Molecular Psychiatry (2006) 11, 837–846 & 2006 Nature Publishing Group All rights reserved 1359-4184/06 $30.00 www.nature.com/mp ORIGINAL ARTICLE Evidence for association of DNA sequence variants in the phosphatidylinositol-4-phosphate 5-kinase IIa (PIP5K2A) with schizophrenia SG Schwab1,2,3, M Knapp4, P Sklar5,6, GN Eckstein7, C Sewekow3, M Borrmann-Hassenbach8, M Albus8, T Becker4, JF Hallmayer9, B Lerer10, W Maier3 and DB Wildenauer2,11 1Western Australian Institute of Medical Research and Center for Medical Research, University of Western Australia, Perth, WA, Australia; 2School of Psychiatry and Clinical Neurosciences, University of Western Australia, Perth, WA, Australia; 3Department of Psychiatry, University of Bonn, Bonn, Germany; 4Institute for Medical Biometry, Informatics and Epidemiology, University of Bonn, Bonn, Germany; 5Program in Medical and Population Genetics, Broad Institute of Harvard and MIT, Cambridge, MA, USA; 6Department of Psychiatry, Psychiatric and Neurodevelopmental Genetics Unit, Center for Human Genetics Research, Harvard Medical School, Massachusetts General Hospital, Charlestown, MA, USA; 7Institute of Human Genetics, GSF-National Research Center for Environment and Health, Munich, Germany; 8Mental State Hospital, Haar, Germany; 9Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Palo Alto, CA, USA; 10Department of Psychiatry, Hadassah-Hebrew University Medical Center, Jerusalem, Israel and 11Center for Clinical Research in Neuropsychiatry, Graylands Hospital, Mt Claremont, WA, Australia

Linkage studies in schizophrenia have identified a candidate region on 10p14–11 as reported for several independent samples. We investigated association of DNA sequence variants in a plausible candidate gene located in this region, the gene for phosphatidylinositol- 4-phosphate 5-kinase IIa (PIP5K2A), in a sample of 65 sib-pair families for which linkage had been reported. Evidence for association was obtained for 15 polymorphisms spanning 73.6 kb in the genomic region of the gene between intron 4 and the 30 untranslated region, a region with high degree of linkage disequilibrium. Single nucleotide polymorphism (SNP) rs10828317 located in exon 7 and causing a non-synonymous amino-acid exchange (asparagine/serine) produced a P-value of 0.001 (experiment-wide significance level 0.00275) for over-transmission of the major allele coding for serine, analysed by transmission disequilibrium test using FAMHAP. Association of this SNP with schizophrenia has been also described in a sample of 273 Dutch schizophrenic patients and 580 controls (P = 0.0004). PIP5K2A is involved in the

biosynthesis of phosphatidylinositol-4,5-bisphosphate (PI(4,5)P2), one of the key metabolic crossroads in phosphoinositide signalling. PI(4,5)P2 plays a role in membrane transduction of neurotransmitter signals as well as in intracellular signalling, pathways that may be impaired in schizophrenia. Molecular Psychiatry (2006) 11, 837–846. doi:10.1038/sj.mp.4001864; published online 27 June 2006 Keywords: linkage disequilibrium; single nucleotide polymorphism; non-synonymous amino-acid exchange; phosphoinositide cell signalling; candidate gene

Introduction The lack of a one to one relationship between phenotype and genotype renders application of Schizophrenia is a devastating mental disorder with classical linkage analysis in complex disorders diffi- complex genetic background.1 Susceptibility alleles, cult.2 Despite this constraint, a number of linkage that is, alleles conferring risk for the disorder and studies have been performed in schizophrenia using having only a moderate effect on phenotype expres- parametric and non-parametric methods.3 Statistical sion, are expected to contribute to the development of evidence for linkage has been reported for several the disorder. This is in contrast to most monogenic regions, distributed over the whole genome. However, disorders where causative DNA sequence variants the majority of these linkage studies did not fulfil (mutations) with usually larger effects are involved. stringent criteria such as genome-wide significance, concentration of elevated LOD scores within a narrow defined peak and replication in independent family Correspondence: Dr DB Wildenauer, Center for Clinical Research samples. Consequently, detecting linkage in schizo- in Neuropsychiatry (CCRN), Graylands Hospital, Private Bag #1, phrenia cannot be taken as proof for existence of a Claremont, WA 6910, Australia. susceptibility gene, but may be suggestive for regions E-mail: [email protected] Received 7 March 2006; revised 30 May 2006; accepted 5 June that can be tested for the existence of candidate 2006; published online 27 June 2006 associated with the disorder. Association of PIP5K2A with schizophrenia SG Schwab et al 838 DNA sequence variants in potential susceptibility approach and investigated association of DNA se- genes associated with schizophrenia have been quence variants in the region, where PIP5K2A is detected in some regions where evidence for linkage located, using the family sample where we had has been reported previously. On chromosome detected linkage. We detected several variants in 8p22–21,4–6 DNA sequence variants in the gene for linkage disequilibrium (LD) with schizophrenia in the neuregulin 1 have been found to be associated with genomic region of PIP5K2A. One of the variants, M41, schizophrenia as single markers and, in combination, that produced nominally significant P-values in this as multilocus haplotypes.7,8 Association of single study is located in exon 7 of the gene and constitutes nucleotide polymorphisms (SNPs) in the genomic a non-synonymous amino-acid exchange N/S in region of dystrobrevin binding protein 1 (dysbindin) position 251 of the protein. Over-transmission of the located on chromosome 6p24–22, another region with allele with major frequency coding for S was observed evidence for linkage to schizophrenia,4,9–15 has been in our family sample. reported.16–20 Further candidate genes identified in areas with reported linkage to schizophrenia include G72 on chromosome 13q32–q34,21 RGS4,22 Materials and methods Capon23 and DISC124,25 on chromosome 1q21–q22, Families TRAR4 on chromosome 6q13–q2626 and Epsin 4 on The sib-pair sample consisted of 65 families with chromosome 5q.27 141 affected offspring. Fifty-six families were from For most of these genes, a role in brain function has Germany and nine from Israel. The sample was taken been proposed,28 but DNA variants (mutations in from 71 families included in the linkage studies coding or regulatory sequences) with impact on brain previously reported.33,37 For six families, the supply function and eventually leading to the development of DNA was not available in sufficient quantity for of schizophrenia have not been identified yet. Thus, these studies. There were 57 families with affected some of the currently proposed susceptibility genes sib pairs, six families with three affected, one with may turn out to be false positives, whereas others still four and one with five. With the exception of three remain to be discovered. The number of true suscept- sib-pair families and one triplet with one parent, but ibility genes involved in development of schizophre- with additional unaffected family members, DNA nia cannot be estimated at the present stage. was available for both parents. All individuals had Evidence from linkage analysis for a locus, con- been interviewed with the Schedule for Affective ferring susceptibility to schizophrenia on chromo- Disorder Schizophrenia-Lifetime Version (SADS-L) some 10p14–p11, has been reported by a number of or the Structured Clinical Interview for DSM-III-R studies.29–36 In addition, evidence for linkage to bipolar Disorders (SCID-II).42 Case records were evaluated by disorder has been reported in this region for the NIMH operational criteria checklist for psychotic illness consortium bipolar sample.30 The highest LOD scores (OPCRIT).43 The complete interview form, family obtained in these studies were concentrating around history information and the medical record of each the microsatellite marker D10S1423 comprising a individual were reviewed by an independent relatively small area. As parental genotype information psychiatrist without prior knowledge of family is available in our sample with linkage,37 we had relationship to the index patient or of morbidity in been able to apply a transmission disequilibrium test the family. The 141 affected offspring consisted of modified for application to sib-pairs in order to test for 121 individuals with a core diagnosis of schizo- association in the presence of linkage.38 We tested all phrenia (DSM-III-R) and 20 individuals with a diag- microsatellite markers from our genome-wide scan nosis of schizoaffective disorder, schizophrenic sub- using the ASPEX program and obtained a nominally type, according to the Research Diagnostic Criteria.44 significant P-value of 0.038 for marker D10S211,37 Written informed consent was obtained from all which is located in the linkage peak. participants after a detailed description of the study, We searched databases for candidate genes in the which was approved by the internal review boards of vicinity of D10S211 and identified approximately the University of Bonn, Germany; the Hadassah- 900 kb downstream the gene for the phos- Hebrew University Medical Center, Israel; and the phatidylinositol-4-phosphate 5-kinase, type II, alpha Mental State Hospital, Haar, Germany. (PIP5K2A) as a potential susceptibility gene for schizophrenia. The enzyme is involved in phospho- Genomic DNA inositide signalling, predominantly catalysing phos- DNA was isolated from whole blood or permanent cell phorylation of phosphatidylinositol-5-phosphate lines derived from Epstein–Barr virus-transformed (PI(5)P) to phosphatidylinositol-4,5-bisphosphate lymphocytes using a Qiagen Blood- and Cell-Culture 39–41 (PI(4,5)P2). The latter compound plays a role in kit according to the instructions of the manufacturer. transmembrane transduction of neurotransmitter sig- nals in connection with intracellular Ca2 þ signalling Identification of SNPs as well as in a number of other intracellular signalling PCR primers were designed covering the 10 exons pathways.39,40 and adjacent intron sequences for amplification of Based on these considerations, we decided to apply 200–300 bp suitable for single-stranded conforma- a combined positional/functional candidate gene tion polymorphism (SSCP) analysis.45 Briefly, PCR

