Polymorphisms in the Trace Amine Receptor 4 (TRAR4) Gene on Chromosome 6q23.2 Are Associated with Susceptibility to Schizophrenia

Abstract

Several linkage studies across multiple population groups provide convergent support for a susceptibility locus for schizophrenia—and, more recently, for bipolar disorder—on chromosome 6q13-q26. We genotyped 192 European-ancestry and African American (AA) pedigrees with schizophrenia from samples that previously showed linkage evidence to 6q13-q26, focusing on the MOXD1-STX7-TRARs gene cluster at 6q23.2, which contains a number of prime candidate genes for schizophrenia. Thirty-one screening single-nucleotide polymorphisms (SNPs) were selected, providing a minimum coverage of at least 1 SNP/20 kb. The association observed with rs4305745 (P=.0014) within the TRAR4 (trace amine receptor 4) gene remained significant after correction for multiple testing. Evidence for association was proportionally stronger in the smaller AA sample. We performed database searches and sequenced genomic DNA in a 30-proband subsample to obtain a high-density map of 23 SNPs spanning 21.6 kb of this gene. Single-SNP analyses and also haplotype analyses revealed that rs4305745 and/or two other polymorphisms in perfect linkage disequilibrium (LD) with rs4305745 appear to be the most likely variants underlying the association of the TRAR4 region with schizophrenia. Comparative genomic analyses further revealed that rs4305745 and/or the associated polymorphisms in complete LD with rs4305745 could potentially affect gene expression. Moreover, RT-PCR studies of various human tissues, including brain, confirm that TRAR4 is preferentially expressed in those brain regions that have been implicated in the pathophysiology of schizophrenia. These data provide strong preliminary evidence that TRAR4 is a candidate gene for schizophrenia; replication is currently being attempted in additional clinical samples.

Introduction

Schizophrenia is a frequently chronic and devastating brain disorder that affects ∼1% of the population worldwide (Jablensky et al. 1992). Typically, it presents in adolescence or young adulthood and is characterized by major disruptions of thinking (delusions and/or disorganization), perception (hallucinations), mood, and behavior (Gottesman and Shields 1982). Schizophrenia is strongly familial, with a heritability of ∼80%, but its etiology is hypothesized to involve both genetic and environmental factors (Sanders and Gejman 2001). Recently, encouraging evidence for several genes potentially involved in the etiology of schizophrenia has been reported—namely, dysbindin (DTNBP1) (Straub et al. 2002; Schwab et al. 2003), neuregulin 1 (NRG1) (Stefansson et al. 2002, 2003; Williams et al. 2003), proline dehydrogenase (oxidase) 1 (PRODH) (Jacquet et al. 2002; Liu et al. 2002), catechol-O-methyltransferase (COMT) (Li et al. 2000; Egan et al. 2001; Shifman et al. 2002), regulator of G-protein signaling 4 (RGS4) (Chowdari et al. 2002; Morris et al. 2004; Williams et al. 2004), D-amino-acid oxidase activator (DAOA [previously called G72]) (Chumakov et al. 2002; Schumacher et al. 2004), and D-amino-acid oxidase (DAO) (Chumakov et al. 2002; Schumacher et al. 2004). Additionally, Brzustowicz et al. (2004) reported linkage disequilibrium between C-terminal PDZ domain ligand of neuronal nitric oxide synthase (CAPON) and schizophrenia, perhaps related to a previous association signal at D1S1679, which is ∼25 kb distal to CAPON (Rosa et al. 2002). Most of the aforementioned genes are positional candidates with likely involvement in dopaminergic or N-methyl D-aspartate (NMDA) brain mechanisms.

We previously reported linkage of schizophrenia to chromosome 6q13-q26 (SCZD5 [MIM 603175]) (Cao et al. 1997), which has accumulated support from a number of studies (Kaufmann et al. 1998; Martinez et al. 1999; Bailer et al. 2000; Levinson et al. 2000; Lindholm et al. 2001; Lerer et al. 2003; Lewis et al. 2003). In the first report of linkage to 6q (Cao et al. 1997), support for linkage was observed in the region from D6S301 (located at 111 cM) to D6S305 (located at 170 cM). Linkage to this region was confirmed in subsequent studies (Kaufmann et al. 1998; Martinez et al. 1999; Levinson et al. 2000). All the families used in this study are from three data sets, which we call “NIMH-IRP” (National Institute of Mental Health Intramural Research Program), “NIMH-GI” (NIMH Genetics Initiative), and “AU/US” (Australia/United States), that previously have been shown to yield evidence for linkage to chromosome 6q13-q26 (SCZD5) (Cao et al. 1997; Martinez et al. 1999; Levinson et al. 2000). An analysis of 12 microsatellite markers in a 50-cM 6q region was performed by Martinez et al. (1999); nonparametric affected sibling pair methods yielded P values of .00018, .00095, and .013 for the NIMH-IRP, NIMH-GI, and AU/US data sets, respectively. In a Palestinian/Israeli pedigree sample (Lerer et al. 2003), the linkage peak with a nonparametric linkage (NPL) score of 4.61 was at D6S292 (∼137 cM), and the region from ∼131 cM to ∼144 cM contained the 1-NPL decrease portion of the linkage peak. Furthermore, another group found evidence for linkage in a Swedish pedigree to the region from ∼170 cM to ∼180 cM (Lindholm et al. 2001). Whether these different results are better explained by the presence of more than one schizophrenia susceptibility gene in 6q or reflect typical peak variability in complex disorders (Hauser and Boehnke 1997; Hsueh et al. 2001) is currently unknown. Recently, bipolar disorder was reported to map to 6q, with one study yielding a maximum LOD score of 2.2 at 113 cM near D6S1021 (Dick et al. 2003) and another study reporting a maximum LOD score of 3.56 at ∼124–126 cM near D6S1639 (Middleton et al. 2004), raising the possibility that a common gene for schizophrenia and bipolar disorder may be located in 6q.

We were interested in a MOXD1-STX7-TRARs gene cluster at 6q23.2 (132.8 cM) that harbors prime candidates for schizophrenia (fig. 1): MOXD1 (monooxygenase, dopamine-β-hydroxylase-like 1) (Chambers et al. 1998), STX7 (syntaxin 7) (Wang et al. 1997), and all known human trace amine receptor (TRAR) genes—namely, TRAR1, TRAR3, TRAR4, TRAR5, PNR (putative neurotransmitter receptor gene) (Zeng et al. 1998; Borowsky et al. 2001; Bunzow et al. 2001; Lee et al. 2001), and three TRAR pseudogenes (TRAR2 , GPR57, and GPR58 [GPR57 and GPR58 are G-protein–coupled receptor pseudogenes]) (Liu et al. 1998; Borowsky et al. 2001; Bunzow et al. 2001; Lee et al. 2001).

Figure 1.

Genomic structure of the 6q23.2 gene cluster and association mapping of the initial screening. The genomic positions are based on the UCSC July 2003 assembly of the human genome (see the UCSC Genome Bioinformatics Web site). a, The relative position of the 6q23.2 gene cluster to the peak markers from various linkage studies: D6S424 (Cao et al. 1997; Martinez et al. 1999), D6S416 (Cao et al. 1997), D6S292 (Lerer et al. 2003), and D6S264 (Lindholm et al. 2001). b, Genes in the 6q23.2 gene cluster. c, The −log transformation of the FBAT P value for the 31 SNP markers analyzed in the initial association screening. Each data point of the markers points to its relative position in the gene cluster shown in panel b. The most significant marker is rs4305745, with a P value of .0014 (see table A5 [online only], for FBAT P values and other detailed information for all the initially selected markers, and table 1, for the single-marker association results for all the additional TRAR4 markers examined in the dense mapping effort).

Trace amines (TAs) are endogenous amine compounds that are chemically similar to classic biogenic amines like dopamine (DA), norepinephrine, serotonin, and histamine. Abnormalities that involve the classic biogenic amines are the basis for biological hypotheses for a wide variety of disorders, including dystonias, Parkinson disease, schizophrenia, drug addiction, and mood disorders. In mammals, TAs are present at low levels, with no apparent dedicated synapses, but a blockage of amine degradation leads to significant accumulations of TAs, suggesting high synthesis and turnover, as reviewed elsewhere (Premont et al. 2001). TAs in mammals include tyramine (TYR), tryptamine, β-phenylethylamine (β-PEA), and octopamine (OCT) (Branchek and Blackburn 2003), and all are synthesized from amino acid precursors by the aromatic amino acid decarboxylase.

TAs were thought to be “false transmitters,” which displace classic biogenic amines from their storage and act on transporters in a fashion similar to the amphetamine (Parker and Cubeddu 1986), but the identification of brain receptors specific to TAs indicates that they also have effects of their own (Borowsky et al. 2001). This might explain the fact that, although TYR, β-PEA, OCT, and amphetamine would require the integrity of vesicular stores of DA if displacement of DA were their only mechanism of action, they (except OCT) are still active when DA is depleted (Baud et al. 1985). TRARs bind amphetamine, MDMA (3,4-methylenedioxymethamphetamine, known as “ecstasy”), and LSD (D-lysergic acid diethylamide) with high affinity. This suggests a direct link between TRARs and mechanisms of psychosis, because the administration of amphetamine can induce a schizophrenia-like psychosis (Connell 1958; Snyder et al. 1967; Angrist et al. 1974; Laruelle and Abi-Dargham 1999). In addition, psychedelic experiences induced by LSD can have a remarkable similarity to schizophrenia (Vardy and Kay 1983; Gouzoulis et al. 1994). Furthermore, LSD can induce habituation deficits (the normal decrease in response magnitude to repeated stimuli over time) that are similar to those exhibited by patients with schizophrenia (Geyer and Braff 1987; Braff et al. 1992).

MOXD1 is a homologue of dopamine-β-hydroxylase that is potentially involved in the biosynthesis of norepinephrine from DA (Chambers et al. 1998). Syntaxin 7 (STX7) is a critical component of the synaptic protein complex SNARE (receptor for soluble N-ethylmaleimide-sensitive factor attachment proteins), which is involved in NMDA receptor and dopaminergic receptor function (Pei et al. 2004); the dysfunction of SNARE has been suggested in schizophrenia (Honer et al. 2002). Specifically, syntaxins mediate vesicle fusion in vesicular transport processes (Teng et al. 2001). We investigated DNA polymorphisms in the MOXD1-STX7-TRARs cluster by use of family-based association methods, and here we present preliminary evidence of association between TRAR4 and schizophrenia.

Material and Methods

Subjects and Phenotypes

Three samples were studied, which we call the “NIMH-IRP,” “NIMH-GI,” and “AU/US” collections. Ascertainment of the NIMH-IRP sample was initially described by Gershon et al. (1988). The full sample—from which the present sample of 67 pedigrees was drawn—was described later (Cao et al. 1997; Gejman et al. 2001). The collection of the NIMH-GI sample was described in a report of a genome scan of 71 pedigrees (Cloninger et al. 1998), and additional NIMH-GI families were subsequently included in the repository-based data set (see the “Electronic-Database Information” section). Of these NIMH-GI pedigrees, 69 were used in the present analysis and for two previous ones (Cao et al. 1997; Martinez et al. 1999). The AU/US sample was described initially in a report of a genome scan of 43 pedigrees (Levinson et al. 1998); full or partial trios for the present study were from 56 of the 71 pedigrees in the expanded sample used in linkage fine-mapping studies (Mowry et al. 2000) and in additional analyses of this data set (Martinez et al. 1999; Levinson et al. 2000). For the present study, we genotyped a total of 827 individuals from 192 families (67 NIMH-IRP families, 69 NIMH-GI families, and 56 AU/US families). Details are provided in table A1 (online only) and in the “Electronic-Database Information” section. Schizophrenia and schizoaffective disorder were diagnosed using the criteria of the DSM-IIIR (American Psychiatric Association 1987). The institutional review board of the Evanston Northwestern Healthcare Research Institute approved this study.

