least 5%.30 As such, we have conducted an in-depth survey in an effort to detect 5 potentially rare polymorphisms that may alter gene function through coding changes or other mechan- isms. It is expected that the polymorphisms that contribute to phenotypic variation are generally rare, for example most coding SNPs that lead to amino acid 0 changes have minor allele frequency under 5%.31,32 Short Long Such low frequency variations are underrepresented Figure 2 Quantitative RT-PCR of average short versus long in the public repositories and thus these data are a forms in brainstem from four non-psychiatric controls (same substantial resource for future genetic studies of controls as above). Each bar represents the lowest mean depression and suicide within the Caucasian popula- cycle number (c(T)) normalized for b-actin at which tion from which our samples were drawn. To this end, fluorescent signals above background were detected for we have estimated haplotype block structure of the the short and long forms with a mean 8.6 fold higher TPH2 gene and generated tag-SNPs using our popula- expression level of long versus short (t = 2.10, P = 0.01). The tion-based controls (Figure 3). Although numerous fold difference is calculated by determining the difference haplotype block definitions are possible, we adopted between the lowest mean cycle number and raising that the more conservative four-gamete test,33 yielding a difference to the second power. The expression of the long form is more abundant than the short form as demonstrated total of nine blocks ranging from 88 to 15 335 bp in by the lower c(T). length. The associations with major depression reported previously19,20 and in the present study, fall entirely within the second block. Within these with either major depression or suicide as affected haplotype blocks, spanning 44 SNPs, we identified also gave significant results (rs1386493—P = 0.003 15 tag-SNPs (Table 3). Despite the block definition and g.22879A > G—P = 0.05, Table 1) with global used, the tag-SNPs are generally consistent over 80% empirical P = 0.04. This is not surprising, since the of the time,34 and we observed this trend in our data major depression model alone and the broader as well when exploring other block definitions (data depression or suicide model have a high degree of not shown). Thus within each block that contains overlap. Therefore, for the subsequent multilocus more than one SNP, only a few tag-SNPs need to be association analyses of the data, we focused on the genotyped to capture the entire block. Thus this study depression alone clinical model. For the loci with provides a haplotype map to guide exploratory observed and reported significant associations (that investigations of other disorders with which TPH2 is, rs1386494, rs1386493, g.22879A > G) we compared is potentially associated. overall estimated haplotype frequencies among cases and controls and found a statistically significant Discussion difference with a global empirical P = 0.01, where the haplotype CGG consisting of the putative risk These results provide evidence for association of variants, was almost three times more frequent in the TPH2 with major depression and/or suicide, since the case sample, with estimated frequencies of 0.065/ majority of our cases had major depression and had 0.024 in case–control samples respectively (Table 2). committed suicide, and therefore support previous To aid the process of a systematic study of this gene, reports of associations of a TPH2 variant (rs1386494) we have characterized the genetic architecture of the with both depression19 and suicide20 in two separate TPH2 gene through systematic sequencing of the samples of depressives and suicides (where depres-

Molecular Psychiatry Novel TPH2 gene variant associated with major depression F Haghighi et al 817 Table 1 Single-locus association results, reporting empirical P-values (P) for allelic test for association

Locus ID (Alleles major > minor) Position CaseMAF ControlMAF P

Depression clinical model (case = 82 and control = 80) rs11316791 [À/T] 15990 0.082 0.123 rs1843809 G > T 16073 0.089 0.115 rs11386496 A > G 18165 0.100 0.137 rs1386494 C > T 19918 0.089 0.138 rs1386493 G > A 22554 0.072 0.213 0.001 g.22879A > G A > G 22879 0.066 0.017 0.05

Depression or suicide clinical model (case = 86 and control = 80) rs11316791 [À/T] 15990 0.096 0.123 rs1843809 G > T 16073 0.102 0.115 rs11386496 A > G 18165 0.115 0.137 rs1386494 C > T 19918 0.097 0.138 rs1386493 G > A 22554 0.082 0.213 0.003 g.22879A > G A > G 22879 0.062 0.017 0.05

The results are presented for three clinical models with the models including major depression yielding the most significant results. For each SNP the position relative to the transcription start site and minor allele frequencies (MAF) for both cases and control groups are given.