Molecular Psychiatry Association of PIP5K2A with schizophrenia SG Schwab et al 839 products (primer information available on request supplied in the AcycloPrime-FP-SNP-Detection kit from the authors) from 40 individuals (20 affected and (Perkin Elmer Life Sciences, Boston, MA, USA). 20 unaffected) were analysed on a Sequigene electro- Briefly, 12.5 ng genomic DNA was amplified by phoresis apparatus (BioRad, Hercules, CA, USA) with PCR using standard conditions and the remaining 0.4 mm non-denaturing polyacrylamide gels prepared nucleotides and primers inactivated by incubation using 0.5 Â MDE (mutation detection enhancement) with shrimp alkaline phosphatase (SAP) and solution from BioWhittaker (Walkersville, MD, USA). exonuclease 1 as supplied with the Cleanup Reagent The gels were run at 8 W at room temperature (Perkin Elmer Life Sciences). Single base extension overnight and the bands detected by silver staining. reaction was performed using the conditions as recommended by the supplier. Fluorescence polari- Sequencing zation was measured using a Tecan 384 Ultra reader PCR products were purified using a QIAquick PCR (Tecan US, Durham, NC, USA) and data analysed Purification Kit (Qiagen, Hilden, Germany). Sequen- with the Easy SNP E 2.0 (Tecan) software. cing was performed using BigDye Terminator Cycle Sequencing kits (Applied Biosystems, Forster City, Allele discrimination by matrix-assisted laser CA, USA) on an ABI Prism 377 DNA Sequencer desorption/ionization time-of-flight analysis. Geno- (primer information available on request from the typing of SNPs by mass spectrometry was performed authors). as described.48 In brief, primers were designed by SpectroDESIGNER software (Sequenom, San Diego, Genotyping CA, USA) and PCR carried out using AmpliTAQ Gold Different methods were used for genotyping DNA (0.1 U, Perkin Elmer, Boston, MA, USA) under sequence variants. The method applied for detection standard conditions. The amplification products of genotypes for the single markers as well as primer were treated with SAP and the MassEXTEND and enzyme information, if available, is shown in reactions performed as described.48 Each reaction Supplementary Table 1. The genotyping methods was loaded on 384-well SpectroCHIPs and analysed applied are described in brief as follows: by a MassARRAY RT mass spectrometer (Bruker- Sequenom49). The resulting spectra were analysed by Restriction enzyme fragment length polymor- SPECTROTYPER software (Sequenom). phism. BstNl (rs1857279), MseI (rs1977372) and NciI (rs3793756) were obtained from New England Masscode SNP allele discrimination. High-through- Biolabs (Ipswich, MA, USA). Buffer and incuba- put SNP genotyping with the Masscode System50 was tion conditions were used as recommended by the conducted at QIAGEN Genomics Inc., Bothell, WA, supplier. The fragments were separated by USA. The method is essentially an allele-specific electrophoresis on 2% agarose gels and visualized PCR SNP discrimination assay using small-molecular- by staining with ethidium bromide. weight tags covalently attached through a photo- Allele discrimination using TaqMan technology. A cleavable linker to an oligonucleotide. After photo- 12.5 ng portion of DNA each was used for TaqMan lysis, the mass of the tags is analysed in a single SNP genotyping assays on demand according to the quadrupole mass spectrometry detection system. protocol of Applied Biosystems. Fluorescence was Genotype determinations are based on the relative read after completion of PCRs by means of a Tecan proportions of the paired allele tags and analysed 384 Ultra reader (Tecan US, Durham, NC, USA). by cluster plots. Excitation and emission wavelengths for the carboxyfluorescein (FAM)-labelled probes were 485 Statistical tests and 535 nm, and for the VIC-labelled probes, 535 and The program FAMHAP version 15 was used to 590 nm, respectively. estimate haplotype frequencies with the EM-algo- rithm.51 The program with the ‘zhao’ option was also Microsatellites. Amplification of microsatellite loci used to calculate unadjusted P-values for association was performed by PCR with fluorescence-labelled of single markers and haplotypes with schizophrenia. primers as described previously.37 Fragment analysis Using this option, an association test based on Monte- was carried out by means of a DNA sequencer type Carlo simulations is performed, in which the set of 377 (Applied Biosystems, Forster City, CA, USA); transmitted and non-transmitted genotypes is ran- genotypes were assigned using genotyping soft- domly permutated in each replication.52 The test has ware Genescan 2.02 and Genotyper 1.1 (Applied been extended to test the null hypothesis of no Biosystems). association in samples of nuclear families with more than a single affected offspring, even in the presence Fluorescence polarization detection of template- of linkage.53 P-values corrected for multiple testing directed dye-terminator incorporation. Fluorescence were obtained as described.54 FAMHAP also calcu- polarization detection of template-directed dye- lates LD measures D055 for non-transmitted (‘pseudo’- terminator incorporation (FP-TDI) was performed controls) and transmitted haplotypes separately. The essentially as described,46,47 except using acyclo- spectral decomposition method56,57 was used to dideoxynucleotides and AcycloPol polymerase as estimate the number of effective loci and to determine