SNP Selection and Genotyping

SNPs were selected from public databases with the help of a bioinformatics tool, SNPper (Riva and Kohane 2002), and novel TRAR4 SNPs were identified by direct sequencing. The DNA samples were genotyped using one of two methods: (1) template-directed dye-terminator incorporation with fluorescence-polarization detection (FP-TDI) (Chen et al. 1999) or (2) the TaqMan assay developed by Applied Biosystems. For the FP-TDI assays, after PCR amplification of genomic DNA, the AcycloPrime-FP SNP detection kit (PerkinElmer) was used for post-PCR cleanup and the single-base extension reaction. We detected FP by use of either an Analyst fluorescence reader (LJL Biosystems) or a Wallac Victor3 (PerkinElmer), and FP data were converted to genotypes with the assistance of an automated genotype-calling spreadsheet (Akula et al. 2002). PCR primers and probes for the FP-TDI assays were designed using Primer3 (Rozen and Skaletsky 2000). For the TaqMan assays, the genomic sequence flanking the SNP was submitted to Applied Biosystems for development of an assay-by-design. Each unique TaqMan minor-groove–binding (MGB) allele-specific probe was labeled with either a 5′-FAM or a 5′-VIC reporter dye. PCR amplification of genomic DNA was performed in a 384-well plate in an ABI Prism 7900 or a DNA Engine Tetrad 2 (MJ Research). After PCR, the allele discrimination was performed on an ABI Prism 7900 Sequence Detection System by use of Sequence Detector Software (SDS), version 2.0. Standard genotype calling was converted by a customized spreadsheet. Nucleotide sequences for the PCR primers, the FP-TDI and TaqMan probes, and related information for each marker can be found in table A2 (online only).

The average completion rate of our experiments was 96%. To empirically check for errors in the genotyping method, we compared the results for marker rs4305745 from both methods; the difference rate between FP-TDI and the TaqMan assay was ∼0.25% (data not shown). We used MERLIN (Multipoint Engine for Rapid Likelihood Inference) (Abecasis et al. 2002) (with all the SNPs at once) to check for Mendelian inconsistencies, blanked them as described below (sometimes for individuals and sometimes for the family, when the error could not be traced to a particular individual), and then addressed all unlikely recombinants. Genotyping errors were detected for 0.17% of genotypes (MERLIN) (95 errors/54,611 nonzero genotypes), including 26 Mendelian inconsistencies (0.047%) and 69 unlikely recombinants (0.12%). We did not change genotypes for unlikely recombinants unless (1) MERLIN estimated a high probability of an individual genotype error, compared with other possible errors (see MERLIN documentation for details), and/or (2) manual rereading of each genotype tracing or other raw genotyping output for the family/marker in question pinpointed a specific error for a particular individual. All genotype errors (Mendelian inconsistencies or specific errors that resulted in unlikely recombinants) were blanked (zeroed) for the involved individuals, and, given our high genotyping completion rate and low genotyping error, we did not perform a second-pass genotyping procedure for these individuals. Genotypes were read blind to psychiatric status. We checked Hardy-Weinberg equilibrium (HWE) for family founders (at least 200 individuals) for all 55 SNPs. Three MOXD1 SNPs were found not to be in HWE (rs2206064, rs1981187, and rs2275394 [although the HWE P value for rs2206064 did not remain significant after the number of markers examined for HWE was taken into account]). The minor-allele frequency of rs2206064 was only 2%, which might explain the lack of HWE. The frequencies of homozygotes for both rs1981187 and rs2275394 were higher than expected (this was not a bias introduced by cleaning—only a handful genotypes were blanked, and they were not primarily heterozygous genotypes). We suspected that our mixed-pedigree sample might have contributed to the absence of HWE, and, indeed, we found that these MOXD1 SNPs were all in HWE when the European-ancestry (EA) founders and the African American (AA) founders were analyzed separately (data not shown).

Intermarker Linkage Disequilibrium (LD) Analysis

LD between the SNPs was estimated with the program ldmax from the GOLD (Graphical Overview of Linkage Disequilibrium) package (Abecasis and Cookson 2000) by use of the genotypes from unrelated founders. The program ldmax estimates haplotype frequencies from genotype data by employing an expectation-maximization algorithm (Excoffier and Slatkin 1995). The standard and normalized Lewontin’s disequilibrium coefficients (D, D′) are derived. Association significance is assessed from a χ² distribution with (n₁-1)(n₂-1) degrees of freedom, where n₁ and n₂ are the number of alleles at each marker locus.

Association Analysis

To detect LD with illness, we used the transmission/disequilibrium test (TDT), as implemented in the Family-Based Association Test (FBAT) program, version 1.5 (Laird et al. 2000; Rabinowitz and Laird 2000). The null hypothesis of interest is the absence of association in the presence of linkage. We thus employed the empirical-variance estimator (the −e flag option in the FBAT program) to account for SNP-genotype correlations among affected siblings that result from linkage. The FBAT test statistic uses a score function, Z=S_j-E(S_j)/Var(S_j), where S_j is the observed number of transmitted marker alleles, j, to affected offspring, and E(S_j) and Var(S_j) are the expected and variance values of S_j under the null hypothesis. Asymptotically, Z is assumed to follow a normal distribution, with a mean and a variance equal to 0 and 1, respectively. The test statistic can also be expressed as Z², which follows a χ² distribution, with 1 df. The FBAT program is able to deal with the transmission of multilocus haplotypes, even when phase is unknown and parental genotypes may be missing. It can use pedigrees as well as nuclear families, but pedigrees are broken down into all individual nuclear families; only informative families (i.e., those contributing to the test statistic) are included. For the analyses of the screening SNPs, alleles and haplotypes were tested for association if they were present in at least 10 informative families; in our data, this corresponds to not testing alleles and haplotypes rarer than 3%. This restriction, however, was not used when the investigation was limited to specific subsets of families in the secondary analyses. For multilocus association analyses, FBAT provides global P values, which assess the significance of transmission distortion for all the tested haplotypes. In the present analyses, we limited the number of multilocus systems tested by use of a stepwise procedure and by restriction of multilocus tests to combinations that included the SNP with highest single Z score value, as further detailed in the “Results” section below. For the FBAT analyses, we assumed an additive model for each SNP, and only one affection status model was used (affected subjects had schizophrenia or schizoaffective disorder). The additive model is expected to perform well even when the true model is nonadditive.

Linkage Analysis

Model-free linkage analyses with the MOXD1-STX7-TRARs gene cluster were performed using the LOD score test from the affecteds-only sharing method (Kong and Cox 1997), as implemented in the MERLIN program (Abecasis et al. 2002). The likelihood of the observed marker information among affected relatives is maximized as a function of the marker-allele–sharing parameter and is compared, using a likelihood-ratio test, with the likelihood of the marker data under the null hypothesis of no linkage. The resulting distribution of the allele-sharing test (T) is χ², with 1 df, and the statistic can also be reported as a LOD score of T/2ln(10).

We performed additional analyses, by race, to account for putative genetic and/or allelic heterogeneity within our family samples. Association and linkage tests were evaluated separately in AA and EA families.

Mutation Detection

Sequencing of TRAR4 was performed on an ABI 3100 Genetic Analyzer. Purified PCR products from various amplicons of relevant genomic DNA fragments were used as templates in sequencing reactions with the chemistry of BigDye 3.1 (Applied Biosystems). PCR primers were designed by Primer3 (Rozen and Skaletsky 2000) and were also used as sequencing primers for forward and reverse sequencing. The primer sequences and product sizes are given in table A3 (online only). We used the software SeqScape, version 2.1 (Applied Biosystems), to assist in mutation detection, and we visually verified each mutation. The reference sequence of TRAR4 used in the analysis was from the University of California–Santa Cruz (UCSC) human genome draft, July 2003 freeze (see the UCSC Genome Bioinformatics Web site).

For the nonhuman primates, DNAs were extracted from peripheral blood samples of two different chimpanzees (PTR-S109 and PTR-S286) from West Africa and from tissue samples of two different lowland gorillas (GGO-S110 and GGO-S249). The forward primer for amplicon 1 and the reverse primer for amplicon 7 were used to PCR-amplify the entire DNA segment by standard methods, with an annealing temperature of 60°C. This product was then sequenced bidirectionally with the seven primer pairs that are detailed in table A3 (online only). PCR products were confirmed by 1.5% agarose gel electrophoresis and were purified using MicroSpin Columns (Amersham Biosciences). The purified PCR products were sequenced using the BigDye Terminator Cycle Sequencing Kit (Applied Biosystems) on an ABI Prism 377/3100 DNA sequencer. Sequence data were assembled by the Phred/Phrap program (Ewing et al. 1998) and also were checked manually using the Consed program (Gordon et al. 1998). Sequence data with reads for both strands and/or with high quality (>30 quality value) were used and deposited into the DDBJ/EMBL/GenBank International Nucleotide Sequence Database (accession numbers AB180397–AB180400).

RT-PCR and Real-Time PCR

Total mRNAs from various brain tissues were purchased from either BD Biosciences or Ambion. Gene expression of TRAR4 was first confirmed with general RT-PCR with primer pairs used previously for amplification of segment 4 of TRAR4 (table A3 [online only]). In brief, total mRNA was reverse-transcribed with TaqMan Reverse Transcription Reagents (Applied Biosystems), and the synthesized first-strand cDNAs were then used as templates to amplify TRAR4 with HotStarTaq polymerase (Qiagen). β-actin was used as an internal control in the RT-PCR.

Reverse-transcribed cDNAs were also used in real-time PCR on an ABI Prism 7900 Sequence Detection System, in accordance with the manufacturer’s protocol. The TaqMan MGB probes and PCR primer pairs for the gene-expression assay of TRAR4, GAPD, (glyceraldehyde-3-phosphate dehydrogenase), and/or TRAR1 were purchased as an Assay-On-Demand from Applied Biosystems. The relative gene expression in different brain tissues was normalized to GAPD expression by use of the standard-curve method, as described by Applied Biosystems.

Bioinformatics Tools for Prediction of Functional Effects of Genetic Polymorphisms

We used SIFT and PolyPhen to predict the potential functional effect of missense polymorphisms (Ramensky et al. 2002; Ng and Henikoff 2003). We used VISTA, which defines the conserved region among genomic sequences from different species (Couronne et al. 2003), to predict the potential regulatory sequences.

Results

We have studied 192 families with previous evidence of linkage. Of the 33 SNPs initially selected for study, 31 were accepted for analysis in the screening experiment, since 2 of the MOXD1 SNPs, rs2206064 and rs7751860, had minor-allele frequencies resulting in <3% informative families. The screening SNPs spanned ∼500 kb of the MOXD1-STX7-TRARs gene cluster, which contains a prime set of positional and pathophysiological candidates for schizophrenia susceptibility. We selected at least one common SNP for each gene, with a minimum coverage of ⩾1 SNP/20 kb (the screening set of SNPs). Linkage analyses confirmed the presence of excess allele sharing in this region with individual SNPs from the MOXD1-STX7-TRARs gene cluster. Nine SNPs showed linkage P values <.05 (table A4 [online only]), with the most significant one being rs6937506 (TRAR4), with a LOD score of 1.76 (P=.002).

