Serine Racemase Binds to PICK1: Potential Relevance to Schizophrenia

Home , Serine racemase

Molecular Psychiatry (2006) 11, 150–157 & 2006 Nature Publishing Group All rights reserved 1359-4184/06 $30.00 www.nature.com/mp ORIGINAL ARTICLE Serine racemase binds to PICK1: potential relevance to schizophrenia K Fujii1,2,8, K Maeda1,3,8, T Hikida4,9, AK Mustafa4,9, R Balkissoon1,9, J Xia4, T Yamada3, Y Ozeki1,2, R Kawahara3, M Okawa2, RL Huganir2,5, H Ujike6, SH Snyder1,4,5,7 and A Sawa1,4,5 1Department of Psychiatry, Johns Hopkins University School of Medicine, Baltimore, MD, USA; 2Department of Psychiatry, Shiga University of Medical Science, Shiga, Japan; 3Division of Neuropsychiatry, Faculty of Medicine, Tottori University, Yonago, Japan; 4Department of Neuroscience, Johns Hopkins University School of Medicine, Baltimore, MD, USA; 5Program in Cellular Molecular Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, USA; 6Department of Neuropsychiatry, Okayama University Graduate School of Medicine and Dentistry, Okayama, Japan and 7Department of Pharmacology and Molecular Science, Johns Hopkins University School of Medicine, Baltimore, MD, USA

Accumulating evidence from both genetic and clinico-pharmacological studies suggests that D-serine, an endogenous coagonist to the NMDA subtype glutamate receptor, may be implicated in schizophrenia (SZ). Although an association of genes for D-serine degradation, such as D-amino acid oxidase and G72, has been reported, a role for D-serine in SZ has been unclear. In this study, we identify and characterize protein interacting with C-kinase (PICK1) as a protein interactor of the D-serine synthesizing enzyme, serine racemase (SR). The binding of endogenous PICK1 and SR requires the PDZ domain of PICK1. The gene coding for PICK1 is located at chromosome 22q13, a region frequently linked to SZ. In a case–control association study using well-characterized Japanese subjects, we observe an association of the PICK1 gene with SZ, which is more prominent in disorganized SZ. Our findings implicating PICK1 as a susceptibility gene for SZ are consistent with a role for D-serine in the disease. Molecular Psychiatry (2006) 11, 150–157. doi:10.1038/sj.mp.4001776; published online 29 November 2005

Keywords: schizophrenia; PICK1; serine racemase; D-serine; yeast two-hybrid; case–control study

Introduction Genetic association studies also implicate D-serine in SZ. In several independent studies, genes coding Glutamate neurotransmission has been implicated in for the D-serine degrading enzyme, D-amino acid schizophrenia (SZ).1–5 Activation of NMDA receptors oxidase (DAO), and its potential regulator G72, are requires stimulation of a glycine site in addition to the associated with SZ.21–27 6 glutamate recognition site. Recent studies establish We have explored a role for D-serine synthesis in D-serine as an endogenous ligand for this site in many SZ. As the gene for SR itself is located to chromosome areas of the brain, while glycine itself may be the 17, which has been neither linked to nor associated endogenous ligand in certain regions.7–13 D-serine is with SZ,28 we focused on binding partners for SR with generated from L-serine by the enzyme serine race- some relationship to SZ. Here, we report binding of mase (SR) that is localized to a population of glial SR to protein interacting with C-kinase (PICK1), astrocytes, which ensheathe synapses in brain regions whose gene is located at chromosome 22q13.1. The enriched in NMDA receptors.14,15 Glutamate released chromosome 22q11–13 region displays linkage to from neurons activates glutamate receptors of the SZ,29–34 and a cluster of genes in the region are AMPA subtype on these astrocytes triggering the associated with SZ, including catechol-O-methyl- release of D-serine.7 Treatment of SZ with D-serine, transferase,35,36 ZDHHC8,37 and proline dehydrogen- D-cycloserine, or glycine together with classic neuro- ase.38–40 We also demonstrate an association of the leptics ameliorates some symptoms.16–20 PICK1 gene with disorganized SZ.