Molecular Psychiatry Association of PIP5K2A with schizophrenia SG Schwab et al 840 the experiment-wide significance threshold required Intermarker LD for SNPs in the region between M28 to keep type I error rate at 5%. and M51 was calculated using the frequency of the non-transmitted alleles as (pseudo-) controls (Table 2 below diagonal). The area between exon 5 (M38) and Results the 30UTR (M51) exhibits extensive LD and appears to The genomic structure of the gene PIP5K2A was not form a haplotype block (Table 2). Recombination hot known when the study commenced. Thus, we spots occur at marker M37 as well as at markers M32 determined exon–intron boundaries by sequencing and M33. The pattern for the transmitted allele BACs (F-factor-based bacterial artificial chromo- (‘cases’; Table 2 above diagonal) was similar, except somes) from coding sequences into introns. BACs a slight increase in LD between markers M41, M42 had been isolated before by screening the CTB and M43 and markers M28–M35. Markers from the library58 with sequences from the published cDNA International HapMap Project Release #19, phase II, sequence.59 The obtained information on the bound- October 2005 (http://www.hapmap.org/index.html.en) aries could be further confirmed when the human analysed by Haploview 3.2 for the Caucasian sample genome sequence became available. SSCP analysis of displayed a similar haplotype block pattern for amplification products of exon and adjacent intron the 63.3 kb region between M50 and M35 (30UTR to sequences revealed a number of sequence variations exon 5) with a recombination hot spot in the area of (Table 1). Amplification products showing changes in M38, 37–39 kb proximal to M50 (Supplementary SSCP analysis were sequenced in order to confirm Figure 1). and identify the position of the variation. We We calculated all possible two- and three-locus identified five polymorphisms, among them an A to haplotypes and estimated the transmission to affected G exchange in exon 7. This variant changes position offspring in comparison to non-transmitted haplotypes. two of the codon for N resulting in a non-synonymous P-values for haplotypes did not improve over the amino-acid exchange with S in amino-acid position values for single markers, which could be explained 251 of the protein. In addition, we identified two by the high degree of LD in the area, and yielded short tandem repeats (STRs) (Table 1) by sequencing. therefore no additional information on the possible These polymorphisms were supplemented with 24 location of a susceptibility allele (data not shown). We SNPs selected from the dbSNP or from the Celera SNP also constructed multilocus haplotypes composed of database, as well as with four STRs (Table 1), in order 12 SNPs (M39–M51) located between intron 6 and the to cover the genomic sequence of the gene as well as 30UTR (Figure 1) spanning 42.9 kb. Transmission of 110.8 kb of the 50 untranslated region (UTR) (M21) 65 alleles from heterozygous parents was counted and 114 kb of the 30UTR (M52) (Table 1). We for the haplotype with highest frequency (H1; Table 3) genotyped these polymorphisms in the sample of 65 as opposed to 37 non-transmitted alleles (P = 0.005 families, where the linkage signal was obtained.33 No uncorrected, P = 0.037 after Bonferroni correction for deviation from Hardy–Weinberg equilibrium was seven haplotypes). This haplotype is composed detected for these markers in the parents. P-values entirely of alleles with major allele frequency. were calculated from the differences between trans- We tested additional SNPs and microsatellite mitted and non-transmitted alleles using the program markers centromeric of the gene (Table 1, M1–M22), FAMHAP51,54 with the ‘zhao’ option. All 65 families covering the promoter region (the gene is on the were included for haplotype frequency estimation, reversed strand) and extending to regions where but for the subsequent testing method only fully according to the UCSC database additional genes are typed families, that is, genotypes of both parents and located. A higher marker density was chosen for siblings, were used (61 families) for calculation of MSRB2 (methionine sulphoxide reductase B2, also P-values. The null hypothesis for this test is no known as PILB), because of its possible role as a association in the presence of linkage.53 Fifteen DNA transcription factor. With the exception of a weak sequence variants located between exon 4 and the nominally significant P-value (0.015) for the micro- 30UTR of the gene and spanning 73.6 kb (Figure 1 and satellite marker D10S1660 (M15; Table 1), no further Table 1) produced nominally significant P-values. A evidence of association was observed for variants P-value of 0.001 was obtained for SNP M41 (N251S) within the genomic region of this gene. However, SNP in exon seven. The value remained nominally M20 (125 kb distal to exon 1 of MSRB2/PILB) significant after applying the spectral decomposition displayed a nominally significant value of P = 0.0009 method,56,57 which reduced the number of effective (Table 1). As it has been reported that LD can extend markers to 21.38. Using this number, the experiment- over considerable distances,61 we tested two and three wide significance threshold was estimated to be marker haplotypes composed of M20 and markers 0.00275. Alleles with major allele frequencies were M39, M41 (exonic variant) and M43 for association over-transmitted for all markers (Table 1). Table 1 also with schizophrenia. These SNPs are located in the shows the transmission ratio T/NT for markers with centre of the associated region and showed nominally P-values < 0.05. This value can be taken as an significant P-values as single markers (Figure 1 and estimator for the relative risk under a multiplicative Table 1). Significant P-values were produced for two- model.60 Accordingly, a relative risk of 1.92 was locus haplotypes composed of M20, M39, M41 and estimated for exonic polymorphism M41 (Table 1). M43 (Table 4). In contrast, haplotypes composed of

Molecular Psychiatry Association of PIP5K2A with schizophrenia SG Schwab et al 841 Table 1 DNA sequence variant IDs, characteristics and P-values

a c Variant ID Distance to Location Polymorphism Frequency T/NT PTDT next locus (ratio)d (kb)b Affected Unaffected