Association results are presented in table 1 and can also be seen in table A5 (online only). Of 31 screening SNPs, 4 SNPs spanning 106 kb, 2 located in TRAR4 and 1 each in STX7 and GPR57, yielded a P value <.05. The most significant one, rs4305745, was found in TRAR4 (P=.0014) (fig. 1c). This SNP was the only one that remained significant after Bonferroni correction for 31 tests (it should be noted that the question of whether Bonferroni corrections are appropriate or overly conservative is open to different interpretations; however, given that LD is likely to be present over short physical distances, the tests cannot be considered independent). The SNP rs4305745 is located 1,214 bp downstream from the stop codon of TRAR4. We concentrated further laboratory efforts on TRAR4. We aimed for a high-density map of >1 SNP/2 kb by searching public SNP databases and sequencing genomic DNA. We sequenced the TRAR4 gene (∼1 kb of the 5′ region, the 1,038-bp CDS (coding DNA sequence), and ∼1.5 kb of the 3′ UTR, which spans rs4305745) in 30 probands selected from the NIMH-GI families: 16 EA families and 14 AA families. We found 10 coding variants (26 total variants [table 2 and online-only table A6]) by sequencing TRAR4. Three of these variants had previously been found in 96 healthy EA individuals (Freudenberg-Hua et al. 2003). Of the seven novel variants, five are missense and are present only in AA individuals, as shown in table 2. From this combined bioinformatics and sequencing effort, 20 SNPs spanning 21.6 kb of the TRAR4 gene (>1 SNP/2 kb) were identified for genotyping. Two additional markers 3′ to rs4305745 showed association with schizophrenia in the whole sample (table 1 and online-only table A7): rs6903874 (P=.0026) and rs6937506 (P=.0052). Two other additional markers (ss28447873 [an insertion/deletion polymorphism] and rs7452939) were found to be in perfect LD with rs4305745 (they span 53 bp) and, hence, were also associated with illness to the same degree as rs4305745.

Table 1.

Single-Marker Association Results for TRAR4

				Nucleotide at				Transmissions
TRAR4 Marker	dbSNPAccessionNo.a	Distance(bp)b	Position(bp)c	Allele 1	Allele 2	Associated Allele	Allele Frequency	Observed	Expected	Variance	Z	P
1	rs2840837	1,753	132,864,153	A	G	1	.281	95	93.60	46.41	.21	.84
2	rs1361280	1,196	132,865,906	A	G	1	.549	186	175.93	72.91	1.18	.24
3	rs4473885	813	132,867,102	C	T	1	.548	205	193.60	82.47	1.26	.21
4	rs4085406	3,403	132,867,915	A	G	1	.545	189	178.60	80.13	1.16	.25
5	rs6907909	437	132,871,318	A	G	1	.636	193	181.03	80.69	1.33	.18
6	ss28447860	385	132,871,755	C	G	1	.848	143	132.20	37.38	1.77	.08
7	ss28447862	383	132,872,140	T	C	2	.009	8	6.50	3.25	.83	.41
8	ss28447876	25	132,872,523	G	A	1	.992	5	4.17	.36	1.39	.17
9	ss28447863	165	132,872,548	A	G	2	.008	10	8.17	3.36	1.00	.32
10	ss28447865	110	132,872,713	C	T	2	.030	20	17.63	21.73	.51	.61
11	rs8192624	79	132,872,823	A	G	2	.933	86	83.12	20.67	.63	.53
12	rs8192625d	104	132,872,902	A	G	2	.912	101	91.53	21.15	2.06	.0396
13	ss28447866	70	132,873,006	G	A	1	.989	12	9.30	2.69	1.65	.10
14	rs7772821	434	132,873,076	G	T	1	.264	102	96.30	45.90	.84	.40
15	ss28447871	772	132,873,510	G	A	2	.698	185	179.67	48.99	.76	.45
16	rs4305745d	716	132,874,282	G	A	2	.526	168	143.42	59.01	3.20	.0014
17	rs7745308	819	132,874,998	T	G	2	.071	30	26.87	16.66	.77	.44
18	rs6912930	1,663	132,875,817	A	C	1	.390	124	119.78	60.11	.54	.59
19	rs6903874	948	132,877,480	T	C	1	.757	204	180.57	60.40	3.02	.0026
20	rs7765655	1,541	132,878,428	G	A	1	.245	73	71.65	39.31	.22	.83
21	rs6937506	3,016	132,879,969	G	A	1	.736	199	177.07	61.65	2.79	.0052
22	rs4129284	2,772	132,882,985	C	T	1	.413	140	131.98	62.72	1.01	.31
23	rs9321354	NA	132,885,757	A	C	1	.165	88	83.43	37.66	.74	.46

^aPreviously known SNPs are indicated by an rs number; novel SNPs are indicated by an NCBI Assay ID number (ss number) that we obtained (data to be released in the next build of dbSNP).

^bDistance is the number of bp to the next SNP. NA = not applicable.

^cPosition in bp was derived from UCSC July 2003 freeze of chromosome 6 (see the UCSC Genome Bioinformatics Web site).

^dTRAR4 screening SNPs.

Table 2.

Coding SNPs Captured by Resequencing TRAR4 in 30 Probands with Schizophrenia (14 AA and 16 EA Individuals)

							Minor-Allele Frequency
dbSNPAccession No.a	Position(bp)b	SNPc	Amino Acid Change	Type of Change	Functiond	Location in Proteine	Average	EAf	AAg
rs8192622	132,872,108	C78T	Pro26Pro	Synonymous	NA	Extracellular	.05	.09	.00
ss28447862	132,872,140	T110C	Ile37Thr	Conservative	Tolerated	TMD 1	.03	.00	.07
ss28447876	132,872,523	G493A	Gly165Ser	Nonconservative	Tolerated	TMD 4	.02	.00	.04
ss28447863	132,872,548	A518G	Tyr173Cys	Nonconservative	Damaging	Extracellular	.03	.00	.07
ss28447864	132,872,660	C630G	Thr210Thr	Synonymous	NA	TMD 5	.03	.03	.04
ss28447865	132,872,713	C683T	Ala228Val	Conservative	Damaging	Cytoplasmic	.10	.00	.21
ss28447867	132,872,774	A744G	Arg248Arg	Synonymous	NA	Cytoplasmic	.02	.00	.04
rs8192624	132,872,823	G793A	Val265Ile	Conservative	Tolerated	TMD 6	.10	.16	.04
rs8192625	132,872,902	G872A	Cys291Tyr	Nonconservative	Tolerated	TMD 7	.13	.16	.11
ss28447866	132,873,006	G976A	Val326Ile	Conservative	Tolerated	Cytoplasmic	.03	.00	.07

^a Previously known SNPs are indicated by an rs number; novel SNPs are indicated by an NCBI Assay ID number (ss number) that we obtained (data to be released in the next build of dbSNP).

^bPosition in bp was derived from the UCSC July 2003 freeze of chromosome 6 (see the UCSC Genome Bioinformatics Web site).

^cSNPs were named according to their relative position to the first letter of start codon ATG and with the first base as the major allele.

^d Function determined by SIFT. NA = not applicable.

^eThe position in the protein was predicted on the basis of 7-transmembrane GPCR structures, as depicted in the GPCR database (see GPCRDB Web site). TMD = transmembrane domain.

^fEA = European ancestry.

^gAA = African American.

Although similar association trends were observed in EA and AA individuals, table A7 (online only) shows that the evidence for association was proportionally stronger in the AA sample. The evidence for association with rs4305745 and rs6937506 was significant in both samples (P=.035 in the AA sample and P=.015 in the EA sample for rs4305745; P=.025 in the AA sample and P=.035 in the EA sample for rs6937506). However, a cluster of three SNPs in the promoter region of TRAR4 yielded significant association in the AA sample—rs4473885 (P=.032), rs4085406 (P=.047), and rs6907909 (P=.019)—but not in the EA sample (table A7 [online only]). These differences raise the possibility of allelic heterogeneity specific to the AA population. All new missense mutations detected by sequencing were also exclusive to the AA sample (table 2). Two mutations, A518G (Tyr173Cys) and C683T (Ala228Val), were predicted by either PolyPhen or SIFT (Ramensky et al. 2002; Couronne et al. 2003; Ng and Henikoff 2003) to be nonconservative (table 2). However, after genotyping the whole sample, we noted that none of the newly found missense SNPs cosegregated with disease in a specific manner (data not shown). Furthermore, all the missense variants, except for A518G (Tyr173Cys), were also found in a set of 48 AA subjects from the Coriell Human Variation AA DNA panel (table A8 [online only]). Additionally, some of these missense variants were homozygous in some control individuals. It is interesting to note that the ratio of missense to synonymous mutations (9:3) is close to what is predicted under neutral expectations (i.e., a pseudogene that has an expected ratio close to 4:1). Indeed, in the gorilla, TRAR4 has already become a pseudogene (with a nonsense mutation at codon 15: Tyr in chimpanzee and human and STOP in gorilla), although the human and chimpanzee TRAR4 versions have not yet become pseudogenes.

The TRAR4 region was found to have two LD blocks, which are depicted in figure 2. The SNP rs4305745 (marker 16 in fig. 2) is in the LD block constituted by 3′-flanking SNPs. The pattern suggests that association for TRAR4 originates from rs4305745. None of the 5′-flanking SNPs are in LD with rs4305745, which instead is in strong LD with markers 19 (rs6903874) and 21 (rs6937506) from the 3′ LD block (and also shows a trend with marker 12 [rs8192625]). The LD pattern generated from the 31 initial screening markers indicated that the whole region of the MOXD1-STX7-TRARs gene cluster is separated into four major strong-LD blocks, whereas the TRAR4 region represented by rs4305745 is not in strong LD with any of the major LD blocks (data not shown).

Figure 2.

Pairwise LD in 192 founders for the TRAR4 region. a, Relative physical position of 23 markers in the TRAR4 region (shown in table 1). The SNPs ss28447862, ss28447876, ss28447865, rs8192624, and ss28447866 were excluded in the following LD measurements because of their low minor-allele frequencies. b, LD pattern in AA subjects (left panel) and EA subjects (right panel). The graph was generated by GOLD (Abecasis and Cookson 2000). In each LD pattern, the D′ values (upper left diagonal) and the P values (lower right diagonal, converted to log P values) are calculated from the program ldmax of the GOLD package (Abecasis and Cookson 2000).

We conducted haplotype association analyses with all TRAR4 two-locus systems (n=17, after exclusion of 5 markers with minor-allele frequencies <3%) derived from rs4305745, which was chosen as the anchor since it had the most significant single-locus association (P=.0014). For each two-locus system, we derived the global χ² value by use of only those haplotypes with frequencies >3% (table A9 [online only]). All 17 of these two-locus systems exhibited P values <.05 (not corrected by multiple testing), and all harbored the “A” allele of rs4305745. None of the two-locus systems showed association stronger than that of rs4305745 alone (table A9 [online only]), and the same results were obtained with up to five multilocus systems, with each haplotype system extended stepwise to contain the most significant previous smaller haplotype (data not shown). This suggests that rs4305745 and/or the other two nearby polymorphisms, ss28447873 and rs7452939, in perfect LD (further confirmed by genotyping the whole sample) with rs4305745 (table A6 [online only]) are the most likely mutations underlying the association of the TRAR4 region with schizophrenia susceptibility.

