Association study on the DUSP6 gene, an affective disorder candidate ...

Molecular Psychiatry (2000) 5, 489–494  2000 Macmillan Publishers Ltd All rights reserved 1359-4184/00 $15.00 www.nature.com/mp

ORIGINAL RESEARCH ARTICLE

Association study on the DUSP6 gene, an affective disorder candidate gene on 12q23, performed by using fluorescence resonance energy transfer-based melting curve analysis on the LightCycler T Toyota1,2, A Watanabe1,2, H Shibuya3, M Nankai4, E Hattori1,3, K Yamada1, A Kurumaji2, JD Karkera5, SD Detera-Wadleigh5 and T Yoshikawa1 1

Laboratory for Molecular Psychiatry, Brain Science Institute, RIKEN, Wako, Saitama 351–0198, Japan; 2Department of Neuropsychiatry, Tokyo Medical and Dental University, Tokyo, Japan; 3National Sanatorium Minami-Hanamaki Hospital, Iwate, Japan; 4Tokyo Metropolitan Police Hospital, Department of Neuropsychiatry, Japan; 5National Institute of Mental Health, Intramural Research Program, Bethesda, MD 20892–4094, USA We introduced a new genotyping method, fluorescence resonance energy transfer-based melting curve analysis on the LightCycler, for the analysis of the gene, DUSP6 (dual specificity MAP kinase phosphatase 6), in affective disorder patients. The DUSP6 gene is located on chromosome 12q22–23, which overlaps one of the reported bipolar disorder susceptibility loci. Because of its role in intracellular signalling pathways, the gene may be involved in the pathogenesis of affective disorders not only on the basis of its position but also of its function. We performed association analysis using a T⬎G polymorphism that gives rise to a missense mutation (Leu114Val). No evidence for a significant disease-causing effect was found in Japanese unipolars (n = 132) and bipolars (n = 122), when compared with controls (n = 299). More importantly, this study demonstrates that melting curve analysis on the LightCycler is an accurate, rapid and robust method for discriminating genotypes from biallelic markers. This strategy has the potential for use in high throughput scanning for and genotyping of single nucleotide polymorphisms (SNPs). Molecular Psychiatry (2000) 5, 489–494. Keywords: chromosome 12q; genotyping; single nucleotide polymorphism (SNP); biallelic polymorphism; unipolar; bipolar

Introduction SNPs are being used increasingly for conducting genetic studies of complex diseases including psychiatric diseases. They are typed by using one, or a combination of the following methods: sequencing,1 PCRsingle strand conformational polymorphism assay (SSCP),2 PCR-restriction fragment length polymorphism analysis (RFLP),3 amplification with sequencespecific primers (SSP)4 and chip hybridization technology.5 Among them, PCR-RFLP is the most frequently employed method because of its convenience and ease. However, sometimes unsatisfactory digestion patterns and difficulty in discriminating genotypes are encountered, leading to the risk of mistyping. In this study, we set out to examine the possible involvement of the gene, DUSP6 (dual specificity MAP kinase phosphatase 6, previously called MKP3:

Correspondence: T Yoshikawa, MD, PhD, Laboratory for Molecular Psychiatry, Brain Science Institute, RIKEN, 2–1 Hirosawa, Wako, Saitama 351–0198, Japan. E-mail: takeo얀brain.riken.go.jp Received 15 December 1999; revised and accepted 28 February 2000

mitogen-activated protein kinase phophatase 3)6 in genetic liability of affective disorder in a case-control design. The reason for choosing this gene is that the DUSP6 is located in 12q22–q23,6 which falls in one of the reported bipolar linkage regions.7–9 DUSP6 is also functionally relevant to neuronal systems because mitogen-activated protein (MAP) kinase pathways have been implicated in the control of a number of cellular processes including neuronal differentiation.10 A missense polymorphism in the DUSP6 gene has been reported at codon 114 in exon 1 (TTG/GTG: Leu/Val) by Furukawa et al.11 Initially, we analyzed this SNP by using PCR-RFLP. However, we could not resolve ambiguities in the digestion patterns to discriminate reliably between different genotypes. Thus, we introduced a new strategy for genotyping, which uses fluorescence resonance energy transfer probes and melting curve analysis on the PCR machine equipped with a fluorescence detection system, LightCycler (Roche). The determination of an optimal condition for the analysis was facile, and typing results were judged to be accurate since they were consistent with results obtained by dye-primer sequencing method. Here, we report the use of the LightCycler in the genetic analysis