M1 rs1414041 GPR158 G/A 0.496 0.569 48/61 0.287 M2 rs1413394 279.1 GPR158 A/G 0.937 0.952 11/13 0.683 M3 rs1492608 282.5 Intergenic C/T 0.78 0.765 32/27 0.515 M4 rs1414457 263.1 Intergenic A/G 0.637 0.702 33/38 0.808 M5 rs1335840 29.2 ARHGAP21 A/G 0.621 0.463 23/19 0.537 M6 rs1411898 230.4 KIAA1217 T/G 0.957 0.917 10/8 0.637 M7 rs1331291 45.4 KIAA1217 T/G 0.885 0.884 12/10 0.827 M8 rs1335830 649.0 Intergenic C/T 0.837 0.815 13/11 0.669 M9 rs1538077 176.5 Intergenic C/T 0.801 0.743 53/39 0.136 M10 rs1325591 367.6 Intergenic C/T 0.726 0.672 34/36 0.808. M11 D10S563 114.3 Intergenic CA repeat — — — 0.960 M12 rs7427 7.7 MSRB2 C/T 0.614 0.745 28/44 0.059 M13 rs1857279 9.7 MSRB2 C/G 0.643 0.599 58/53 0.635 M14 rs1977372 1.3 MSRB2 G/A 0.689 0.713 48/52 0.689 M15 D10S1660 2.9 MSRB2 GT repeat — — — 0.015 M16 rs2296467 0.8 MSRB2 A/T 0.741 0.809 26/42 0.052 M17 rs3793756 7.3 MSRB2 A/C 0.736 0.819 27/46 0.026 M18 rs11817587 6.1 Intergenic G/T 0.608 0.578 48/40 0.394 M19 rs1060761 106.6 ARMC3 G/A 0.56 0.548 72/51 0.079 M20 rs1417374 144.6 Intergenic C/T 0.787 0.641 58/27 (2.15) 0.0009f M21 rs946961 33.6 Intergenic C/G 0.807 0.659 28/16 0.093 M22 rs1891878 77.2 Intergenic T/C 0.637 0.705 21/20 0.876 M23 rs1132816 14.3 PIP5K2A, exon 1 — 0.709 0.781 37/38 0.908 M24 PIK196_197e 30.4 PIP5K2A, intron 1 GT repeat — — — 0.024 M25 rs1343504 0.8 PIP5K2A, intron 1 A/G 0.759 0.698 59/43 0.069 M26 D10S553 11.2 PIP5K2A, intron 1 CA repeat — — — 0.057 M27 rs1171506 8.8 PIP5K2A, intron 1 G/A 0.714 0.677 49/31 0.045 M28 rs1171496 5.9 PIP5K2A, intron 1 C/T 0.563 0.61 47/40 0.227 M29 D10S1662 9.8 PIP5K2A, intron 1 CA repeat — — — 0.055 M30 rs1409395 3.6 PIP5K2A, intron 1 A/G 0.716 0.646 66/40 (1.65) 0.010 M31 rs943196 3.9 PIP5K2A, intron 1 C/T 0.74 0.669 39/33 0.332 M32 rs1624192 4.8 PIP5K2A, intron 1 C/G 0.824 0.881 17/19 0.739 M33 rs943194 13.8 PIP5K2A, intron 1 A/C 0.655 0.612 64/45 0.064 M34 rs2256542 14.8 PIP5K2A, intron 3 — 0.634 0.627 50/37 0.064 M35 rs1409396 15.9 PIP5K2A, intron 4 T/C 0.586 0.501 68/44 0.023 M36 rs11013052e 16.5 PIP5K2A, intron 4 G/T 0.774 0.622 72/40 (1.8) 0.005 M37 rs7916957e 0.4 PIP5K2A, intron 5 G/A 0.807 0.845 31/32 0.898 M38 rs2296624 5.1 PIP5K2A, intron 5 C/T 0.734 0.611 37/23 0.071 M39 rs7081744 9.0 PIP5K2A, intron 6 G/A 0.663 0.523 64/37 (1.73) 0.007 M40 rs943190 5.5 PIP5K2A, intron 6 A/G 0.679 0.543 56/32 (1.75) 0.003 M41 rs10828317e 2.7 PIP5K2A, exon 7 A/G 0.784 0.603 73/38 (1.92) 0.001f M42 rs10828316 1.2 PIP5K2A, intron 7 A/C 0.77 0.622 60/33 (1.82) 0.005 M43 rs10430590 5.3 PIP5K2A, intron 7 A/T 0.779 0.631 66/37 (1.78) 0.003 M44 rs746203 2.5 PIP5K2A, intron 8 A/G 0.675 0.577 72/38 (1.89) 0.001f M45 PIK53_54e 1.7 PIP5K2A, intron 9 CT repeat — — — 0.026 M46 rs1053454e 2.7 PIP5K2A, 3’UTR G/T 0.715 0.607 81/50 (1.62) 0.015 M47 rs3215826e 0.8 Intergenic À/C 0.719 0.607 79/46 (1.72) 0.005 M48 rs8341 0.1 Intergenic C/T 0.709 0.583 75/40 (1.87) 0.0009f M49 rs2364115 7.4 Intergenic G/C 0.724 0.633 57/30 (1.9) 0.001f M50 rs2148363 1.9 Intergenic A/G 0.741 0.617 61/35 (1.74) 0.005 M51 rs1536332 10.8 Intergenic G/C 0.802 0.736 52/34 0.037 M52 rs1342317 90.2 Intergenic A/T 0.973 0.972 6/7 0.782 M53 rs1536429 220.5 Intergenic C/G 0.671 0.696 32/44 0.238 M54 D10S211 586.4 MLLT10 CA repeat — — — 0.038 aNCBI dbSNP accession no. (rs-numbers), short tandem repeat (STR) loci, novel variants (PIK). bCalculated from the location given in Browser Gateway (UCSC) assembly May 2004. cMajor/minor allele; type of STR. dTransmitted/non-transmitted alleles corresponding to major/minor alleles; in parentheses is the ratio for nominally significant P-values < 0.01 as estimator for the genotype relative risk. eIdentified by SSCP and sequencing. fExperiment-wide significance threshold for type I error rate of 5% = 0.00275.

Molecular Psychiatry Association of PIP5K2A with schizophrenia SG Schwab et al 842

Figure 1 Genomic structure of PIP5K2A and position of DNA sequence variants. Location of variants is indicated by arrows. The length of the arrows corresponds to P-values. The order of the marker is the same as in Table 1. The gene maps at chromosome 10p12.32 on the reverse strand.

Table 2 Intermarker LD (D0 values) for SNPs in the genomic region of the gene for cases (transmitted alleles) and (pseudo-) controls (non-transmitted alleles)

Abbreviations: LD, linkage disequilibrium; SNP, single nucleotide polymorphism. Note: Above diagonal are D0 values for cases, below diagonal are D0 values for (pseudo-)controls. Grey cells indicate areas of high LD (D0 between 0.8 and 1.0); ID of DNA sequence variants (SNPs) as indicated in Table 1.