Table 2 Multilocus association including rs1386494, more common allele would be even more common rs1386493, and g.22879A > G, comparing overall estimated in the disease group, we computed the power of haplotype frequencies for the case and control groups with observing such a disease model in our sample. For the the clinical model being major depression depression clinical model the power is 0.1% for Type I error rate of 0.001, assuming a lifetime prevalence of Haplotype Case frequency (s.e.) Control frequency (s.e.) 5% for major depression35 (a conservative lower bound estimate used for the remaining power ana- CGA 0.84 (0.011) 0.79 (0.0132) CGG 0.06 (0.008) 0.02 (0.009) lyses for this discussion), and two-fold increase in CAA 0.011 (0.006) 0.04 (0.01) relative risk for depression with a dominant model. CAG p0.0002 p0.0002 Considering a recessive model with the same assump- TGA 0.014 (0.004) 0.001 (0.003) tions, the power only increases to 2.6% (the other TAA 0.07 (0.007) 0.14 (0.010) clinical models follow similar trends). For such a model the sample necessary for 80% power is The estimated haplotype frequencies and s.e. are reported B189 500 cases and the equivalent number of for each group. controls, assuming a dominant model (755 is the estimate for the recessive model). Clearly, both this study, and the previously published studies,19,20 sion status of the suicide subjects were not reported). are underpowered in detecting such common risk We also examined our data in a suicide alone clinical variants contributing to disease susceptibility given model and found no significant association after the available sample sizes. The low power raises the multiple testing correction (data not shown). This possibility that the observed findings for these loci might suggest that TPH2 plays a role in major might be spurious. In contrast, the novel variant depression, but not suicide. g.22879A > G yields a power of 41% assuming a Our findings with a downstream novel variant dominant model for relative risk of disease (using (g.22879A > G) suggest a potentially more important the same model parameters above). To detect such association. This is because a more critical evaluation a variant with 80% power, a total of 215 cases and indicates that the association with rs1386493 might matched controls would be required, a far more be spurious. Specifically, the direction of association feasible sample size than the previously published for marker rs1386493 in this study as well as SNPs. Therefore, our findings with g.22879A > G are previously reported studies19,20 indicates the poten- very promising and warrant investigation in a larger tial presence of a common risk allele, as observed sample. Additionally, the comprehensive genetic in the pattern of allelic and genotypic frequencies characterization of the TPH2 candidate gene reveals in cases versus controls, where the major allele is that future studies of major depression should focus more common in cases than controls. This trend is on the second haplotype block containing such observed for both major depression and suicide (for variants of interest (Figure 3). example, 0.93/0.79 corresponding to cases/controls Cross-species examination revealed that human, within the depression clinical model). In a model for chimpanzee, dog, mouse and rat show conserved disease susceptibility one would not predict that the non-coding sequences spanning the human TPH2

Molecular Psychiatry Novel TPH2 gene variant associated with major depression F Haghighi et al 818 a

E-4 b B1

2 rs1386496 B2 r rs2129575 rs1386494 B1 c rs11316791 E-5 rs1386493 B2 g.228879A>G B3 g.29627delinsTT rs1843809 rs1386492 rs1386491 B4 g.34762A>G E-6 g.34766C>G B3 rs1843810 rs7305115 rs1843811 B5 rs1023990 B6 rs1843812 B7 E-7 rs11179031 rs1352250 rs1023989 B8 B5 g.65164T>C rs10506645 rs1352251 rs1007023 E-8 B9 rs11179045 g.56119C>T B4 B7 rs9325202 rs4760754 rs11179046 rs1386497 rs12228278 rs10506646 g.65025C>T |D’| rs1487275 rs10506648 g.68122G>A B6 rs1386486 rs1487278 rs1386485 g.79764G>A rs12231356 rs1386483 E-9 rs10879353 B8 B9 g.79885C>T rs110879354 rs1386482 rs11615016 E-10