Correspondence: Dr A Sawa, Department of Psychiatry, Johns Materials and methods Hopkins University School of Medicine, 600 North Wolfe St, Baltimore, MD 21287, USA. Yeast two-hybrid screening E-mail: [email protected] Yeast two-hybrid screening was performed using a 8These authors contributed equally to the work. 9These authors contributed equally to the work. human adult whole brain cDNA library subcloned Received 13 April 2005; revised 7 September 2005; accepted 4 into the pPC86 vector, which contains the GAL4 October 2005; published online 29 November 2005 transactivation domain.41 The bait was the full length PICK1 and schizophrenia K Fujii et al 151 of human SR subcloned into pDBlue vector, which follows: a polyclonal rabbit anti-HA antibody (Medi- contains the GAL4 DNA binding domain. Over one cal and Biological Laboratories) and a monoclonal million independent clones were screened, and mouse anti-myc antibody (Calbiochem), which are positively interacting proteins were identified by commercially available as specific antibodies. Sec- selecting for Leu þ , Trip þ , His þ growth phenotype. ondary antibodies used for fluorescence were follows: Positive clones were further evaluated for b-galacto- a Rhodamine Red-X conjugated anti-rabbit IgG anti- sidase expression. All of the constructs were from body and a cyanine Cy2 conjugated anti-mouse PCR products subcloned in frame into pPC86 or antibody (Jackson ImmunoResearch). pDBlue vectors and were confirmed by sequencing. PICK1 K27A/D28A mutant was generated as des- Subjects in association study cribed previously.42 The subjects in the present case–control association study were 200 unrelated Japanese SZ patients (108 Cell culture and transfection men and 92 women; mean age 46.379.9 years) Cultured astrocytes were prepared from the cerebral fulfilling the international statistical classification of cortex of 2- to 6-day-old postnatal mice as described.43 the disease, revision 10 (ICD-10) diagnostic criteria for Cultured astrocytes and HEK293 cells were main- SZ, and 200 age-matched Japanese normal controls tained in DMEM media supplemented with 2 mML- (91 men and 109 women; mean age 43.3710.9 years) glutamine and 10% fetal bovine serum. Transfection who had no positive personal or familial history of of hemagglutinin (HA)-tagged SR and/or myc-tagged major psychiatric disorders. Patients in this study PICK1 was carried out by calcium phosphate method. retained their diagnosis by the ICD-10 for at least 3 years indicating a high reliability of their initial In vitro binding assay diagnosis. Trained psychiatrists who did not partici- PICK1, PICK1 mutants or GAPDH was fused to pate in the genetic association study diagnosed glutathione-S-transferase (GST) by subcloning into a patients. Population stratification was discussed in pGEX vector (Amersham Biosciences). Extracts of the previous studies using the same sample set.44,45 Escherichia coli strain BL21 were used as a source of For further analysis, the SZ patients were subdivided GST fusion proteins. Lysates of HEK293 cells tran- into 90 patients of disorganized type and 110 of siently transfected with a HA-tagged SR cDNA and nondisorganized type according to the criteria set out purified GST-fusion protein were incubated in 50 mM in the Diagnostic and Statistical Manual of Mental Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM phenylmethane Disorders, Fourth Edition (DSM-IV). Disorganized sulfonylfluoride (PMSF), and 1 mM 1,4-dithiothreitol, type (DSM-IV: 295.10) were previously diagnosed as and a protease inhibitor mixture (Roche) for 45 min. hebephrenic type (ICD-10: F20.1), and nondisorga- HA-tagged SR bound to GST-fusion protein was nized type as paranoid (F20.0), catatonic (F20.2), or precipitated with glutathione sepharose beads (Amer- undifferentiated type (F20.3), respectively.44,45 Asses- sham Biosciences). The protein precipitates were sment for diagnosis and subtype of SZ were per- analysed with SDS-PAGE, followed by Western blot- formed by trained psychiatrists who did not partici- ting with a polyclonal antibody against HA (Medical pate in the genetic analyses, on the basis of all and Biological Laboratories). available information including hospital notes. After being provided with a complete description of the Co-immunoprecipitation study, written informed consent to participate was Cultures astrocytes were homogenized in buffer obtained from all participants prior to examination. containing 50 mM Tris-HCl pH 7.8, 10% glycerol, The genetic association study was approved by the 1mM PMSF and a protease inhibitor cocktail (Sigma). Ethics Committee in Okayama and Tottori Universi- Lysates were centrifuged at 10 000 g, and the super- ties in Japan. natant containing 500 mg of proteins was precleared with mouse immunoglobulin with Pansorbin (EMD Sequencing and genotyping Biosciences). Cleared lysates were immunoprecipi- Genomic DNA was extracted from peripheral leuko- tated with an anti-SR monoclonal antibody (BD cytes using a standard phenol/CHCL3 procedures.44,45 Biosciences) overnight at 41C. The immunoprecipi- First, five SZ and five control subjects were randomly tates and input were analyzed in SDS-PAGE followed selected for direct sequencing of all known coding by immunoblotting with anti-SR and anti-PICK1 exons of PICK1 as well as 50 bp of intronic sequences antibodies (Santa Cruz biotechnology). Both the flanking each exon. By sequencing, we identified a anti-SR and anti-PICK1 antibodies recognize the total of four intronic single-nucleotide polymorph- single band at the expected size of the proteins in isms (SNPs) including two previously reported and the Western blotting (Figure 2b). two novel polymorphisms. Of these, we chose to genotype the two previously reported common SNPs Immunocytochemistry (rs713729 and rs2076369) likely to be informative for Immunofluorescent cell staining was carried out genetic association testing. These SNPs were selected as described.41 In brief, cells were fixed with 3.7% for genotyping because they crossed our threshold of paraformaldehyde in PBS, and permealized with 15% for the minor allele frequency in the Japanese 0.1% Triton X-100. Primary antibodies used were population. The two novel SNPs we identified were

