Novel Polymorphisms in the Somatostatin Receptor 5 (SSTR5

Home , Somatostatin receptor 5

Molecular Psychiatry (2002) 7, 745–754  2002 Nature Publishing Group All rights reserved 1359-4184/02 $25.00 www.nature.com/mp ORIGINAL RESEARCH ARTICLE Novel polymorphisms in the somatostatin receptor 5 (SSTR5) gene associated with bipolar affective disorder M Nyegaard1, AD Børglum2, TG Bruun3, DA Collier4, C Russ4, O Mors5, H Ewald5 and TA Kruse1

1Department of Clinical Biochemistry and Genetics, Odense University Hospital, Denmark; 2Institute of Human Genetics, Aarhus University, Denmark; 3Mood Disorders Research Unit, Psychiatric Hospital in Aarhus, Denmark; 4Department of Psychological Medicine, Institute of Psychiatry, London, UK; 5Institute for Basic Psychiatric Research, Psychiatric Hospital in Aarhus, Denmark

The somatostatin receptor 5 (SSTR5) gene is a candidate gene for bipolar affective disorder (BPAD) as well as for other neuropsychiatric disorders. The gene is positioned on chromosome 16p13.3, a region that has been implicated by a few linkage studies to potentially harbor a disease susceptibility gene for BPAD. Recent evidence shows that the dopamine D2 receptor (DRD2) and SSTR5 interact physically to form heterodimers with enhanced functional activity. Brain D2 dopamine receptors are one of the major targets of neuroleptic treatments in psychiatric disorders. In this study we systematically screened the promoter and coding region of the SSTR5 gene for genetic variation that could contribute to the development of neuropsychiatric disorders. Eleven novel single nucleotide polymorphisms (SNPs) were identified including four missense SNPs, Leu48Met, Ala52Val, Pro109Ser and Pro335Leu. We carried out an association study of BPAD using 80 Danish cases and 144 control subjects, and replication analysis using 55 British cases and 88 control subjects. For the Danish population, association was suggested between silent SNP G573A and BPAD (P = 0.008). For the British population we found association to BPAD with missense mutation Leu48Met (P = 0.003) and missense mutation Pro335Leu (P = 0.004). The statistical significance of the association was, however, greatly reduced after correcting for multiple testing. When combining genotypes from Leu48Met and Pro335Leu into haplotypes, association to BPAD was found in the British population (P = 0.0007). This haplotype association was not replicated in the Danish population. Our results may indicate that the SSTR5 gene is involved in the etiology of BPAD or may exist in linkage disequilibrium with a susceptibility gene close to SSTR5. However, given the marginal statistical significance and the potential for false-positive results in association studies with candidate genes, further studies are needed to clarify this hypothesis. Molecular Psychiatry (2002) 7, 745–754. doi:10.1038/sj.mp.4001049 Keywords: SSTR5; chromosome 16p13.3; candidate gene; bipolar disorder; DNA mutational analysis; genetic variation; SNP; linkage disequilibrium; haplotype analysis; association study

