Molecular Psychiatry (2002) 7, 104–109  2002 Nature Publishing Group All rights reserved 1359-4184/02 $15.00 www.nature.com/mp ORIGINAL RESEARCH ARTICLE Identification of genomic organisation, sequence variants and analysis of the role of the human dishevelled 1 gene in late onset Alzheimer’s disease C Russ, S Lovestone and JF Powell

Department of Neuroscience, Institute of Psychiatry, London SE5 8AF, UK

Keywords: polymorphism; genomic; DVL1; association; kinases, such as c-Jun N-terminal kinase (JNK), and SNP glycogen synthase kinase 3 beta (GSK3␤) have been Alzheimer’s disease (AD) is a disorder characterised by shown to phosphorylate tau in vitro, in cell at sites that a progressive deterioration in memory and other cogni- are phosphorylated in PHF.4–7 JNK and GSK3␤ are tive functions. Neurofibrillary tangles (NFT) are a major regulated by the dishevelled 1 gene (DVL1), the latter pathological hallmark of AD, these are aggregations of as part of the wnt signalling pathways.8–11 Activation paired helical filaments (PHF) comprised of the hyper- of the wnt signalling pathway is thought to cause DVL1 phosphorylated microtubule associated protein tau. to inactivate GSK3␤ through complex-formation with Several kinases, such as glycogen synthase kinase 3 ␤ ␤ adenomatous poliposis coli (APC), axin and -catenin beta (GSK3 ) and c-Jun N-terminal kinase (JNK), phos- proteins,8 whereas JNK is activated by DVL1.10,11 The phorylate tau at sites that are phosphorylated in PHF. Dishevelled 1 (DVL1) is thought to act as a positive mechanism by which this JNK activation occurs is regulator of the wnt signalling pathway, and inhibits unclear, however, it has been demonstrated that a DEP GSK3␤ activity preventing ␤-catenin degradation and (dishevelled, egl-10 and pleckstrin) protein binding thus allowing wnt target gene expression. JNK acti- domain found at the C-terminal of DVL1 is essential vation is also regulated by DVL1, however it is unclear for JNK activation.10 It has been suggested that the wnt if this is via the wnt signalling pathway. These obser- signalling pathway is altered in AD,8,9 therefore DVL1 vations suggest a central role for DVL1 in tau phos- is a very strong candidate gene for AD. Furthermore a phorylation and AD and led us to investigate DVL1 as recent genome scan has shown suggestive linkage (LOD a candidate gene for this disorder. We determined the score 1.33, markers D1S548 and D1S1592) to the distal genomic structure of the DVL1 gene by sequencing and end of chromosome 1 p to AD, where DVL1 is data mining and searched for sequence variations in the located.12 coding sequences and flanking introns. The DVL1 gene 13 spans a region of approximately 13.8 kb (not including DVL1 is a cytoplasmic phosphoprotein encoding a the 5Ј untranslated region) and is encoded by 15 exons. 670-amino acid polypeptide mapping to chromosome Analysis of over 4.3 kb of sequence, including 98% of 1p36.14 In Drosophila it regulates cell proliferation, act- exonic sequences and introns 2, 3, 6, 7, 9, 10, 11 and 12, ing as a transducer molecule for developmental pro- revealed there to be six rare (Յ6%) sequence variations. cesses, including segmentation and neuroblast speci- None of these had any association with late onset AD. fication.14 DVL1 is part of a multi-gene family of This would suggest that polymorphic variations in the proteins that