Molecular Psychiatry (2007) 12,87–93 & 2007 Nature Publishing Group All rights reserved 1359-4184/07 $30.00 www.nature.com/mp ORIGINAL ARTICLE Genome-wide scan supports the existence of a susceptibility locus for schizophrenia and bipolar disorder on 15q26 G Vazza1, C Bertolin1, E Scudellaro1, A Vettori1, F Boaretto1, S Rampinelli2, G De Sanctis3, G Perini3, P Peruzzi2 and ML Mostacciuolo1 1Department of Biology, University of Padova, Padova, Italy; 2Department of Mental Health, ULSS 14 Chioggia, Chioggia, Italy and 3Department of Neurosciences, University of Padova, Padova, Italy

Schizophrenia (SZ) and bipolar disorder (BPD) are two severe psychiatric diseases with a strong genetic component. In agreement with the ‘continuum theory’, which suggests an overlap between these disorders, the existence of that affect simultaneously susceptibility to SZ and BPD has been hypothesized. In this study we performed a 7.5 cM genome scan in a sample of 16 families affected by SZ and BPD, all originating from the same northeast Italian population. Using both parametric and non-parametric analyses we identified linkage peaks on four regions (1p, 1q, 4p and 15q), which were then subjected to a follow-up study with an increased marker density. The strongest linkage was obtained on chromo- some 15q26 with a non-parametric linkage of 3.05 for marker D15S1014 (nominal P = 0.00197). Interestingly, evidence for linkage with the same marker has been reported previously by an independent study performed on SZ and BPD families from Quebec. In this region, the putative susceptibility ST8SIA2 (also known as SIAT8B) was recently associated with SZ in a Japanese sample. However, our allele frequency analyses of the two single-nucleotide polymorphisms (SNPs) with putative functional outcome (rs3759916 and rs3759914) suggest that these polymorphisms are unlikely to be directly involved in SZ in our population. In conclusion, our results support the presence of a gene in 15q26 that influences the susceptibility to both SZ and BPD. Molecular Psychiatry (2007) 12, 87–93. doi:10.1038/sj.mp.4001895; published online 12 September 2006 Keywords: schizophrenia; bipolar disorder; genome scan; linkage analysis; gene mapping

Introduction widely accepted that many genes and environmental factors contribute to their predisposition.2,3 Schizophrenia (SZ) and bipolar disorder (BPD) are Traditionally, according to ‘Kraepelinian dicho- two severe and debilitating mental disorders. SZ is tomy’ SZ and BPD have been classified as two characterized by delusions, hallucinations and diffi- distinct clinical entities and most of the clinical, culties in performing everyday activities and inap- epidemiological and genetic studies have been car- propriate display or total lack of emotions; on the ried out maintaining this rigorous classification.4 other hand, BPD is a distinct neuropsychiatric illness More recently, an alternative approach based on a characterized by a significant mood disturbance with continuum in which SZ and BPD overlap with the a recurrent alternation of elevated and low moods. affective spectrum, has been considered in psychia- Together these illnesses represent one of the top 10 tric research.5,6 According to this model, the genetic causes of disability in the world, each of them component could be conceptualized as a set of affecting about 1% of the human population and susceptibility genes conferring risk along different presenting a frequency of suicide among the affected domains of the spectrum. In other words, some individuals of about 10%.1 genes would be specific for SZ, others for BPD Several studies confirm the genetic component in and others would simultaneously confer risk across their etiology: family, twin and adoption studies the spectrum.5 evidence a high heritability (up to 80%) and it is Whole-genome linkage studies for SZ and BPD have identified significant linkage signals on several chromosome regions and some of them (13q, 22q, 18 Correspondence: Dr G Vazza, Department of Biology, University of and 6q) clearly overlap in both disorders.7 In Padova, Via Ugo Bassi 58/b, 35131 Padova, Italy. E-mail: [email protected] addition, several association and expression studies Received 22 March 2006; revised 13 July 2006; accepted 9 August supported the involvement of the same genes in SZ 2006; published online 12 September 2006 and BPD, including DAOA, DISC1, NRG1, COMT7 Genome scan for schizophrenia and bipolar disorder G Vazza et al 88 and GRK3,33 further supporting the genetic conver- for each participant and affected subjects were gence between these psychiatric disorders. diagnosed according to Diagnostic and Statistical In the past 2 years, linkage studies tried to identify Manual (DSM)-IV criteria. susceptibility genes shared by SZ and BPD using In addition all patients have been hospitalized and different approaches: Sklar et al.8 studied the occur- medical records were available for all of them. The rence of psychosis in SZ and BPD families and initial diagnosis was then reviewed and confirmed by reported evidence for linkage on chromosome 5q, a second