Molecular Psychiatry PICK1 and schizophrenia K Fujii et al 152 rare and unlikely to be informative for genetic a Bait Prey positive 6 association (sequence data not shown). The third SR PICK1 5 / 10 6 SNP we genotyped (rs3952) is an intronic SNP that SR GRIP 1 / 10 was previously reported by Hong et al.46 Specific marker locations were determined using the May b 1 PDZ domain CC domain 416 2004 Freeze of the UCSC Genome Browser (http:// PICK1 KD SR genome.ucsc.edu/). All samples were genotyped by PICK1 + + using the Taqman assay and the ABI 7900HT PICK1(1-386) PICK1 ∆(139-166) - sequence detection system (Applied Biosystems). AA All probes and primers were designed by the Assay- PICK1 K27A/D28A - by-Designt service of Applied Biosystems. After we genotyped rs713729 and rs2076369, we verified the c 1 341 SR PICK1 accuracy of our genotyping method by comparing the SVSV Taqman genotypes to the direct sequencing results we SR + SVSV Y - obtained previously. We found that there was 100% SR(+Y) agreement for both markers between their Taqman Figure 1 Protein interaction of serine racemase (SR) with genotypes and our sequencing data, indicating a high PICK1 in the yeast two-hybrid system. (a) PICK1 and GRIP degree of accuracy of the Taqman genotyping method. were obtained as possible interactors to SR from human The high degree of agreement between direct sequen- brain cDNA library. (b) SR interacts PICK1 via the PDZ cing and Taqman genotyping was also demon- domain and the coiled-coil domain. The results from b- strated.47 It is consistently reported that the Taqman galactosidase assays are indicated. (c) The C-terminal amino acids in SR are crucial for the interaction with PICK1. assay produces greater than 99.5% concordance with independent methods.

Statistics to SR, establishing the importance of the PDZ domain The statistical significance of differences in the for the binding interaction. Deletion of the coiled-coil genotype distribution and allele frequency between domain of PICK1 also abolishes binding to SR. patients and controls was assessed by a w2 test or To examine the possibility that PICK1 binds to the Fisher’s exact test at a significance level of 0.05. C-terminus of SR, we generated a mutant of SR with Statview software (SAS Institute Inc.) was used for tyrosine added to the C-terminus to obscure the C- these analyses. Marker-to-marker linkage disequili- terminal valine. This modification abolishes binding brium (LD), haplotype block structure, and associa- of SR and PICK1, establishing that the C-terminus of tion tests for single SNPs and haplotypes of the PICK1 SR is critical for its binding to PICK1. gene were analyzed by Haploview, version 3.2.48 To confirm the interactions of SR and PICK1 Single markers and haplotypes were permuted 1000 biochemically, we transfected HEK293 cells with times to correct for bias due to multiple testing. HA-tagged SR (HA-SR) and demonstrate its binding Haplotype association results were confirmed using in cell lysates to PICK1 tagged with glutathione-S- COCAPHASE (Dudbridge 2003; http://www.hgmp. transferase (GST-PICK1) and assessed the binding of mrc.ac.uk/Registered/Option/unphased.html). We also GST-PICK1 and HA-SR (Figure 2a). The PDZ domain used COCAPHASE to obtain global P-values for 2- and is critical, as mutation of the lysyl-aspartate of PICK1 3-marker haplotype tests. abolishes the binding. The importance of the coiled- coil domain is confirmed, as its deletion from PICK1 also abolishes binding. To establish that PICK1 and Results SR interact physiologically in brain tissue, we Binding of SR to PICK1 utilized primary glial cultures and demonstrate We used human SR as a bait in yeast two-hybrid coimmunoprecipitation of the two proteins at endo- screening with a human adult hippocampal cDNA genous level (Figure 2c). library (Figure 1). Out of one million independent For SR and PICK1 to interact physiologically, they clones, we observe five positive clones for PICK1, in should have similar localizations. We transfected SR addition to one positive clone for glutamate receptor and PICK1 into HEK293 cells and examined their interacting protein (GRIP), which we previously localization by immunofluorescence (Figure 3). Both reported to bind SR in rodent cDNA library.49 PICK1 SR and PICK1 occur exclusively in the cytoplasm in a contains a prominent PDZ domain, a motif associated similar granular pattern. Staining of primary glial with protein–protein interactions. The PDZ domain of cultures with antibodies to native SR and PICK1 PICK is responsible for its binding to protein kinase resembles the results obtained in the HEK293 cells C.42,50 PICK1 also contains a coiled-coil domain. To (data not shown). ascertain domains of PICK1 responsible for binding to SR, we employed yeast two-hybrid analysis. Mutation Association of PICK1 with disorganized SZ of a lysyl-aspartate dipeptide within the PDZ domain In an independent study, Hong et al.46 conducted a of PICK1, which is critical for protein kinase C case–control study in a Chinese population genotyp- binding, to alanyl-alanine abolishes binding of PICK1 ing a SNP that occurs within the PICK1 gene (rs3952)