contain three highly conserved species of coding sequences of DVL1 are not important in AD. protein binding domains: an N-terminal DIX However further analysis of regulatory regions may lead (dishevelled and axin) domain; a central PDZ (PSD-95, to the identification of other sequence variations which may be implicated in AD. discs-large and ZO-1) domain; and a DEP domain. Molecular Psychiatry (2002) 7, 104–109. DOI: 10.1038/ DVL1 is highly expressed in both adult and fetal sj/mp/4000941 tissues such as skeletal muscle and pancreas but is also found in the brain and neural tube.14 In contrast to JNK Late onset Alzheimer’s disease (AD) is the most com- activation, all three protein binding domains are mon form of progressive neurodegenerative dementia required for GSK3␤ inactivation and thus ␤-catenin afflicting 5–10% of the population aged over 65 years.1 upregulation.10,15 AD is characterised by two major neuropathological A chromosomal deletion that includes the DVL1 features, neuritic plaques and neurofibrillary tangles gene causes severe morphological abnormalities in (NFT). NFT are intraneuronal inclusions of paired heli- children.16 However knockout mice lacking DVL1 are cal filaments (PHF), mostly comprised of highly phos- viable, fertile, and structurally normal but show abnor- phorylated microtubule associated protein tau. Tau mal social behaviour and several neurological promotes microtubule (MT) assemble in vitro2 and defects.17 Recent studies demonstrated that over- stabilises MT in vivo3 and it is thought that phos- expression of mouse DVL1 causes inhibition of GSK3␤ phorylation inhibits these processes and leads to PHF mediated tau phosphorylation.18 Furthermore Strovel formation. Several mitogen activated protein (MAP) et al19 demonstrated the inability of producing stable Role of DVL1 in late onset AD C Russ et al 105 cell lines overexpressing DVL1. They showed that We examined each SNP in larger sample populations DVL1 causes cell death due to the induction of of ADs and controls to determine whether these alter- apoptosis in an APC-dependent manner. This led us ations were potential disease-related or non-patho- to hypothesise that a subtle variation in the coding or genic. Statistical analysis showed that all of the poly- regulatory region of DVL1 may have an influence later morphisms examined are in Hardy–Weinberg in life and in AD. equilibrium. Haplotype analysis showed that SNPs We therefore sought to sequence the DVL1 gene 98876 G/A and 102296 C/T were in strong linkage dis- including some of its introns in a series of AD cases equilibrium (DЈ = 1.00, df = 3, P Ͻ 10−7). When compar- and controls to identify polymorphisms and then ing ADs to controls no statistically significant associ- investigate these in AD. ation of any SNP was found for individual alleles (see We systematically sequenced 1979 bp of the DVL1 Table 3) and genotypes (data not shown). coding sequence and 2337 bp of the non-coding As the APOE ⑀4 allele has repeatedly been associated sequence in 24 control and 24 AD samples to identify with risk for AD, this association should be seen in our differences between the two populations. Sequencing Caucasian AD case-control sample. Indeed, statistical and BLAST20 analysis of the cDNA (Acc. No. analysis of the sample showed there to be a highly sig- AF006011) sequence against the NCBI (National Centre