psychiatrist using all data available such as whereas using similar diagnostic criteria in BPD interviews, lifetime medical and psychiatric informa- pedigrees Park et al.9 and Cheng10 reported significant tion. Families were ascertained through probands logarithm of odds (LOD) scores in 9q31 and 8p21, and affected by SZ spectrum disorders, bipolar spectrum in 16p12.3, respectively. Moreover, a genome-wide disorders (BPD) or SCA and were included in the search in families that were ascertained through a study in case at least an additional member of the proband with schizoaffective disorder (SCA) (a form family (first or second-degree relative) met DSM-IV of illness with features of both SZ and BPD), was criteria for SZ, BPD or SCA. performed by Hamshere,11 who reported a significant The sample included a total of 57 subjects available linkage at 1q42. Finally, by combining SZ and BPD for genotyping; among them 51 individuals families Maziade et al.12 identified significant linkage (26 females and 25 males) were diagnosed as follows: peaks on 18q and 15q26. In several 33 SZ, 14 BPD and four SCA. Based on clinical of these loci association studies identified putative assessment the families could be classified into: seven common susceptibility genes, such as DISC1 on SZ (with only SZ patients), two BPD (with only BPD chromosome 1q42,13 FZD314 and NRG115 on 8p2, patients) and seven mixed pedigrees (with SZ þ BPD, Epsin 417,18 and GABAA receptor19 on chromosome SZ þ SCA or SZ þ BPD þ SCA patients). 5q. In addition, a new gene, ST8SIA2 (also known as Considering the whole sample, 16 different SIAT8B), which maps on chromosome 15q26, pedigrees were studied: nine nuclear families has been recently associated with SZ in a Japanese (seven siblings pairs, one sibling trio and one sibling sample. Interestingly, this gene has been identified by quartet), two families including half-sibs, one family a positional-independent approach, considering its with avuncular pairs and three extended families functional involvement in the neuronal development (containing eight, seven and five affected subjects, via modification of the neural cell adhesion molecule respectively) with two or more sibships and complex 1(NCAM1) molecule.16 Among the genes reported relationships. above, a number of studies have reported positive For each family, pedigree structure was traced association with SZ and BPD only for DISC120,21 and confidentially by a genealogist back to four/five NRG1,22,23 whereas the involvement of the other genes generations using both church registers and demo- in both disorders is still uncertain. graphical records. For all the families their origin These findings could help unraveling the overlap from the old Chioggia population was confirmed and between SZ and BPD, but their replication in other no inter-family marriage or inbreeding loops were populations by independent groups is required to found in the sample. assess their validity. The number of regions involved Using the same approach, a group of 50 healthy in SZ and BPD by linkage studies and the failure in subjects all originating from Chioggia and unrelated replicating most of them suggests the existence to each other and the family sample was collected. of a wide genetic heterogeneity and raises the concern In addition a second sample of 50 unrelated healthy about studying samples of different ethnic origin. subjects randomly chosen from the wide Italian Several reports suggest that using samples from small population (defined as non-Chioggia sample) was populations could decrease allelic and locus hetero- available for the study. Controls from Chioggia were geneity; besides, the variability owing to environ- used to derive appropriate markers alleles frequencies mental, cultural and life style factors would also be for linkage analysis (see linkage analysis section) and lower than in large and mixed populations.24 to test allele frequencies of single-nucleotide poly- In this study we present the results of a genome- morphism (SNPs) in ST8SIA2 gene (see SNPs analysis wide scan and a follow-up study aimed at localizing on ST8SIA2 gene section); the last analysis was also susceptibility genes shared by SZ and BPD in a set of carried out in the non-Chioggia sample. families originating from the population of Chioggia, a town located in an island of the Venetian lagoon. Genotyping Blood samples were collected from all participants and DNA was isolated using a standard phenol– Materials and methods chloroform extraction procedure. The genome-wide Sample collection and ascertainment scan was performed using a set of 465 autosomal The Medical Ethic Committee of Chioggia in accor- microsatellites including 382 markers from the ABI dance with the current legislation approved the Prism Linkage Mapping Set, MD-10 and additional 83 present study, as well as the sample collection markers chosen from the Linkage Mapping Set, MD-5. procedures. All subjects were interviewed by an Overall the average marker spacing was 7.5 cM with expert psychiatrist, informed consent was obtained an average heterozigosity of 0.75.