Introduction in a genome-wide scan in two Costa Rican families with BPAD.4 Furthermore an increased prevalence of In 1995 Ewald and co-workers1 found evidence for the psychiatric diagnosis especially mood disorder was existence of a bipolar susceptibility locus on chromo- found in 13 patients with Rubinstein–Taybi syndrome, some 16p13.3 in two Danish families segregating a multiple congenital anomaly syndrome mapped to bipolar affective disorder (BPAD). The highest lodscore 16p13.3.5 A number of studies have, however, not been (2.52) was obtained for marker D16S510 (9 cM on able to replicate linkage between BPAD and markers on Genethon map)2 when combining the two families and 16p13.3 and some studies have even found evidence assuming a recessive mode of inheritance.1 Positive against linkage to BPAD on 16p13.3.6–12 A number of lodscores were also obtained with surrounding mark- these studies have, however, not reported results ers supporting that a susceptibility locus for bipolar assuming a recessive mode of inheritance.13 The soma- disorder may exist distal on chromosome 16p. Weak tostatin receptor 5 (SSTR5) gene is a positional and support for this finding was found in a study of 97 functional candidate gene for BPAD. The gene has been bipolar families of the NIMH Genetics Initiative,3 and mapped to chromosome band 16p13.3 by in situ hybridization14 and radiation hybrid mapping.15 The SSTR5 mediates the effects of somatostatin, a Correspondence: M Nyegaard, MSc, Department of Clinical Bio- multi-functional neuroendocrine hormone originally chemistry and Genetics, KKA, Odense University Hospital, 5000 Odense C, Denmark. E-mail: mette.nyegaardȰouh.fyns-amt.dk isolated more than 20 years ago from the hypothalamus Received 9 July 2001; revised 9 August 2001; accepted 15 Nov- as a peptide that inhibited the release of growth hor- ember 2001 mone.16 Somatostatin is widely distributed throughout SSTR5 polymorphisms and bipolar disorder M Nyegaard et al 746 the central nervous system, and plays a major role in an association study with BPAD in two case-control neurotransmission, glandular secretion, smooth mus- samples. cle contractility, and cell proliferation.17 Somatostatin is known to function as a neuropeptide with secretion intimately related to the regulation of other classical neuropeptides and neurotransmitters. It increases the Materials and methods turnover of cholinergic and monoaminergic neurotransmitters and additionally modulates their release. Detection of genetic variation In turn, somatostatin release is stimulated by dopa- Five affected individuals with bipolar I disorder mine, and appears to be stimulated by acetylcholine according to DSM-IV24 from two Danish families trans- and inhibited by gamma-amino-butyric acid (GABA).18 mitting BPAD were selected for mutation screening. It has been proposed that the somatostatin system These individuals were IV-20, VI-1, III-25, and V-2 may be involved in neurodegenerative and/or affective from family 4 and III-2 from family 2.1 The affected disorders. In animals a wide spectrum of behavioral individuals were selected from different sub branches and physiological functions is related to somatostatin, of the pedigrees, as genetic heterogeneity and phe- including learning, regulation of slow wave sleep and nocopies may exist even within each pedigree. Positive REM sleep, food consumption, and locomoter linkage was obtained using a recessive mode for dis- activity,18 all of which are often altered in manic or ease transmission, and therefore ‘married-into’ chro- depressed patients. In addition, several investigators mosomes that were transmitted to affecteds were con- have reported a reduction in somatostatin concen- sidered important to screen for variations. Two healthy tration in the cerebrospinal fluid of patient with neur- individuals recruited among laboratory staff were opsychiatric disorders, particularly Alzheimer’s dis- included in addition. The 5Ј-promoter region (2258 bp ease and depression, but also Parkinson’s disease, immediately upstream of the ATG start codon, Gen- senile dementia, multiple sclerosis, and schizo- bank accession No. U18976) and the complete coding phrenia.18,19 region (1092 bp, Genbank accession No. D16827) of the SSTR5 is a member of the somatostatin receptor fam- SSTR5 gene were systematically screened for genetic ily encompassing four other subtypes SSTR1–4. All variation using direct cycle sequencing of genomic receptor subtypes are 7-transmembrane G-protein DNA. For each individual two overlapping fragments coupled receptors having a high degree of sequence (fragments A and B) were amplified from genomic DNA similarity within the 7-membrane spanning alpha- using primers and reaction conditions described helix segment but diverge in their amino- and carboxy- below. Direct cycle sequencing was performed by the termini. The receptors display an overlapping but dideoxy nucleotide chain termination method25 using characteristic pattern of expression that is sub-type and the Dye Termination Cycle Sequencing kit (Applied tissue specific.17 The SST5 receptor is expressed in the Biosystems, Foster City, CA, USA). Sequencing was brain, in the pituitary gland in adults and in the pitu- performed on both strands. itary and hypothalamus in the fetus.20 The SSTR5 gene contains no introns and was cloned in 1993.21,22 Recently, it has been shown that the somatostatin PCR amplification receptor 5 interacts physically with the long form of Fragment A and fragment B, together containing the the dopamine D2 receptor through hetero-oligomeriz- total set of SSTR5 SNPs, were separately amplified ation to create a novel receptor with enhanced func- using PCR. Each PCR reaction mix contained genomic tional activity.23 Hetero-oligomerization leads to syn- DNA (50 ng), 4 pmol of each primer, 0.35 mM dNTP, ergy such that the binding affinity for the second ligand Expand Long Template enzyme mixture (1 U, Roche is increased as a result of receptor occupancy of the Diagnostics, Mannheim, Germany) and buffer system 1 first ligand. This receptor complex, simultaneously supplied by the manufacturer, in a total volume of occupied by two ligands, appears to be the most acce- 10 ␮l. Primer sequences for fragment A were Afor: 5Ј- lerating signaling form. This newly characterized func- tctctggaccttgtgccag (nt pos 1–20 Genbank accession No. tional interaction between dopaminergic and somato- AF152962) and Arev: 5Ј-ggccactgccaggttgagaa (nt pos statinergic neurotransmission increases the interest for 286–305 Genbank accession No. D16827). The primer SSTR5 as a candidate gene for affective disorders and sequences for fragment B were Bfor: 5Ј- other neuropsychiatric disorders, as dopamine D2 atgtgctggttcagggactc (nt pos 2178–2197 Genbank receptor (DRD2) antagonists are currently being used in accession No. AF152962) and Brev: 5Ј- the treatment of manic symptoms and schizophrenia. tgactggcaggtcatgggtg (nt pos 1213–1232 Genbank In this study the following investigations were per- accession No. D16827). PCR was performed using 30 formed: (1) we determined the position of SSTR5 on cycles of 94°C for 45 s, 60°C for 40 s and 68°C for the genetic map and compared this with the region sug- 2 min. After PCR, 6 ␮l of fragment A and 6 ␮l of frag- gested by linkage findings on BPAD; (2) we searched ment B were pooled and purified from unincorporated for single nucleotide polymorphisms (SNPs) in the dNTPs and primers adding 1 U of SAP (USB, Cleve- SSTR5 gene; (3) we performed haplotype analysis in land, OH, USA) and 2 U of ExoI (New England Biolabs, two Danish families transmitting BPAD with sugges- Beverly, MA, USA). The reactions were incubated at tive linkage to region 16p13.3; and (4) we carried out 37°C for 60 min followed by 85°C for 15 min.