nificant association between APOE ⑀4 allele frequency for Biotechnology Information) htgs (unfinished high and AD (0.36) vs controls (0.16) (P = 0.001; OR = 3.02, throughput genomic sequences) nucleotide database, CI 1.44–6.40).22,23 As all SNPs were found to be rare led us to determine the genomic structure of the DVL1 (Յ6%), we did not perform any further statistical gene (see Figure 1 and Table 1). DVL1 is encoded by analysis of the sample, such as stratification by APOE 15 exons similar to that of the mouse DVL1 gene21 rang- ⑀4 status, sex, age of onset and ‘probable’ or ‘poss- ing in size from 68 bp (exon 10) to 374 bp (exon 15) ible’ AD. spanning a region of approximately 13.8 kb (not To ensure that our analyses were specific to DVL1 including the 5Ј untranslated region) with the largest we performed radiation hybrid mapping24 (data not and smallest introns being 6139 bp (intron 1) and 69 bp shown) using several primer pairs. This revealed link- (intron 7) respectively. We fully sequenced exons 2–14 age of our PCR fragments to chromosome 1p36 with and partially sequenced exons 1 and 15. Due to sequen- LOD scores greater than 7, therefore confirming the cing primer design constraints, the first 21 bp of exon specificity of our investigation. No other chromosomal 1 and the last 12 bp of exon 15 were not sequenced. location showed any significant linkage. We also fully sequenced introns 2, 3, 6, 7, 9, 10, 11, 12 In this study we have elucidated the genomic organ- and partially sequenced other introns. The other isation of the human DVL1 gene and have investigated introns and the 5Ј and 3Ј untranslated region were not the frequency of polymorphisms in the DVL1 gene and sequenced due to time constraint. The coding sequence analysed their possible involvement in AD. Statistical we elucidated is identical to the Semenov sequence analysis showed there to be no significant association (Acc.No. AAB65242). The DIX domain is located in between any of the detected polymorphisms and AD. exons 1–3, the PDZ domain is in exons 7–10 and the However, control populations are an intrinsic problem DEP domain is encoded by exons 12 and 13. in AD genetic studies and this is as true for the control Sequencing analysis detected six single nucleotide group we used as for those used in other studies. We polymorphisms (SNPs). We identified a rare silent included only controls with no evidence of cognitive coding SNP in exon 1 at codon 22 position nt 69 C/T impairment (MMSE Ͼ 25) but it remains possible that (relative to Acc.No. AF006011) with rare allele fre- we included a few individuals in the early, perhaps quency of 1%. Furthermore we identified five rare non- presymptomatic, stages of dementia. However it is coding SNPs at positions nt 98876 G/A (intron 2), nt unlikely that this would have affected our results. As 99188 C/A (intron 3), nt 99890 G/A (intron 5), nt we have investigated effectively 96 chromosomes for 101121 A/T (intron 6) and 102296 C/T (intron 11) with the DVL1 gene, we are confident that we have detected rare allele frequencies of Յ6% (relative to Acc. No. all common SNPs present in regions examined in the AL139287.6) (see Table 2). Caucasian population. The power in this study for

Figure 1 Graphical representation of genomic organisation of DVL1 gene (not to scale). Exon/intron organisation, sequencing primer pairs and SNPs are shown. SNP positions are relative to Acc. No. AL139287.6. Table 1 shows the sizes of exons and introns.