Molecular Psychiatry Genome scan for schizophrenia and bipolar disorder G Vazza et al 89 Polymerase chain reaction (PCR) amplifications In the genome-wide linkage analysis, markers allele were performed under standard conditions using frequencies were estimated from the entire sample appropriate primers with fluorescent dye (FAM, using the quasi likelihood method implemented in HEX and NED). PCR products were then pooled into CC-QLS (case-control quasi-likelihood score test) multiplex panels and separated on an ABI PRISM program that take into account the kinship between 3700 DNA Sequencer (Applied Biosystems, Foster subjects of the same family.29 In the candidate regions, City, CA). Genotypes were assigned by two expert allele frequencies of all the markers were directly readers using GeneScan v2.0 and Genotyper v2.0 estimated from the sample of 50 unrelated controls software (Applied Biosystem). from Chioggia. No statistically significant differences In chromosomal regions showing a non-parametric were found between frequencies derived from linkage (NPL) > 1.5 in the first scan (i.e. 1p, 1q and, the family sample and those obtained from the sample 15q) we carried out a follow-up study using a of controls. denser marker map. Microsatellites were chosen from To establish the genome-wide empirical threshold, the UCSC Database given priority 100 replicates of whole genome scan were simulated to markers with the highest heterozygosity. Overall under the hypothesis of no linkage using Merlin 24 additional markers (D1S434, D1S228, D1S507, program.30 Each replicate matched the observed D1S2672, D1S436, D1S1592, D1S2644, D1S229, data set in family structure, marker map, alleles D1S2833, D1S251, D1S2709, D1S2850, D1S409, frequencies and missing data etc. Each replicate was D4S2912, D4S1587, D4S2971, D4S428, D15S207, then analyzed with Genehunter and empirical geno- D15S657, D15S1014, D15S212, D15S966, D15S203 mic P-value was obtained evaluating the false posi- and N15S1Q) were genotyped in all the families using tive rate of a given LOD score. For the follow-up the same procedure described previously. Marker study empirical thresholds were re-computed using N15S1Q is a new polymorphic dinucleotyde repeat the same procedure and incorporating the new that we identified starting at position 99425150 on markers analyzed. chromosome 15q26. It was amplified using primers 50-GCCAGTCTCCTCTGTGAAGG-30 and 50-TCCACCC SNPs analysis on ST8SIA2 gene TCTTATGCTCCAG-30 under standard PCR condi- The allele frequencies distribution of two SNPs tions. The marker’s order and genetic distances (SNPs 1: À1126T > C, rs3759916 and SNP3: À851T > C, were based on the UCSC Human Genome Database rs3759914) located in the promoter region of the gene (http://genome.ucsc.edu, hg17 release) and on the ST8SIA2 and directly associated to SZ in a Japanese genetic Rutgers map (http://compgen.rutgers.edu/ sample was evaluated in the Chioggia and non- maps/, build 34).25 Chioggia control samples. Genotyping was performed by PCR amplification Linkage analysis statistical methods and direct sequencing of a fragment containing both Errors in the allele calling were checked using SNPs. PCR reactions were performed under standard PedCheck program26 and mendelian inconsistencies conditions using the primers 50-AGGACATTCCATGT were resolved; the overall error rate was very low and AGCTTCTCCAA-30 and 50-CCAGGTGCTGATCTTTA was estimate in about 0.25%. Considering this data we TCTCCATTT-30. Sequencing was carried out on an performed genome-wide linkage study using a multi- ABI PRISM 3100 Genetic Analyzer using the Big Dye point approach that is theoretically more sensitive to terminator chemistry. genotyping errors but in contrast more powerful and less prone to random stochastic fluctuation than single point.27 Genome-wide analysis was performed with Results Genehunter version 2.128 using both