rs4290270 E-11 rs1487279

Figure 3 (a) Haplotype blocks representing the four most frequent haplotypes color coded by frequencies of each haplotype, where each SNP corresponds to a letter (A, G, T, or C) or an insertion/deletion polymorphism (I or D). Letters grouped by parentheses are SNPs within the same block. These blocks were determined using the program HaploBlockFinder38 with haplotypes inferred from 28 control subjects with the statistical software package PHASE, version 2.1.1,26 using default settings for 49-SNPs with minor allele frequency X0.01. Haplotype block boundaries were defined based on the four-gamete test, where a block is defined as a region in which fewer than four gametes are observed for all possible pairs of SNPs examined (D0 = 1—reflecting evidence for potential absence of historical recombination).33 (b) Shown as vertical bars consecutively to the left of the LD matrix (with pairwise |D0| and r2 values) from top to bottom, these blocks are also depicted in (a) with SNP groupings denoted by B1–B9 for blocks with at least two SNP loci. (c) Schematic representation of the SNP coverage across the TPH2 gene with boxes denoting exons ranging from E4 to E11 with connecting lines corresponding to the intronic regions, and tag-SNPs within blocks bold and italicized.

intron 5 region (as seen by the conservation track from site. Such a truncated isoform might regulate TPH2 UCSC Genome Browser based on the May 2004 draft activity via post-transcriptional interactions with the of the ). This confers evidence for long isoform or with other mediators with which the presence of conserved elements with regulatory it may compete, given that the first 5 exons of the potential, and of particular interest is our finding that short and long TPH2 isoforms are identical. There this newly discovered variant resides in a conserved are precedents for such interactions of truncated genomic region. This variant and others in this region products.36,37 Together the short and long forms may may affect generation of short and long isoforms of regulate local levels of serotonin in serotonergic the enzyme with functional consequences for catalytic terminal field targets. Further investigations of poly- activity. We have demonstrated experimentally that morphisms at the splice junction flanking exon 6 or in both the short and long TPH2 isoforms are present the coding/30UTR regions of the short transcript such in all brain regions tested including the brainstem. The as g.22879A > G, may reveal variant isoforms with role of these TPH2 transcripts in serotonergic target potentially aberrant functional effect. Thus, the possi- regions is unknown. One possibility is that the short bility that the short isoform might play a critical role form acts in a regulatory manner and has limited or no in the rate of brain serotonin synthesis and homeo- enzymatic function since it has a truncated catalytic stasis, should motivate exploration of gene expression

Molecular Psychiatry Novel TPH2 gene variant associated with major depression F Haghighi et al 819 Table 3 Distribution of known and novel SNPs characterized in 28 control samples with assignment of haplotype blocks and block lengths denoted in base pairs following the block numbers