Molecular Psychiatry Role of DVL1 in late onset AD C Russ et al 106 % % % % % % % % % % % % % % % ends of each Ј and 3 sequence Ј Ј CTCGTGGCGCATGCCCCAGC ‡ 247 GTACAGTGGTCATGTCTGCC 225 GTGTGTCTGTGGGGGCTGCG 6139 GTCAGTCGGCCGCGAGGCCT 1163 GTGTGTGGGAGTGTCTGTGG 1461 GTGAGAACGCCGCCCCTACC + + + + + UTR* 863 Ј ends of each intron, three amino acids of the 5 sequence (bp) 5 Ј Ј GAC TTC GG 1 GTC TCC TGGCGC GAG GAG 2 GGCG GAC 4 CGG 204 GTGAGCCCCTGTGTCTTGGC 6 93 GTAGGCAGCCGGGTGGAGCA TCC TTC CA 3CTC AAC ATG G 206TTG CTG CAG GTAAGGACACCGTGGGCGCT 7 8 69 GTGAGGCCTCGGGGCGAGGG ATC ATG TGA 5 ACG AGC AG 5 GTC CCA CGG GCCC CGC TAC GGCC 10 GTC ATC G 11 142 12 GTGAGTGGCCCTCGGGGCAC 708 GTACGAGTCCCTGCTCCAGC 85 GTGAGTGGCCGTCCCCGCGG CAG ACG GG 9CTC TGC AGC ACAG 86 AGT GAA G GTGAGGGCTGGGGGCGCTGG 13 14 85 GTGAGTGGGGGCTTCGGGCT and 3 % % % % % % % % % % % % % % % Ј 70 23 94 31 70 23 6878 22 26 77 26 170 57 104 34 122 41 140 47 374 124 139 47 132 44 168207 56 69 % % % % % % % % % % % % % % % sequence (bp) amino acids 3 Ј PNV REE VVKLSS VSW ADR PIS VPR MAELVLASS DFG SFQ LNM VND QTG NLA QSE GSK IM– GAD LCS ERHELE LLQ AVI AA R T SR ADP PRY 1 ATG GGC GAG † Genomic organisation of the human dishevelled 1 gene, showing 20 bp of the 5 untranslated region (UTR) size not determined. untranslated region. Ј Ј sequence 5 5 3 CTCCGCGCCGCCGCCGCGCC ACGGCAGCATATGCCCACAGCATGCCTGCTCCCTCCACAG 2ACCCTGCTCTTGTTGCACAG G 3 GTG GTG AAG CTG 4 GTC CTG CTCTGACAGTCTGTTCCCAG C CCG AATACGCCCCCTCTCCTCCATAG GTG 6GCTCATGGCTCTTCCTGCAG G 7 CTC AGC AGC GCC 8 TCC TCC AA AGA CAT CATCCTGTTCTCACTCCCAG 15 GG AGC AAA CCAACCTGCTGCCCCCACAG 5 CC GCC CGG GGGCTGTCTGTGCTTTGCAG 12 AG CTG GAA CCTAATCCCCACCCCTGCAGTGTGACCACCGTTCCTGCAG 9GCCACGTGGCGGCCACCCAG 10 GTG AAT GAC G 11 CCC ATC AGC TGACCGCTGGTGCCGGCCAG CT GAC CCG CACCTACCCTCCCCCGCCAG 13 GG 14 GCG GAC AT CTC GCC Ј Size determined from genomic clone (Acc. No. AL139287.6). *Size determined from mRNA† (Acc. No. AF006011). Table 1 exon, and size of the exons and introns Intron3 % Exon Exon Length Number of Exon Intron Length+ Intron ‡ % % % % % % % % % % % % % %

Molecular Psychiatry Role of DVL1 in late onset AD C Russ et al 107 Table 2 Details of PCR/sequencing primers and SNPs located within their products. Positions of primers and SNPs relative to AL139287.6. The start codon is at nt 92391 and the stop codon is at nt 105313

Primer Primer sequence 5Ј–3Ј Position relative to PCR SNP relative to name AL139287.6 (bp) AL139287.6

1F TGGGCGAGACCAAGATTATC 92392–92411 439 C to T* 1R AGTCTGACACGGGCCAAG 92830–92813 92456 2F GGTGTCCACAGAGTGACATTG 98583–98603 396 N/A 2R GACCAGCTGTGGAGGGAG 98978–98961 3F TGGTGAAGGAGGAGATCTTTG 98701–98721 362 G to A 3R GCCTGTCCGCTCAAGAGG 99062–99045 98876 4F CTGGTCCTGGCTGAGGGT 98973–98990 432 C to A 4R CCTCCTCGCGGTTCCGGC 99404–99387 99188 5F CGAATGTGGCCAGCAGCC 99303–99320 129 N/A 5R ACAGCTGGGCAGACATGAC 99431–99413 6F CCCATGACCTCTGGGAGAC 99599–99617 388 G to A 6R CACAGTGAAAACACCTCATGTG 99986–99965 99890 7F CTGAGCTCTGACAGTCTGTTCC 100929–100950 117 N/A 7R GGTCCGCCTGCCGAAGGC 101045–101028 8F AGGCTCAGCAGCTCCACG 100952–100969 213 A to T 8R GGTTATGCTGCTGAAGGAGG 101164–101145 101121 9F AAGTCAGTGTCGGGCAGC 101087–101104 218 A to T 9R CTGATGCCCAGAAAGTGATG 101304–101285 101121 10F GGCTCATGGCTCTTCCTG 101259–101276 202 N/A 10R ACCCGCAGCTACATGTCAC 