Screening the Human Protocadherin 8 (PCDH8) Gene in Schizophrenia

Screening the human protocadherin 8 (PCDH8) gene in schizophrenia

N. J. Bray,* G. Kirov, R. J. Owen, N. J. Jacobsen, tion and social functioning. Despite evidence for a large heri- L. Georgieva, H. J. Williams, N. Norton, table component, the genetic loci that contribute to the dis- G. Spurlock, S. Jones, S. Zammit, order remain to be identified, and the neurobiological mech- M. C. O’Donovan and M. J. Owen anisms involved are largely unknown. In addition to possible disturbances in monoaminergic function, investigators have Department of Psychological Medicine, University of Wales Col- postulated aberrant development and plasticity of synaptic lege of Medicine, Heath Park, Cardiff, CF14 4XN, UK. connections within the prefrontal cortex and temporo-limbic *Corresponding author: Nicholas Bray, Department of Psycho- structures such as the hippocampus in the etiology of the logical Medicine, University of Wales College of Medicine, Heath disorder (Weinberger 1999). Park, Cardiff, CF14 4XN, UK. E-mail: [email protected] It has been suggested that the recently cloned protocadherin family of transmembrane adhesion/recognition molecules Abnormalities in synaptic connectivity and plasticity plays a particularly important role in the formation of neuronal have been implicated in the pathophysiology of schizo- circuitry (Kohmura et al. 1998; Hilschmann et al. 20 01). One phrenia. Molecules involved in the development and maintenance of neural circuitry include the recently protocadherin that has been directly implicated in synaptic cloned protocadherins. Human protocadherin 8 function is the rat ‘arcadlin’ (Yamagata et al. 1999). Arcadlin is (PCDH8) is homologous to ‘arcadlin’, a molecule localised to the synapses and soma of neurons, and is ob- shown to play a role in hippocampal synaptic function served in structures subserving specific neural circuits during in the rat. The gene encoding PCDH8 maps to a region development, including the hippocampus, amygdala and an- on chromosome 13 where linkage to schizophrenia has terior thalamic nuclei of the limbic system. Rat arcadlin (GI: been reported. In this study, the entire expressed se- 5478348) shares 85% amino acid identity with full-length hu- quence of the PCDH8 gene and over 800bp of the 5ø man protocadherin 8 (PCDH8; Strehl et al. 1998;GI: 6513852, flanking region were screened for polymorphisms in 30 see Footnote to introduction), and is therefore likely to be its DSM-IV schizophrenia individuals using Denaturing High Performance Liquid Chromatography (DHPLC). A ortholog. An alternatively spliced form of PCDH8 submitted to total of nine single nucleotide polymorphisms were GenBank (GI: 14589936), which is missing 97 amino acid resi- identified, including three in the first exon that are pre- dues from the cytoplasmic domain, shows 94% homology to dicted to change the amino acid sequence. One poly- arcadlin at the amino acid level. Human PCDH8 is found ex- morphism, causing the Trp7Arg change in the putative clusively in fetal and adult brain, and shows a similar pattern of signal peptide, showed a trend towards excess of the developmental expression to that of arcadlin (Strehl et al. arginine encoding allele in a case-control sample con- 1998). The full-length PCDH8 protein is predicted to consist of sisting of 520 DSM-IV schizophrenia patients and 535 1070 amino acids, and possess six cadherin ectodomains. matched controls from the UK (c2 3.72, P [1df] Ω Ω Strehl et al. (1998) report that the gene encoding PCDH8 has 0.054). However, this polymorphism did not show only three exons; a first exon calculated at over 2800bp en- preferential transmission to schizophrenic individuals in a separate sample of 203 proband–parent trios from coding the extracellular domain, the transmembrane region Bulgaria. A second, rare single nucleotide variation, and part of the cytoplasmic tail, and the two smaller exons en- predicting the non-conservative amino acid change coding the remaining cytoplasmic region and 3øUTR. The Glu39Ala, was found in one schizophrenic individual PCDH8 gene was found to span less than 5kb of genomic and their affected sibling but not in a further 352 DNA, with intron sizes of less than 500bp. affected individuals, nor 357 controls. These results The human PCDH8 gene has been mapped to chromosome suggest that any contribution of PCDH8 polymorph- 13q14.3 (Strehl et al. 1998), between the markers D13S284 isms to schizophrenia susceptibility is likely to be and D13S163 (Genemap ø99), approximating a region for weak, although the existence of rare variations of which several research groups report linkage to schizophrenia. stronger effect cannot be excluded. In short, Lin et al. (1995) reported positive lod scores for the markers D13S119 and D13S144 using a narrow diagnostic Keywords: Association, gene, neurodevelopment, poly- model. In a somewhat larger sample, Shaw et al. (1998) re- morphism, protocadherin, schizophrenia ported linkage to the nearby marker D13S168, whilst Riley et al. Received 28 March 2002, revised 3 May 2002, accepted (1998) found positive lod scores between the flanking markers for publication 13 May 20012 D13S126and D13S119in a sample of Bantu-speaking families. Protocadherin 8 was selected as candidate gene for schizophrenia on the basis of its predicted role in synaptic Schizophrenia is a common yet severe psychiatric condition, development and plasticity, and its localisation to a region on characterised by profound disturbances of cognition, emo- chromosome 13 where linkage to schizophrenia has been