Molecular Psychiatry SSTR5 polymorphisms and bipolar disorder M Nyegaard et al 747 Single basepair extension verted to genetic distances in Morgans using the Kos- Primers for single basepair extension (SBE) were ambi mapping function. All distances are sex-aver- designed to terminate on the base 5Ј to the SNP and to aged distances. avoid containing any neighboring SNPs. The length of the primers was adjusted to 18, 22, 24, 26, 30, 32, 34, Association study 36, 38, 42, by adding a random sequence to the 5Ј-end Two case-control samples were included in the study. (Table 1). Multiplex SBE were carried out in a volume The first sample (DK) consisted of 80 Danish Caucasian of 5 ␮l using 2.5 ␮l SNaPshot mix (Applied bipolar patients and 144 ethnically matched controls. Biosystems), 2 ␮l of pooled purified fragment A and The Danish patients were recruited from the Psychi- fragment B, 1 pmol of SBE primer SNP1, SNP8, atric Hospital in Aarhus, and interviewed using the 0.3 pmol of SNP4, SNP5, SNP6, SNP7, SNP9, SNP10, semi-structured interview ‘Schedules for Clinical SNP11, and 0.05 pmol of SNP3. Thirty cycles of 94°C Assessment in Neuropsychiatry’. All cases were diag- for 10 s, 50°C for 8 s, and 60°C for 30 s were performed. nosed as bipolar 1 according to DSM-IV.24 The 144 To remove excess ddNTPs, 1 U of SAP (USB) was Danish control individuals were blood donors assigned added and the reactions were incubated at 37°C for through the blood bank in Odense. They were not 60 min followed by 85°C for 15 min. Samples were interviewed or screened to exclude subjects with a his- assayed by electrophoresis on a ABI377 sequencer tory of psychiatric illness prior to the assessment. The using a 12% denaturing polyacrylamide gel. second sample (UK) consisted of 55 British Caucasian bipolar patients and 88 ethnically matched controls. SNP2 detection The British patients were recruited from three Lithium For detection of SNP2 (1 bp deletion), primer R11 (5Ј- therapy clinics at the Institute of Psychiatry, London, catttggcttctgtctaagg nt pos 196–215 Genbank accession Epsom General Hospital and Northern General Hospi- No. AF152962) was end-labeled using polynucleotide tal, Sheffield. The patients were given a best-estimate kinase (Amersham Pharmacia Biotech, Uppsala, lifetime diagnosis according to DSM-IV criteria on the Sweden) and [␥-33P]ATP. PCR was performed using basis of hospital case-notes and structured interviews primer set Afor/R11 and the fragments (215 vs 214 bp) (SADS-L, SCAN). Control subjects were Caucasian vol- were resolved on a 6% denaturing polyacrylamide gel. unteers from UK recruited from general practices and Autoradiography was performed for 1–2 days at room were screened with a brief semi-structured interview to temperature. ascertain the absence of psychiatric illness. All control individuals scored higher than 25 points in the mini- Genetic fine mapping mental state examination. For linkage mapping, family members from seven informative pedigrees (CEPH pedigree Nos 102, 884, Statistical analysis 1331, 1347, 1362, 1413, and 1416) were genotyped for Chi-square statistics were used to test the null hypoth- SNP2 and SNP11. To increase marker informativity, esis of no difference in the allele frequency for patients genotypes from the two polymorphic sites were com- and controls. The chi-square values and corresponding bined into haplotypes. Multipoint linkage analysis was two-sided P-values were calculated using the program carried out using the program CMAP from the LINK- CONTING v.2.30.27 When expected numbers were less AGE package,26 and data on chromosome 16 markers than 5, Fishers exact test was applied, and the two- from the CEPH database 8.2 (www.cephb.fr/cephdb/). sided P-values were calculated using the program After analysis, recombination frequencies were con- 2BY2 v.1.50.27 For interesting two-locus (SNP6–