101460–101442 11F ACTGAGCTGCACATCCTGC 101597–101615 275 N/A 11R TGGGACGGTGAAGTAGCTTC 101871–101852 12F AGCCTCACTGTGGCCAAGT 101815–101833 271 N/A 12R CAGGGCTCCTGTCAGTGC 102085–102068 13F CGCCTACCTAGCCTGAAGG 101918–101936 273 N/A 13R CCTCAGGAGCTGCCACTTAC 102190–102171 14F CACTGACAGGAGCCCTGC 102069–102086 858 C to T 14R CATTGGCGATGGTGATCTTG 102926–102907 102296 15F GCAGAGCTGGAAGAGGCG 102800–102817 387 N/A 15R CTGCAGAGATCCCCGAAGAC 103186–103167 16F GGACGTGGTGGACTGGCT 103025–103042 242 N/A 16R GGGAGGGTAGGTGAGGGC 103266–103249 17F GGCCCTCACCTACCCTCC 103248–103265 302 N/A 17R GAGCAGCAGTGGAGTGGG 103549–103532 18F CTTCCCCATCCTGTTCTCAC 104916–104935 408 N/A 18R GCCACGAGTCACATGATGTC 105323–105304

*Coding silent SNP. detecting SNPs with frequencies of 10% is Ͼ95% and polymorphic variation we have detected in the DVL1 for detecting SNPs with frequencies of 5% is slightly gene. The possible disruption of the wnt signalling less at ෂ90%, based on the number of chromosomes pathway seen in AD could be due to polymorphism in examined. Future analysis may reveal further SNPs in other genes in this signalling pathway. We have the promoter, intron 1 and other introns. The number recently shown that polymorphic variations in the of SNPs detected in non-coding regions examined was GSK3␤ gene and its promoter are also not associated one SNP per 511 bp, corresponding to a nucleotide with AD.23 diversity of ␪ = 0.000453 ± 0.00011. We also detected one synonymous SNP per 1979 bp coding region Methods screened, corresponding to a nucleotide diversity of ␪ = 0.000105 ± 0.000022. Although these values are Subjects lower than average nucleotide diversities previously All subjects used were Caucasians. The test samples described,25–27 they are consistent with the nucleotide consisted of: 52 community-based late onset AD cases diversity values for some of the genes investigated in over the age of 60 years (mean age of onset 76 (SD = these studies.25,27 7.2)) who fulfilled the NINCDS-ADRDA criteria28 for The pathological changes seen in AD, such as JNK ‘probable’ (70%) and ‘possible’ (30%) AD, recruited and GSK3␤ mediated phosphorylation of tau and the from the Camberwell Dementia Case Register;22 49 disruption of ␤-catenin turnover, which are probably community dwelling controls aged over 75 years (mean very relevant to the pathogenesis of AD, are not due to age 80.1 (SD = 3.8)) who scored 25 points or above on

Molecular Psychiatry Role of DVL1 in late onset AD C Russ et al 108 Table 3 Details of SNP frequencies in AD and controls. RFLP analyses of SNPs showing name and position of SNP, restriction enzyme used and restriction fragments created for each allele. Association statistics of alleles are also shown. Positions of SNPs are relative to AL139287.6 (AUG at nt 92391)

SNPs SNPs Frequency of SNPs Restriction Common Rare Association relative to relative to enzyme allele RFLP allele RFLP statistics of AL139287.6 exons/intronsAD Controls products products alleles (bp) (bp)