Genotyping of the DUSP6 with the LightCycler T Toyota et al

490

of the DUSP6 gene polymorphism to test for association in Japanese affective disorder patients.

Materials and methods Patients and controls Mood disorder patients met the diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), and were diagnosed as either unipolar or bipolar by the consensus of at least two psychiatrists. They consisted of both outpatients and inpatients. The unipolar patients consisted of 52 males (mean 56 ± 14 years), and 80 females (mean age 58 ± 13 years). About 30% of the unipolars had a single episode and the rest had recurrent episodes. The bipolar patients consisted of 70 males (mean 50 ± 13 years), and 52 females (mean 52 ± 12 years). Sixty per cent of the bipolars were diagnosed as bipolar I and 40% were bipolar II. Control subjects were recruited from hospital staff documented to be free of psychoses, and company employees who did not manifest psychiatric problems in brief interviews by psychiatrists. They included 177 males (mean 46 ± 10 years) and 122 females (mean 45 ± 15 years). The present study was approved by the Ethics Committee of RIKEN and Tokyo Medical and Dental University, and all participants gave written informed consent. PCR amplification of the genomic fragment encompassing the missense polymorphic site, restriction analysis and sequencing of PCR products Genomic DNA was extracted using the DNA Extraction kit (Stratagene, La Jolla, CA, USA). The upstream primer used was 5⬘-ATGGCGATCAGCAAGACGGTG (5⬘ end at nt 391 counted from the 5⬘ end of exon 1), and the downstream primer was 5⬘-GCCC TGCGCCCCTAGCCGTGC (3⬘ end at nt 30 counted from

the 5⬘ end of intron 1). The missense polymorphism is located at the 691st nucleotide counting from the first nucleotide in exon 1 (T⬎G: Leu114Val). PCR was performed with an initial 1-min denaturation at 94°C, followed by 30 cycles of 94°C for 15 s, 55°C for 30 s, 72°C for 1 min, and a final extension period at 72°C for 5 min, using Ex Taq (Takara, Tokyo, Japan) and the GeneAmp PCR System 9700 thermocycler (PE Applied BioSystems, Norwalk, CT, USA). A 20-␮l aliquot of the amplified mixture was incubated with 0.8 unit of Bsp1286I (New England Biolabs, Beverly, MA, USA) for 2 h to overnight, and then separated on a 2% Nusieve 3:1 agarose gel. For sequencing, the template was prepared with the −21M13 sequence (5⬘-TGTAAAACGACGGCC AGT)-tailed upstream primer at its 5⬘-end, and the downstream primer. Sequencing reaction was conducted using the BigDye primer cycle sequencing kit (−21M13) (PE Applied BioSystems), and products were run on an ABI 377 DNA sequencer (PE Applied BioSystems). Genotyping methodology using fluorescence resonance energy transfer (FRET) Fluorescence monitoring using hybridization probes is based on the concept that a fluorescence signal is generated if fluorescence resonance energy transfer occurs between two adjacent fluorophores (Figure 1). The first hybridization (detection) probe, which is labelled with fluorescein (FITC) as the donor fluorophore on its 3⬘end, can hybridize to a second hybridization (anchor) probe that is in close proximity and is labelled with the acceptor fluorophore LightCycler RED 640 (LC-RED) at its 5⬘-end and is blocked from extension at its 3⬘-terminus. The fluorescein is excited by the LED light source and emits light with a wavelength that excites the acceptor fluorophore. The acceptor fluorophore then emits light of a longer wavelength that can be