187 Bray et al reported. In this study, the entire expressed sequence and http://www.cgi). All PCRs were carried out in a total reaction predicted proximal promoter of the PCDH8 gene were volume of 12ml, containing 24ng genomic DNA template, 1X screened for polymorphisms. Nine single nucleotide poly- PCR buffer, dNTPs at 0.1mM, primers at 0.12–0.24mM and morphisms were identified, including three that are predicted 0.3 m ‘Hot Star’ taq polymerase (Qiagen, Crawley, UK). In to alter the amino acid sequence. These variants were geno- some instances, 6% DMSO was added. typed in case-control and proband–parent trio samples in order to test for association with schizophrenia. Mutation screening The human PCDH8 gene was screened for polymorphisms Footnote to introduction using Denaturing High Performance Liquid Chromatography In the original paper describing arcadlin (Yamagata et al. (DHPLC), as described by Jones et al. (1999). 1999), it is noted that whilst the extracellular domain shares 91% amino acid identity with the predicted amino acid se- Sequencing of PCR fragments quence for human PCDH8 (Strehl et al. 1998), the cyto- Fluorescent sequencing of PCR fragments was carried out plasmic region is distinct. However, this discrepancy can be using the Big Dye Terminator Cycle Sequencing kit (PE Ap- explained by a sequencing error in the original description of plied Biosystems, Warrington, UK), as described by the human PCDH8, which resulted in a frameshift and truncation manufacturer. of the predicted protein. A corrected sequence for human PCDH8 has subsequently been submitted to GenBank by the Genotyping original authors and forms the basis of the present study. Individual genotyping of PCDH8 polymorphisms was carried out using RFLP assays. All restriction endonucleases were sup- Materials and Methods plied by New England Biolabs (Hitchin, UK), with digests performed using conditions recommended by the manufacturer. Subjects The mutation screening set consisted of 30 UK Caucasians Statistical analyses meeting DSM-IV (American Psychiatric Association 1994) The null hypothesis of no association between schizophrenia criteria for schizophrenia. All had an affected first-degree and PCDH8 polymorphisms was tested using the CLUMP relative, mainly a sibling, with schizophrenia. Patients within analysis package (Sham & Curtis 1995a). TDT analyses the mutation screening set and case-control sample were re- (Spielman et al. 1993) were performed using the programme cruited from psychiatric services in the UK via consultant re- ETDT developed by Sham and Curtis (1995b). All tests were ferral and underwent a semi-structured interview by a trained two-tailed. psychiatrist or psychologist using PSE-9 (Wing et al.1974) or SCAN (Wing et al. 1990). Control individuals were drawn from over 1400 blood donors recruited from the local branch Results of the National Blood Transfusion Service (Wales). Whilst controls were not specifically screened for the absence of Genomic characterisation psychiatric illness, the policy of the National Blood Trans- BLAST (/http://www.ncbi.nlm.nih.gov/BLAST/) alignments fusion service is not to receive donations from individuals between updated, full-length human PCDH8 cDNA se- who are on medication. The case-control sample consisted quence (GI: 6513852) and a chromosome 13 genomic clone of 552 UK Caucasians meeting DSM-IV criteria for schizo- (RP11–103K9, GI: 13785010) confirmed that PCDH8 is en- phrenia (398 males, mean age 42years ∫ 13) and 552 con- coded by three exons, as reported by Strehl et al. (1998). trol individuals, matched for age, sex and ethnicity (393 Calculated exon/intron sizes differed slightly from those in the males, mean age 41years ∫ 13), of which a ‘stage 1’ subset original report, in accordance with the corrected sequence. of 184 schizophrenic (134 males, mean age 43years ∫ 13 ) Thus, exon 1 was calculated as 2834bp, including a 5øUTR and 184 control individuals (129 males, mean age 42years of 203bp, exon 2 as 208bp and exon 3 at 958bp, including ∫ 12) was used for initial tests of association. Members of a3øUTR of 588bp. Intron sizes were calculated at 173bp and the proband–parent trio sample were recruited in Bulgaria in 491bp for introns 1 and 2, respectively. Intron-exon bound- a collaborative project organised by G. Kirov, and consisted aries were found to conform to the GT-AG rule (Shapiro & of 203 DSM-IV schizophrenic probands (101 males, mean Senapathy 1987). The use of 5øRACE by Strehl et al. to obtain age 33years ∫ 8) and their parents. The same diagnostic the 5øend of PCDH8 cDNA provided a demarcation between and assessment methods were used both in Bulgaria and 5øUTR and 5øflanking/proximal promoter regions in the pres- the UK. Approval to conduct genetic association studies was ent alignment. In accordance, the BLAST alignment revealed obtained in all local Ethics Committees where patients were a putative TATA box 26 bp upstream from the predicted tran- recruited. All patients, parents and controls were given an scription start site. information sheet and signed a consent form. Identification of PCDH8 polymorphisms PCR amplification All 4kb PCDH8 expressed sequence, approximately 890bp 5ø PCR primers were designed using the Primer3 programme flanking sequence and 268bp adjacent intronic sequence was (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_ amplified using 16 primer pairs (sequences available on