To explore the possible functional effects of associated SNPs and their haplotypes, we first defined the conserved noncoding sequence (considered as a potential functional region) by comparative genomic analysis of TRAR4 genomic sequences of human, mouse, and rat by use of VISTA (Couronne et al. 2003). The cluster of three polymorphisms (rs4305745, ss28447873, and rs7452939—all equally implicated as candidates by the association analysis) that exhibit the most significant association is very close to two conserved regions right after the stop codon (sequence identity >70% among human, mouse, and rat genomes). The sequence identity immediately around this SNP (rs4305745) is ∼50% (fig. A1 [online only]), suggesting that this SNP and/or other polymorphisms in perfect LD with rs4305745 may ultimately affect gene expression, a hypothesis we are currently testing. RT-PCR from various brain tissues also confirmed that rs4305745 was flanked by the 3′ UTR of TRAR4 (data not shown), suggesting that rs4305745 or one of its haplotypes may affect gene expression at the posttranscriptional level. Another significant SNP, rs6903874, is also within a conserved region (fig. A1 [online only]), and it is possible that this SNP and/or other SNPs nearby might be functional, although this SNP is much farther than rs4305745 from the stop codon of TRAR4.

A comprehensive gene-expression analysis of TRAR4 will help to elucidate its potential functional roles in the pathophysiology and pharmacology of schizophrenia. We investigated TRAR4 expression in various human tissues by RT-PCR and found that TRAR4 was expressed at low abundance in various human brain tissues as well as in human fetal liver but not in the cerebellum or placenta (fig. 3a). A quantitative real-time PCR, shown in figure 3b, further revealed that TRAR4 has comparable levels of expression in basal ganglia, frontal cortex, substantia nigra, amygdala, and hippocampus, with the highest expression in hippocampus and the lowest expression in basal ganglia. These results are consistent with a previous expression study that included TRAR4 (Borowsky et al. 2001). These regions have been implicated in the pathophysiology and pharmacology of schizophrenia (Grossberg 2000; Freedman 2003). The tissue distribution of TRAR4 gene expression is similar to that of the only well-characterized TRAR, TRAR1 (Borowsky et al. 2001; Bunzow et al. 2001). However, further comparison of gene expression of TRAR4 and TRAR1 indicated that TRAR4 is, overall, more abundant than TRAR1, particularly in basal ganglia (∼14-fold), frontal cortex (∼21-fold), and substantia nigria (∼14-fold) (fig. 3c), which suggests that TRAR4 may play a more important role than TRAR1 in those regions.

Figure 3.

Expression pattern of TRAR4 in human tissues. a, TRAR4 expression pattern in various human brain regions. Lane 1 is a 100-bp molecular weight standard ladder (Promega). Lanes 2–13 are human brain, human fetal brain, cerebellum, fetal liver, placental, spinal cord, control (no reverse transcriptase added), basal ganglia, frontal cortex, substantia nigra, amygdala, and hippocampus. RT-PCR products from total RNAs are displayed in the photograph of the ethidium bromide–stained agarose gel; β-actin was used as internal control. b, Quantitative real-time PCR determined the relative abundance of the TRAR4 transcript in various human brain regions. c, Comparison of gene expression of TRAR4 with TRAR1. Samples S1–S6 in panels b and c are basal ganglia, frontal cortex, substantia nigra, amygdala, hippocampus, and cerebellum.

Discussion

TA receptor genes have been proposed as candidate genes for schizophrenia on the basis of highly provocative pathophysiological evidence (Boulton 1980; Premont et al. 2001; Branchek and Blackburn 2003) as well as linkage mapping data (Cao et al. 1997). Here, we present preliminary evidence that TRAR4, a gene that belongs to the TRAR family, contributes to susceptibility to schizophrenia in three data sets, with evidence of genetic linkage to 6q. Furthermore, the TRARs gene cluster at chromosome 6q23 is contained within a wide area of linkage detected in several other clinical samples (Bailer et al. 2000; Levinson et al. 2000; Lindholm et al. 2001; Lerer et al. 2003; Lewis et al. 2003). The linkage evidence for schizophrenia in 6q is not population specific—it has been gathered from multiple population groups: AA individuals, EA individuals, and Jews and Arabs from Israel. However, the evidence for association of TRAR4 in our samples, although present in both EA and AA individuals, appears higher in AA subjects.

The most significantly associated SNPs within TRAR4 are located in the 3′ UTR of TRAR4, and these polymorphisms may affect the gene expression at the posttranscriptional level. There are several possible mechanisms by which TRAR4 could contribute to the pathogenesis of schizophrenia. First, the to-be-determined ligand(s) of TRAR4 may be neurotransmitter(s) specifically implicated—by a new signaling pathway—in schizophrenia. Second, TRAR4 may be activated strongly by β-PEA, an endogenous chemical analogue of amphetamine—or perhaps by amphetamine itself. It has been documented that amphetamine can produce a paranoid schizophrenia syndrome in humans by inducing DA release in the striatum, and the DA release induced by amphetamine (or β-PEA) is increased in patients with schizophrenia (Laruelle et al. 2003). Pharmacological profiling has shown that β-PEA and amphetamine can directly activate TRAR1 (Borowsky et al. 2001; Bunzow et al. 2001). Finally, TRAR4 may crosstalk with the COMT system, which has been implicated in the pathogenesis and etiology of schizophrenia (Li et al. 2000; Egan et al. 2001; Shifman et al. 2002). It has been shown that rat TRAR1 can be activated with higher potency and efficacy by 3-methyltyramine, a major metabolite of DA produced by COMT, than by the precursor catecholamines (Bunzow et al. 2001). If TRAR4 were shown to act like TRAR1 in this regard, some “inactive” catecholamine metabolites might act as endogenous agonists of TRAR4, which could predispose to psychosis. Moreover, we have shown that TRAR4 is more abundant than TRAR1 (fig. 3), which suggests that TRAR4 may play a more important role in areas of the brain that are relevant to the dopaminergic system. It can be seen that the plausible contributions of TRAR4 to schizophrenia susceptibility are closely related to the DA hypothesis of schizophrenia, a major hypothesis that has been intensively researched over the past 50 years. Given the increased understanding about the interactions of the DA system with the glutamate system in the pathogenesis of schizophrenia (Laruelle et al. 2003), our finding about TRAR4 may prompt new investigations of the interactions among these neurotransmitter systems.

Regulatory sequence disruption can affect protein expression and cause disease (Mitchison 2001). The associated SNPs in the 3′ UTR of TRAR4 may contribute to the susceptibility for schizophrenia by affecting the gene expression at the posttranscriptional level. Our RT-PCR experiment indicated that the TRAR4 3′ UTR spanned the most associated SNP, rs4305745; therefore, it is possible that TRAR4 gene expression was affected at the posttranscriptional level by these 3′ UTR SNPs (rs4305745 and/or ss28447873 and rs7452939 [two SNPs in perfect LD with rs4305745]). The chimpanzee and gorilla sequencing results indicated that the ancestral allele for rs4305745 is A (table A6 [online only]). Because of the perfect LD among rs4305745, ss28447873, and rs7452939, the associated allele A of rs4305745 actually represents a human haplotype of A-A-A, spanning rs4305745, ss28447873, and rs7452939 (fig. A1c [online only]). (Also note that the ancestral haplotype appears to be A-A-G, as seen in table A6 [online only].) It is interesting that the predicted TRAR4 mRNA structure exhibited a significant change for the overtransmitted haplotype A-A-A, compared with haplotype G-del-G; the same structure change can be generated by allele A of rs4305745 alone (data not shown), suggesting that rs4305745 is most likely the causative SNP.

We have found that the mutation rate in the coding region for TRAR4 (1 mutation/100 bp) is well above the average (1 mutation/346 bp) (Cargill et al. 1999). In addition, there are more missense mutations in TRAR4 than synonymous mutations (nine vs. three) (table 2 and online-only table A8), suggesting that TRAR4 may be becoming a pseudogene. However, an expressed pseudogene sometimes regulates the mRNA stability of a homologous gene (Hirotsune et al. 2003). We did not find evidence of association with schizophrenia for missense SNPs; however, some missense mutations—particularly A518G (Tyr173Cys)—may be pharmacologically important. A518G is located in the putative extracellular domain of the receptor and, hence, may affect ligand binding (table 2). Besides the changes in protein structure, these missense mutations may also alter the gene expression by affecting mRNA folding structures, as described for dopamine D2 receptor (DRD2) (Duan et al. 2003). Actually, A518G was predicted to have a remarkable effect on TRAR4 mRNA folding, as predicted in silico by Mfold (Zuker et al. 1999) (data not shown). It is also notable that this missense SNP and several others were only found in AA subjects, and 173Cys was only detected in AA probands with schizophrenia. It will be important to examine the functional effects of some of these missense SNPs and to test the association with schizophrenia in additional samples.

Molecular genetics studies of schizophrenia have found several replicated linkages to various chromosomal regions (see review by Owen et al. [2004]), and association studies have recently pointed to several genes at some of those linkage regions with independent confirmations, including NRG1 at 8p21-p12 (Stefansson et al. 2002, 2003; Williams et al. 2003), DTNBP1 at 6p22.3 (Straub et al. 2002; Schwab et al. 2003), COMT at 22q11.21 (Li et al. 2000; Egan et al. 2001; Shifman et al. 2002), RGS4 at 1q23.3 (Chowdari et al. 2002; Morris et al. 2004; Williams et al. 2004), and DAOA at 13q33.2 (Chumakov et al. 2002; Schumacher et al. 2004). Although nonreplications have been reported, it would be extremely unlikely that all the aforementioned results will end as false positives. We now propose that TRAR4 is also a susceptibility gene for schizophrenia. There are also two other weak association peaks at the MOXD1-STX7-TRARs gene cluster, in addition to the TRAR4 peak shown in figure 1: one is in STX7 and the other is in the pseudogene, GPR5. Therefore, we cannot exclude the possibility that those loci may also contribute to susceptibility for schizophrenia. The identification of TRAR4 as a susceptibility gene for schizophrenia, which is consistent with human and animal models of toxic psychosis and is in agreement with the expression pattern of TRAR4 (expressed in frontal cortex, amygdala, and hippocampus), appears to substantiate the dopaminergic hypothesis of schizophrenia, but the exact mechanisms of disease mediated by TRAR4 remain to be elucidated. The two reports suggesting linkage of the same 6q chromosomal area to bipolar disorder raise the possibility that TRAR4 might be involved in the pathophysiology of both schizophrenia and bipolar disorder, and there is a precedent for a gene potentially involved in both disorders (Chumakov et al. 2002; Hattori et al. 2003; Chen et al. 2004; Schumacher et al. 2004). We would like to note a couple of caveats: first, our combined sample has a rather modest size in the context of complex genetics, and, second, additional gene(s) in the region might account for our association signal. However, although it is certainly true that gene-prediction programs have their limitations and different methods sometimes produce different results, at least one of them has been able to pick up all the TRARs genes and pseudogenes we know about (see the UCSC Genome Bioinformatics Web site [July 2003 freeze]). Although it is possible that gene-prediction programs have missed a G-protein–coupled receptor (GPCR) pseudogene, there is little chance they have missed an actual GPCR gene in this cluster since the GPCR motif is so robust. Other non-GPCR genes, however, might have been missed. Although the identification of TRAR4 as a susceptibility gene for schizophrenia is encouraging, it is important that our observations should be considered as preliminary until replication has been completed in additional samples.