parametric and Multipoint parametric and non-parametric LOD model-free statistical methods. To minimize the error scores obtained from the 7.5 cM genome scan are owing to multiple testing, heterogeneity parametric presented in Figure 1. LOD scores (HLOD) were calculated under one simple As shown by the NPL and HLOD plots, our findings dominant and one recessive model. are quite concordant for positive chromosomal re- According to the multidimensional model for these gions. The parametric analysis seems less powerful psychiatric illnesses, we used a single affection status than NPL, probably because of the sensitiveness definition grouping together patients with SZ, BPD of the multipoint analysis to the misspecification of and SCA phenotypes. the inheritance models. For this reason, we focused This analysis was conducted using an ‘affected only our attention on NPL statistics for P-value estimation. approach’ thus coding as phenotype ‘unknown’ all According to our genome-wide simulations, the the unaffected relatives. For both the models a NPL threshold for ‘suggestive’ linkage was 2.1. As phenocopy rate of 0.01 was assumed and the disease none of the initial linkage signals reached this value, allele frequency was set at 0.01 and 0.15 for dominant we chose to analyze additional markers in the regions and recessive models, respectively. NPL analysis was exceeding the arbitrary threshold of NPL > 1.5. performed using the Sall statistic that looks for excess Overall, four regions met these criteria and were allele sharing combining information from all affected selected for the follow-up study: chromosome 1p36 subjects within a pedigree. (NPL = 1.69 at marker D1S2697), 1q43 (NPL = 1.51 at

Molecular Psychiatry Genome scan for schizophrenia and bipolar disorder G Vazza et al 90 marker D1S2842), 4p14 (NPL = 1.71 for marker D4S405) and 15q26 (NPL = 1.93 for marker D15S120). All families were then genotyped with a total of 24 additional markers (seven for chromosome 1p, six in 1q, four in 4p and seven in 15q), achieving an average intermarker distance in these regions of 1.9, 3.5, 4.0 and 2.4 cM, respectively. Both initial and follow-up NPL results for chromo- somes 1, 4 and 15 are reported on panels in Figure 2. Overall, NPL on chromosome 1p weakly increased from 1.69 to 1.89, whereas on regions 1q and 4p we observed weakened linkage signals (with a NPL value decreasing from 1.51 to 1.38 and from 1.71 to 1.54, respectively) (Figure 2). The best score was obtained on chromosome 15q, with a NPL strongly increasing from 1.93 up to 3.05 (Table 1). Positive values in the same region were also observed by parametric analysis, yielding a maximum recessive HLOD of 2.43 (alpha = 0.91) and a dominant HLOD of 2.12 (alpha = 0.82). Genome-wide simulations were carried out once more, using the denser marker set. Results suggested that a NPL of 2.2 could be considered suggestive for linkage, whereas a NPL of 3.15 was needed to establish significant linkage. Based on these thresh- olds, linkage peaks on chromosome 1 and 4 fell into ‘background signals’, whereas our results on chromo- some 15 met the statistical criteria for suggestive linkage (genomic P-value = 0.07). The highest NPL score on chromosome 15q26 was observed at position 117 cM, close to markers D15S1014 and D15S212. Assuming max NPL-1 as the approximation of the 95% confidence interval (CI), the critical region spanned about 15 cM (from 107 to 122 cM) and was flanked by the external markers Figure 1 Genome-wide multipoint linkage results under D15S130 and D15S203. dominant and recessive models and for the non-parametric A thorough inspection of the LOD scores by family analysis. evidenced that the linkage signal on chromosome

Figure 2 Comparison of the multipoint NPL curves obtained in the original scan (dotted line) and after the follow-up study (solid line) for regions on chromosomes 1p, 1q, 4p and 15q.