188 Genes, Brain and Behavior (2002) 1: 187–191 Screening PCDH8 in schizophrenia

Table 1: Conditions used for ampliﬁcation and RFLP genotyping of PCDH8 variants. HGVbase (release 14.0) identiﬁers in bold and parentheses

PCDH8 variant Primer sequences Amplicon size Annealing temperature Restriction endonuclease

ª331C/T F: agctcccagggcttaagaag 448bp 58æC Mnl-1 (SNP000430916) R: cacagggctccaggagtaaa

222T/C (Trp7Arg) F: gagaggcgcagagtgagg 376bp 56æC BsaH1 (SNP001493931) R: gaagcttgtgtcacccgata

319A/C (Glu39Ala) F: gagaggcgcagagtgagg 376bp 56æC Aci-1 (SNP001493932) R: gaagcttgtgtcacccgata

2431T/C (Val743Ala) F: ctctcggctcggggtgtc 359bp 62æC Hae-II (SNP001493935) R: cctctgagcccggcacttc

request) and screened for polymorphisms in 30 schizophrenic differences were not statistically significant (genotypic: c2Ω individuals using DHPLC. Nine variants in or flanking the 3.72, p [2df]Ω0.156; allelic: c2Ω3.72, p [1df]Ω0.054). The PCDH8 gene were identified in total: two SNPs in the 5øflank- odds ratio for the C-allele was 1.24 (95% CIΩ0.996–1.543). ing region (ª637C/T and ª331C/T), five SNPs in exon 1 Genotype counts for the 222T/C polymorphism in the entire (222T/C, 319A/C, 1343G/T, 1889C/T and 2431T/C), one SNP case-control sample are shown in Table2. in intron 1 (2834π65T/C) and one synonymous SNP in exon 2 On the basis of the stage 1 case-control data, the 222T/C (2837C/T) (numbering relative to the transcription start site ac- polymorphism was also genotyped in a sample of 203 schizo- cording to GI: 6513852). Three SNPs in exon 1 were predicted phrenia proband–parent trios from Bulgaria. Allele trans- to alter amino acids. These were: 222T/C, encoding Trp7Arg, mission to affected offspring of heterozygous parents is shown 319A/C, encoding Glu39Ala and 2431T/C, encoding Val743A- in Table3. Overall, the frequency of the C-allele in the parents la. Identified polymorphisms have been submitted to the Hu- was 0.180, and in the affected probands, 0.182. Although the man Genome Variation database (http://hgvbase.cgb.ki.se). C-allele was transmitted slightly more often than non-transmitted to schizophrenia probands, the TDT pattern did not ap- Genotyping of PCDH8 polymorphisms proach statistical significance (c2Ω0.03, P [1df]Ω0.86). As the most likely to impact on PCDH8 function, the non-syn- The 319A/C transversion was initially identified as a hetero- onymous 222T/C, 319A/C and 2431T/C variants within exon 1, zygote in one member of the mutation screening set. The C and the ª331C/T polymorphism within the 5øflanking/proxi- allele is predicted to result in a non-conservative amino acid mal promoter region, were individually genotyped to test for change from glutamic acid to alanine at amino acid residue association with schizophrenia. A sequential genotyping strat- 39 (Glu39Ala), located within the first cadherin ectodomain. egy was employed, whereby variants were initially genotyped P – BLAST analysis of PCDH8 amino acid sequence indicated in the stage 1 subset of the case-control sample, with those showing suggestive differences (i.e. PϽ0.1) genotyped in the remainder of the sample. Conditions used for amplification Table 2: Genotype counts for the 222 T/C polymorphism in the and genotyping of these variants are shown in Table1. entire case-control sample The 222T/C polymorphism is predicted to result in a non- Genotype Schizophrenia (nΩ520) Control (nΩ535) conservative amino acid change from tryptophan to arginine (Trp7Arg), and occurs in the putative signal peptide of the T/T 330 (63.5%) 369 (69%) PCDH8 molecule. The 222C allele had a frequency of 0.144 T/C 167 (32%) 148 (27.7%) in the control group, and 0.212 in the schizophrenia group, in C/C 23 (4.5%) 18 (3.4%) the stage 1 sample. This excess of the C (arginine-encoding) P-value (genotypic) 0.156 allele in the cases was found to be significant in terms of P-value (allelic) 0.054 both genotype (c2Ω6.19, P [2df]Ω0.042) and allele (c2Ω 5.39, P [1df]Ω0.02) frequencies. As an initially significant finding, the 222T/C polymorphism Table 3: Allele frequency distribution for the 222T/C polymorph- was genotyped in the remainder of the case-control sample. A ism in parents and probands within the Bulgarian trio sample and slight excess of the arginine encoding C allele was also ob- TDT results: the number of times each allele was transmitted and served in this second sample set, although differences were non-transmitted from heterozygous parents (nΩ122). not statistically significant (allelic: c2Ω0.52, P [1df]Ω0.47). In Allele Parents ProbandsTrans- Non- c2 TDT P- the combined case-control sample, the overall frequency of (nΩ406) (nΩ203) mitted trans- value the 222C allele was 0.205 in cases and 0.172 in controls, with mitted genotype frequencies conforming to Hardy–Weinberg expectation. Whilst a trend towards excess of the 222C allele in T 666 (82.0%) 332 (81.8%) 60 62 0.03 0.86 cases was therefore retained in the entire sample, observed C 146 (18.0%) 74 (18.2%) 62 60 0.03 0.86