Figure 1 Position of the FITC-labelled detection probe and the LC-RED-labelled anchor probe. Both probes hybridize to their complementary antisense strand. The proximity of the FITC and LC-RED results in fluorescence resonance energy transfer that is monitored during melting curve analysis. The detection probe perfectly matches the T allele, but has a T-C mismatch with the G allele. This mismatch destabilizes the hybrid which results in a decrease in the probe melting temperature. In contrast, complete matching of detection probe and antisense strand of the T allele produces a higher melting temperature of the hybrid. The arrows show the PCR primers used to amplify the fragment. Molecular Psychiatry


a

measured by an optical unit (photohybrid). Genotyping using two hybridization probes is possible with a shorter ‘detection probe’ spanning the polymorphic site and a longer ‘anchor probe’ recognizing an adjacent sequence. The greater stability of the longer ‘anchor probe’ causes the ‘detection probe’ to melt off the template at lower temperature so that polymorphic alleles can be distinguished by the melting temperature (Tm) of the ‘detection probe’. Continuous fluorescence monitoring of the reaction as the temperature is raised from annealing to denaturation results in a sharp decrease in fluorescence at the temperature at which the ‘detection probe’ dissociates from the template.12,13 The single base change caused by the DUSP6 polymorphism results in a decrease in the fluorescence of the ‘detection probe’ that can easily be distinguished with the LightCycler (Roche).

b

5

Figure 2 PCR-RFLP patterns and sequence electropherograms of the missense polymorphism (Leu114Va1). (a) Digestion patterns of the missense polymorphism done by Bsp1286I restriction enzyme. Lanes 3 and 6 were verified as T/T and G/G genotypes, respectively, by dye primer sequencing. M: 100-bp DNA ladder (Gibco-BRL). (b) Electropherograms of the dye primer sequencing. The sequencing templates of lane ‘n’ correspond to those shown in (a).

491

Design of fluorogenic probes and fluorescence analysis protocol The detection probe was a 20-mer oligonucleotide, labelled at the 3⬘-end with FITC. The sequence was 5⬘GAGTCGTTGCTCGGGCTGCT (based on the sense strand), with the polymorphic site at the 7th base from the 5⬘-end (underlined nt) (Figure 1). The anchor probe was 5⬘-LC-RED labelled 24-mer, 5⬘-CTCAAGAA GCTCAAGGACGAGGGC, whose 3⬘-end was phosphorylated to block extension by Taq polymerase. The 5⬘-end of the anchor probe was located two bases downstream of the 3⬘-end of the detection probe (Figure 1). Both fluorophore-labelled probes were synthesized and purified by reverse-phase HPLC by Nihon Gene Research Labs Inc, Japan. Two different protocols were employed for the whole fluorescence analysis: (1) a single-step method, that is, the PCR amplification and subsequent melting curve analysis in one tube; (2) a two-step method consisting of separate PCR amplification and FRET analysis. In the first protocol, PCR was performed by rapidcycling in a reaction volume of 20 ␮l containing 0.5 ␮M of each primer, 200 ␮M of dNTPs, 4 pmol of FITC-labelled detection probe, 8 pmol of LC-RED-labelled anchor probe, BSA (final 500 ␮g ml−1), MgCl2 (final 5 mM), 0.7 unit of Ex Taq (Takara) DNA polymerase, respectively, and 50 ng of genomic DNA. Thirty-two samples were loaded into individual glass capillary cuvettes (Roche), centrifuged to place the samples at the capillary tip, and were put in a carousel. The carousel was set to the LightCycler. After an initial denaturation step at 95°C for 1 min, amplification was performed using 35 cycles of denaturation (95°C for 0 s), annealing (55°C for 10 s) and extension (72°C for 15 s). The temperature transition rates were programmed at 20°C s−1 from denaturation to annealing, 20°C s−1 from annealing to extension and 20°C s−1 from extension to denaturation. After amplification was completed in approximately 30 min, a melting curve was recorded under the following process: the samples were first denatured at 95°C for 0 s, then cooled down to 40°C at 20°C s−1, held at 40°C for 2 min, and then heated slowly at 0.1°C s−1 up to 80°C. In the second protocol, ampliMolecular Psychiatry