Appendix A: Supplementary Material

Figure A1.

Conserved noncoding regions defined by VISTA (Couronne et al. 2003) and the relative position to associated markers. a, TRAR4 3′-flanking conserved regions, generated by comparing human genome with mouse, rat, and chimpanzee genomes (upper, middle, and lower plots, respectively). The region with sequence similarity reaching 70% is defined as a conserved region, and such peak areas are shaded pink. b, Transformed −log FBAT P values of TRAR4 SNPs versus their relative genomic positions. Panels a and b are aligned according to the genomic position (UCSC Genome Bioinformatics Web site [July 2003 genome draft]). c, The local genomic sequence alignment (human-mouse) around rs4305745 and two other polymorphisms (ss28447873 and rs7452939) in perfect LD with rs4305745.

Table A1.

Subjects and Phenotypes

	Entire Sample									Subjects with DNA Available
	No. of Subjects						% of Subjects, by Race			No. of		No. with Diagnosisa
Data Set	With DNA Available	Without DNA Available	Total	No. of Families	No. of Full Triosb	No. of Founders with DNA	EA	AA	Other Race	Males	Females	SZ	SA	UD	Average Age (Years) at Psychosis Onset [Range]
NIMH-IRP	325	117	442	67	39	108	88	4	7	180	145	136	24	165	20.4 [6–45]
NIMH-GI	321	39	360	69	38	106	55	32	13	159	162	137	22	162	18.5 [5–38]
AU/US	181	204	385	56	47	103	84	13	4	100	81	79	…	102	23.1 [14–42]
All data sets	827	360	1,187	192	124	317	75	17	8	439	388	352	46	429	20.3 [5–45]

^a SZ = schizophrenia; SA = schizoaffective disorder; UD = unknown diagnosis.

^bFull trios consist of an affected offspring and both parents; the remainder of the families have partial trios with supplemental information from additional family members.

Table A2.

Assay Information^[Note]

				Nucleotide at		PCR Primer		Probe			Probe Used for Detection for
SNP No.	Screening SNP No.	Gene	dbSNP Accession No.a	Allele 1	Allele 2	Forward (5′→3′)	Reverse (5′→3′)	FP (5′→3′)	TaqMan VIC (5′→3′)	TaqMan FAM (5′→3′)	Allele 1	Allele 2
1	1	MOXD1	rs2206064	C	T	ACAGTAATTTCTATAATTATCAAATAATCTCAGCTGAAGAAT	GAGAGAGAGGGAGAGAGAAATGACA	…	CATTAATCTTCACATTTTT	TACATTAATCTTCATATTTTT	VIC	FAM
2	2	MOXD1	rs599660	G	A	CATCTCTGATGTCACTTTTATGATAC	TGAGTGTCGCTACAACACGAAA	GAAAAAGGTAAAAACCATTTCCATTT	…	…	R110	TAMRA
3	3	MOXD1	rs7751860	C	T	TTGTAATTCACAACCTGCATAAGTTTTCAA	TTTGTTAGATTTCTTCCTTTATCTGCCAAGA	…	ATTGGCCACGTTCCT	TGGCCACATTCCT	VIC	FAM
4	4	MOXD1	rs1981187	C	T	CTCGAATTCTCCTCCCAAGATG	AAAAACTCTGGCAATTTGTATC	TGCACATGTGTGCAATTAAGAGG	…	…	R110	TAMRA
5	5	MOXD1	rs1338387	C	T	GGCTCTACAGTAATCTGGCTTCAAG	GGGAGTGTTTCCTAGAACATGAAAATTTAAG	…	TTGAATAGCCGTGGAATG	TGAATAGCCATGGAATG	VIC	FAM
6	6	MOXD1	rs1538308	A	G	GATTTTCAATATGTAAATGAAATAACAATGTCTGTAGACAA	GAAATATGAGTTCAGTGAAAAGGCACAG	…	TTAACTGATCTAAATTTCTCA	TTAACTGATCTAAATCTCTCA	VIC	FAM
7	7	MOXD1	rs2275394	C	T	GGAAATCTTCCTCAGGCATACA	AAGCTCAGCTAGAGTCTTTCAAGC	AAGAGAAATACATCAACGCCTAGTA	…	…	R110	TAMRA
8	8	MOXD1	rs6937815	G	A	GGCTGCTGTGGTTTGCA	CTTCCAGGTGCAGGAGACA	…	CCAGAGTCTATTCGCCTGG	CCAGAGTCTATTCACCTGG	VIC	FAM
9	9	MOXD1	rs3823288	A	T	GCCTGTAGGCCGCTATGAC	CCTTAGAAGAGGAGTGAGAGGAAGA	…	AAGGATTAAAGATAATTTTT	AAGGATTAAAGATATTTTTT	VIC	FAM
10	10	STX7	rs1856352	A	G	ACTTTCCAAAAGTAATTTTGACCTAAAAATATTAACCATT	TCATCTGGTTTTGTGGGCCATTATT	…	TCTATAAGACATAAAGGAAGGTA	AAGACATAAGGGAAGGTA	VIC	FAM
11	11	STX7	rs3757299	A	C	TGCATTTTCCACATTGGCTTCTATG	TGTTACTGGAGCATCAATCCATGTT	…	CTATCTGTAAAATAAAACAC	CTATCTGTAAAATCAAACAC	VIC	FAM
12	12	STX7	rs1002799	A	G	AAGGACCCAGGGCTGAAAATC	TCCACTTTCCTCTTGCCTTCTTTC	…	TGGGTTATCCCAAACAG	TGGGTTATCCCGAACAG	VIC	FAM
13	13	STX7	rs2788942	G	A	TCTTCCAGAAAGAGATTATTGAACAAAGCA	TGGGCATGGGAGGAGAGT	…	AAATCACTGGGTGTACTT	TAAAATCACTGGATGTACTT	VIC	FAM
14	14	STX7	rs2842884	G	C	AAAGGACAGAGTCATGTTAAACACA	GGGCCCTGCCTTATAATTGTT	AAATTGCATATTTCCTTTGAAAAC	…	…	R110	TAMRA
15	15	STX7	rs1591811	G	A	GCGGCATTCCTGGACTTG	AGAGCAAGAAAAACAAATAAAATGCATACAGA	…	ATCTGCCACGTAACTT	AATCTGCCATGTAACTT	VIC	FAM
16	16	TRAR3	rs2842899	A	T	TGGAAACTTACTGGTCATGATTGCT	GCAATCAGAAAGTTTGTAGGTGTGT	…	CTTCACTTCAAACAAC	CCTTCACTTCTAACAAC	VIC	FAM
17	17	TRAR3	rs2788935	C	T	TTCCTATACTTTATGAATCAAGTAATTTGATGTTTACTGA	GTAAGGCACTGTGCTAGGTATTGAA	…	CTTAGCAGCAAGTGAAAA	TTAGCAGCAAATGAAAA	VIC	FAM
18	18	TRAR5	rs1933988	A	C	ATCCTTGACTCTTACCTGCATCATG	ACCTTGCTTCTCTATATTCTGAAATGTGATAC	…	CCAAAAGCATTCAGGGTT	AAAAGCATGCAGGGTT	VIC	FAM
19	19	TRAR5	rs8192627	A	C	GCTCTATTTTATCCTTGGTTTAGGAAAGC	TCTAAAAATAAACTAATGGTTGATGAACTAGCCT	…	TTATTTTAAGTGGAGATGTTTTA	TTAAGTGGAGCTGTTTTA	VIC	FAM
20	20	TRAR5	rs2840836	G	A	TCAAACCATCTCCCCCACTT	AGAAGGCCCAGCTCACTTCA	CAATTATAAAAGCTCAACTGAACAC	…	…	R110	TAMRA
21	…	TRAR4	rs2840837	A	G	CTGACTGTCAGGAAAGGACTTAGAG	GTTTTTAAAGAAATGTCAGGTCTAAGATGGG	…	CAAAAGCATGGAAGACAA	AAGCATGGAAGGCAA	VIC	FAM
22	…	TRAR4	rs1361280	A	G	TCCCTGTGGCAATAAGAGATAGGA	GGTAAGTTTTATAGGCTAGGAGGACATG	…	TCAGTGCATTCATAGAAG	CAGTGCATTCGTAGAAG	VIC	FAM