Genes, Brain and Behavior (2002) 1: 187–191 189 Bray et al that this residue is conserved in numerous other human pro- significant excess of the C allele in terms of both genotype tocadherins, including all 52 genes in the PCDH a, b and g and allele frequencies. However, whilst a trend was retained clusters. In order to assess its frequency, the Glu39Ala variant in the entire case-control sample, observed differences were was genotyped in 368 schizophrenic and 368 control individ- not statistically significant. Moreover, TDT analysis of the uals from the case-control sample. Of the 709 samples suc- Trp7Arg polymorphism using the Bulgarian proband–parent cessfully genotyped, the C (alanine encoding) allele was ob- trios did not provide convincing evidence for the preferential served in only one case, as a heterozygote, which corre- transmission of the C allele to schizophrenic probands. sponded to the affected sibling of the screening set patient It is however, possible that the effect of the 222C allele is in whom the mutation was initially identified. too small to be reliably detected in samples even of the size In order to assess the transmission of the Glu39Ala muta- used in this study. On the basis of the case-control data, the tion, DNA was obtained from the father, mother and unaf- odds ratio for the C allele was calculated at 1.24. If genuine, fected sibling of the affected brothers. Neither the father nor such an effect size would require approximately 1100 cases mother were available for psychiatric assessment, although and an equivalent number of controls for 80% power to be both were reportedly unaffected. Genotyping of these ad- detected at the 0.05 probability level. Similarly, a substantially ditional family members showed that the Ala-encoding C al- larger family trio sample than was used would be necessary lele was inherited from the heterozygous father, whilst the to reliably detect association at such an effect size. Thus, a unaffected sibling was not found to carry the mutation. weak contribution of the 222C allele to schizophrenia suscep- No significant differences were observed in the distribution tibility cannot be excluded. of the ª331T/C polymorphism, for which the C allele showed A second interesting finding is that of the rare 319A/C tranv- a frequency of 0.120 and 0.106 in the stage 1 affected and ersion causing a non-conservative amino acid change from control groups, respectively (genotype: c2Ω1.0 01, P [2df]Ω glutamic acid (acidic) to alanine (neutral, hydrophobic). Of over 0.71; allele: c2Ω0.357, P [1df]Ω0.56). Genotype frequencies 350 schizophrenic and 350 control individuals screened, the conformed to Hardy–Weinberg expectation. Glu39Ala mutation was detected in only one schizophrenic The 2431T/C polymorphism, encoding Val743Ala, also did member of the screening set and their schizophrenic sibling not show association with schizophrenia in the stage 1 (as heterozygotes). Subsequent genotyping of other family sample (genotype: c2Ω0.357, P [2df]Ω0.553; allele: c2Ω members revealed that the variant was inherited from the 0.198, P [1df]Ω0.755). The distribution of genotypes in the father, for whom there were no clinical data, but was not stage 1 sample conformed to Hardy–Weinberg expectation, passed to the unaffected brother of the two affected siblings. with