Table 1 SNPs identiﬁed in the 5Ј-upstream and coding region of the SSTR5 gene and oligonucleotide primers used for single base extension (SBE) analysis

Name Region Sequence change Amino acid SBE primer substitution

SNP1 5Ј-upstream T-2187G 5Ј-ccattcaccaattctctc SNP2 5Ј-upstream −2134delT SNP3 5Ј-upstream A-1665G 5Ј-actagtgctctgaggtgcctcc SNP4 5Ј-upstream C-801G 5Ј-cttaggtagtcacacagggcgaagga SNP5 5Ј-upstream T-461C 5Ј-aatcgccatctctctccactgcca SNP6 Coding (TMI) C142A Leu48Met 5Ј-aattgtcaggtgctggtgcccgtgctgtac SNP7 Coding (TMI) C155T Ala52Val 5Ј-gacccttaatcgtgttcccgcccagcccggcc SNP8 Coding (ECLI) C325T Pro109Ser 5Ј-caggtcattacgcgcgtccttctggcccttcggc SNP9 Coding G573A 5Ј-gttaagagtcatccgtacctgcaacgccagctggcc SNP10 Coding C633T 5Ј-ctaggaaaggggtatgacacggccgtgctgggcttctt SNP11 Coding (C-term) C1004T Pro335Leu 5Ј-atgctcagacacaataagcgaaggacgctgacgccacggagc

Each SNP was assigned a nucleotide position relatively to the ﬁrst base in the coding sequence (position 1).

Molecular Psychiatry SSTR5 polymorphisms and bipolar disorder M Nyegaard et al 748 SNP11) haplotype, CLUMP v.1.6 was used to estimate matches in nucleotide sequences using a library of 280 significance.28 CONTING, 2BY2, and CLUMP were known transcription factor binding sites. SNP3 and found at URL:http://linkage.rockefeller.edu/soft/ (2BY2 SNP5 were located in the core sequences of two differ- and CONTING were found under LINKAGE UTILITY ent promoter element-binding sites, while no known PROGRAMS). promoter element sites were predicted to include SNP1, SNP2, and SNP4. The T-allele (allele 1) of SNP3 Linkage disequilibrium and Hardy–Weinberg (A-1665G)/(T-1665C) (+strand/ −strand) interrupts the equilibrium core sequence (CGGA) of the CETS1P54 binding site The degree of linkage disequilibrium between pairs of on the anti-sense (−) strand. The C-allele (allele 2) of SNPs was calculated for the Danish control sample SNP5 (T-461C) was found to interrupt the core (144 subjects). For each set of SNPs, genotypes were sequence (CCAT) of the YY1 binding site on the sense combined into haplotypes and haplotype frequencies (+) strand. were calculated using the program EH v.1.14 (URL: http://linkage.rockefeller.edu/soft/).27 From haplotype Genetic fine mapping frequencies, the degree of linkage disequilibrium (D) A reference map of 16 markers covering the telomeric and the fraction (DЈ), called the standardized linkage region of chromosome 16 p was used including the fol- disequilibrium, of the maximum possible value of D at lowing six markers: telomere-D16S521—5.0 cM— the given allele frequencies, was calculated.29 We used D16S3024–0.6 cM—D16S3070—0.8 cM—D16S3072— the chi-square values for allelic association, calculated 1.6 cM—D16S510—0.1 cM—D16S423 (Figure 1). The by the EH-program, to determine if the DЈ value was order of this map is supported by odds of at least statistically significant (if the null hypothesis of no 1000:1.2 A CMAP multipoint analysis placed the linkage disequilibrium could be rejected). We used a SSTR5 gene in the 5.0 cM interval between D16S521 significance level of 0.05. Deviation from Hardy–Wein- and D16S3024. A peak location score of 95.6 was seen berg (HW) equilibrium was tested using the program between these two markers, corresponding to a multi- HWI v.2.10 (URL: http://linkage.rockefeller.edu/ point lodscore of 20.8. This position was favored over soft/).27 When expected numbers were less than 5, the the second most likely position (telomeric to D16S521) exact significance probability was used, calculated by by odds 1540:1. All other locations could be ruled out Stata (URL: http://stata.com/). A significance level of by odds greater than 100 000:1. 0.05 was used. Discrepancies were found in the 5Ј-upstream region HW and linkage disequilibrium compared to accession No. U18976 during sequence Deviation from HW equilibrium was found for SNP8 analysis. A revised 5Ј-upstream sequence was submit- in the DK case sample and for SNP3, SNP4, and SNP10 ted to GenBank under accession No. AF152962. During the completion of this manuscript, the Human Genome Project identified two SNPs in the SSTR5 gene as base pair discrepancies during alignment of overlapping clones (SNP rs169068 and rs642249 at http://www.ncbi.nlm.nih.gov/SNP/). The SNPs are not yet validated. SNP rs169068 is identical to SNP11 in this manuscript.