Molecular Psychiatry Genome scan for schizophrenia and bipolar disorder G Vazza et al 91 Table 1 NPL scores observed on chromosome 15q26 in the basis of its functional features and the published first genome scan and after the follow-up study association study. We then performed a preliminary analysis to obtain the frequencies of the two SZ- Marker Genetic map Genome Follow-up Nominal associated SNPs in the population of Chioggia location (cM) scan NPL study NPL P-value (SNPs 1: À1126T > C, rs3759916 and SNP3: À851T > C, rs3759914). In addition, we also genotyped 50 healthy D15S127 94.64 0.29 0.36 0.35039 subjects coming from other Italian regions, in order to D15S130 106.11 1.21 1.99 0.02664 D15S207 112.37 — 2.30 0.01361 evaluate the allele distribution of these SNPs in the D15S657 113.82 — 2.35 0.01211 general Italian population. As reported in Table 2, no D15S1014 116.86 — 3.05 0.00197 differences were observed between the two samples. D15S212 118.52 — 3.00 0.00233 In both cases, SNP1 resulted to be a rare poly- D15S966 121.79 — 2.65 0.00587 morphism (minor allele frequency (MAF), lower D15S120 122.34 1.93 2.52 0.00811 than 5%), whereas for SNP3 only one allelic variant D15S203 123.54 — 1.95 0.02968 (T allele) was detected in all 100 subjects analyzed. N15S1Q 127.50 — 1.31 0.09776 A preliminary analysis was also carried out in the index cases of the family sample, showing compar- Abbreviation: NPL, non-parametric linkage. able allele frequencies with the control groups. Additional markers used for the follow-up study are in bold. Marker location on genetic map are given according to the Rutgers genetic map. Nominal P-value are calculated by Discussion GENEHUNTER program and are referred to markers after follow-up study. Our study presents a genome-wide search and a follow-up linkage study on a sample of families originating from Chioggia, a town in an island of the Table 2 Genotypes and alleles frequencies of the ST8SIA2 Venetian lagoon with a current population of 52,000. polymorphisms (SNP1 and SNP3) in the Chioggia and non- Based on its history, geographical position and social Chioggia control samples aspects, the population of Chioggia can be considered Chioggia Non-Chioggia a culturally ‘closed’ community, characterized by 31 sample sample (n = 50) a low number of family names. This latter peculi- (n = 50) arity, which seems to reflect the low migration and high endogamy of the town, suggests that the affected Genotypes individuals may share a small number of suscept- SNP1 T/T 48 (0.96) 48 (0.96) ibility alleles. À1126T > C T/C 2 (0.04) 2 (0.04) Following the hypothesis of a genetic overlap (rs 3759916) C/C 0 (0.00) 0 (0.00) between SZ and BPD, in this study we included families with SZ, families with BPD as well as Alleles families in which SZ, BPD and SCA occurred together T 98 (0.98) 98 (0.98) in the same pedigree. C 2 (0.02) 2 (0.02) A preliminary 7.5 cM autosomal scan detected Genotypes weak linkage peaks in four regions (1p, 1q, 4p and SNP3 T/T 50 (1.00) 50 (1.00) 15q), which were then subjected to a follow-up study À851T > C T/C 0 (0.00) 0 (0.00) with an increased marker density. (rs 3759914) T/C 0 (0.00) 0 (0.00) Overall, the best evidence of linkage was observed on a region of about 12 cM on chromosome 15q26, Alleles with a peak of NPL of 3.05 (nominal P = 0.00197). T 100 (1.00) 100 (1.00) These findings were also supported by the parametric C 0 (0.00) 0 (0.00) analysis that yielded a maximum HLOD of 2.43 (alpha = 0.91, under a recessive model of inheritance). Abbreviation: SNP, single-nucleotide polymorphism. These results should be taken into consideration even if the maximum NPL does not reach the 15q26 was supported by almost all the families (14/16 threshold for significance (genomic P-value of 0.07), pedigrees showed increased allele sharing in this as other groups reported evidence for linkage in region), with a contribution of both BPD, SZ and the same region. In particular, in a study aiming mixed families. at identifying susceptibility genes shared by SZ and BPD, Park et al.9 reported a suggestive multipoint NPL ST8SIA2 gene analysis of 2.62 for maker D15S652 in BPD families that had The ST8SIA2 gene, which has been recently asso- been reclassified on the basis of the presence of ciated to SZ in the Japanese population, maps psychotic features. More recently, a genome-wide on chromosome 15q26, about 5 Mb proximally to search on a set of 21 multigenerational families from our linkage peak. the Eastern Quebec population affected by SZ, BPD Even though the gene is located slightly outside of qor both disorders was carried out by the group of our 95% CI, it remains a plausible candidate on the Maziade et al.12 Using a ‘common locus’ phenotype