Molecular Psychiatry PICK1 and schizophrenia K Fujii et al 153 and observed a mild association of the SNP and SZ. according to the criteria that their minor allele Here, we have conducted an association study of frequencies are 15% or greater in our Japanese three independent SNPs, including the SNP utilized population. These two SNPs (rs713729 and by Hong et al.,46 in Japanese SZ (Figure 4). To identify rs2076369) occur in the PDZ domain as does the appropriate SNPs for analysis, first we sequenced all SNP examined by Hong et al.46 As indicated by the 13 coding exons of PICK1 along with 50 base Haploview, all the SNPs follow Hardy–Weinberg adjacent intronic sequences, and selected two SNPs, equilibrium (rs713729, P = 0.666; rs3952, P = 1.0; rs2076369, P = 0.802). There is almost complete LD (disequilibrium coefficient D0 = 0.99, 0.07 > r2 > 0.40) a in the region spanning approximately 7000 bases, Binding including these three SNPs, indicating that the markers all cosegregate through recombination Input events. HA-SR ++++ All case–control association tests were performed GST-PICK1 + -- - using 1000 permutations in Haploview, which rando- GST-PICK1 ∆(139-166) - + - - mizes affection status for single markers and haplo- GST-PICK1 K27A/D28A - - + - types in blocks with each permutation to control for GST-GAPDH - - - + the possibility of bias due to multiple testing. In our tests of individual markers, we find weak evidence for an overall difference in allele distribution between b (kDa) 188 our 200 cases and 200 controls for one SNP, rs2076369 (P = 0.080) in intron 4 of the PICK1 gene (Table 1). Given the risk allele (‘T’ allele) frequency for the 62 rs2076369 SNP of 0.382, our sample of 200 cases and 200 controls has the ability to detect an effect size 38 with a relative risk (RR)1 (heterozygote) of 2.0 and a RR2 (homozygote) of 3.1 with 80% power, assuming a disease prevalence of 1%, which was estimated by the 17 Genetic Power Calculator available on-line (http:// PICK1 SR statgen.iop.kcl.ac.uk/gpc/).51 We subdivided our cases further into ‘disorganized’ c Mouse IgG SR antibody (n = 90) and ‘nondisorganized’ (n = 110), and per- Input IP formed single-marker analyses for each group as PICK1 compared all controls (n = 200). This is the only SR subdivision of the cases in this study, based on our hypothesis that ‘disorganized’ SZ may have a pre- Figure 2 Characterization of SR/PICK1 binding. (a) In vitro ferential susceptibility to genetic risk factors related protein binding confirms the interaction of SR and PICK1 to glutamatergic neurotransmission. The subset dis- that requires both the PDZ and coiled-coil domains of plays poorer prognosis with more prominent negative PICK1. (b) The specificity of the antibodies against SR and symptoms. In the ‘disorganized’ cases, we observe a PICK1 is tested in Western blotting, respectively. Both statistically significant difference in allele distribu- antibodies recognize the single bands of the expected tion for two SNPs, rs3952 (P = 0.050) and rs2076369 molecular weight in the Western blotting. (c) Co-immuno- (P = 0.012) in introns 3 and 4 of the PICK1 gene, precipitation of endogenous PICK1 with SR from primary respectively. We find no evidence in support of a culture glia. SR was immunoprecipitated from glial lysates and co-immunoprecipitated PICK1 was detected with a difference in allele distribution between ‘nondisorga- PICK1-specific antibody. Loading control showing SR nized’ cases and controls for any of our three markers. immunoprecipitation with a monoclonal antibody to SR. For the full sample (200 cases, 200 controls), the Input lane contains 1% of lysates used for co-immunopre- T-A-T haplotype (rs713729–rs3952–rs2076369) is dif- cipitation. ferentially distributed with moderate significance

Figure 3 SR and PICK1 distributions in cultured cells. Co-transfection of SR and PICK1 into HEK293 cells shows colocalization of SR and PICK1.

Molecular Psychiatry PICK1 and schizophrenia K Fujii et al 154

Figure 4 The PICK1 gene and the SNPs analyzed in the present study. (a) The PICK1 gene is located on the q arm of human chromosome 22; 22q13.1. This map was taken from the May 2004 UCSC Genome Browser. (b) Location of rs713729, rs3952, and rs2076369 relative to the 13-exon structure of the PICK1 gene (based on GenBank NM_012407) and the protein it encodes. The SNPs we studied were all contained within the PDZ domain of the PICK1 protein. (c) Haploview output indicating D0 measure of LD and the corresponding LD block structure. The SNPs we studied were in complete LD. Arrows indicate individual SNP locations within the PICK1 gene.