Locus ID Block ID position Locus position MAF rs2129575 B1-8543 70626340 7448 0.143 rs11316791 70634882 15990 0.232 rs1843809 70634965 16073 0.232 rs1386496 B2-11589 70637057 18165 0.232 rs1386494 70638810 19918 0.232 rs1386493 70641446 22554 0.232 g.22879A > G 70641771 22879 0.018 g.25627D > I 70644519 25627 0.018 rs1386492 70648532 29640 0.268 rs1386491 70648645 29753 0.161 g.34762A > G B3-88 70653654 34762 0.161 g.34766G > C 70653658 34766 0.179 rs1843810 70653687 34795 0.214 rs1843811 70653741 34849 0.411 rs1843812 70653922 35030 0.232 rs7305115 70659129 40237 0.375 rs1023990 70668514 49622 0.179 rs11179031 B4-15335 70668583 49691 0.018 rs1023989 70668596 49704 0.161 rs10506645 70671767 52875 0.143 rs1007023 70674641 55749 0.214 g.56119T > C 70675011 56119 0.018 rs4760754 70675106 56214 0.429 rs1386497 70678557 59665 0.196 rs10506646 70682759 63867 0.161 g.65025T > C 70683917 65025 0.018 rs1352250 B5-111 70684051 65159 0.411 g.65164C > T 70684056 65164 0.018 rs1352251 70684161 65269 0.411 g.68122A > G B6-105 70687014 68122 0.036 rs1487278 70687118 68226 0.143 rs11179045 B7-322 70693670 74778 0.036 rs9325202 70693744 74852 0.375 rs11179046 70693968 75076 0.375 rs12228278 70693991 75099 0.036 rs12231356 B8-235 70695815 76923 0.107 rs10879353 70696047 77155 0.446 rs110879354 70696049 77157 0.446 rs1487275 70696559 77667 0.321 rs10506648 B9-7656 70697918 79026 0.018 rs1386486 70698487 79595 0.429 rs1386485 70698634 79742 0.429 g.79764A > G 70698656 79764 0.018 rs1386483 70698761 79869 0.429 g.79885T > C 70698777 79885 0.054 rs1386482 70698839 79947 0.429 rs11615016 70702261 83369 0.089 rs4290270 70702502 83610 0.446 rs1487279 70705573 86681 0.411

The tag-SNPs are denoted in bold italics. For each SNP chromosome-12 positions are designated using the May 2004 draft of the human genome, along with minor allele frequencies (MAF). differences within brains of case and control subjects, functional studies investigating the role of TPH2 in which will be investigated in future studies. Taken disease etiology. In addition, the collection of 16 novel together these results provide evidence for TPH2 as variants discovered in this study will significantly a candidate gene for depression and/or suicide for contribute towards future studies of such disorders as future investigations. To this end, we have constructed major depression, bipolar disorder, schizophrenia, a haplotype map of the TPH2 gene and identified tag- alcoholism, drug abuse, aggression and suicide, with SNPs that will be valuable in numerous genetic and which TPH2 is associated.