Molecular Psychiatry Genome scan for schizophrenia and bipolar disorder G Vazza et al 92 definition including SZ, BPD and SCA phenotypes, suggestive overlap with loci reported by other studies. the Authors reported a significant linkage at marker Namely, the peak on chromosome 1q is close to the D15S1014 (multipoint mod score of 4.55, significant DISC1 gene and the 1p region is slightly distal to the threshold = 4.0). MTHR gene, which has been recently associated both It is important to note that the marker D15S652, to SZ and BPD.34 These findings seem to support the reported in the work of Park et al., is only 0.63 cM role of these genes as common susceptibility factors to centromeric to our positive marker; moreover, the SZ and BPD, however they have to be interpreted significant linkage reported by Maziade et al. per- with caution given their low statistical significance. fectly overlaps with our peak and their highest LOD Recently, Arai et al.16 identified a new gene in the score was obtained for the same marker (D15S1014). proximity of our linkage peak in 15q26, named Following the Lander and Kruglyak32 criteria, a ST8SIA2, which has been indicated as a possible nominal P-value of 1 Â 10À2 is needed in order to susceptibility gene for SZ. The gene codes for a replicate linkage in a region that received significant poly-sialil-tranferase that catalyzes the transfer of scores in a previous genome-wide search. Consider- poly-sialic acid (PSA) to the NCAM1. Interestingly, ing the findings reported by Maziade et al., our results the authors considered ST8SIA2 a good candidate for marker D15S1404 (P-value of 0.00197) confirm the gene using a functional approach, as PSA-dependent linkage in 15q26. Moreover, nominal P-values < 0.01 modification of NCAM1 plays an important role in were observed for three additional consecutive mar- neuronal development and the perturbation of this kers flanking the D15S1014 (D15S112 P = 0.0023, process might therefore be an etiologic factor for D15S966 P = 0.0059 and D15S120 P = 0.0081). It is psychiatric disorders. In a case-control study per- important to notice that the statistical significance of formed on a Japanese sample the Authors identified our data was not hampered by the relatively small size significant allelic association for two SNPs (SNPs 1: of our sample, as we applied empirical P-values and À1126T > C, rs3759916 and SNP3: À851T > C, rs3759914) statistical thresholds via simulation. Also, we chose a located in the promoter of the ST8SIA2 gene. More- multipoint approach, rather than the single point one, over, an in vitro promoter assay performed to evaluate as the former is less sensitive to false positives. the functional role of these SNPs showed an increased Despite the many genome-wide searches for SZ or transcriptional activity of the risk-conferring haplo- BPD published so far, it is intriguing to notice that type, suggesting an involvement of the SNP1-T only the studies designed to identify genes shared by and SNP3-T alleles in the developing SZ. As no data SZ and BPD could find linkage signals on chromo- are reported in the dbSNP database (http://www. some 15q26.9,12 Indeed, our confirmatory linkage on ncbi.nlm.nih.gov/SNP) about alleles frequencies of locus 15q26 has also been obtained using a ‘common these SNPs in the European population, we per- locus’ model similar to that adopted by Maziade et al. formed a preliminary analysis on their frequencies in Altogether, our findings suggest that investigating SZ, a sample of 100 Italian subjects (50 individuals from BPD and SCA families in the same study may be a Chioggia and 50 randomly chosen from other Italian valid approach, particularly when dealing with regions). The SNPs analyzed resulted to be rare samples with a common ethnic origin. (SNP1) or not polymorphic at all (SNP3), both in The fact that many studies failed to find interesting Chioggia and in the general Italian population. This linkage signal on chromosome 15q268,10,11 could be means that whereas SNP3 could not be used for ascribed to the high locus heterogeneity that char- association studies, for SNP1 a sample of over 2000 acterize SZ and BPD. However, it should be noted that cases and 2000 controls it would be required to detect the 15q26 region is near to the telomere and therefore significant association in the Italian population the final result of a linkage analysis can be deeply (and probably also in the