(P = 0.056–0.092) between cases and controls in every by PICK1 and GRIP is dependent on phosphorylation 2- and 3-marker combination. Our haplotype analyses status of the receptor.52 of ‘disorganized’ cases controls reveal a statistically We found an association of three SNPs in PICK1 significant association for the same risk haplotype T- with ‘disorganized’ but not with ‘nondisorganized’ SZ A-T (P = 0.010–0.015) for every combination of mar- (Figure 5). The selectivity of association with only kers as well as another differentially distributed one subtype of SZ, comprising about half the patients haplotype T-G-G with mild significance (P = 0.050– studied here, suggests the specificity of this relation- 0.087). We do not detect association of haplotypes in ship. ‘Disorganized’ patients generally represent a ‘non-disorganized’ cases as compared to controls. more severely ill population. Taken together, these results suggest that the In an independent investigation, Hong et al.46 genetic variation in the PICK1 gene may primarily observed a mild association of SZ with a single SNP reflect ‘disorganized’ cases. (P = 0.025). We examined the SNP employed by Hong et al.46 as well as two additional SNPs. In the total population of SZ we studied, the SNP employed by 46 Discussion Hong et al. did not show a statistically significant association with SZ, while there was a significant In the present study we have demonstrated a association with ‘disorganized’ patients (P = 0.050) physiologic protein interaction of PICK1 with SR. after correction for multiple testing. It should be The only other known protein interactor with SR is noted that the ‘risk’ allele over-represented in our GRIP whose binding to SR elicits a five-fold augmen- cases differs from that identified in the study by Hong tation of D-serine synthesis and release.49 Physiologic et al.46 In their study, Hong et al.46 indicate that the ‘G’ activation of AMPA receptors on glia markedly allele may be a risk factor for SZ. In our study, enhances SR activity and D-serine release. How PICK1 however, the ‘A’ allele is over-represented in SZ cases. and GRIP interact to regulate SR is unclear. In Despite these discrepancies, we speculate that the neurons, coordinated regulation of AMPA receptors rs3952 polymorphism may serve as a marker of a

Molecular Psychiatry PICK1 and schizophrenia K Fujii et al 155 Table 1 Case-control results for PICK1; 22q13.1 dbSNP-ID Locationa SNPb MAc Freq.c Casesc Cont.c w2 P-value Significant haplotypes

(a) All cases (n = 200) p-telomere rs713729 Intron3 T/A A 0.15 0.130 0.162 1.692 0.433 T T T rs3952 Intron3 A/G G 0.34 0.320 0.362 1.607 0.493 A A A rs2076369 Intron4 G/T T 0.42 0.462 0.382 5.246 0.080 T T T q-telomere Cases 0.550 0.460 0.462 0.460 Controls 0.475 0.382 0.382 0.380 w2 4.503 4.928 5.246 5.254 Global P-value 0.095 0.081 0.063 0.140 Haplotype P-value 0.092 0.062 0.056 0.081

(b) Disorganized cases only (n = 90) p-telomere rs713729 Intron3 T/A A 0.15 0.128 0.162 1.163 0.062 T T T T T T rs3952 Intron3 A/G G 0.33 0.261 0.362 5.762 0.050 AA AGG G rs2076369 Intron4 G/T T 0.42 0.489 0.382 8.416 0.012 TTT GGG q-telomere Cases 0.611 0.511 0.511 0.511 0.261 0.261 0.361 0.261 Controls 0.475 0.382 0.382 0.382 0.362 0.362 0.455 0.360 w2 9.210 8.416 8.416 8.416 5.762 5.762 4.476 5.494 Global P-value 0.009 0.011 0.015 0.024 0.009 0.011 0.015 0.024 Haplotype P-value 0.013 0.015 0.010 0.015 0.050 0.057 0.087 0.070

(c) Nondisorganized cases only (n = 110) p-telomere rs713729 Intron3 T/A A 0.15 0.132 0.162 1.039 0.605 rs3952 Intron3 A/G G 0.37 0.368 0.362 0.020 0.999 rs2076369 Intron4 G/T T 0.40 0.423 0.382 0.960 0.631 q-telomere

Results are presented here for single SNPs and 2- and 3-marker haplotypes for (a) all cases, (b) ‘disorganized’ cases and (c) ‘nondisorganized’ cases in a Japanese population for the PICK1 gene. Association was determined by Haploview Version 3.2 (Barrett) using 1000 permutations and confirmed by COCAPHASE (Dudbridge). P-values p0.05 are presented in bold type. Global P-values were obtained using COCAPHASE. aRelative to GenBank NM_012407. bOn human PICK1 coding strand. cMinor allele.

22q13 13q33 our study (P = 0.080 total samples; P = 0.012 ‘disorganized’ cases). PICK1 G72 Besides its interactions with protein kinase C,42,50 AMPA receptors,53,54 and metabotropic glutamate SR D-serine DAAO receptors55–58 and SR, PICK1 binds to the neuronal 17p13 12q24 membrane transporters for dopamine and norepi- nephrine.59 Coexpression of PICK1 with dopamine transporters leads to increased numbers of plasma NMDA-R membrane dopamine transporters and augmented dopamine uptake. Dopamine neurotransmission has been extensively implicated in SZ, as classic neuro- Figure 5 Association of SZ with genes that encodes leptics act by blocking dopamine D2 receptors while proteins involved in synthesis and degradation of D-serine. release of dopamine by amphetamines is associated The synthesizing and degradation enzymes for D-serine as with a worsening of symptoms.60–62 well as their potential modulators, such as PICK1 and G72, Most studies of neurotransmitters in mental illness are listed together with the chromosomal loci of the genes have focused upon their neuronal disposition. SR and that encode the proteins. Although the gene for SR is not D-serine are formed primarily in glia that ensheath located in the linkage spot for SZ, the genes for the other glutamatergic synapses.7 In the present study we proteins are in the linkage spot and are associated with the disease. demonstrate PICK1 in glia, where it interacts with SR. Many recent studies have emphasized the importance of glia for regulation of neurotransmission in the disease mutation rather than being a causative factor. brain.63,64 Our results suggest that glial disposition of This idea is consistent with the higher significance neurotransmitters may be relevant to understanding observed with another SNP (rs2076369) employed in the pathophysiology of SZ.