the frequency of the alanine encoding C allele observed Of course, the probability that the observed pattern of trans- at 0.065 and 0.057 for cases and controls, respectively. mission occurred by chance is high (one in four). In the absence of any other families found to carry the mutation, only by demonstrating that the amino acid change is of functional Discussion consequence might a case be made for its involvement in pro- moting schizophrenia in the two affected individuals in which In this study, the human PCDH8 gene was further character- it was identified. The altered amino acid occurs next to another ised at the genomic level and screened for polymorphisms in glutamic acid residue in the EC1 domain that is conserved in all a set of schizophrenic patients. Nine single nucleotide vari- cadherin-like proteins (Wu & Maniatis 1999), and which, in the ants were identified, including three in the first exon that are present BLAST analyses, itself appears to be conserved in nu- predicted to alter amino acids. Genotyping of PCDH8 poly- merous other protocadherins. Moreover, the EC1 domain ap- morphisms in case-control and proband-parent trio samples pears to be important for cadherin homophilic binding (Shapiro revealed no strong association with schizophrenia, although et al. 1995), and thus heterozygous changes in this region the possible role of a rare mutation in predisposing to the might be particularly detrimental to protein function. disorder in one family cannot be excluded. The 2431T/C polymorphism was the third variant identified Of the common nucleotide variants identified, the 222T/C in the large first exon of the PCDH8 gene that is predicted to polymorphism appears to be the most interesting. This SNP result in an amino acid change. This polymorphism, encoding is predicted to cause a non-conservative amino acid change Val743Ala, was not found to be associated with schizophrenia from tryptophan (non-polar) to arginine (basic) in the putative in the stage one case-control sample. However, this conserva- signal peptide of the PCDH8 molecule (Trp7Arg). Signal pep- tive amino acid change occurs near the transmembrane do- tides are short sequences of amino acids that determine the main in an extracellular region that shows less homology to rat eventual location of a protein. Yamagata et al. (1999) detail arcadlin, suggesting poorer conservation, and might therefore only one (conservative) amino acid variation in the putative have less potential to impact on protein function. 30 amino acid signal peptide between rat arcadlin and hu- Whereas Yamagata et al. (1999) note disparity between rat man PCDH8, suggesting strong evolutionary conservation. arcadlin and human PCDH8 throughout the cytoplasmic re- The Trp7Arg polymorphism was found to be relatively gion, updated PCDH8 sequence, confirmed in this study, ac- common in the present samples, however, with the arginine- tually predicts strong homology between rat and human in encoding C allele observed in controls within the overall case- this region. Mutation screening identified only one coding control sample at a frequency of approximately 0.17. In the SNP in this region (2837C/T), occurring in exon 2, which is stage 1 subset, the schizophrenic group showed a modestly predicted to be silent. No polymorphisms were detected in