European population; data affected by markers analyzed. For instance, the most calculated with Power Calculator program assuming telomeric marker used by Sklar et al.8 for the long arm MAF of 0.03, power of 70% and relative risk 1.5). of (D15S675) was centromeric to Moreover, the alleles of the at-risk haplotype with our linkage peak; had we not used more telomeric putative functional outcome in the Japanese sample markers, in the first scan we would have obtained a (SNP1-T and SNP3-T) are unlikely to be directly NPL of 1.4, which would have been not considered involved in SZ in our population, as these variants eligible for subsequent study. This observation raises are the most common ones in the 100 healthy subjects the issue of comparing genome-wide scan results analyzed (SNP1-T f = 0.98, SNP3-T f = 1.00). The obtained with different markers sets and supports the association with the ST8SIA2 gene cannot be ex- relevance of using appropriate markers to replicate cluded, but further studies with different SNPs across previous findings. This factor should be taken into the gene and on a larger sample will be required. consideration especially when analyzing telomeric In summary, here we described a genome-wide regions, in which the presence of recombination search in a small but significant set of families hot spots can affect the linearity between physical with subjects affected by SZ, BPD and SCA. Further and genetic maps. analyses of positive regions supported evidence for The other regions highlighted in our genome linkage on chromosome 15q26. These results, together scan (i.e. 1p36 and 1q43) did not yield significant with the results from other independent studies, data in our follow-up study; yet, they showed strongly support the presence in this region of a gene

Molecular Psychiatry Genome scan for schizophrenia and bipolar disorder G Vazza et al 93 that influences the susceptibility to both SZ and BPD. schizophrenia in Chinese Han population. Am J Med Genet This would be in agreement with the hypothesized (Neuropsychiatr Genet) 2004; 129: 16–19. existence of susceptibility genes that affect the 15 Stefansson H, Sigurdsson E, Steinthorsdottir V, Bjornsdottir S, Sigmundsson T, Ghosh S et al. Neuregulin 1 and susceptibility to clinical spectrum of functional psychoses. schizophrenia. Am J Hum Genet 2002; 71: 877–892. 16 Arai M, Yamada K, Toyota T, Obata N, Haga S, Yoshida Y et al. Association between polymorphisms in the promoter region of Acknowledgments the sialyltransferase 8B (ST8SIA2) gene and schizophrenia. Biol Psychiatry 2006; 59: 652–659. We thank all the families for participating to this 17 Pimm J, McQuillin A, Thirumalai S, Lawrence J, Quested D, research. We are grateful to Dr A Coppe for ‘alacon- Bass N et al. The Epsin 4 gene on chromosome 5q, which encodes soool’ program and to Dr Vitiello for help in manu- the clathrin-associated protein enthoprotin, is involved in the script editing. This study was conducted for the genetic susceptibility to schizophrenia. Am J Hum Genet 2005; 76: 902–907. Regione Veneto, Giunta Regionale, Ricerca Sanitaria 18 Tang RQ, Zhao XZ, Shi YY, Tang W, Gu NF, Feng GY et al. Family- Finalizzata, Venezia Italia, Grant No 149/03; main based association study of Epsin 4 and Schizophrenia. Mol responsible for the project: Dr P Peruzzi. Financial Psychiatry 2006; 11: 395–399. support was also given by MIUR (National funds of 19 Petryshen TL, Middleton FA, Tahl AR, Rockwell GN, Purcell S, Ministry of Education, University and Research) Aldinger KA et al. Genetic investigation of chromosome 5q GABAA receptor subunit genes in schizophrenia. Mol Psychiatry and by the Foundation of Cassa di Risparmio di 2005; 10: 1074–1088. Padova e Rovigo. 20 Hodgkinson CA, Goldman D, Jaeger J, Persaud S, Kane JM, Lipsky RH et al. Disrupted in schizophrenia 1 (DISC1): association with schizophrenia, schizoaffective disorder, and bipolar disorder. Am J References Hum Genet 2004; 75: 862–872. 21 Thomson PA, Wray NR, Millar JK, Evans KL, Hellard SL, Condie A 1 American Psychiatric Association. Diagnostic and Statistical et al. Association between the TRAX/DISC locus and both bipolar Manual of Mental Disorders, 4th edn. American Psychiatric disorder and schizophrenia in the Scottish population. Mol Association: Washington, DC, 1994. Psychiatry 2005; 10: 657–668. 