Molecular Psychiatry PICK1 and schizophrenia K Fujii et al 156 Acknowledgments 19 Goff DC, Tsai G, Levitt J, Amico E, Manoach D, Schoenfeld DA et al. A placebo-controlled trial of D-cycloserine added to We thank Drs Ann Pulver, David Valle, and Peter conventional neuroleptics in patients with schizophrenia. Arch Zandi for critical reading of the manuscript. We thank Gen Psychiatry 1999; 56: 21–27. 20 Heresco-Levy U, Javitt DC, Ermilov M, Mordel C, Silipo G, Ms Yukiko Lema for preparation of the figures. We Lichtenstein M. Efficacy of high-dose glycine in the treatment appreciate Dr Yukihiko Shirayama, Ms Pratima Dullor of enduring negative symptoms of schizophrenia. Arch Gen and Mr John Kleiderlein for technical assistance. This Psychiatry 1999; 56: 29–36. work was supported by US Public Health Service 21 Chumakov I, Blumenfeld M, Guerassimenko O, Cavarec L, Palicio Grant MH-069853, foundation grants from NARSAD, M, Abderrahim H et al. Genetic and physiological data implicating the new human gene G72 and the gene for D-amino acid oxidase in S-R and Stanley (to AS); US Public Health Service schizophrenia. Proc Natl Acad Sci USA 2002; 99: 13675–13680. Grant MH-18501 and Research Scientist Award 22 Hattori E, Liu C, Badner JA, Bonner TI, Christian SL, Maheshwari DA00074 (to SHS); National Institutes of Health Grant M et al. Polymorphisms at the G72/G30 gene locus, on 13q33, are NS036715 and the Howard Hughes Medical Institutes associated with bipolar disorder in two independent pedigree series. Am J Hum Genet 2003; 72: 1131–1140. (to RLH). TH is supported by fellowships from the 23 Chen YS, Akula N, Detera-Wadleigh SD, Schulze TG, Thomas J, Sankyo Foundation of Life Science and Uehara Potash JB et al. Findings in an independent sample support an Memorial Foundation. association between bipolar affective disorder and the G72/G30 locus on chromosome 13q33. Mol Psychiatry 2004; 9: 87–92; image 85. 24 Schumacher J, Jamra RA, Freudenberg J, Becker T, Ohlraun S, Otte AC et al. Examination of G72 and D-amino-acid oxidase as genetic References risk factors for schizophrenia and bipolar affective disorder. Mol Psychiatry 2004; 9: 203–207. 1 Goff DC, Coyle JT. The emerging role of glutamate in the 25 Addington AM, Gornick M, Sporn AL, Gogtay N, Greenstein D, pathophysiology and treatment of schizophrenia. Am J Psychiatry Lenane M et al. Polymorphisms in the 13q33.2 gene G72/G30 are 2001; 158: 1367–1377. associated with childhood-onset schizophrenia and psychosis not 2 Javitt DC, Zukin SR. Recent advances in the phencyclidine model otherwise specified. Biol Psychiatry 2004; 55: 976–980. of schizophrenia. Am J Psychiatry 1991; 148: 1301–1308. 26 Wang X, He G, Gu N, Yang J, Tang J, Chen Q et al. Association of 3 Tsai G, Coyle JT. Glutamatergic mechanisms in schizophrenia. G72/G30 with schizophrenia in the Chinese population. Biochem Annu Rev Pharmacol Toxicol 2002; 42: 165–179. Biophys Res Commun 2004; 319: 1281–1286. 4 Sawa A, Snyder SH. Schizophrenia: diverse approaches to a 27 Korostishevsky M, Kaganovich M, Cholostoy A, Ashkenazi M, complex disease. Science 2002; 296: 692–695. Ratner Y, Dahary D et al. Is the G72/G30 locus associated with 5 Sawa A, Snyder SH. Schizophrenia: neural mechanisms for novel schizophrenia? Single nucleotide polymorphisms, haplo- therapies. Mol Med 2003; 9: 3–9. types, and gene expression analysis. Biol Psychiatry 2004; 56: 6 Johnson JW, Ascher P. Glycine potentiates the NMDA response in 169–176. cultured mouse brain neurons. Nature 1987; 325: 529–531. 