M21, which did not show association as single block (Table 4) that contains the schizophrenia- marker, did not produce significant P-values for associated variants. association with schizophrenia (Table 4). In addition, we calculated pairwise intermarker LD between M20 and M39, M41 and M43, as well as between Discussion 0 M21 and three SNPs. D values around 0.5 were Our study suggests potential involvement of PIP5K2A obtained between M20 and the three SNPs in the in the development of schizophrenia: PIP5K2A gene using the non-transmitted alleles as (pseudo-) controls. Conversely, no LD was observed 1. The gene is located in an area where linkage with between M21 and the three SNPs. This suggests that schizophrenia has been reported independently in SNP M20 is in long-range LD with the haplotype several family samples.29–36

Molecular Psychiatry Association of PIP5K2A with schizophrenia SG Schwab et al 843 Table 3 Transmission of 12-locus haplotypes (SNPs) to affected offspring

M39 M40 M41 M42 M43 M44 M46 M47 M48 M49 M50 M51 T a NT b Pc

H1 1 1 1 1 1 1 1 1 1 1 1 1 65 37 0.037 H2221111111111221.0 H3 2 2 2 1 1 1 2 1 1 1 1 2 3 2 0.65 H4 2 2 2 2 2 2 2 2 2 2 2 2 21 45 0.028 H5222222222211150.1 H6122111222221221.0 H7 1 2 2 1 1 1 2 2 2 2 2 1 5 6 0.76

Abbreviation: SNP, single nucleotide polymorphism. Note: DNA sequence variants (SNPs) are indicated in Table 1. The rows represent haplotypes (H1–H7) composed of alleles with major (1) or minor (2) allele frequency. aTransmitted from heterozygous parents. bNon-transmitted from heterozygous parents. cP-values corrected for number of haplotypes.

Table 4 LD and P-values for selected two marker haplotypes enzyme catalyses preferentially phosphorylation of 41 PI(5)P in the 4th position to PI(4,5)P2. PI(4,5)P2 0 0 Marker Distance D D PTDT* plays a key role in phosphoinositide signalling path- combination between (cases) (controls) ways. Upon activation by G-proteins, PI(4,5)P2 is markers (bp) cleaved by the action of phospholipase C into inositoltrisphosphate (IP3) and diacylglycerol. IP3 M20–M39 280 582 0.473 0.444 0.0147 activates release of Ca2 þ from intracellular sequester- M20–M41 288 847 0.492 0.395 0.0014 M20–M43 295 412 0.492 0.403 0.0037 ing compartments, whereas diacylglycerol activates M21–M39 246 935 0.251 0.285 > 0.05 kinase C, which is involved in intracellular regulatory M21–M41 255 200 0.029 0.034 > 0.05 processes via protein phosphorylation. A number of M21–M43 261 765 0.029 0.048 > 0.05 neurotransmitter receptors that have been attributed to schizophrenia are connected to these pathways, for Abbreviation: LD, linkage disequilibrium. example, the presynaptic dopamine D3 receptor, the *P-values for transmitted versus non-transmitted haplo- muscarinic acetylcholine receptor, the serotonergic types (uncorrected). receptor 5HT2. DNA sequence variants within the coding or regulatory sequences of the PIP5K2A gene may have impact on enzyme activity and interfere 2. Association/LD has been detected for single SNPs with phosphoinositide pathways, possibly inducing and haplotypes located within the gene as de- an enhanced risk for development of schizophrenia. scribed in the present study. There are two types of PIPK involved in phospho- 3. The gene encodes an enzyme that participates in inositide signalling, type I kinases phosphorylate phosphoinositide pathways. These pathways play PI(4)P in the 5th position, whereas PIPKs from type a role in transmembrane signal transduction of II phosphorylate PI(5)P in the 4th position. These neurotransmitter receptors, in intracellular Ca2 þ kinases have been shown to be differentially loca- signalling as well as in a number of other signalling lized, and may participate in the synthesis of PI(4,5)P2 pathways.39 at distinct sites.39 PIP5K2A belongs to the type II 4. Interference with phosphoinositide pathways has kinases that have been found in cytosol, nucleus and been proposed for the three drugs used as mood associated with the endoplasmic reticulum.39 The stabilizers in psychiatric disorders, lithium, val- physiological role of type II kinases is not well proic acid and carbamazepine.62 defined yet, but they may be involved in tumour necrosis factor (TNF)a-mediated signalling.66 Interest- Additional evidence comes from an independent ingly, there are some reports on association of SNPs in study following our preliminary reports communicated the TNFa gene with schizophrenia,67,68 suggesting at the IXth and XIth World Congress for Psychiatric that impairment of TNFa-mediated signalling may Genetics.63,64 Bakker and co-workers65 obtained nomin- contribute to development of schizophrenia. More- ally significant association (P = 0.0004) with the exonic over, there are also reports on a possible involve- SNP rs10828317 (M41, N251S) in a sample of 273 ment of PIP5K2A in the regulation of gene expression, Dutch schizophrenia patients and 580 controls. In both pre-mRNA processing and mRNA transport in studies, the same allele (A) was found to be associated general.40,69,70 with schizophrenia. Another function of type II kinases, that is, regula-

PIP5K2A was originally reported to phosphorylate tion of PI(5)P levels for synthesis of PI(3,4,5)P3, was 59 71 PI(4)P to PI(4,5)P2. Later, it has been shown that the recently discovered. PI(3,4,5)P3 is involved in the