190 Genes, Brain and Behavior (2002) 1: 187–191 Screening PCDH8 in schizophrenia the 3øUTR of the gene, which, along with the predicted Jones, A.C., Austin, J., Hansen, N., Hoogendoorn, B., Oefner, P.J., Cheadle, 5øUTR, was screened entirely. J.P. & O’Donovan, M.C. (1999) Optimal temperature selection for mutation It is likely that all but relatively rare polymorphisms occurring detection by denaturing HPLC and comparison to single-stranded confor- in PCDH8 expressed sequence were detected in this study. mation polymorphism and heteroduplex analysis. Clin Chem 45, 1133– 114 0. Under the null hypothesis that no PCDH8 polymorphisms are Kohmura, N., Senzaki, K., Hamada, S., Kai, N., Yasuda, R., Watanabe, M., Ishii, associated with schizophrenia, the screening set of 30 patients H., Yasuda, M., Mishina, M. & Yagi, T. (1998) Diversity revealed by a novel had 95% power to detect polymorphisms occurring in the family of cadherins expressed in neurons at a synaptic complex. Neuron general UK Caucasian population with a frequency of 0.05. 20, 1137–1151. However, whilst over 800bp 5øflanking sequence was screen- Lin, M.W., Curtis, D., Williams, N., Arranz, M., Nanko, S., Collier, D., McGuffin, ed for polymorphisms, we cannot exclude the presence of sus- P., Murray, R., Owen, M., Gill, M. & Powell, J. (1995) Suggestive evidence ceptibility variants in distal regulatory elements. for linkage of schizophrenia to markers on chromosome 13q14.1-q22. The PCDH8 gene should be screened for rare mutations Psychiatric Genet 5, 117–126. other than the Glu39Ala variant, which might also change an Riley, B.P., Lin, M.W., Mogudi-Carter, M., Jenkins, T., Williamson, R., Powell, amino acid, and which could potentially promote a neuropsy- J.F., Collier, D. & Murray, R. (1998) Failure to exclude a possible schizophrenia susceptibility locus on chromosome 13q14.1-q32 in Southern chiatric phenotype. However, identifying such variants might African Bantu- speaking families. Psychiatric Genet 8, 155–162. require screening the entire gene in very large sample num- Sham, P.C. & Curtis, D. (1995a) Monte–Carlo test for association between bers. disease and alleles at highly polymorphic loci. Ann Human Genet 59,97– PCDH8 maps to chromosome 13q14.3, a region for which 105. there is modest evidence for linkage to schizophrenia (e.g. Sham, P. & Curtis, D. (1995b) An extended transmission disequilibrium test Lin et al. 1995; Shaw et al. 1998), and perhaps also bipolar (TDT) for multiallele marker loci. Ann Human Genet 59, 323–336. affective disorder (Badenhop et al. 2001). Interestingly, as Shapiro, L., Fannon, A.M., Kwong, P.D., Thompson, A., Lehmann, M.S., Grub- noted by Yagi and Takeichi (2000), many other cadherin-re- el, G., Legrand, J.F.., Als-Nielsen, J., Colman, D.R. & Hendrickson, W.A. lated genes, including several protocadherins, map within (1995) Structural basis of cell-cell adhesion by cadherins. Nature 374, 327–337. chromosomal regions for which there is evidence for linkage Shapiro, M.B. & Senapathy, P. (1987) RNA Splice junctions of different to schizophrenia and/or bipolar disorder. These include classes of eukaryotes – sequence statistics and functional implications in PCDH9, also located on chromosome 13q, the PCDH a, b gene- expression. Nucl Acids Res 15, 7155–7174. and g clusters on chromosome 5q31 and PCDH7 on chromo- Shaw, S.H., Kelly, M., Smith, A.B., Shields, G., Hopkins, P.J., Loftus, J., Laval, some 4p15. Although the number and breadth of putative