28 Yamada K, Ohnishi T, Hashimoto K, Ohba H, Iwayama-Shigeno Y, 7 Schell MJ, Molliver ME, Snyder SH. D-serine, an endogenous Toyoshima M et al. Identification of multiple serine racemase synaptic modulator: localization to astrocytes and glutamate- (SRR) mRNA isoforms and genetic analyses of SRR and DAO stimulated release. Proc Natl Acad Sci USA 1995; 92: 3948–3952. in schizophrenia and D-serine levels. Biol Psychiatry 2005; 57: 8 Schell MJ, Brady Jr RO, Molliver ME, Snyder SH. D-serine as a 1493–1503. neuromodulator: regional and developmental localizations in 29 Pulver AE, Karayiorgou M, Wolyniec PS, Lasseter VK, Kasch L, rat brain glia resemble NMDA receptors. J Neurosci 1997; 17: Nestadt G et al. Sequential strategy to identify a susceptibility gene 1604–1615. for schizophrenia: report of potential linkage on chromosome 9 Snyder SH, Ferris CD. Novel neurotransmitters and their neuro- 22q12–q13.1: Part 1. Am J Med Genet 1994; 54: 36–43. psychiatric relevance. Am J Psychiatry 2000; 157: 1738–1751. 30 Pulver AE, Karayiorgou M, Lasseter VK, Wolyniec P, Kasch L, 10 Snyder SH, Kim PM. D-amino acids as putative neurotransmitters: Antonarakis S et al. Follow-up of a report of a potential linkage for focus on D-serine. Neurochem Res 2000; 25: 553–560. schizophrenia on chromosome 22q12–q13.1: Part 2. Am J Med 11 Mothet JP, Parent AT, Wolosker H, Brady Jr RO, Linden DJ, Ferris Genet 1994; 54: 44–50. CD et al. D-serine is an endogenous ligand for the glycine site of 31 Karayiorgou M, Gogos JA. A turning point in schizophrenia the N-methyl-D-aspartate receptor. Proc Natl Acad Sci USA 2000; genetics. Neuron 1997; 19: 967–979. 97: 4926–4931. 32 Pulver AE. Search for schizophrenia susceptibility genes. Biol 12 Baranano DE, Ferris CD, Snyder SH. Atypical neural messengers. Psychiatry 2000; 47: 221–230. Trends Neurosci 2001; 24: 99–106. 33 Liu H, Abecasis GR, Heath SC, Knowles A, Demars S, Chen YJ 13 Boehning D, Snyder SH. Novel neural modulators. Annu Rev et al. Genetic variation in the 22q11 locus and susceptibility to Neurosci 2003; 26: 105–131. schizophrenia. Proc Natl Acad Sci USA 2002; 99: 16859–16864. 14 Wolosker H, Blackshaw S, Snyder SH. Serine racemase: a glial 34 Karayiorgou M, Gogos JA. The molecular genetics of the 22q11- enzyme synthesizing D-serine to regulate glutamate-N-methyl-D- associated schizophrenia. Brain Res Mol Brain Res 2004; 132: aspartate neurotransmission. Proc Natl Acad Sci USA 1999; 96: 95–104. 13409–13414. 35 Karayiorgou M, Gogos JA, Galke BL, Wolyniec PS, Nestadt G, 15 Wolosker H, Sheth KN, Takahashi M, Mothet JP, Brady Jr RO, Antonarakis SE et al. Identification of sequence variants and Ferris CD et al. Purification of serine racemase: biosynthesis of the analysis of the role of the catechol-O-methyl-transferase gene neuromodulator D-serine. Proc Natl Acad Sci USA 1999; 96: in schizophrenia susceptibility. Biol Psychiatry 1998; 43: 721–725. 425–431. 16 Javitt DC, Zylberman I, Zukin SR, Heresco-Levy U, Lindenmayer 36 Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, JP. Amelioration of negative symptoms in schizophrenia by Straub RE et al. Effect of COMT Val108/158 Met genotype on glycine. Am J Psychiatry 1994; 151: 1234–1236. frontal lobe function and risk for schizophrenia. Proc Natl Acad 17 Heresco-Levy U, Javitt DC, Ermilov M, Mordel C, Horowitz A, Sci USA 2001; 98: 6917–6922. Kelly D. Double-blind, placebo-controlled, crossover trial of 37 Mukai J, Liu H, Burt RA, Swor DE, Lai WS, Karayiorgou M et al. glycine adjuvant therapy for treatment-resistant schizophrenia. Evidence that the gene encoding ZDHHC8 contributes to the risk Br J Psychiatry 1996; 169: 610–617. of schizophrenia. Nat Genet 2004; 36: 725–731. 18 Tsai G, Yang P, Chung LC, Lange N, Coyle JT. D-serine added to 38 Gogos JA, Santha M, Takacs Z, Beck KD, Luine V, Lucas LR et al. antipsychotics for the treatment of schizophrenia. Biol Psychiatry The gene encoding proline dehydrogenase modulates sensorimo- 1998; 44: 1081–1089. tor gating in mice. Nat Genet 1999; 21: 434–439.