Molecular Psychiatry Association of PIP5K2A with schizophrenia SG Schwab et al 844 activation of AKT (protein kinase B). There is participating in development of schizophrenia. The additional evidence for involvement of AKT- here reported association of DNA sequence variants mediated signalling pathways in development of in the PIP5K2A gene has been replicated in an schizophrenia. Impaired phosphorylation and expres- additional, independent sample. However, before a sion of AKT72 as well as association of SNPs in the role in ethiology of schizophrenia can be postulated, AKT gene with schizophrenia have been reported.72,73 impact of the N251S amino-acid exchange variant on Additional support for a role of AKT pathways in PIP5K2A activity has to be shown in future functional schizophrenia comes from the use of lithium, valproic studies. acid and carbamazepine as mood stabilizers in major psychoses. These drugs have been found to interfere Acknowledgments with AKT-mediated pathways.62 The tertiary structure of PIP5K2B located on We are extremely thankful to Dr SC Bakker (Depart- chromosome 17q12, an isomer to PIP5K2A with ment of Psychiatry and Department of Biomedical approximately 78% homology, has been published.74 Genetics, University Medical Center, Utrecht, The There are highly conserved regions shared by the two Netherlands) for communicating data before publica- isozymes, that is, the activation loop, the ATP binding tion. We are very grateful to the patients and their sites and several sites forming a-helices. The se- family members without whose support and interest quence of a-helix 8 reads SKKNFLEKLKRDVEFLA for this study would not have been possible. We grate- PIPK2B and NKKVFLEKLKRDVEFLA for PIP5K2A.74 fully acknowledge support from the German–Israeli The coding variant SNP M41 leads to a non-synon- Foundation (to BL and DBW), Deutsche Forschungs- ymous exchange of N to S in the first position of the gemeinschaft SFB 400 (to SGS, MK, WM and DBW) highly conserved a-helix 8 of PIP5K2A. Helix 8 does and DFG Grant Kn378/1-3 (to MK). DBW is recipient not seem to be directly involved in function, but of a NARSAD Independent Investigator Award. SGS the substitution of N with S, a possible target for would like to thank Professor Eric Lander for phosphorylation, may influence the structure of the hospitality during the initial phase of this work. protein with consequences for function. The high frequency of the associated major allele (0.603 in non- References affected, 0.784 in affected; Table 1) as well as the relatively low attributable risk of 1.92 as estimated 1GottesmanI.Schizophrenia Genesis. WH Freeman: New York, 1991. from the ratio transmitted to non-transmitted60 would 2 Risch NJ. Searching for genetic determinants in the new be compatible with the common disease/common millennium. Nature 2000; 405: 847–856. 75,76 3 Riley BP, McGuffin P. Linkage and associated studies of allele hypothesis. However, the variant could also schizophrenia. Am J Med Genet 2000; 97: 23–44. reflect LD with another, not yet identified risk 4 Schizophrenia Linkage Collaborative Group for 3a. conferring allele, rather than being a disease-causing Additional support for schizophrenia linkage on chromosomes 6 susceptibility allele by itself. Studying the impact of and 8: a multicenter study. Schizophrenia Linkage Collaborative the amino-acid exchange on enzyme activity, that is, Group for Chromosomes 3, 6 and 8. Am J Med Genet 1996; 67: 580–594. enzymatic phosphorylation of PIP and correlation 5 Kendler KS, MacLean CJ, O’Neill FA, Burke J, Murphy B, Duke F with the genotype, may help to clarify this question. et al. Evidence for a schizophrenia vulnerability locus on In addition to the Dutch study mentioned above,65 chromosome 8p in the Irish Study of High-Density Schizophrenia this polymorphism was tested in another case–control Families. Am J Psychiat 1996; 153: 1534–1540. 6 Blouin JL, Dombroski BA, Nath SK, Lasseter VK, Wolyniec PS, sample comprising 176 patients with schizophrenia Nestadt G et al. Schizophrenia susceptibility loci on chromosomes and 178 controls from North America with European 13q32 and 8p21. Nat Genet 1998; 20: 70–73. background and from Israel, as well as in a sample 7 Stefansson H, Sigurdsson E, Steinthorsdottir V, Bjornsdottir S, comprising 198 bipolar patients and 212 controls Sigmundsson T, Ghosh S et al. Neuregulin 1 and susceptibility to from USA and the Czech Republic,77 following up our schizophrenia. Am J Hum Genet 2002; 71: 877–892. 8 Stefansson H, Sarginson J, Kong A, Yates P, Steinthorsdottir V, preliminary report on association of the N251S Gudfinnsson E et al. Association of neuregulin 1 with schizo- polymorphism in schizophrenia at the IXth World phrenia confirmed in a Scottish population. Am J Hum Genet Congress for Psychiatric Genetics 2001.64 Association 2003; 72: 83–87. was not detected with this particular variant, but a 9 Schwab SG, Albus M, Hallmayer J, Honig S, Borrmann M, Lichtermann D et al. Evaluation of a susceptibility gene for modest difference was found in the distribution of schizophrenia on chromosome 6p by multipoint affected sib-pair alleles from the CT repeat pik53 þ 54 (M45), which linkage analysis. Nat Genet 1995; 11: 325–327. was also associated in our sample (Table 1, P = 0.026). 10 Straub RE, MacLean CJ, O’Neill FA, Burke J, Murphy B, Duke F In conclusion, we have obtained evidence for et al. A potential vulnerability locus for schizophrenia on association of schizophrenia with DNA sequence chromosome 6p24–22: evidence for genetic heterogeneity. Nat Genet 1995; 11: 287–293. variants in the gene for the enzyme PIP5K2A, which 11 Turecki G, Rouleau GA, Joober R, Mari J, Morgan K. Schizophrenia is located in an area on chromosome 10p where seve- and chromosome 6p. Am J Med Genet 1997; 74: 195–198. ral groups earlier reported linkage. We have identified 12 Wang S, Sun CE, Walczak CA, Ziegle JS, Kipps BR, Goldin LR et al. a non-synonymous amino-acid exchange variant Evidence for a susceptibility locus for schizophrenia on chromosome 6pter–p22. Nat Genet 1995; 10: 41–46. eventually interfering with functioning of phospho- 13 Moises HW, Yang L, Kristbjarnarson H, Wiese C, Byerley W, inositide signalling pathways. Malfunctioning of Macciardi F et al. An international two-stage genome-wide search these pathways might be a plausible mechanism for schizophrenia susceptibility genes. Nat Genet 1995; 11: 321–324.