23	…	TRAR4	rs4473885	C	T	TCAATTTTATACCAACAAAAATTCCACAGGTT	TCTTTAGTCCATTCCATTTCTTTTGAGATTTCA	…	CATGCCGATTTCT	CATGCCAATTTCT	VIC	FAM
24	…	TRAR4	rs4085406	A	G	ATGTGAGATACTGCTACACCCTGTA	CTAAGCCCTCTTCAACACTTGGTA	…	ACAACAAAAAGAAATGGCAG	CAACAAAAAGAGATGGCAG	VIC	FAM
25	…	TRAR4	rs6907909	A	G	CCAATGCAGATGGAAAATTGATTCTTAACA	AGCCACCCGGGTTTTTGTT	…	TATTGATGCTTACTATTTACA	TTGATGCTTACTGTTTACA	VIC	FAM
26	…	TRAR4	ss28447860	C	G	GTTGCACAGAGAAACTCAAAAGGTAAAAATA	ATTTGACAAAAATTATATTGCACAAGATTATTGGAGA	…	ATGCTCTTAATTTGATAAAA	ATGCTCTTAATTTCATAAAA	VIC	FAM
27	…	TRAR4	ss28447862	T	C	TGAATGGGTCCTGTGTGAAAATCC	TCACCAGGAGGTTTCCAAACAC	…	AAAGCCAAACACTATGTAC	AAGCCAAACACTGTGTAC	VIC	FAM
28	…	TRAR4	ss28447876	G	A	TCAGCGTGTCCTGGATCCT	GATAATTCCTCCAGCCCATCGT	…	ATGTACAGCGGTGCTG	ATGTACAGCAGTGCTG	VIC	FAM
29	…	TRAR4	ss28447863	A	G	GATCCTGCCCCTCATGTACAG	CCTCCTATACAGTTTAGGGCATCAG	…	CACAGGTGTCTATGACGAT	CAGGTGTCTGTGACGAT	VIC	FAM
30	…	TRAR4	ss28447865	C	T	ACTGGGTGTTGACAGATTTTCTATCC	GAGGATGATTCTGTCTTGCTACCA	…	TTTTCTATCTTTTTCGCCTGTCG	TTTCTATCTTTTTCACCTGTCG	VIC	FAM
31	…	TRAR4	rs8192624	A	G	CAGGAGAGAGAGAAAAGCAGCTAAA	GGCATCAATTAATGAATCAATGCTATATGGT	…	TCACAGTGATAGCATTT	ACAGTGGTAGCATTT	VIC	FAM
32	21	TRAR4	rs8192625	A	G	GATTCATTAATTGATGCCTTTATGGGCTTT	GCTGAGTTATAATAAGCACACCAACAG	…	CCCTGCCTATATTTAT	CCTGCCTGTATTTAT	VIC	FAM
33	…	TRAR4	ss28447866	G	A	CCTTTGATTTATGCTTTATTTTACCCATGGT	CAAATTCATGGTTGCTGAACTGTTCT	…	CTGACCAGTTACAATAA	CTGACCAGTTATAATAA	VIC	FAM
34	…	TRAR4	rs7772821	C	A	AGCAACCATGAATTTGTTTTCTGAACA	CGCTTGGTAATTTTAAAGGTATCCTGAAC	…	TTCGTCTATCCAACTGC	TTCGTCTATACAACTGC	VIC	FAM
34	…	TRAR4	rs7772821	C	A	AGAACAGTTCAGCAACCATGAATTTG	GAAGAGCAATTTATTTGCTATTCATTCATAGTCTT	…	TTCGTCTATCCAACTGC	TTCGTCTATACAACTGC	VIC	FAM
35	…	TRAR4	ss28447871	G	A	CAGCCTGCCAAAAATTTCA	CAACAGCCTGGTCAAGATGA	TTGGTTTCTTTTGCTTTGGTTATTTT	…	…	R110	TAMRA
36	22	TRAR4	rs4305745	C	T	TCTATGTCCTTTCTTCCCCAAA	TTTCCTGTAGATCATGACAGTTTCT	GCAGAATATTCCCGATAAAGTTT	…	…	R110	TAMRA
36	22	TRAR4	rs4305745	C	T	TCCTTTCTTCCCCAAATCCATAACC	CCTGTAGATCATGACAGTTTCTCATACTTTATT	…	AAAATGTCAGACGAAACT	ATGTCAGACAAAACT	VIC	FAM
37	…	TRAR4	ss28447873	A	del	CAAATCCATAACCCCCATGTAGTCA	TCCTGTAGATCATGACAGTTTCTCATACT	…	ACTTTATCGAGGAATATT	CTTTATCGGGAATATT	VIC	FAM
38	…	TRAR4	rs7452939	G	A	TCCTTTCTTCCCCAAATCCA	CGGATTTCCTCACTTCCCCTA	CATAATATTTGATCAGTATTTCTCAAAACT	…	…	R110	TAMRA
39	…	TRAR4	rs7745308	T	G	AAGGAGATGTATTAAAGAAGTGGCATTCA	TCCCATCCTTTCCCTCCTACAA	…	AAATCCTTCTTAAAATTTAAG	AATCCTTCTTAAACTTTAAG	VIC	FAM
40	…	TRAR4	rs6912930	A	C	ACCTGTCCAATTTTGTTTAAATGTGCAT	TGGTGTTCCTTAAATATTTTCCTTGGACAA	…	TCATATCAAAAGAGTACCCATG	TCATATCAAAAGAGTCCCCATG	VIC	FAM
41	…	TRAR4	rs6903874	T	C	GTTTCTCTCATAAGTCAGTGGTTCTTGA	GTGATTGTGTGCATTTTTCTAGAGAAAGA	…	ATTTCAATGTGACTATGATCT	TCAATGTGACTGTGATCT	VIC	FAM
42	…	TRAR4	rs7765655	G	A	AAAACTTCTTACCAAATTCTCTCACAGGT	CCTGTTATTTTTTCCATTCTCTTTTTCTCAGT	…	TGTATAAATCCATAAGGAAGTG	TATAAATCCATAAGAAAGTG	VIC	FAM
43	…	TRAR4	rs6937506	G	A	TCAGGAGCTGTGCACTGG	TGAAAAACCCTGTGCGAGTCT	…	TATCCCATCTGTTGTACTAG	CTATCCCATCTATTGTACTAG	VIC	FAM
44	…	TRAR4	rs4129284	C	T	CCCCAAACATGCCCATATCCTA	GGTCCTTTGACCAAGCAAGGTAATA	…	TTCCCAGAACCTGTTACT	ATTCCCAGAACTTGTTACT	VIC	FAM
45	…	TRAR4	rs9321354	A	C	GCTGCTGTTAATGTGACCTCTTGAT	CTGTATCGCTTCTGATTCACTCTGT	…	CTCTCTACAGAACTAAC	CTCTACCGAACTAAC	VIC	FAM
46	23	PNR	rs3813354	G	A	TGTGGCATTTGCTGTGTCCTA	GAGAGCAGCAGGAAGTTGGT	…	CTTCAAAGCGCTTCAC	CTTCAAAGCACTTCAC	VIC	FAM
46	23	PNR	rs3813354	G	A	TGTGGCATTTGCTGTGTCCTA	GGGCCAGGGAGAGCAG	…	CTTCAAAGCGCTTCAC	CTTCAAAGCACTTCAC	VIC	FAM
47	24	TRAR2	rs4144146	G	A	CATGGAGAGGATCAAGAAGTTGGT	CTGGGCAACATGATCGTAATGATTT	…	ATCGCTCACTTCAAGCA	TCGCTCACTTTAAGCA	VIC	FAM
48	25	TRAR2	rs4467795	G	C	AGTAACTGTAGGAGGCATCTACCTC	CTGTTCTGTGCTTCTAGGTTCTGAT	…	CCACTACTTCTTTCCC	CACTACTTGTTTCCC	VIC	FAM
49	26	GPR57	rs4421218	C	T	CTTCTCTGTTTGCTACTCTCTTCCT	AGGGAAGGAGCATGAAGAGTAGAG	…	CCCTCTGCCTCTCC	CCCTCTGTCTCTCC	VIC	FAM
50	27	GPR57	rs1081074	A	C	AAGAGAAGCTTGACATTGTGCCTAT	TCTCCTCTGTCTGATTCACTCAAGA	…	CACAGAAAAGAAGGCCCA	ACAGAAAAGCAGGCCCA	VIC	FAM
51	28	GPR58	rs8192646	G	A	TCCACCAAAATAACTATTCCAGTCATTAAAAGA	ACCCCGAAGGCAAATGCT	…	CTTCTATGTTGGTCGGTCC	TACTTCTATGTTAGTCGGTCC	VIC	FAM
52	29	GPR58	rs4451148	T	C	GTGTAACTCAGAGCCCTATGCAA	CTCTGCTCATAGCCTTGTGTGT	…	CAACTTGTCAAACTGTCA	AACTTGTCAAGCTGTCA	VIC	FAM
53	30	GPR58	rs4380767	T	C	CTAGCAGCCTGGGAAAACAGT	CCCTTAGGGTCTTGTCTTTTGGTTA	…	CTTTGGATTCAGTCCTCAGAGA	TTGGATTCAGTCCCCAGAGA	VIC	FAM
54	31	TRAR1	rs7739700	T	C	GCAACAAAGAGAACCGGGATTTTAA	CCTGGCATATGGAGACAAGCT	…	TTTGGAACCACGTCAGC	CTTTGGAACCATGTCAGC	VIC	FAM
55	32	TRAR1	rs8192620	C	T	GATTAAATGTAGAGTTCAAGTAGCCAAACC	ACAGTCATGGACCCTTTTCTTCAC	…	TTTGAATGATGTGTTGATTT	CTTTGAATGATGTATTGATTT	VIC	FAM
56	33	TRAR1	rs4897595	G	A	TCCTCTCACTTCTAAAACATTTACCTATTACTCT	AATGTCCAGCTCCAGGTAAAAGTTT	…	TTGGAAGGGACTTTAT	TTTTGGAAGGAACTTTAT	VIC	FAM