492

a

fication was done using the GeneAmp PCR System 9700 thermocycler (PE Applied BioSystems). A 1-␮l aliquot of the amplified mixture was transferred to a 19 ␮l of reaction mixture for the melting curve analysis, which contained 4 pmol of FITC-labelled detection probe, 8 pmol of LC-RED-labelled anchor probe, BSA (final 500 ␮g ml−1), and MgCl2 (final 5 mM). The conditions for melting analysis were the same as in the first protocol. In both protocols, fluorescence was measured continuously during the slow temperature rise to monitor the dissociation of the FITC-labelled detection probe. The fluorescence signal (F) was plotted in realtime against temperature (T) to produce melting curves for each sample (F vs T). Melting curves were then converted to melting peaks by plotting the negative derivative of the fluorescence with respect to temperature against temperature (-dF/dT vs T). The step for melting curve analysis was completed in approximately 7 min. Statistical analyses for genotype results Tests for Hardy–Weinberg equilibrium and allele comparisons were made by the use of ␹2 statistics with the significance level at 5%. Differences of genotype distributions were assessed by the Monte Carlo method using CLUMP program.14

b Results

Figure 3 Genotyping of the DUSP6 gene by melting curve analysis on the LightCycler. (a) Melting curves of the three different genotypes; fluorescence signal (F) vs temperature is plotted. The temperature transition was programmed at 0.4°C s−1, and the fluorescence from LC-RED was continuously monitored. (b) Derivative melting curve; the melting curve plot of fluorescence signal (F) vs temperature (T) was transformed into a derivative melting curve plot with −dF/dT vs temperature (T).

Molecular Psychiatry

A T⬎G transversion in exon 1 of DUSP6 giving rise to Leu111Val substitution was described previously.11 We sequenced all the three exons including the flanking splice junctions in 20 unrelated unipolar and 20 unrelated bipolar disorder samples in an effort to detect new SNPs but found no additional variants. The T⬎G SNP in exon 1 of DUSP6 can be genotyped by using the restriction enzyme Bsp1286I. Examples of genotyping results determined by PCR-RFLP are shown in Figure 2a. Theoretically, the T/T genotype gives two bands of 249 and 162 bp (lane 3 in Figure 2a). In the G/G genotype, the enzyme further digests the 1620-bp band into 108-bp and 54-bp fragments to yield three bands in total (lane 6 in Figure 2a). Therefore, a heterozygote displayed both 162- and 108-bp bands, the former slightly brighter in ethidium bromide staining than the latter (lanes 1 and 2 in Figure 2a, and Figure 2b). However, in our analysis, some of seemingly heterozygous patterns turned out to be homozygous when checked by the dye primer sequencing method that is, either G/G, probably due to incomplete digestion (lane 5 in Figure 2a, and Figure 2b), or T/T (lane 4 in Figure 2a, and Figure 2b), whose cause could not be easily explained. These complications could not be resolved by using different amounts and sources of the enzyme, or by varying incubation time and ionic strength of buffers (data not shown). We decided to employ another method that can unambiguously distinguish each genotype. By exploiting the principle behind the LightCycler method, we surmised that the different genotypes may exhibit distinctive melting profiles. In the LightCycler analysis, melting of the sample homozygous for T/T


Table 1

493

Distribution of the missense polymorphism of the DUSP6 gene

Subjects

n

Genotype counts (frequency) G/G

Mood disorder total 254 Unipolar 132 Bipolar 122 Controls 299

64 34 30 61

(0.25) (0.26) (0.25) (0.20)

G/T 122 65 57 162

(0.48) (0.49) (0.46) (0.54)

P value

Allele counts (frequency)