Molecular Psychiatry Association of PIP5K2A with schizophrenia SG Schwab et al 845 14 Nurnberger Jr JI, Foroud T. Chromosome 6 workshop report. Am J pedigrees: chromosomes 1, 6, 8, 10, and 12. Am J Med Genet Med Genet 1999; 88: 233–238. 1997; 74: 247–253. 15 Hwu HG, Lin MW, Lee PC, Lee SF, Ou-Yang WC, Liu CM. 33 Schwab SG, Hallmayer J, Albus M, Lerer B, Hanses C, Kanyas K Evaluation of linkage of markers on chromosome 6p with et al. Further evidence for a susceptibility locus on chromosome schizophrenia in Taiwanese families. Am J Med Genet 2000; 96: 10p14–p11 in 72 families with schizophrenia by nonparametric 74–78. linkage analysis. Am J Med Genet 1998; 81: 302–307. 16 Straub RE, Jiang Y, MacLean CJ, Ma Y, Webb BT, Myakishev MV 34 Straub RE, MacLean CJ, Martin RB, Ma Y, Myakishev MV, Harris- et al. Genetic variation in the 6p22.3 gene DTNBP1, the human Kerr C et al. A schizophrenia locus may be located in region ortholog of the mouse dysbindin gene, is associated with 10p15–p11. Am J Med Genet 1998; 81: 296–301. schizophrenia. Am J Hum Genet 2002; 71: 337–348. 35 DeLisi LE, Shaw SH, Crow TJ, Shields G, Smith AB, Larach VW 17 Van Den Bogaert A, Schumacher J, Schulze TG, Otte AC, Ohlraun et al. A genome-wide scan for linkage to chromosomal regions in S, Kovalenko S et al. The DTNBP1 (dysbindin) gene contributes to 382 sibling pairs with schizophrenia or schizoaffective disorder. schizophrenia, depending on family history of the disease. Am J Am J Psychiat 2002; 159: 803–812. Hum Genet 2003; 73: 1438–1443. 36 Holliday E, Mowry B, Chant D, Nyholt D. The importance of 18 Schwab SG, Knapp M, Mondabon S, Hallmayer J, Borrmann- modelling heterogeneity in complex disease: application to NIMH Hassenbach M, Albus M et al. Support for association of Schizophrenia Genetics Initiative data. Hum Genet 2005; 117: schizophrenia with genetic variation in the 6p22.3 gene, dysbin- 160–167. din, in sib-pair families with linkage and in an additional sample 37 Schwab SG, Hallmayer J, Albus M, Lerer B, Eckstein GN, of triad families. Am J Hum Genet 2003; 72: 185–190. Borrmann M et al. A genome-wide autosomal screen for 19 Kirov G, Ivanov D, Williams NM, Preece A, Nikolov I, Milev R schizophrenia susceptibility loci in 71 families with affected et al. Strong evidence for association between the dystrobrevin siblings: support for loci on chromosome 10p and 6. Mol Psychiat binding protein 1 gene (DTNBP1) and schizophrenia in 2000; 5: 638–649. 488 parent–offspring trios from Bulgaria. Biol Psychiat 2004; 55: 38 Risch N, Spiker D, Lotspeich L, Nouri N, Hinds D, Hallmayer J 971–975. et al. A genomic screen of autism: evidence for a multilocus 20 Funke B, Finn CT, Plocik AM, Lake S, DeRosse P, Kane JM et al. etiology. Am J Hum Genet 1999; 65: 493–507. Association of the DTNBP1 locus with schizophrenia in a U.S. 39 Doughman RL, Firestone AJ, Anderson RA. Phosphatidylinositol population. Am J Hum Genet 2004; 75: 891–898. phosphate kinases put PI4,5P(2) in its place. J Membr Biol 2003; 21 Chumakov I, Blumenfeld M, Guerassimenko O, Cavarec L, Palicio 194: 77–89. M, Abderrahim H et al. Genetic and physiological data impli- 40 Anderson RA, Boronenkov IV, Doughman SD, Kunz J, Loijens JC. cating the new human gene G72 and the gene for D-amino acid Phosphatidylinositol phosphate kinases, a multifaceted family of oxidase in schizophrenia. Proc Natl Acad Sci USA 2002; 99: signaling . J Biol Chem 1999; 274: 9907–9910. 13675–13680. 41 Rameh LE, Tolias KF, Duckworth BC, Cantley LC. A new pathway 22 Chowdari KV, Mirnics K, Semwal P, Wood J, Lawrence E, Bhatia T for synthesis of phosphatidylinositol-4,5-bisphosphate. Nature et al. Association and linkage analyses of RGS4 polymorphisms in 1997; 390: 192–196. schizophrenia. Hum Mol Genet 2002; 11: 1373–1380. 42 Fyer A, Endicott J, Mannuzza S, Klein D. Schedule for Affective 23 Brzustowicz LM, Simone J, Mohseni P, Hayter JE, Hodgkinson KA, Disorders and Schizophrenia-Lifetime Version (SADS-LA).New Chow EW et al. Linkage disequilibrium mapping of schizophrenia York State Psychiatric Institute: New York, 1985. susceptibility to the CAPON region of chromosome 1q22. Am J 43 McGuffin P, Farmer A, Harvey I. A polydiagnostic application of Hum Genet 2004; 74: 1057–1063. operational criteria in studies of psychotic illness. Development 24 Millar JK, Wilson-Annan JC, Anderson S, Christie S, Taylor MS, and reliability of the OPCRIT system. Arch Gen Psychiat 1991; 48: Semple CA et al. Disruption of two novel genes by a transloca- 764–770. tion co-segregating with schizophrenia. Hum Mol Genet 2000; 9: 44 Spitzer R, Endicott J, Robins E. Diagnostic Criteria for a Selected 1415–1423. Group of Functional Disorders. New York State Psychiatric 25 Hennah W, Varilo T, Kestila M, Paunio T, Arajarvi R, Haukka J Institute: New York, 1978. et al. Haplotype transmission analysis provides evidence of 45 Orita M, Iwahana H, Kanazawa H, Hayashi K, Sekiya T. Detection association for DISC1 to schizophrenia and suggests sex- of polymorphisms of human DNA by gel electrophoresis as single- dependent effects. Hum Mol Genet 2003; 12: 3151–3159. strand conformation polymorphisms. Proc Natl Acad Sci USA 26 Duan J, Martinez M, Sanders AR, Hou C, Saitou N, Kitano T et al. 1989; 86: 2766–2770. Polymorphisms in the trace amine receptor 4 (TRAR4) gene 46 Chen X, Levine L, Kwok PY. Fluorescence polarization in on chromosome 6q23.2 are associated with susceptibility to homogeneous nucleic acid analysis. Genome Res 1999; 9: schizophrenia. Am J Hum Genet 2004; 75: 624–638. 492–498. 27 Pimm J, McQuillin A, Thirumalai S, Lawrence J, Quested D, Bass 47 Hamilton SP, Slager SL, Helleby L, Heiman GA, Klein DF, Hodge N et al. The Epsin 4 gene on chromosome 5q, which encodes SE et al. No association or linkage between polymorphisms in the the clathrin-associated protein enthoprotin, is involved in the genes encoding cholecystokinin and the cholecystokinin B genetic susceptibility to schizophrenia. Am J Hum Genet 2005; 76: receptor and panic disorder. Mol Psychiat 2001; 6: 59–65. 902–907. 48 Sklar P, Gabriel SB, McInnis MG, Bennett P, Lim YM, Tsan G et al. 28 Harrison PJ, Weinberger DR. Schizophrenia genes, gene expres- Family-based association study of 76 candidate genes in bipolar sion, and neuropathology: on the matter of their convergence. disorder: BDNF is a potential risk locus. Brain-derived neutrophic Mol Psychiat 2005; 10:804. factor. Mol Psychiat 2002; 7: 579–593. 29 Faraone SV, Matise T, Svrakic D, Pepple J, Malaspina D, Suarez B 49 Buetow KH, Edmonson M, MacDonald R, Clifford R, Yip P, Kelley J et al. Genome scan of European-American schizophrenia pedi- et al. High-throughput development and characterization of grees: results of the NIMH Genetics Initiative and Millennium a genomewide collection of gene-based single nucleotide poly- Consortium. Am J Med Genet 1998; 81: 290–295. morphism markers by chip-based matrix-assisted laser desorption/ 30 Foroud T, Castelluccio PF, Koller DL, Edenberg HJ, Miller M, ionization time-of-flight mass spectrometry. Proc Natl Acad Sci Bowman E et al. Suggestive evidence of a locus on chromosome USA 2001; 98: 581–584. 10p using the NIMH genetics initiative bipolar affective disorder 50 Kokoris M, Dix K, Moynihan K, Mathis J, Erwin B, Grass P et al. pedigrees. Am J Med Genet 2000; 96: 18–23. High-throughput SNP genotyping with the Masscode system. Mol 31 Levinson DF, Holmans P, Straub RE, Owen MJ, Wildenauer DB, Diagn 2000; 5: 329–340. Gejman PV et al. Multicenter linkage study of schizophrenia 51 Becker T, Knapp M. Maximum-likelihood estimation of haplo- candidate regions on chromosomes 5q, 6q, 10p, and 13q: type frequencies in nuclear families. Genet Epidemiol 2004; 27: schizophrenia linkage collaborative group III. Am J Hum Genet 21–32. 2000; 67: 652–663. 52 Zhao H, Zhang S, Merikangas KR, Trixler M, Wildenauer DB, 32 Rice JP, Goate A, Williams JT, Bierut L, Dorr D, Wu W et al. Sun F et al. Transmission/disequilibrium tests using multiple Initial genome scan of the NIMH genetics initiative bipolar tightly linked markers. Am J Hum Genet 2000; 67: 936–946.