C to T Exon 1 C = 0.99 (69) C = 0.99 (97) BtgI C T P = 0.80 92456 codon 22, T = 0.01 (1) T = 0.01 (1) (64, 87, (88, 151, 300) + 66 bp from 88, 200) 5Ј end G to A Intron 2, G = 0.96 (81) G = 1.00 (96) HpyCH4V G A P = 0.10 98876 + 108 bp from A = 0.04 (3) (362) (174, 188) 5Ј end C to A Intron 3, C = 0.98 (104) C = 0.98 (63) Cac8I C A P = 0.68 99188 + 94 bp from T = 0.02 (2) T = 0.02 (1) (18, 21,24, (21, 24, 65, 5Ј end 65, 124, 180) 142, 180) G to A Intron 5, G = 0.94 (77) G = 0.94 (83) HpyCH4IV G A P = 0.58 99890 + 100 bp from A = 0.06 (5) A = 0.06 (5) (14, 29, (14, 29, 345) 5Ј end 55, 290) A to T Intron 6, A = 0.99 (103) A = 0.99 (93) SSCP N/A N/A P = 0.73 101121 + 74 bp from T = 0.01 (1) T = 0.01 (1) (12% gel) 5Ј end C to T Intron 11, C = 0.96 (79) C = 1.00 (92) Tsp509I C T P = 0.10 102296 + 201 bp from, T = 0.04 (3) (858) (225,633) 5Ј end

the Mini Mental State Examination with no evidence phism (RFLP) or single stranded conformational poly- of cognitive impairment, recruited from general prac- morphism (SSCP) analysis in larger cohorts of AD and tices in the UK (in collaboration with the investigators control samples. SSCP analysis used 12% non-denatur- of the MRC trial of assessment and management of eld- ing PAGE at a constant 10 V cm−1 for 14 h at 4°C. Prior erly people in the community). to loading, samples were mixed 1:1 with formamide loading buffer and denatured at 95°C for 10 min and PCR and sequencing analysis chilled on ice. Details of individual RFLP analyses are We amplified approximately a total of 4.3 kb of DVL1 shown in Table 3. genomic sequence from lymphocyte-derived genomic DNA. Primers (MWG-Biotech, Milton Keynes, UK) Statistics used are described in Table 2. PCR reactions were per- Linkage disequilibrium was calculated using the EH formed in a final volume of 40 ␮l containing 50 ng program.33 Values of DЈ, the normalised linkage dis- DNA, 0.3 ␮M each primer and 1 U Taq DNA poly- equilibrium statistic were calculated using published merase (Hybaid, UK). The PCR conditions consisted of methods.34 DЈ values are between 0 and 1 with larger 35 cycles of 45 s at 94°C, 45 s at 50–60°C and 45 s at values suggesting stronger linkage disequilibrium. 72°C, preceded by 5 min at 95°C and followed by Statcalc35 was used for power calculations and for the 10 min at 72°C. Products for sequencing were then pur- comparison allelic distribution of SNPs between con- ified using the Hybaid DNA purification Kit II (Hybaid, trols and ADs. Nucleotide diversity (␪) was calculated Ashford, UK). Twenty-four controls and 24 ‘probable’ as described by Halushka et al.25 AD samples were sequenced for each of the 15 exons and introns 2, 3, 6, 7, 9, 10, 11, and 12. PCR products Acknowledgements were sequenced in both directions using the Big Dye Terminator sequencing kit (Applied Biosystems, War- This study was supported by the Wellcome Trust rington, UK) and relevant PCR primers. Sequencing (049542/Z). The authors wish to thank Dr Clive Holmes was performed on an ABI377 automated sequencer. for the use of the clinical and genotypic data for the The sequence data were reanalysed by the Phred pro- Camberwell Dementia Case Register. Samples from gram,29,30 assembled by the Phrap program, possible participants in the MRC trial of assessment and man- polymorphic bases identified by the PolyPhred pro- agement of elderly people in the community (MRC Eld- gram31 and traces viewed using the program Consed.32 erly Study) were collected in collaboration with the Possible SNPs detected by sequencing were further trial investigators: Professor Astrid Fletcher, London investigated by restriction fragment length polymor- School of Hygiene & Tropical Medicine (Principal