T/T 68 27 35 76

produced a rapid decrease in fluorescence at 70–71°C (Figure 3a). In contrast, the melting of the G/G homozygote occurred at 64–65°C. The heterozygous sample exhibited two distinct decreases in fluorescence, corresponding to the presence of amplicons derived from both alleles. By plotting the negative derivatives of the fluorescence signal with temperature vs temperature (−dF/dT vs T), peaks were obtained at the respective melting temperatures (Figure 3b). Accordingly, the melting peak of the sample homozygous for the T/T allele was at 70°C, whereas the sample homozygous for the G/G allele produced a melting peak at 64°C (the T allele perfectly matches the detection probe sequence, whereas the G allele does not). The heterozygous sample contained both types of targets and thus generated both peaks. The genotypes judged by melting curve analysis perfectly matched those obtained by dye primer sequencing (data not shown), which proved the reliability of the LightCycler method. The genotype and allelic distributions of the missense polymorphism obtained by melting curve analysis are shown in Table 1. We analyzed affective disorder patients by dividing them into three groups, unipolar, bipolar and all affective (unipolar plus bipolar) groups. In all three groups and in the controls, the genotypic frequencies were in Hardy–Weinberg equilibrium (␹2 = 0.03, P = 0.86, df = 1 for unipolar; ␹2 = 0.50, P = 0.48, df = 1 for bipolar; ␹2 = 0.39, P = 0.53, df = 1 for all affective; ␹2 = 2.2, P = 0.14, df = 1 for controls). The comparison of genotypic frequencies showed that none of affective disorder groups differed from controls (P = 0.44 for unipolar; P = 0.39 for bipolar; P = 0.31 for all affective) (Table 1). Likewise, allele frequencies were also not significantly different between affective groups and controls (␹2 = 0.61, P = 0.43, df = 1 for unipolar; ␹2 = 0.02, P = 0.90, df = 1 for bipolar; ␹2 = 0.33, P = 0.57, df = 1 for all affective). These analyses suggest that the DUSP6 gene on 12q22– q23 does not have a major effect on the pathogenesis of affective disorders in these Japanese cohorts.

Discussion Chromosome 12q23–24.1 has been implicated in affective disorder by linkage studies. Co-segregation between affective disorder and Darier’s disease (keratosis follicularis), a rare autosomal dominant skin disorder linked to the same region on 12q, has also been

(0.27) (0.25) (0.29) (0.26)

G 0.31 0.44 0.39

250 133 117 284

(0.49) (0.50) (0.48) (0.47)

P value

T 258 131 127 314

(0.51) (0.50) (0.52) (0.53)

0.67 0.43 0.90

reported.15,16 Recently, the Darier’s disease gene has been identified to be ATP2A2 encoding a Ca2+ pump.17 Jacobsen et al found a ‘non-random clustering of mutations in the 3⬘ end of the gene’ in neuropsychiatric samples.9 The phospholipase A2 gene (PLA2A) which maps to this region has been analyzed as a candidate gene by linkage analysis18 and association study.19 However, the results so far have not favored its relevance to disease susceptibility. Another positional and functional candidate gene may be the DUSP6 gene. In addition to its involvement in intracellular signalling pathways, the location of the DUSP6 has been shown to display frequent allelic loss in pancreatic cancer.20 It is known that depression and anxiety occur more frequently in patients with cancer of the pancreas than in patients with other forms of cancer. The etiology of depression in pancreatic cancer may be traced to other factors beyond the disease’s poor prognosis, the pain it causes, or existential issues related to death and dying.21 In this study, we examined the possible role of the DUSP6 gene using the missense polymorphism (Leu114Val) in affective disorder susceptibility. We first adopted a conventional PCR-RFLP approach as a genotyping method. However, it produced inconsistent results. There are other PCR-based methods for scoring biallelic markers. PCR with allele-specific oligonucleotide (ASO) dot-blot hybridization, for example, is timeconsuming and results are crucially dependent on the quality of PCR products and hybridization conditions.22 PCR-SSP is an increasingly popular method for genotyping biallelic systems.23 But most protocols to date require the use of separate tubes for specifically amplifying each allele, adding time and expense. In addition to the inherent risk of false positive amplification, the performance of the sequence-specific primers has to be assured by amplifying internal control fragments. In the present study, we used probe hybridization and fluorescence-based methods by melting-curve analysis on the LightCycler to determine DUSP6 genotypes. Our 20-mer detection probe showed a 6°C difference in Tm between the two alleles, indicating that the design of our detection probe was useful to discriminate between genotypes at that locus. This protocol enabled the fluorescence genotyping of 32 samples in 7 min without need for electrophoresis. It was also possible to PCR-amplify DNA and perform melting-curve analysis in the same tube consecutively. Molecular Psychiatry