Results Genetic variation Eleven previously unknown SNPs in the 5Ј-upstream (2258 bp) and the coding region (1092 bp) of the SSTR5 gene were identified from five psychiatric patients and two unrelated healthy individuals (Table 1). Nine of the SNPs were found in the affected individuals. In the 5Ј-flanking region, that formerly has been sequenced and shown to contain a functional promoter,30 five SNPs were found: T-2187G, −2134delT, A-1665G, C- 801G, and T-461C. In the coding region four SNPs introducing an amino acid change were found: Leu48- Met, Ala52Val, Pro109Ser and Pro335Leu. In addition Figure 1 Genetic2 and radiation hybrid map15 of chromo- two silent SNPs were found: G573A and C633T. Ј some 16p13.3 incorporating the SSTR5 gene. Vertical lines To elucidate whether the SNPs in the 5 -upstream are drawn for markers that have been assigned a position in region were located in transcription factor binding both maps. Linkage analysis placed the SSTR5 gene between sites, we ran the complete 5Ј-upstream region through genetic markers D16S521 and D16S3124. Distances are indi- the program MatInspector 2.1 (URL: http://www.gfs. cated in centiMorgans (cM) on the genetic map and in centi- de.biodv/software).31 The program located consensus rays (cR) on the radiation hybrid map.

Molecular Psychiatry SSTR5 polymorphisms and bipolar disorder M Nyegaard et al 749

Figure 2 The degree of linkage disequilibrium (DЈ) between pairs of SNPs in the SSTR5 gene, with dark colors indicating the highest degree of LD. Statistically non-significant DЈ values are indicated in italic. in the DK control sample (significance level 0.05). In addition deviation from HW equilibrium was found for SNP7 and SNP8 in the UK case samples. We performed 46 (2 × 11 + 2 × 10) tests for deviation from HW equilibrium (SNP2 was not analyzed in the UK cases and UK controls). The tests performed were not independent due to linkage disequilibrium, although for compari- son, in 46 independent tests approximately two Type 1 errors are expected at a 5% significance level. Using the Bonferroni correction for multiple testing = − − n Pcon 1 (1 P) , where n is the number of independent tests carried out, the most extreme P-value in the DK control sample (0.001 for SNP4) increased to 0.05 in 46 tests. Hence, the deviation from HW equilibrium for SNP4 was significant after correction. It should be noted that the majority of SNPs found to be out of HW equilibrium were SNPs with a minor allele frequency of less than 7–8%. Complete linkage disequilibrium was found between approximately half of the pairs of SNPs in the DK control sample. The other half of SNPs displayed a lower Figure 3 The degree of linkage disequilibrium (͉DЈ͉)asa degree of linkage disequilibrium and for a small subset function of distance between pairs of SNPs in the SSTR5 of loci (SNP4/SNP6 SNP4/SNP7, SNP4/SNP8, gene. Only statistically significant DЈ values are shown. SNP5/SNP7, and SNP5/SNP8), almost complete linkage equilibrium was found (Figure 2). For all pairs of SNPs the statistical significance of the DЈ value was calculated. Approximately half of the DЈ values were stat- model), we also counted the number of chromosomes istically significant. In Figure 3, significant DЈ values and haplotypes which were married-into the family (ie are shown as a function of distance in base pairs non-IBD) and present among affected family members. between SNPs. It is clear that even though strong sig- Chromosomes IBD were discarded to avoid accumu- nificant linkage disequilibrium exists between SNPs in lation of haplotypes in affecteds due to familial the SSTR5 gene, linkage disequilibrium is not com- relationship. From this analysis, four haplotypes (H1, plete across the region. H3, H7, and H8) were repeatedly found in affecteds. Together these four haplotypes account for 12 out of 16 Haplotype segregation in families transmitting BPAD ‘married-into’ chromosomes in affecteds. We compared Family members, from whom DNA was available, were the haplotype distribution in affecteds with the haplo- genotyped for SNP1–11 and 11-locus haplotypes were type distribution in the background population. The constructed. Haplotypes, which could not be deter- haplotype frequencies for the background population mined unambiguously, were considered unknown and were estimated from the DK control sample (144 con- left out of the analysis. By visual inspection of the trol individuals) using the program EH v.1.14 pedigrees and considering both healthy and affected (Table 2).27 The absolute number of haplotypes in the individuals, nine different haplotypes (named H1–H9) DK control sample was calculated by multiplying rela- were found to segregate within the two families. Eight tive frequency with the number of control chromo- of these haplotypes were found in affecteds. As the dis- somes (288) (Table 2). A significant difference was = ease mechanism is expected to be recessive (highest obtained (P1 0.038) when the distribution of haplo- lodscores were obtained in the recessive disease types in affected family members and in the back-