Molecular Psychiatry Novel TPH2 gene variant associated with major depression F Haghighi et al 820 Acknowledgments 17 Bicalho MA, Pimenta GJ, Neves FS, Correa H, de Moraes EN, De Marco L et al. Genotyping of the G1463A (Arg441His) TPH2 This study was supported by PHS Grants MH40210, polymorphism in a geriatric population of patients with major MH64168, MH63749, HG002915 and MH62185. We depression. Mol Psychiatry 2006; 11: 799–800. thank Drs A Dumas, B Mancevski and T Serafimova 18 Kelly TM, Mann JJ. Validity of DSM-III-R diagnosis by psycho- for contributing to collection of specimens and logical autopsy: a comparison with clinician ante-mortem diagnosis. Acta Psychiatr Scand 1996; 94: 337–343. psychological autopsies. 19 Zill P, Baghai TC, Zwanzger P, Schule C, Eser D, Rupprecht R et al. SNP and haplotype analysis of a novel tryptophan hydroxylase isoform (TPH2) gene provide evidence for association with major References depression. Mol Psychiatry 2004; 9: 1030–1036. 20 Zill P, Buttner A, Eisenmenger W, Moller HJ, Bondy B, Ackenheil 1 Milak MS, Parsey RV, Keilp J, Oquendo MA, Malone KM, Mann JJ. M. Single nucleotide polymorphism and haplotype analysis of Neuroanatomic correlates of psychopathologic components of a novel tryptophan hydroxylase isoform (TPH2) gene in suicide major depressive disorder. Arch Gen Psychiatry 2005; 62: 397–408. victims. Biol Psychiatry 2004; 56: 581–586. 2 Stockmeier CA. Involvement of serotonin in depression: evidence 21 Keilp JG, Waniek C, Goldman RG, Zemishlany Z, Alexander GE, from postmortem and imaging studies of serotonin receptors and Gibbon M et al. Reliability of post-mortem chart diagnoses of the serotonin transporter. J Psychiatr Res 2003; 37: 357–373. schizophrenia and dementia. Schizophr Res 1995; 17: 221–228. 3 Mann J, Arango V. Abnormalities of Brain Structure and Function 22 First MB, Spitzer RL, Gibbon M, Williams JMG, Benjamin L. in Mood Disorder. Oxford University Press: San Francisco, 2003. Structured Clinical Interview for DSM-IV Axis II Personality 4 Sullivan PF, Neale MC, Kendler KS. Genetic epidemiology of Disorders (SCID II), Version 2.0. Biometrics Research Department, major depression: review and meta-analysis. Am J Psychiatry New York State Psychiatric Institute: New York, 1996. 2000; 157: 1552–1562. 23 Goldsmith SK, Pellmar TC, Kleinman AM, Bunney WE (eds). 5 Statham DJ, Heath AC, Madden PA, Bucholz KK, Bierut L, Reducing Suicide: A National Imperative. National Academies Dinwiddie SH et al. Suicidal behaviour: an epidemiological and Press: Washington, D.C., 2002. genetic study. Psychol Med 1998; 28: 839–885. 24 Spitzer RL, Williams JBW, Gibbon M, First MB. Structured Clinical 6 Higley JD, Thompson WW, Champoux M, Goldman D, Hasert MF, Interview for DSM-III-R. Patient edition (SCID-P). American Kraemer GW et al. Paternal and maternal genetic and environ- Psychiatric Press: Washington, D.C., 1990. mental contributions to cerebrospinal fluid monoamine metabo- 25 Emahazion T, Feuk L, Jobs M, Sawyer SL, Fredman D, St Clair D lites in rhesus monkeys (Macaca mulatta). Arch Gen Psychiatry et al. SNP association studies in Alzheimer’s disease highlight 1993; 50: 615–623. problems for complex disease analysis. Trends Genet 2001; 17: 7 Rogers J, Martin LJ, Comuzzie AG, Mann JJ, Manuck SB, Leland M 407–413. et al. Genetics of monoamine metabolites in baboons: overlapping 26 Stephens M, Smith NJ, Donnelly P. A new statistical method for sets of genes influence levels of 5-hydroxyindolacetic acid, haplotype reconstruction from population data. Am J Hum Genet 3-hydroxy-4-methoxyphenylglycol, and homovanillic acid. Biol 2001; 68: 978–989. Psychiatry 2004; 55: 739–744. 27 Purcell S, Cherny SS, Sham PC. Genetic Power Calculator: design 8 Mann JJ, Arango V, Marzuk PM, Theccanat S, Reis DJ. Evidence for of linkage and association genetic mapping studies of complex the 5-HT hypothesis of suicide. A review of post-mortem studies. traits. Bioinformatics 2003; 19: 149–150. Br J Psychiatry 1989; 155(Suppl 8): 7–14. 