494

Owing to the rapid amplification on the LightCycler, the whole process was completed within 30 min for 32 samples.13 There were two reasons why we tested a two-step protocol, that is, performing PCR amplification and fluorescence analysis in separate tubes. First, the quality of our sample DNAs was variable, and PCR yield of amplicons did not reach a satisfactory level for some of the samples. Thus, it was prudent to check the PCR amplification before proceeding to melting-curve analysis. Second, conducting these reactions separately allowed us to PCR in larger reaction volumes, sufficient for both the melting curve analysis and sequencing reaction. In this protocol we learned that the melting curve analysis needs optimal template concentrations although the window is not narrow (approximately two-fold range is permissible). To our knowledge, this is the first report on the use of the two-step protocol, and this may add broader versatility to the LightCycler. Although one-step analysis is more efficient, the twostep approach still affords a much higher genotyping throughput with accuracy than the combination of enzyme digestion and subsequent electrophoresis. In summary, although we examined only one polymorphic site in the DUSP6 gene, the results excluded the possibility of a substantial disease-causing effect of DUSP6 in the Japanese affective disorder cohorts. For genotyping, we introduced a melting-curve analysis on the LightCycler, and showed that it can be used for rapid, precise and unambiguous detection and analysis of biallelic SNPs. We believe that this approach could be extended to molecular scanning of candidates for disease-related mutations. Considering that the field of psychiatric genetic research is rapidly expanding along with systematic identification of human genes and SNPs,24 it is imperative to incorporate high throughput, precise, robust and reliable genetic screening methods for SNP scoring.

References 1 Yoshikawa T, Padigaru M, Karkera JD, Sharma M, Berrettini W, Esterling LE et al. Genomic structure and novel variants of myoinositol monophosphatase 2 (IMPA2). Mol Psychiatry 2000; 5: 165–171. 2 Hinney A, Bornscheuer A, Depenbusch M, Mierke B, To¨lle A, Middeke K et al. No evidence for involvement of the leptin gene in anorexia nervosa, bulimia nervosa, underweight or early onset extreme obesity: identification of two novel mutations in the coding sequence and a novel polymorphism in the leptin gene linked upstream region. Mol Psychiatry 1998; 3: 539–543. 3 Laruelle M, Gelernter J, Innis RB. D2 receptors binding potential is not affected by Taq1 polymorphism at the D2 receptor gene. Mol Psychiatry 1998; 3: 261–265. 4 Cavanagh G, Dunn AN, Chapman CE, Metcalfe P. HPA genotyping by PCR sequence-specific priming (PCR-SSP): a streamlined method for rapid routine investigations. Transfus Med 1997; 7: 41–45. 5 Hacia JG, Brody LC, Collins FS. Applications of DNA chips for genomic analysis. Mol Psychiatry 1998; 3: 483–492.