Molecular Psychiatry Role of DVL1 in late onset AD C Russ et al 109 investigator) and co-investigators Dr Dee Jones, Univer- 17 Lijam N, Paylor R, McDonald MP, Crawley JN, Deng CX, Herrup K sity of Wales College of Medicine and Professor Chris et al. Social interaction and sensorimotor gating abnormalities in mice lacking Dvl1. Cell 1997; 90:895–905. Bulpitt, Imperial College. We also thank Aaron Jeffries 18 Wagner U, Brownless J, Irving NG, Lucas FR, Salinas PC, Miller for his Unix skills. CC. Overexpression of the mouse dishevelled-1 protein inhibits GSK-3beta-mediated phosphorylation of tau in transfected mam- malian cells. FEBS Lett 1997; 411:369–372. References 19 Strovel ET, Sussman DJ. Transient overexpression of murine dis- hevelled genes results in apoptotic cell death. Exp Cell Res 1999; 1 Clark R, Goate A. Molecular genetics of Alzheimer’s disease. Arch 253:637–648. Neurol 1993; 50: 1164–1172. 20 Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local 2 Kanai Y, Takemura R, Oshima T, Mori H, Ihara Y, Yanagisawa M. alignment search tool. J Mol Biol 1990; 215:403–410. Expression of multiple tau isoforms and microtubule bundle for- 21 Lijam N, Sussman DJ. Organization and promoter analysis of the mation in fibroblasts transfected with a single tau cDNA. J Cell Biol mouse dishevelled-1 gene. Genome Res 1995; 5: 116–124. 1989; 109: 1173–1184. 22 Holmes C. The Camberwell Dementia Case Register. Int J Geriatr 3 Drubin DG, Kirschner MW. Tau protein function in living cells. J Psychiatry 1996; 11:369–375. Cell Biol 1986; 103: 2739–2746. 23 Russ C, Lovestone S, Powell JF. Identification of sequence variants 4 Reynolds CH, Betts JC, Blackstock WP, Nebreda AR, Anderton BH. and analysis of the role of the glycogen synthase kinase 3b gene and Phosphorylation sites on tau identified by nanoelectrospray mass promoter in late onset Alzheimer’s disease. Mol Psychiatry 2001; 6: spectrometry: differences in vitro between the mitogen-activated 320–324. protein kinases ERK2, c-Jun N-terminal kinase and P38, and glyco- 24 Gyapay G, Schmitt K, Fizames C, Jones H, Vega-Czarny N, Spillett gen synthase kinase-3beta. J Neurochem 2000; 74: 1587–1595. D et al. A radiation hybrid map of the human genome. Hum Mol 5 Lovestone S, Reynolds CH, Latimer D, Davis DR, Anderton BH, Genet 1996; 5:339–346. Gallo JM et al. Alzheimer’s disease-like phosphorylation of the 25 Halushka MK, Fan JB, Bentley K, Hsie L, Shen N, Weder A et al. microtubule-associated protein tau by glycogen synthase kinase-3 Patterns of single-nucleotide polymorphisms in candidate genes for in transfected mammalian cells. Curr Biol 1994; 4: 1077–1086. blood-pressure homeostasis. Nat Genet 1999; 22:239–247. 6 Anderton BH, Brion JP, Couck AM, Davis DR, Gallo JM, Hanger 26 Nickerson DA, Taylor SL, Weiss KM, Clark AG, Hutchinson RG, DP et al. Modulation of PHF-like tau phosphorylation in cultured Stengard J et al. DNA sequence diversity in a 9.7-kb region of the neurones and transfected cells. Neurobiol Aging 1995; 16:389–397; human lipoprotein lipase gene. Nat Genet 1998; 19: 233–240. discussion 398–402. 27 Cambien F, Poirier O, Nicaud V, Herrmann SM, Mallet C, Ricard 7 Sperber BR, Leight S, Goedert M, Lee VM. Glycogen synthase kin- S et al. Sequence diversity in 36 candidate genes for cardiovascular ase-3 beta phosphorylates tau protein at multiple sites in intact disorders. Am J Hum Genet 1999; 65: 183–191. cells. Neurosci Lett 1995; 197:149–153. 