Molecular Psychiatry SSTR5 polymorphisms and bipolar disorder M Nyegaard et al 750 Table 2 Different haplotypes segregating within two families transmitting bipolar affective disorder

H1 H2 H3 H4 H5 H6 H7 H8 H9 Other Total

Haplotypes observed in fam 2 and 4 (healthy and affecteds) SNP1 1 1 1 222111 SNP2 1 1 1 111221 SNP3 1 1 1 222111 SNP4 1 1 1 111111 SNP5 1 1 1 112222 SNP6 1 1 2 111111 SNP7 1 1 1 111112 SNP8 1 1 1 111112 SNP9 1 1 1 111111 SNP10 1 2 1 111111 SNP11 1 1 1 122211

Haplotypes observed in affected family members No. of chr IBS fam 4 2 0 2 1 1 0 3 1 1 11 No. of chr IBS fam 2 1 1 1 0 0 0 0 2 0 5 No. of chr IBS (fam 2 + 4) 3 1 3 110331016 Frequence (fam 2 + 4) 0.1875 0.0625 0.1875 0.0625 0.0625 0 0.1875 0.1875 0.0625 0 1

Haplotypes observed in background population Total No. of chr in pop 52 16 14 6 31 49 48 26 2 44 288 Pop. freq. if no LD 0.0603 0.004 0.0036 0.0106 0.0118 0.0124 0.0315 0.0282 0.0001 0.8375 1 Pop. freq. observed 0.1800 0.0543 0.0502 0.0209 0.1096 0.1700 0.1657 0.0899 0.0054 0.1540 1

Haplotype frequencies in affected family members and the background population are shown. Because of a putative recessive disease mechanism, only haplotypes ‘married-into’ the families were considered. Four different haplotypes were repeatedly observed in affecteds (indicated in bold).

ground population was compared using CLUMP By sequencing the promoter and coding region of the v.1.6.28 SSTR5 gene (3350 bp in total) on 14 chromosomes, we have identified 11 previously unknown sequence vari- Association study ants. Two additional polymorphic sites in the 5Ј- The DK case-control sample was genotyped with upstream region have been described by Sasi et al.32 respect to SNP1–SNP11 and a significant difference These polymorphisms were not found in our screen- (significance level 0.05) in allele frequencies between ing sample. cases and controls was found for SNP9 (P = 0.008). SNP9 is a silent variation in the coding region. The Linkage disequilibrium UK case-control sample was genotyped with respect to To estimate linkage disequilibrium between pairs of SNP1 and SNP3–11. Differences in allele frequencies SNPs we used Lewontin’sDЈ.29 DЈ has been shown to were found for SNP6 (P = 0.003) and SNP11 (P = 0.004) be reasonably robust towards differences in allele fre- (Table 3), both missense mutations. We combined quencies in contrast to D.33 Large differences in the genotypes from SNP6 and SNP11 into haplotypes for degree of linkage disequilibrium were found within the UK and DK cases and controls. For the UK population, SSTR5 gene, with ͉DЈ͉ ranging from 1 to 0.01. A cluster the following numbers of haplotypes were observed of SNPs displayed strong linkage disequilibrium (haplotype 1–1, 1–2, 2–1, and 2–2): 34(37%), 57(62%), within the coding region. However, only approxi- 1(1%) and 0(0%) for cases and 74(45%), 71(43%), mately half of the DЈ values calculated in this study 19(12%), and 0(0%) for controls. This difference was were statistically significant (the H null hypothesis of = 28 significant (P1 0.0007) using CLUMP. This haplo- no linkage disequilibrium could be rejected). This is type association was not replicated in the Danish case- most likely due to the low allele frequency for the control sample, showing the following haplotype dis- minor allele for many of the SNPs identified in this tribution: 62(39%), 83(52%), 11(7%), and 2(1%) for study. For pairs of polymorphisms where either one or cases and 82(40%), 108(53%), 14(7%) and 0(0%) for both loci have a very rare allele, a much larger number controls. of observations is needed to reject the null-hypothesis of no linkage disequilibrium, than for sets of loci with more common alleles.33 It is clear that a substantial Discussion decline in linkage disequilibrium has occurred within This is the first report of a systematic search for poly- the SSTR5 gene. It may therefore seem reasonable to morphisms in the SSTR5 gene in psychiatric patients. include a number of SNPs in future association studies

Molecular Psychiatry SSTR5 polymorphisms and bipolar disorder M Nyegaard et al 751 Table 3 Genotype and allele distribution of SSTR5 SNPs in Danish and UK bipolar affected patients and controls

Bipolar affective disorder Controls P-value (2-sided) Genotype Allele Genotype Allele