Note.— The SNPs rs3813354, rs4305745, and rs7772821 were genotyped by both FP and TaqMan.

^aPreviously known SNPs are indicated by an rs number; novel SNPs are indicated by an NCBI Assay ID number (ss number) that we obtained (data to be released in the next build of dbSNP).

Table A3.

Primer Sequences for TRAR4 Amplicons

		PCR Primer
Amplicon	Size(bp)	Forward (5′→3′)	Reverse (5′→3′)
1	521	GGCTCACTGCCCATTTGT	CAATGGGAAAATCCCTCGAT
2	573	GAATAACTTTCCTAGTAATCACTGTTG	GGGATGAATTGCTGCTCATAA
3	570	CAACAAGGACAAAACTTCTCCA	AGCCCATCGTCATAGACACC
4	578	AGGAATTTGCATCAGCGTGT	AAAACAAATTCATGGTTGCTGA
5	517	TTTACCCATGGTTTAGGAAAGC	CCAATAATTTGATAAAGGCTATTCAC
6	526	GACTCTTCCCTCTGCTCTGG	GAAAAGAAAGCTAGAGACTGCAC
7	647	TCCTTTAGGGAGGATATCTTTCAA	TCCTCACTTCCCCTAATGTCC

Table A4.

Linkage Analyses

				Results for Data Set
				All (179 Familiesa)		AA (27 Familiesa)		EA (123 Familiesa)		NIMH-IRP (67 Familiesa)		NIMH-GI (57 Familiesa)		AU/US (55 Familiesa)
SNP No.	Screening SNP No.	Gene	dbSNP Accession No.b	LOD	P	LOD	P	LOD	P	LOD	P	LOD	P	LOD	P
1	1	MOXD1	rs2206064	.26	.14	.27	.13	−.01	.6	−.01	.6	.21	.2	.20	.2
2	2	MOXD1	rs599660	1.01	.02	.47	.07	.84	.02	.23	.2	.87	.02	.05	.3
3	3	MOXD1	rs7751860	.06	.3	.04	.3	.02	.4	−.01	.6	.31	.12	.00	.5
4	4	MOXD1	rs1981187	.04	.3	.45	.08	.17	.2	.02	.4	.00	.5	.14	.2
5	5	MOXD1	rs1338387	.11	.2	.06	.3	.30	.12	.01	.4	.06	.3	.24	.15
6	6	MOXD1	rs1538308	.03	.3	−.05	.7	.22	.2	.03	.4	.01	.4	.01	.4
7	7	MOXD1	rs2275394	1.27	.008	1.15	.011	.62	.05	.26	.14	1.11	.012	.02	.4
8	8	MOXD1	rs6937815	−.13	.8	−.15	.8	−.11	.8	.00	.5	−.49	.9	.00	.4
9	9	MOXD1	rs3823288	.25	.14	.49	.07	.00	.5	.18	.2	.06	.3	.03	.4
10	10	STX7	rs1856352	−.14	.8	−.08	.7	−.04	.7	−.09	.7	−.15	.8	−.04	.7
11	11	STX7	rs3757299	−.03	.6	−.02	.6	−.08	.7	−.03	.6	−.02	.6	.13	.2
12	12	STX7	rs1002799	−.01	.6	.01	.4	−.11	.8	.00	.4	−.04	.7	.03	.4
13	13	STX7	rs2788942	−.03	.6	−.01	.6	−.01	.6	−.09	.7	.10	.2	.05	.3
14	14	STX7	rs2842884	−.04	.7	−.02	.6	−.07	.7	−.01	.6	−.14	.8	.20	.2
15	15	STX7	rs1591811	.12	.2	−.02	.6	.00	.5	.06	.3	.06	.3	−.01	.6
16	16	TRAR3	rs2842899	.01	.4	−.22	.8	.00	.5	.18	.2	−.21	.8	.01	.4
17	17	TRAR3	rs2788935	.13	.2	.07	.3	.01	.4	.21	.2	.00	.5	.12	.2
18	18	TRAR5	rs1933988	.20	.2	.52	.06	.00	.5	.20	.2	.01	.4	.14	.2
19	19	TRAR5	rs8192627	.04	.3	.09	.3	.03	.4	.03	.3	.00	.4	.01	.4
20	20	TRAR5	rs2840836	−.02	.6	−.13	.8	−.08	.7	.03	.4	−.18	.8	−.03	.6
21	…	TRAR4	rs2840837	.06	.3	.45	.08	−.06	.7	.19	.2	−.03	.6	.11	.2
22	…	TRAR4	rs1361280	.17	.2	.15	.2	−.02	.6	.26	.14	.00	.5	.32	.11
23	…	TRAR4	rs4473885	.27	.13	.10	.2	−.01	.6	.48	.07	.00	.6	.32	.11
24	…	TRAR4	rs4085406	.22	.2	.19	.2	−.01	.6	.27	.13	.00	.5	.32	.11
25	…	TRAR4	rs6907909	.13	.2	.10	.2	.00	.5	.23	.15	.00	.5	.22	.2
26	…	TRAR4	ss28447860	−.02	.6	−.02	.6	.00	.5	.05	.3	−.22	.8	.25	.14
27	…	TRAR4	rs8192624	.27	.13	.00	.5	.04	.3	.25	.14	.04	.3	.01	.4
28	21	TRAR4	rs8192625	.93	.02	−.07	.7	.86	.02	.71	.04	.23	.2	.01	.4
29	…	TRAR4	rs7772821	.19	.2	.88	.02	−.04	.7	-0.04	.7	.72	.03	.16	.2
30	…	TRAR4	ss28447871	.08	.3	.00	.5	.02	.4	.26	.14	−.05	.7	−.02	.6
31	22	TRAR4	rs4305745	.23	.2	.86	.02	.00	.6	.03	.3	.25	.14	.01	.4
32	…	TRAR4	rs7745308	.55	.06	.27	.13	.65	.04	.37	.1	.14	.2	.10	.3
33	…	TRAR4	rs6912930	.07	.3	.09	.3	.00	.5	.88	.02	−.42	.9	.09	.3
34	…	TRAR4	rs6903874	1.22	.009	.73	.03	.36	.1	.19	.2	1.12	.011	.13	.2
35	…	TRAR4	rs7765655	.00	.5	.09	.3	−.05	.7	.32	.11	−.29	.9	.08	.3
36	…	TRAR4	rs6937506	1.76	.002	.78	.03	.50	.07	.39	.09	1.51	.004	.13	.2
37	…	TRAR4	rs4129284	.13	.2	.16	.2	−.02	.6	1.18	.01	−.39	.9	.09	.3
38	…	TRAR4	rs9321354	.49	.07	.00	.5	.32	.11	.77	.03	.02	.4	.02	.4
39	23	PNR	rs3813354	.04	.3	.00	.4	−.03	.6	.40	.09	−.15	.8	.21	.2
40	24	TRAR2	rs4144146	.79	.03	.00	.5	.34	.11	.79	.03	.10	.3	.01	.4
41	25	TRAR2	rs4467795	.33	.11	1.06	.013	.00	.5	.57	.05	.02	.4	.01	.4
42	26	GPR57	rs4421218	−.02	.6	.24	.15	−.03	.7	.00	.5	−.11	.8	.20	.2
43	27	GPR57	rs1081074	.05	.3	.09	.3	−.01	.6	−.02	.6	.16	.2	.12	.2
44	28	GPR58	rs8192646	.91	.02	.14	.2	.76	.03	.35	.1	.48	.07	.11	.2
45	29	GPR58	rs4451148	.06	.3	.43	.08	.46	.07	.06	.3	−.01	.6	.30	.12
46	30	GPR58	rs4380767	.89	.02	−.04	.7	.82	.03	.70	.04	.20	.2	.10	.3
47	31	TRAR1	rs7739700	1.26	.008	.17	.2	1.03	.015	1.72	.002	.05	.3	−.02	.6
48	32	TRAR1	rs8192620	.40	.09	.44	.08	.57	.05	.44	.08	.00	.5	.23	.2
49	33	TRAR1	rs4897595	−.11	.8	.15	.2	−.06	.7	−.10	.7	−.05	.7	.05	.3

^aFamilies counted are those that were informative for linkage (i.e., those with at least two affected members genotyped).

^bPreviously known SNPs are indicated by an rs number; novel SNPs are indicated by an NCBI Assay ID number (ss number) that we obtained (data to be released in the next build of dbSNP).

Table A5.

SNP Markers in Initial Screening and FBAT Results

		Nucleotide at
Gene	dbSNP Accession No.	Allele 1	Allele 2	MAFa	Distanceb(bp)	Positionc(bp)	Position in Gene	Overtransmitted Allele	Z	P
MOXD1	rs599660	G	A	.45	38,891	132,622,245	Intron 8 (boundary)	2	1.482	.14
MOXD1	rs1981187	C	T	.31	7,476	132,661,136	Intron 3	2	1.037	.30
MOXD1	rs1338387	C	T	.27	10,414	132,668,612	Intron 3	1	.450	.65
MOXD1	rs1538308	A	G	.07	12,744	132,679,026	5′ flanking	1	1.162	.25
MOXD1	rs2275394	C	T	.30	34,763	132,691,770	5′ flanking	1	.382	.70
MOXD1	rs6937815	G	A	.04	16,432	132,726,533	5′ flanking	1	.853	.39
MOXD1	rs3823288	A	T	.24	18,648	132,742,965	5′ flanking	1	.475	.63
STX7	rs1856352	A	G	.05	4,181	132,761,613	3′ flanking	1	.632	.53
STX7	rs3757299	A	C	.22	16,651	132,765,794	Intron 8	2	.524	.60
STX7	rs1002799	A	G	.25	12,162	132,782,445	Intron 2	1	.344	.73
STX7	rs2788942	G	A	.36	10,942	132,794,607	Intron 2	1	1.148	.25
STX7	rs2842884	G	C	.36	17,417	132,805,549	Intron 1	2	2.369	.0178
STX7	rs1591811	G	A	.35	17,213	132,822,966	5′ flanking	1	.263	.79
TRAR3	rs2842899	A	T	.28	3,755	132,840,179	Lys61Stop	2	.507	.61
TRAR3	rs2788935	C	T	.29	8,289	132,843,934	3′ flanking	1	.332	.74
TRAR5	rs1933988	A	C	.36	3,161	132,852,223	Promoter	2	.048	.96
TRAR5	rs8192627	A	C	.06	4,541	132,855,384	Asp328Ala	1	.361	.72
TRAR5	rs2840836	G	A	.23	12,977	132,859,925	3′ flanking	2	.316	.75
TRAR4	rs8192625	A	G	.09	1,380	132,872,902	Cys291Tyr	2	2.058	.0396
TRAR4	rs4305745	C	T	.47	16,922	132,874,282	3′ flanking	2	3.190	.0014
PNR	rs3813354	G	A	.08	5,759	132,891,204	Ala64Ala	2	1.357	.17
TRAR2	rs4144146	G	A	.08	4,812	132,896,963	Phe62Phe	1	1.241	.21
TRAR2	rs4467795	G	C	.45	9,887	132,901,775	Promoter	2	1.002	.32
GPR57	rs4421218	C	T	.05	365	132,911,662	Promoter	2	.141	.89
GPR57	rs1081074	A	C	.07	7,385	132,912,027	Promoter	2	2.084	.0372
GPR58	rs8192646	G	A	.03	1,451	132,919,412	Trp123Stop	1	.789	.43
GPR58	rs4451148	T	C	.26	6,429	132,920,863	Promoter	1	.355	.72
GPR58	rs4380767	T	C	.31	8,241	132,927,292	Promoter	2	1.518	.13
TRAR1	rs7739700	T	C	.34	11,316	132,935,533	3′ flanking	2	.920	.36
TRAR1	rs8192620	C	T	.02	1,092	132,946,849	Val288Val	1	1.720	.09
TRAR1	rs4897595	G	A	.07	NA	132,947,941	Promoter	1	.409	.68

^aMAF = minor-allele frequency.

^bDistance is the number of bp to the next SNP. NA = not applicable.

^cPosition in bp was derived from UCSC July 2003 freeze of chromosome 6 (see the UCSC Genome Bioinformatics Web site).

Table A6.

Mutations Detected in TRAR4 by Sequencing 30 Probands with Schizophrenia

dbSNP Accession No.a	SNPb	Ancestral Allelec	Local Positiond(bp)	Positione(bp)	TRAR4 Position	MAFf	Sequence Context(20 bp-SNP-20 bp)
rs6907909	A-713G	A	289	132,871,318	5′ flanking	.47	GCCAAGTATTGATGCTTACT[A/G]TTTACACCCTATTGTATCTT
ss28447859	G-285A	G	717	132,871,746	5′ flanking	.04	GGATATTTAAAATCAAAAG[G/A]AATTTTATCAAATTAAGAGC
ss28447860	C-276G	C	726	132,871,755	5′ flanking	.14	AAATCAAAAGNAATTTTAT[C/G]AAATTAAGAGCATGAGACAT
ss28447861	A-240C	A	762	132,871,791	5′ flanking	.04	ACATTTATCAGTTGAAACA[A/C]TCTCCAATAATCTTGTGCAA
ss28447875	T-10A	T	992	132,872,021	5′ flanking	.11	AAACTTCTCCATATGTAAN[A/T]AACAGCGTTATGAGCAGCAA
rs8192622	C78T	C	1,079	132,872,108	Pro26Pro	.05	GGTCCTGTGTGAAAATCCC[C/T]TTCTCGCCGGGATCCCGGGT
ss28447862	T110C	T	1,111	132,872,140	Ile37Thr	.03	ATCCCGGGTGATTCTGTACA[T/C]AGTGTTTGGCTTTGGGGCTG
ss28447876	G493A	G	1,494	132,872,523	Gly165Ser	.02	TCCTGCCCCTCATGTACAGC[G/A]GTGCTGTGTTCTACACAGGT
ss28447863	A518G	A	1,519	132,872,548	Tyr173Cys	.03	TGTGTTCTACACAGGTGTCT[A/G]TGACGATGGGCTGGAGGAAT
ss28447864	C630G	C	1,631	132,872,660	Thr210Thr	.03	CTATCCTTCTTTATACCTAC[C/G]TTTATTATGATAATTCTGTA
ss28447865	C683T	C	1,684	132,872,713	Ala228Val	.10	TCTTGTGGCTAGACGACAGG[C/T]GAAAAAGATAGAAAATACTG
ss28447867	A744G	A	1,745	132,872,774	Arg248Arg	.02	TCAGAGAGTTACAAAGCCAG[A/G]GTGGCCAGGAGAGAGAGAAA
rs8192624	G793A	G	1,794	132,872,823	Val265Ile	.10	AAACCCTGGGGGTCACAGTG[G/A]TAGCATTTATGATTTCATGG
rs8192625	G872A	A	1,873	132,872,902	Cys291Tyr	.13	GGGCTTTATAACCCCTGCCT[G/A]TATTTATGAGATTTGCTGTT
ss28447866	G976A	G	1,977	132,873,006	Val326Ile	.03	GGAAAGCAATAAAAGTTATT[G/A]TAACTGGTCAGGTTTTAAAG
rs7772821	T1046G	G	2,047	132,873,076	3′ flanking	.23	TGAACATATATAAGCAGTTG[T/G]ATAGACGAAGTTCAGGATAC
rs7752618	C1083G	C	2,084	132,873,113	3′ flanking	.04	ATACCTTTAAAATTACCAAG[C/G]GAAATGAGTTTTTAAAAATC
ss28447868	C1350G	G	2,351	132,873,380	3′ flanking	.27	ATTTTTCCTAAAAATATTT[G/C]TNTTTTTTTTTTTATTTATT
ss28447869	G1352T	T	2,353	132,873,382	3′ flanking	.42	TTTTTCCTAAAAATATTTNT[T/G]TTTTTTTTTTTATTTATTCC
ss28447872	T1437A	T	2,438	132,873,467	3′ flanking	.03	CCAAAAATTTCATTTGTGAA[T/A]AGCCTTTATCAAATTATTGG
ss28447871	A1480G	A	2,481	132,873,510	3′ flanking	.23	TCTTTTGCTTTGGTTATTTT[A/G]CCACAGGAGTCCTTTTAGGT
ss28447870	T1544C	T	2,545	132,873,574	3′ flanking	.10	GGGAGAGATCTCAGGGTGTA[T/C]GGGGCAATTTGCAAATGAAG
ss28447874	G1994C	G	2,995	132,874,024	3′ flanking	.03	TTGTATGGAAATCAGTGNTA[G/C]ATGCCTTAGACACAGGCATA
rs4305745g	A2252G	A	3,253	132,874,282	3′ flanking	.39	TCATGAGAAAAATGTCAGAC[A/G]AAACTTTATCGNGGAATATT
ss28447873g	A2263del	A	3,264	132,874,294	3′ flanking	.39	ATGTCAGACNAAACTTTATCG[A/-]GGAATATTCTGCATAATATT
rs7452939g	A2305G	G	3,306	132,874,335	3′ flanking	.39	GATCAGTATTTCTCAAAACT[A/G]TCAGTCATCAAAAATAAAGT