28 Hubbard T, Barker D, Birney E, Cameron G, Chen Y, Clark L et al. 9 Bach-Mizrachi H, Underwood MD, Kassir SA, Bakalian MJ, Sibille The Ensembl genome database project. Nucleic Acids Res 2002; E, Tamir H et al. Neuronal tryptophan hydroxylase mRNA 30: 38–41. expression in the human dorsal and median raphe nuclei: major 29 Ota T, Suzuki Y, Nishikawa T, Otsuki T, Sugiyama T, Irie R et al. depression and suicide. Neuropsychopharmacology 2006; 31: Complete sequencing and characterization of 21 243 full-length 814–824. human cDNAs. Nat Genet 2004; 36: 40–45. 10 Underwood MD, Khaibulina AA, Ellis SP, Moran A, Rice PM, 30 Kruglyak L, Nickerson DA. Variation is the spice of life. Nat Genet Mann JJ et al. Morphometry of the dorsal raphe nucleus seroto- 2001; 27: 234–236. nergic neurons in suicide victims. Biol Psychiatry 1999; 46: 31 Cargill M, Altshuler D, Ireland J, Sklar P, Ardlie K, Patil N et al. 473–483. Characterization of single-nucleotide polymorphisms in coding 11 Boldrini M, Underwood MD, Mann JJ, Arango V. More tryptophan regions of human genes. Nat Genet 1999; 22: 231–238. hydroxylase in the brainstem dorsal raphe nucleus in depressed 32 Halushka MK, Fan JB, Bentley K, Hsie L, Shen N, Weder A et al. suicides. Brain Res 2005; 1041: 19–28. Patterns of single-nucleotide polymorphisms in candidate genes 12 Zhang X, Beaulieu JM, Sotnikova TD, Gainetdinov RR, Caron MG. for blood-pressure homeostasis. Nat Genet 1999; 22: 239–247. Tryptophan hydroxylase-2 controls brain serotonin synthesis. 33 Wang N, Akey JM, Zhang K, Chakraborty R, Jin L. Distribution of Science 2004; 305: 217. recombination crossovers and the origin of haplotype blocks: the 13 Zhang X, Gainetdinov RR, Beaulieu JM, Sotnikova TD, Burch LH, interplay of population history, recombination, and mutation. Am Williams RB et al. Loss-of-function mutation in tryptophan J Hum Genet 2002; 71: 1227–1234. hydroxylase-2 identified in unipolar major depression. Neuron 34 Zeggini E, Barton A, Eyre S, Ward D, Ollier W, Worthington J et al. 2005; 45: 11–16. Characterisation of the genomic architecture of human chromo- 14 Glatt CE, Carlson E, Taylor TR, Risch N, Reus VI, Schaefer CA. some 17q and evaluation of different methods for haplotype block Response to Zhang et al. (2005): loss-of-function mutation in definition. BMC Genet 2005; 6:21. tryptophan hydroxylase-2 identified in unipolar major depression. 35 Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M, Neuron 45 11–16. Neuron 2005; 48: 704–705; author reply Eshleman S et al. Lifetime and 12-month prevalence of 705–706. DSM-III-R psychiatric disorders in the United States. Results from 15 Van Den Bogaert A, De Zutter S, Heyrman L, Mendlewicz J, the National Comorbidity Survey. Arch Gen Psychiatry 1994; 51: Adolfsson R, Van Broeckhoven C et al. Response to Zhang et al. 8–19. (2005): loss-of-function mutation in tryptophan hydroxylase-2 36 Lopez AJ. Developmental role of transcription factor isoforms identified in unipolar major Depression. Neuron 45, 11–16. generated by alternative splicing. Dev Biol 1995; 172: 396–411. Neuron 2005; 48: 704; author reply 705–706. 37 Stasiv Y, Regulski M, Kuzin B, Tully T, Enikolopov G. The 16 Zhou Z, Peters EJ, Hamilton SP, McMahon F, Thomas C, McGrath Drosophila nitric-oxide synthase gene (dNOS) encodes a family of PJ et al. Response to Zhang et al. (2005): loss-of-function mutation proteins that can modulate NOS activity by acting as dominant in tryptophan hydroxylase-2 identified in unipolar major depres- negative regulators. J Biol Chem 2001; 276: 42241–42251. sion. Neuron 45, 11–16. Neuron 2005; 48: 702–703; author reply 38 Zhang K, Jin L. HaploBlockFinder: haplotype block analyses. 705–706. Bioinformatics 2003; 19: 1300–1301.

Molecular Psychiatry