11 12 22 1 2 11 12 22 1 2

Danish sample set SNP1 14 12 0 40 12 (0.23) 62 67 8 191 83 (0.30) 0.29 SNP2 9 14 3 32 20 (0.38) 66 66 11 198 88 (0.37) 0.27 SNP3 46 32 1 124 34 (0.21) 66 69 6 20 81 (0.29) 0.10 SNP4 68 10 1 146 12 (0.08) 125 11 4 261 19 (0.07) 0.75 SNP5 14 43 23 71 89 (0.56) 33 60 36 126 132 (0.51) 0.37 SNP6 66 13 0 145 13 (0.08) 117 15 0 249 15 (0.06) 0.31 SNP7 73 7 0 153 7 (0.04) 127 10 0 264 10 (0.04) 0.71 SNP8 67 4 2 138 8 (0.05) 125 10 0 260 10 (0.04) 0.40 SNP9 66 7 0 139 7 (0.05) 100 1 0 201 1 (0.00) *0.008 SNP10 67 11 0 145 11 (0.07) 94 9 2 197 13 (0.06) 0.74 SNP11 16 42 22 74 86 (0.54) 27 46 33 100 112 (0.53) 0.86

UK sample set SNP1 23 13 2 59 17 (0.22) 36 13 0 85 13 (0.13) 0.12 SNP2 – SNP3 28 20 3 76 26 (0.25) 56 27 1 139 29 (0.17) 0.10 SNP4 42 6 0 90 6 (0.06) 74 7 0 155 78 (0.04) *0.38 SNP5 7 21 21 35 63 (0.64) 17 42 29 76 100 (0.57) 0.23 SNP6 46 1 0 93 1 (0.01) 66 17 1 149 19 (0.11) 0.003 SNP7 42 2 2 86 6 (0.07) 80 4 0 164 4 (0.02) *0.10 SNP8 33 2 2 68 6 (0.08) 43 4 0 90 4 (0.04) *0.19 SNP9 45 3 0 93 3 (0.03) 83 3 0 169 3 (0.02) *0.67 SNP10 48 8 0 104 8 (0.07) 77 9 0 163 9 (0.05) 0.51 SNP11 11 23 23 45 69 (0.61) 30 38 18 98 74 (0.43) 0.004

Allele frequencies are given in parenthesis. *Fishers exact test. involving SSTR5. In contrast it should not be necessary minor protein chemical changes. Ala and Val residues to analyze the complete set of SNPs. The pattern of are small hydrophobic amino acids, and Leu and Met linkage disequilibrium across the SSTR5 gene is com- residues are large hydrophobic amino acids. It can, parable to what is seen for the Dystrophin (DMD44) however, not be excluded that minor protein chemical gene and lipoprotein lipase (LPL) gene.34 It is not yet changes may have an effect on binding afﬁnity or fully understood which underlying population genetic receptor stability. In the presenilin-1 gene, two differ- mechanisms are responsible for the extent of linkage ent Ala to Val substitutions were identiﬁed in four disequilibrium within genes. Genetic drift, selection unrelated patients with early onset Alzheimer’s dis- pressure, the temporal history of mutation events, vari- ease.35 In these patients, the Ala to Val substitutions ation in recombination rates in different genomic were positioned in the N terminal region and in a trans- regions, and sampling, which might by chance remove membrane (TM) region respectively, the latter certain haplotypes, are components that affect linkage mutation predicted to interfere with the alpha helix disequilibrium. structure. The Leu48Met (SNP6) and the Ala52Val (SNP7) missense substitutions in the SSTR5 gene are Promoter SNPs both located in the predicted TM domains (Figure 4). Two SNPs in the 5Ј-upstream region (SNP3 and SNP5) We found association to BPAD with SNP6 in the UK were found to be located in the core sequences of pro- population. The other two missense mutations, Pro10- moter element-binding sites CETS1P54 and YY1 9Ser (SNP8) and Pro335Leu (SNP11) substitution are respectively. Polymorphisms in promoter element- non-conservative amino acid changes. Pro has a rigid binding sites may affect the transcription level of the cyclic structure, and is often observed in turns between SSTR5 gene. None of these SNPs showed association helices. As Pro is often of structural importance and to BPAD. has no structural similar counterpart, it is possible that the Pro substitutions involve a stability or structure Missense SNPs change of the receptor. The Pro109Ser substitution was Of the four missense substitutions, the Leu48Met found in extra-cellular loop 1 (ECLI) and the Pro335- (SNP6) and the Ala52Val (SNP7) substitutions are Leu substitution was found in the C-terminal intra-

Molecular Psychiatry SSTR5 polymorphisms and bipolar disorder M Nyegaard et al 752 gene for these disorders. In the study of panic disorder the highest lodscore on chromosome 16 was obtained with marker D16S423 (9 cM on Genethon map),2 which was the most telomeric marker analyzed in that study.38 In a study of autism, positive lodscores were obtained in the region D16S423–D16S405 (30 cM on Genethon map).2,39 The position of SSTR5 on the genetic map and on Genemap99 (marker STSG39114 on the the GB4 radiation hybrid map, http://www.ncbi.nlm.nih. gov/genemap99/)15 is in agreement.