28 McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan 8 De Ferrari GV, Inestrosa NC. Wnt signaling function in Alzheimer’s EM. Clinical diagnosis of Alzheimer’s disease: report of the disease. Brain Res Brain Res Rev 2000; 33:1–12. NINCDS-ADRDA Work Group under the auspices of Department of 9 Anderton BH, Dayanandan R, Killick R, Lovestone S. Does dysreg- Health and Human Services Task Force on Alzheimer’s Disease. ulation of the Notch and wingless/Wnt pathways underlie the Neurology 1984; 34: 939–944. pathogenesis of Alzheimer’s disease? Mol Med Today 2000; 6: 29 Ewing B, Green P. Base-calling of automated sequencer traces using 54–59. phred. II. Error probabilities. Genome Res 1998; 8: 186–194. 10 Li L, Yuan H, Xie W, Mao J, Caruso AM, McMahon A et al. Dishev- 30 Ewing B, Hillier L, Wendl MC, Green P. Base-calling of automated elled proteins lead to two signaling pathways. Regulation of LEF- sequencer traces using phred. I. Accuracy assessment. Genome Res 1 and c-Jun N-terminal kinase in mammalian cells. J Biol Chem 1998; 8:175–185. 1999; 274:129–134. 31 Nickerson DA, Tobe VO, Taylor SL. PolyPhred: automating the 11 Boutros M, Paricio N, Strutt DI, Mlodzik M. Dishevelled activates detection and genotyping of single nucleotide substitutions using JNK and discriminates between JNK pathways in planar polarity fluorescence-based resequencing. Nucleic Acids Res 1997; 25: and wingless signaling. Cell 1998; 94:109–118. 2745–2751. 12 Kehoe P, Wavrant-De Vrieze F, Crook R, Wu WS, Holmans P, Fen- 32 Gordon D, Abajian C, Green P. Consed: a graphical tool for ton I et al. A full genome scan for late onset Alzheimer’s disease. sequence finishing. Genome Res 1998; 8:195–202. Hum Mol Genet 1999; 8: 237–245. 33 Terwilliger JD, Ott J. Linkage disequilibrium between alleles at 13 Klingensmith J, Nusse R, Perrimon N. The Drosophila segment marker loci. In: Terwilliger JD, Ott J (eds). Handbook of Human polarity gene dishevelled encodes a novel protein required for Genetic Linkage, 1st edn. Johns Hopkins University Press: Balti- response to the wingless signal. Genes Dev 1994; 8: 118–130. more, 1994, pp 188–198. 14 Pizzuti A, Amati F, Calabrese G, Mari A, Colosimo A, Silani V et al. 34 Klitz W, Stephens JC, Grote M, Carrington M. Discordant patterns cDNA characterization and chromosomal mapping of two human of linkage disequilibrium of the peptide-transporter loci within the homologues of the Drosophila dishevelled polarity gene. Hum Mol HLA class II region. Am J Hum Genet 1995; 57: 1436–1444. Genet 1996; 5:953–958. 35 Lee AJ, McInerney PJ, Mullins PR. STATCALC: an integrated stat- 15 Kishida S, Yamamoto H, Hino S, Ikeda S, Kishida M, Kikuchi A. istics system for the Apple II microcomputer. Comput Programs DIX domains of Dvl and axin are necessary for protein interactions Biomed 1984; 18:265–272. and their ability to regulate beta-catenin stability. Mol Cell Biol 1999; 19:4414–4422. Correspondence: JF Powell, Department of Neuroscience, Institute of 16 Bedell JA, Wagner-McPherson CB, Bengtsson U, Handa K, Dumars Psychiatry, De Crespigny Park, London SE5 8AF, UK. E-mail: john. KW, Marsh JL et al. A 1p deletion syndrome patient is hemizygous powellȰiop.kcl.ac.uk for a human homologue of the Drosophila dishevelled gene Received 24 February 2001; revised 2 April 2001; accepted 5 April (Abstract). Am J Hum Genet 1996; 59: A298. 2001

Molecular Psychiatry