^aPreviously known SNPs are indicated by an rs number; novel SNPs are indicated by an NCBI Assay ID number (ss number) that we obtained (data to be released in the next build of dbSNP).

^bSNPs were named according to their relative position to the first letter of start codon ATG and with the first base as the major allele; CDS is 1,038 bp.

^cAncestral allele was determined according to the sequence comparison between human and two chimpanzees (GenBank accession numbers AB180397 and AB180398) and two gorillas (GenBank accession numbers AB180399 and AB180400).

^dLocal position refers to the 3,539-bp sequenced fragment of TRAR4.

^ePosition in bp was derived from UCSC human genome draft (July 2003 freeze) for chromosome 6 (see the UCSC Genome Bioinformatics Web site).

^fMAF = minor-allele frequency.

^gThe SNP rs4305745 was in perfect LD with ss28447873 and rs7452939.

Table A7.

Single-Marker Association via FBAT for All Markers with ⩾10 Informative Families in the Whole Sample^[Note]

					Results for Data Set
					All Families			AA Families			EA Families
SNP No.	Screening SNP No.	Gene	dbSNP Accession No.a	Associated Allele	No. of Informative Families	O/Eb	Pc	No. of Informative Families	O/Eb	Pc	No. of Informative Families	O/Eb	Pc
1	2	MOXD1	rs599660	2	80	1.07	.14	8	1.59	.10	65	1.07	.14
2	4	MOXD1	rs1981187	2	56	1.08	.30	9	.83	.06	43	1.23	.06
3	5	MOXD1	rs1338387	1	73	1.03	.65	16	.92	.48	48	1.15	.12
4	6	MOXD1	rs1538308	1	32	1.16	.25	9	1.01	.94	18	1.45	.029
5	7	MOXD1	rs2275394	1	52	1.04	.70	7	.83	.18	34	1.17	.22
6	8	MOXD1	rs6937815	1	13	1.06	.39	2	1.00	1.00	7	1.18	.21
7	9	MOXD1	rs3823288	1	72	1.02	.63	19	.99	.92	46	1.07	.17
8	10	STX7	rs1856352	1	28	1.09	.53	7	1.29	.29	18	.88	.46
9	11	STX7	rs3757299	2	65	1.05	.60	12	1.06	.78	43	1.03	.77
10	12	STX7	rs1002799	1	62	1.01	.78	9	1.01	.92	43	1.05	.43
11	13	STX7	rs2788942	2	74	1.05	.36	13	1.13	.24	54	1.03	.59
12	14	STX7	rs2842884	2	63	1.17	.018	8	1.32	.13	50	1.17	.022
13	15	STX7	rs1591811	1	71	1.02	.79	11	.92	.59	49	1.06	.53
14	16	TRAR3	rs2842899	2	72	1.04	.61	5	.73	.13	58	1.10	.19
15	17	TRAR3	rs2788935	1	86	1.03	.70	17	1.16	.34	56	1.03	.73
16	18	TRAR5	rs1933988	2	98	1.00	.96	17	1.32	.043	68	.95	.48
17	19	TRAR5	rs8192627	1	26	1.02	.72	4	.83	.25	20	1.07	.36
18	20	TRAR5	rs2840836	2	63	1.03	.75	2	.86	.78	52	1.07	.38
19	…	TRAR4	rs2840837	1	70	1.01	.84	12	1.22	.28	47	1.05	.56
20	…	TRAR4	rs1361280	1	91	1.06	.24	19	1.17	.15	60	1.10	.07
21	…	TRAR4	rs4473885	1	97	1.06	.21	22	1.21	.032	63	1.07	.19
22	…	TRAR4	rs4085406	1	92	1.06	.25	22	1.23	.047	59	1.06	.30
23	…	TRAR4	rs6907909	1	85	1.07	.18	19	1.28	.019	56	1.05	.36
24	…	TRAR4	ss28447860	1	54	1.08	.08	9	1.18	.06	41	1.07	.22
25	…	TRAR4	rs8192624	2	31	1.03	.53	5	1.07	.49	23	.99	.89
26	21	TRAR4	rs8192625	2	37	1.10	.040	9	1.24	.011	26	1.05	.46
27	…	TRAR4	rs7772821	1	65	1.06	.40	20	.89	.30	37	1.21	.041
28	…	TRAR4	ss28447871	2	76	1.03	.45	9	.96	.67	56	1.06	.24
29	22	TRAR4	rs4305745	2	78	1.17	.0014	17	1.20	.035	53	1.17	.015
30	…	TRAR4	rs7745308	2	21	1.12	.44	10	1.23	.25	10	.91	.71
31	…	TRAR4	rs6912930	1	86	1.04	.59	15	1.17	.24	60	1.00	.96
32	…	TRAR4	rs6903874	1	75	1.13	.0026	18	1.27	.0082	49	1.09	.07
33	…	TRAR4	rs7765655	1	64	1.02	.83	10	1.29	.16	49	.98	.81
34	…	TRAR4	rs6937506	1	77	1.12	.0052	18	1.24	.025	51	1.11	.035
35	…	TRAR4	rs4129284	1	86	1.06	.31	17	1.21	.09	58	1.01	.88
36	…	TRAR4	rs9321354	1	60	1.05	.46	16	.97	.77	31	1.12	.32
37	23	PNR	rs3813354	2	41	1.14	.17	9	1.01	.96	22	1.18	.26
38	24	TRAR2	rs4144146	1	37	1.06	.21	11	1.06	.49	19	1.03	.65
39	25	TRAR2	rs4467795	2	83	1.06	.32	16	.95	.60	60	1.12	.11
40	26	GPR57	rs4421218	2	23	1.03	.89	5	1.57	.14	14	.65	.14
41	27	GPR57	rs1081074	2	26	1.12	.037	16	1.20	.025	6	1.02	.74
42	28	GPR58	rs8192646	1	14	1.07	.43	3	1.00	1.00	11	1.08	.40
43	29	GPR58	rs4451148	1	63	1.03	.72	10	1.28	.41	47	.97	.78
44	30	GPR58	rs4380767	2	80	1.07	.13	5	.86	.17	67	1.07	.13
45	31	TRAR1	rs7739700	2	90	1.06	.36	15	.98	.91	65	1.08	.26
46	32	TRAR1	rs8192620	1	66	1.17	.09	12	1.05	.85	50	1.19	.10
47	33	TRAR1	rs4897595	1	33	1.02	.68	6	.94	.62	23	1.03	.63

Note.— For the analyses of these SNPs, alleles were tested for association if there were at least 10 informative families; in our data, this corresponds to not testing alleles and haplotypes rarer than 3%. This restriction, however, was not used when the investigation was limited to specific subsets of families in the secondary analyses, (i.e., the AA and EA families in this table).

^aPreviously known SNPs are indicated by an rs number; novel SNPs are indicated by an NCBI Assay ID number (ss number) that we obtained (data to be released in the next build of dbSNP).

^bRatio of the observed number of transmissions to the expected number of transmissions.

^cP values <.05 are in bold italics.

Table A8.

Coding Variants Detected in AA Probands with Schizophrenia and AA Controls^[Note]

		No. of Minor Alleles/Total Chromosomes Assayed (Frequency [%]) in
		Missense Variants										Synonymous Variants
Subject Group (No.)	Method	T110C(Ile37Thr)	G162A(Met54Ile)	G493A(Gly165Ser)	A518G(Tyr173Cys)	C683T(Ala228Val)	T692C(Ile231Thr)	G793A(Val265Ile)	G872A(Cys291Tyr)	G976A(Val326Ile)	Total	C78T(Pro26Pro)	C630G(Thr210Thr)	A744G(Arg248Arg)	Total
Probands with Schizophrenia (n = 14)	Sequencing	2/28 (7)	0/28 (0)	1/28 (4)	2/28 (7)	6/28 (21)	0/28 (0)	1/28 (4)	3/28 (11)	2/28 (7)	17/28 (61)	0/28 (0)	1/28 (4)	1/28 (4)	2/28 (7)
Probands with Schizophrenia (n = 18)	Genotyping	1/34 (3)	NA	1/36 (3)	1/34 (3)	8/32 (25)	NA	1/36 (3)	5/34 (15)	0/34 (0)	17/36 (47)	NA	NA	NA	NA
Controls (n = 48)	Sequencing	2/96 (2)	1/96 (1)	6/94 (6)	0/96 (0)	13/96 (14)	1/96 (1)	9/96 (9)	10/96 (10)	2/92 (2)	44/96 (46)	0/96 (0)	0/96 (0)	1/96 (1)	1/96 (1)

Note.— The minor-allele count is the numerator, and the total number of chromosomes assayed is the denominator, leading to a percentage (in parentheses). NA = not assayed.

Table A9.

Two-Marker Haplotype Association Analysis for TRAR4, for Haplotypes Including rs4305745^[Note]

Second Markera	χ2	Global P
rs2840837	10.906	.0122
rs1361280	10.365	.0157
rs4473885	12.720	.0053
rs4085406	10.701	.0135
rs6907909	11.983	.0074
ss28447860	12.824	.0050
rs8192624	11.318	.0101
rs8192625	12.493	.0059
rs7772821	13.302	.0040
ss28447871	9.825	.0201
rs7745308	11.689	.0085
rs6912930	11.470	.0094
rs6903874	9.110	.0105
rs7765655	10.686	.0136
rs6937506	10.261	.0165
rs4129284	13.055	.0045
rs9321354	9.847	.0199

Note.— The χ² test had 3 df for all markers, except for rs6903874, which had 2 df. Only markers with a minor-allele frequency >3% were included, and only haplotypes with at least a 3% frequency were analyzed.

^aPreviously known SNPs are indicated by an rs number; novel SNPs are indicated by an NCBI Assay ID number (ss number) that we obtained (data to be released in the next build of dbSNP).