Molecular Psychiatry

6 Smith A, Price C, Cullen M, Muda M, King A, Ozanne B et al. Chromosomal localization of three human dual specificity phosphatase genes (DUSP4, DUSP6, and DUSP7). Genomics 1997; 42: 524–527. 7 Ewald H, Degn B, Mors O, Kruse TA. Significant linkage between bipolar affective disorder and chromosome 12q24. Psychiatr Genet 1998; 8: 131–140. 8 Morissette J, Villeneuve A, Bordeleau L, Rochette D, Laberge C, Gagne B et al. Genome-wide search for linkage of bipolar affective disorders in a very large pedigree derived from a homogeneous population in Quebec points to a locus of major effect on chromosome 12q23–q24. Am J Med Genet 1999; 88: 567–587. 9 Jacobsen NJO, Lyons I, Hoogendoorn B, Burge S, Kwok P-Y, O’Donovan MC et al. ATP2A2 mutations in Darier’s disease and their relationships to neuropsychiatric phenotypes. Hum Mol Genet 1999; 8: 1631–1636. 10 Cowley S, Paterson H, Kemp P, Marshall CJ. Activation of MAP kinase kinase is necessary and sufficient for PC12 differentiation and for transformation of NIH 3T3 cells. Cell 1994; 77: 841–852. 11 Furukawa T, Yatsuoka T, Youssef EM, Abe T, Yokoyama T, Fukushige S et al. Genomic analysis of DUSP6, a dual specificity MAP kinase phosphatase, in pancreatic cancer. Cytogenet Cell Genet 1998; 82: 156–159. 12 Aslanidis C, Schmitz G. High-speed apolipoprotein E genotyping and apolipoprotein B3500 mutation detection using real-time fluorescence PCR and melting curves. Clin Chem 1999; 45: 1094– 1097. 13 Nauck MS, Gierens H, Nauck MA, Marz W, Wieland H. Rapid genotyping of human platelet antigen 1 (HPA-1) with fluorophore-labelled hybridization probes on the LightCycler. Br J Haematol 1999; 27: 234–238. 14 Sham PC, Curtis D. Monte-Carlo tests for association between disease and alleles at highly polymorphic loci. Ann Hum Genet 1995; 59: 97–105. 15 Craddock N, Owen M, Burge S, Kurian B, Thomas P, McGuffin P. Familial cosegregation of major affective disorder and Darier’s disease (keratosis follicularis). Br J Psychiatry 1994; 164: 355–358. 16 Wakem P, Ikeda S, Haake A, Polakowska R, Ewing N, Sarret Y et al. Localization of the Darier disease gene to a 2-cM portion of 12q23–24.1. J Invest Dermatol 1996; 106: 365–367. 17 Sakuntabhai A, Ruiz-Perez V, Carter S, Jacobsen N, Burge S, Monk S et al. Mutations in ATP2A2, encoding a Ca2+ pump, cause Darier disease. Nat Genet 1999; 21: 271–277. 18 Dawson E, Parfitt E, Roberts Q, Daniels J, Lim L, Sham P et al. Linkage studies of bipolar disorder in the region of the Darier’s disease gene on chromosome 12q23–24.1. Am J Med Genet 1995; 60: 94–102. 19 Jacobsen NJO, Franks EKE, Owen MJ, Craddock NJ. Mutational analysis of phospholipase A2A: a positional candidate susceptibility gene for bipolar disorder. Mol Psychiatry 1999; 4: 274–279. 20 Kimura M, Furukawa T, Abe T, Yatsuoka T, Youssef EM, Yokoyama T et al. Identification of two common regions of allelic loss in chromosome arm 12q in human pancreatic cancer. Cancer Res 1998; 58: 2456–2460. 21 Passik SD, Breitbart WS. Depression in patients with pancreatic carcinoma. Diagnostic and treatment issues. Cancer 1996; 78: 615–626. 22 Kennedy JL, Bradwejn J, Koszycki D, King N, Crowe R, Vincent J et al. Investigation of cholecystokinin system genes in panic disorder. Mol Psychiatry 1999; 4: 284–285. 23 Skogen B, Bellissimo DB, Hessner MJ, Santoso S, Aster RH, Newman PJ et al. Rapid determination of platelet alloantigen genotypes by polymerase chain reaction using allele-specific primers. Transfusion 1994; 34: 955–960. 24 Wang DG, Fan JB, Siao CJ, Berno A, Young P, Sapolsky R et al. Large-scale identification, mapping, and genotyping of singlenucleotide polymorphisms in the human genome. Science 1998; 280: 1077–1082.