Molecular Psychiatry PICK1 and schizophrenia K Fujii et al 157 39 Liu H, Heath SC, Sobin C, Roos JL, Galke BL, Blundell ML et al. 51 Purcell S, Cherny SS, Sham PC. Genetic Power Calculator: design Genetic variation at the 22q11 PRODH2/DGCR6 locus presents an of linkage and association genetic mapping studies of complex unusual pattern and increases susceptibility to schizophrenia. traits. Bioinformatics 2003; 19: 149–150. Proc Natl Acad Sci USA 2002; 99: 3717–3722. 52 Chung HJ, Xia J, Scannevin RH, Zhang X, Huganir RL. Phosphor- 40 Jacquet H, Raux G, Thibaut F, Hecketsweiler B, Houy E, ylation of the AMPA receptor subunit GluR2 differentially Demilly C et al. PRODH mutations and hyperprolinemia regulates its interaction with PDZ domain-containing proteins. in a subset of schizophrenic patients. Hum Mol Genet 2002; 11: J Neurosci 2000; 20: 7258–7267. 2243–2249. 53 Dev KK, Nishimune A, Henley JM, Nakanishi S. The protein 41 Ozeki Y, Tomoda T, Kleiderlein J, Kamiya A, Bord L, Fujii K et al. kinase C alpha binding protein PICK1 interacts with short but not Disrupted-in-Schizophrenia-1 (DISC-1): mutant truncation pre- long form alternative splice variants of AMPA receptor subunits. vents binding to NudE-like (NUDEL) and inhibits neurite out- Neuropharmacology 1999; 38: 635–644. growth. Proc Natl Acad Sci USA 2003; 100: 289–294. 54 Xia J, Zhang X, Staudinger J, Huganir RL. Clustering of AMPA 42 Staudinger J, Lu J, Olson EN. Specific interaction of the PDZ receptors by the synaptic PDZ domain-containing protein PICK1. domain protein PICK1 with the COOH terminus of protein kinase Neuron 1999; 22: 179–187. C-alpha. J Biol Chem 1997; 272: 32019–32024. 55 Boudin H, Doan A, Xia J, Shigemoto R, Huganir RL, Worley P et al. 43 Banker G, Goslin K (eds). Culturing Nerve Cells, 2nd edn. MIT Presynaptic clustering of mGluR7a requires the PICK1 PDZ Press: Cambridge, MA, 1998. domain binding site. Neuron 2000; 28: 485–497. 44 Nakata K, Ujike H, Sakai A, Takaki M, Imamura T, Tanaka Y et al. 56 Hirbec H, Perestenko O, Nishimune A, Meyer G, Nakanishi S, The human dihydropyrimidinase-related protein 2 gene on Henley JM et al. The PDZ proteins PICK1, GRIP, and syntenin bind chromosome 8p21 is associated with paranoid-type schizophre- multiple glutamate receptor subtypes. Analysis of PDZ binding nia. Biol Psychiatry 2003; 53: 571–576. motifs. J Biol Chem 2002; 277: 15221–15224. 45 Uchida N, Ujike H, Nakata K, Takaki M, Nomura A, Katsu T et al. 57 Dev KK, Nakanishi S, Henley JM. Regulation of mglu(7) receptors No association between the sigma receptor type 1 gene and by proteins that interact with the intracellular C-terminus. Trends schizophrenia: results of analysis and meta-analysis of case– Pharmacol Sci 2001; 22: 355–361. control studies. BMC Psychiatry 2003; 3:13. 58 Dev KK, Nakajima Y, Kitano J, Braithwaite SP, Henley JM, 46 Hong CJ, Liao DL, Shih HL, Tsai SJ. Association study of PICK1 Nakanishi S. PICK1 interacts with and regulates PKC phosphor- rs3952 polymorphism and schizophrenia. Neuroreport 2004; 15: ylation of mGLUR7. J Neurosci 2000; 20: 7252–7257. 1965–1967. 59 Torres GE, Yao WD, Mohn AR, Quan H, Kim KM, Levey AI et al. 47 De la Vega FM, Lazaruk KD, Rhodes MD, Wenz MH. Assessment of Functional interaction between monoamine plasma membrane two flexible and compatible SNP genotyping platforms: TaqMan transporters and the synaptic PDZ domain-containing protein SNP Genotyping Assays and the SNPlex Genotyping System. PICK1. Neuron 2001; 30: 121–134. Mutat Res 2005; 573: 111–135. 60 Creese I, Burt DR, Snyder SH. Dopamine receptor binding predicts 48 Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and clinical and pharmacological potencies of antischizophrenic visualization of LD and haplotype maps. Bioinformatics 2005; 21: drugs. Science 1976; 192: 481–483. 263–265. 61 Seeman P, Lee T, Chau-Wong M, Wong K. Antipsychotic drug 49 Kim PM, Aizawa H, Kim PS, Huang AS, Wickramasinghe SR, doses and neuroleptic/dopamine receptors. Nature 1976; 261: Kashani AH et al. Serine racemase: activation by glutamate 717–719. neurotransmission via glutamate receptor interacting protein and 62 Carlsson A. The current status of the dopamine hypothesis of mediation of neuronal migration. Proc Natl Acad Sci USA 2005; schizophrenia. Neuropsychopharmacology 1988; 1: 179–186. 102: 2105–2110. 63 Sawa A, Pletnikov MV, Kamiya A. Neuron–glia interactions clarify 50 Staudinger J, Zhou J, Burgess R, Elledge SJ, Olson EN. PICK1: a genetic-environmental links in mental illness. Trends Neurosci perinuclear binding protein and substrate for protein kinase C 2004; 27: 294–297. isolated by the yeast two-hybrid system. J Cell Biol 1995; 128: 64 Newman EA. New roles for astrocytes: regulation of synaptic 263–271. transmission. Trends Neurosci 2003; 26: 536–542.

Molecular Psychiatry