S.H., Vita, A., De Hert, M., Cardon, L.R., Crow, T.J., Sherrington, R. & DeLisi, linkage regions for schizophrenia would predict a degree of L.E. (1998) A genome-wide search for schizophrenia susceptibility genes. overlap, these genes might also be considered suitable can- Am J Med Genet 81, 364–376. didates for susceptibility under a neurodevelopmental model. Spielman, R.S., McGinnis, R.E. & Ewens, W.J. (1993) Transmission test for This study provides a comprehensive screen for poly- linkage disequilibrium – the insulin gene region and insulin-dependent morphisms in the gene encoding PCDH8, the human homo- diabetes-mellitus (IDDM). Am J Human Genet 52, 5 0 6–516. Strehl, S., Glatt, K., Liu, Q.M., Glatt, H. & Lalande, M. (1998) Characterization logue of the arcadlin molecule, which is involved in of two novel protocadherins (PCDH8 and PCDH9) localized on human hippocampal synaptic function in the rat. The identification chromosome 13 and mouse chromosome 14. Genomics 53, 81–89. of several single nucleotide polymorphisms across the gene, Weinberger, D.R. (1999) Cell biology of the hippocampal formation in including three that predict variation in amino acid sequence, schizophrenia. Biol Psychiatry 45, 395–402. should be useful in exploring the role of PCDH8 in suscepti- Wing, J.K., Babor, T., Brugha, T., Burke, J., Cooper, J.E., Giel, R., Jablenski, bility to schizophrenia in other ethnic groups, in general cog- S., Regier, D. & Sartorius, N. (1990) SCAN – Schedules for clinical-as- nitive function, and in predisposition to other neuropsychiatric sessment in neuropsychiatry. Arch General Psychiatry 47, 589–593. conditions where disturbances in synaptic development and Wing, J.K., Cooper, J.E. & Sartorius, N. (1974) The Description and Classi- plasticity have been implicated. fication of Psychiatric Symptoms. An Instruction Manual for the PSE and Catego Systems. Cambridge University Press, Cambridge. Wu, Q. & Maniatis, T. (1999) A striking organization of a large family of human neural cadherin-like cell adhesion genes. Cell 97, 779–790. References Yagi, T. & Takeichi, M. (2000) Cadherin superfamily genes: functions, genomic organization, and neurologic diversity. Genes Dev 14, 116 9–118 0. American Psychiatric Association (1994) Diagnostic and Statistical Manual Yamagata, K., Andreasson, K.I., Sugiura, H., Maru, E., Dominique, M., Irie, of Mental Disorders, 4th edn. (DSM-IV). American Psychiatric Press, Y., Miki, N., Hayashi, Y., Yoshioka, M., Kaneko, K., Karto, H. & Worley, Washington DC. P.F. (1999) Arcadlin is a neural activity-regulated cadherin involved in Badenhop, R.F., Moses, M.J., Scimone, A., Mitchell, P.B., Ewen, K.R., Rosso, long term potentiation. J Biol Chem 274, 19473–19479. A., Donald, J.A., Adams, L.J. & Schofield P.R. (2001) A genome screen of a large bipolar affective disorder pedigree supports evidence for a suscep- Acknowledgments tibility locus on chromosome 13q. Mol Psychiatry 6, 396–403. Hilschmann, N., Barnikol, H.U., Barnikol-Watanabe, S., Gotz, H., Kratzin, H. & This work was funded by an MRC (UK) studentship to N. J. Bray, a Medical Thinnes, F.P. (2001) The immunoglobulin-like genetic predetermination of Research Council (UK) programme grant and Co-Operative Group Grant the brain: the protocadherins, blueprint of the neuronal network. Natur- (M. J. Owen, M. C. O’Donovan) and by a grant to G. Kirov from the Janssen wissenschaften 88, 2–12. Research Foundation (Belgium).

Genes, Brain and Behavior (2002) 1: 187–191 191