2 Cardno AG, Gottesman II. Twin studies of schizophrenia: from 22 Stefansson H, Sarginson J, Kong A, Yates P, Steinthorsdottir V, bow-and-arrow concordances to star wars Mx and functional Gudfinnsson E et al. Association of neuregulin 1 with schizo- genomics. Am J Med Genet 2000; 97: 12–17. phrenia confirmed in a Scottish population. Am J Hum Genet 3 Smoller JW, Finn CT. Family, twin, and adoption studies of 2003; 72: 83–87. bipolar disorder. Am J Med Genet (Semin Med Genet) 2003; 123: 23 Green EK, Raybould R, Macgregor S, Gordon-Smith K, Heron J, 48–58. Hyde S et al. Operation of the schizophrenia susceptibility gene, 4 Moller HJ. Bipolar disorder and schizophrenia: distinct illnesses neuregulin 1, across traditional diagnostic boundaries to increase or a continuum? J Clin Psychiatry 2003; 64(Suppl 6): 23–27. risk for bipolar disorder. Arch Gen Psychiatry 2005; 62: 642–648. 5 Craddock N, Owen MJ. The beginning of the end for the 24 Peltonen L, Palotie A, Lange K. Use of population isolates for Kraepelinian dichotomy. Br J Psychiatry 2005; 186: 364–366. mapping complex traits. Nat Rev Genet 2000; 1: 182–190. 6 Berrettini W. Evidence for shared susceptibility in bipolar disorder 25 Kong X, Murphy K, Raj T, He C, White PS, Matise TC. A combined and schizophrenia. Am J Med Genet (Semin Med Genet) 2003; 123: linkage-physical map of the human genome. Am J Hum Genet 59–64. 2004; 75: 1143–1148. 7 Craddock N, O’Donovan MC, Owen MJ. The genetics of schizo- 26 O’Connell JR, Weeks DE. PedCheck: a program for identification of phrenia and bipolar disorder: dissecting psychosis. J Med Genet genotype incompatibilities in linkage analysis. Am J Hum Genet 2005; 42: 193–204. 1998; 63: 259–266. 8 Sklar P, Pato MT, Kirby A, Petryshen TL, Medeiros H, Carvalho C 27 Sullivan PF, Neale BM, Neale MC, van den Oord E, Kendler KS. et al. Genome-wide scan in Portuguese Island families identifies Multipoint and single point non-parametric linkage analysis with 5q31–5q35 as a susceptibility locus for schizophrenia and imperfect data. Am J Med Genet (Neuropsychiatr Genet) 2003; 121: psychosis. Mol Psychiatry 2004; 9: 213–218. 89–94. 9 Park N, Juo SH, Cheng R, Liu J, Loth JE, Lilliston B et al. Linkage 28 Kruglyak L, Daly MJ, Reeve-Daly MP, Lander ES. Parametric and analysis of psychosis in bipolar pedigrees suggests novel putative nonparametric linkage analysis: a unified multipoint approach. loci for bipolar disorder and shared susceptibility with schizo- Am J Hum Genet 1996; 58: 1347–1363. phrenia. Mol Psychiatry 2004; 9: 1091–1099. 29 Bourgain C, Hoffjan S, Nicolae R, Newman D, Steiner L, Walker K 10 Cheng R, Juo SH, Loth JE, Nee J, Iossifov I, Blumenthal R et al. et al. Novel case–control test in a founder population identifies Genome-wide linkage scan in a large bipolar disorder sample from P-selectin as an atopy-susceptibility locus. Am J Hum Genet 2003; the National Institute of Mental Health genetics initiative suggests 73: 612–626. putative loci for bipolar disorder, psychosis, suicide, and panic 30 Abecasis GR, Cherny SS, Cookson WO, Cardon LR. Merlin – rapid disorder. Mol Psychiatry 2006; 11: 252–260. analysis of dense genetic maps using sparse gene flow trees. Nat 11 Hamshere ML, Bennett P, Williams N, Segurado R, Cardno A, Genet 2002; 30: 97–101. Norton N et al. Genomewide linkage scan in schizoaffective 31 Barrai I, Rodriguez-Larralde A, Mamolini E, Scapoli C. Isonymy disorder: significant evidence for linkage at 1q42 close to DISC1, and isolation by distance in Italy. Hum Biol 1999; 71: 947–961. and suggestive evidence at 22q11 and 19p13. Arch Gen Psychiatry 32 Lander E, Kruglyak L. Genetic dissection of complex traits: 2005; 62: 1081–1088. guidelines for interpreting and reporting linkage results. Nat 12 Maziade M, Roy MA, Chagnon YC, Cliche D, Fournier JP, Genet 1995; 11: 241–247. Montgrain N et al. Shared and specific susceptibility loci for 33 Niculescu III AB, Segal DS, Kuczenski R, Barrett T, Hauger RL, schizophrenia and bipolar disorder: a dense genome scan in Kelsoe JR. Identifying a series of candidate genes for mania and Eastern Quebec families. Mol Psychiatry 2005; 10: 486–499. psychosis: a convergent functional genomics approach. Physiol 13 Hennah W, Varilo T, Kestila M, Paunio T, Arajarvi R, Haukka J Genomics 2000; 4: 83–91. et al. Haplotype transmission analysis provides evidence of 34 Kempisty B, Mostowska A, Gorska I, Luczak M, Czerski P, association for DISC1 to schizophrenia and suggests sex- Szczepankiewicz A et al. Association of 677C > T polymorphism dependent effects. Hum Mol Genet 2003; 12: 3151–3159. of methylenetetrahydrofolate reductase (MTHFR) gene with 14 Zhang Y, Yu X, Yuan Y, Ling Y, Ruan Y, Si T et al. Positive bipolar disorder and schizophrenia. Neurosci Lett 2006; 400: association of the human frizzled 3 (FZD3) gene haplotype with 267–271.

Molecular Psychiatry