Molecular Psychiatry Association of PIP5K2A with schizophrenia SG Schwab et al 846 53 Knapp M, Becker T. Family-based association analysis with tightly tumor necrosis factor receptor and phosphatidylinositol-4- linked markers. Hum Hered 2003; 56: 2–9. phosphate 5-kinase. J Biol Chem 1997; 272: 5861–5870. 54 Becker T, Knapp M. A powerful strategy to account for multiple 67 Schwab SG, Mondabon S, Knapp M, Albus M, Hallmayer J, testing in the context of haplotype analysis. Am J Hum Genet 2004; Borrmann-Hassenbach M et al. Association of tumor necrosis 75: 561–570. factor alpha gene ÀG308A polymorphism with schizophrenia. 55 Hedrick PW. Gametic disequilibrium measures: proceed with Schizophr Res 2003; 65: 19–25. caution. Genetics 1987; 117: 331–341. 68 Boin F, Zanardini R, Pioli R, Altamura CA, Maes M, Gennarelli M. 56 Li J, Ji L. Adjusting multiple testing in multilocus analyses Association between ÀG308A tumor necrosis factor alpha gene using the eigenvalues of a correlation matrix. Heredity 2005; 95: polymorphism and schizophrenia. Mol Psychiat 2001; 6: 79–82. 221–227. 69 York JD, Odom AR, Murphy R, Ives EB, Wente SR. A phospho- 57 Nyholt DR. A simple correction for multiple testing for single- lipase C-dependent inositol polyphosphate kinase pathway nucleotide polymorphisms in linkage disequilibrium with each required for efficient messenger RNA export. Science 1999; 285: other. Am J Hum Genet 2004; 74: 765–769. 96–100. 58 Kim UJ, Birren BW, Slepak T, Mancino V, Boysen C, Kang HL et al. 70 Boronenkov IV, Loijens JC, Umeda M, Anderson RA. Phosphoi- Construction and characterization of a human bacterial artificial nositide signaling pathways in nuclei are associated with nuclear chromosome library. Genomics 1996; 34: 213–218. speckles containing pre-mRNA processing factors. Mol Biol Cell 59 Boronenkov IV, Anderson RA. The sequence of phosphatidylino- 1998; 9: 3547–3560. sitol-4-phosphate 5-kinase defines a novel family of lipid kinases. 71 Carricaburu V, Lamia KA, Lo E, Favereaux L, Payrastre B, Cantley J Biol Chem 1995; 270: 2881–2884. LC et al. The phosphatidylinositol (PI)-5-phosphate 4-kinase 60 Altshuler D, Hirschhorn JN, Klannemark M, Lindgren CM, Vohl type II enzyme controls insulin signaling by regulating PI-3,4,5- MC, Nemesh J et al. The common PPARgamma Pro12Ala trisphosphate degradation. Proc Natl Acad Sci USA 2003; 100: polymorphism is associated with decreased risk of type 2 diabetes. 9867–9872. Nat Genet 2000; 26: 76–80. 72 Emamian ES, Hall D, Birnbaum MJ, Karayiorgou M, Gogos JA. 61 Hinds DA, Stuve LL, Nilsen GB, Halperin E, Eskin E, Ballinger DG Convergent evidence for impaired AKT1-GSK3beta signaling in et al. Whole-genome patterns of common DNA variation in three schizophrenia. Nat Genet 2004; 36: 131–137. human populations. Science 2005; 307: 1072–1079. 73 Schwab SG, Hoefgen B, Hanses C, Hassenbach MB, Albus M, Lerer 62 Harwood AJ. Lithium and bipolar mood disorder: the inositol- B et al. Further evidence for association of variants in the AKT1 depletion hypothesis revisited. Mol Psychiat 2005; 10: 117–126. gene with schizophrenia in a sample of European sib-pair families. 63 Sewekow CA, Schwab SG, Knapp M, Hallmayer J, Eckstein GN, Biol Psychiat 2005; 58: 446–450. Gabel S et al. Association of SNPs with schizophrenia on 74 Rao VD, Misra S, Boronenkov IV, Anderson RA, Hurley JH. chromosome 10p, a region with previously detected linkage. Structure of type IIbeta phosphatidylinositol phosphate kinase: a Am J Med Genet 2003; 122B: 244. protein kinase fold flattened for interfacial phosphorylation. 64 Schwab SG, Sklar P, Hallmayer J, Albus M, Rietschel M, Lerer B et Cell 1998; 94: 829–839. al. Association of SNPs with schizophrenia on chromosome 10p, a 75 Cargill M, Daley GQ. Mining for SNPs: putting the common region with previously detected linkage. Am J Med Genet 2001; variants – common disease hypothesis to the test. Pharmaco- 105: 562. genomics 2000; 1: 27–37. 65 Bakker SC, Hoogendoorn CLC, Hendriks J, Verzijlbergen K, Caron 76 Lander ES. The new genomics: global views of biology. Science S, Verduijn W et al. The PIP5K2A and RGS4 genes are differe- 1996; 274: 536–539. ntially associated with deficit and non-deficit schizophrenia. 77 Stopkova P, Saito T, Fann CS, Papolos DF, Vevera J, Paclt I et al. Genes Brain Behav, 13 April 2006 [E-pub ahead of print]. Polymorphism screening of PIP5K2A: a candidate gene for 66 Castellino AM, Parker GJ, Boronenkov IV, Anderson RA, Chao MV. chromosome 10p-linked psychiatric disorders. Am J Med Genet A novel interaction between the juxtamembrane region of the p55 B Neuropsychiatr Genet 2003; 123: 50–58.

Supplementary Information accompanies the paper on the Molecular Psychiatry website (http:// www.nature.com/mp)

Molecular Psychiatry