Figure 4 Schematic representation of somatostatin receptor Haplotypes in families transmitting BPAD type 5, illustrating transmembrane topology, intracellular and When evaluating the segregation of SSTR5 haplotypes extracellular domains. Black spots indicate missense in the disease families, we used a family-based associ- mutations identified in a group of patients from two families transmitting bipolar affective disorder. ation approach. This was done as the linkage results pointing to a candidate gene for BPAD at chromosome region 16p13.3 were obtained in a recessive mode of cellular part of the receptor (Figure 4). According to an transmission (although the region received some sup- alignment of the five different somatostatin receptors port in a dominant mode of transmission as well).40 (SSTR1–5),36 a proline in position 335 in the C-ter- When evaluating haplotypes it was clear that affected minal part is conserved in four of the five receptor sub- family members did not consistently carry two ident- types (SSTR1, 3, 4 and 5). The Pro335Leu (SNP11) ical haplotypes. This was albeit not expected for a com- showed association to BPAD in the UK population. plex heterogenic disorder. In the study we considered Mutation analysis of SSTR5 involving C-tail truncation haplotypes that had been passed on to affected off- has shown that a positive molecular internalization sig- springs, discarded haplotypes IBD. This was done to nal is contained within the C-terminal segment, includ- avoid overrepresentation of certain haplotypes solely ing Pro335.37 due to familial relationship, but unfortunately this also reduced the number of haplotypes within the study. Fine mapping Sixteen non-IBD chromosomes were present in affec- SSTR5 is a positional and functional candidate gene teds in the two families. Four haplotypes (H1, H3, H7, for bipolar affective disorder. Ewald and coworkers1 and H8) were present three times each in affecteds. For found in two Danish bipolar pedigrees suggestive link- haplotypes H3 and H8, this was more than twice as age to a 10-cM broad chromosome region on 16p13.3. often as expected (19% and 19% in affecteds vs 5% In these two families a combined lodscore of 2.65 was and 9% in the background population), whereas the found for marker D16S510 (9 cM on Genethon map).2 frequencies of haplotype H1 and H7 were similar to the In the larger family alone, lodscores of 1.60 and 1.76 background population (19% and 19% in affecteds vs were found with D16S85 (HBA1) (0 cM, not on Gene- 18% and 17% in the background population). thon map) and D16S510 respectively. The lodscore at Together the four haplotypes (H1, H3, H7 and H8) D16S521 (0 cM on Genethon map),2 which was pos- accounted for 12 of the 16 non-IBD chromosomes seen itioned in between D16S85 and D16S510 on the map in affecteds. Interestingly, the four haplotypes were used by Ewald et al (1995), dropped to −0.57. Interest- pair wise very similar. H1 and H3 differed only at one ingly, a BLAST search against the finished human SNP. H7 and H8 differed at another SNP. It is possible genome sequences (http://www.ncbi.nlm.nih.gov/ that H1 and H3 originally were the same haplotype, genome/seq/) using D16S85 and D16S521 suggested and that a sequence change took place on a H1 haplo- that the order of these two markers are reversed com- type, thus forming a new and rare haplotype (H3). The pared to the map used by Ewald et al.* This revised same applies for H7 and H8. If a possible susceptibility marker order gives consistent positive lodscores within variation were present on a certain haplotype before the 16p13.3 region. The genetic mapping result these sequence changes took place, several haplotypes presented here showed that the SSTR5 gene is located in a population would carry that susceptibility gene or 3 cM centromeric to D16S521 and 2 cM telomeric to mutation. This may be the case for H1, H3, H7 and H8 D16S3024 (Figure 1). This position is within the region in the two Danish disease families. The difference in implicated by linkage studies, and close to marker haplotype distribution between affecteds and the back- = D16S85. ground population was significant (CLUMP P1 0.038). The fine mapping results showed in addition that SSTR5 is located in a region with suggestive linkage to Association study panic disorder,38 and autism,39 and may be a candidate In the DK case-control sample allelic association to BPAD was found with SNP9 (allele 2), a synonymous = * D16S85 was found on compound Z69706 and D16S521 was sequence change (P 0.008). The association was not found on compound Z69719, both on contig NT 000655, at URL: replicated in the UK sample. Allele 2 was initially http://www.ncbi.nlm.nih.gov/genome/guide/HsChr16.shtml. identified in a control individual, and not found in any

Molecular Psychiatry SSTR5 polymorphisms and bipolar disorder M Nyegaard et al 753 of the family members from the two Danish disease influenced by the presence of DRD2 mutations. It is families. This indicates that the significant association possible that SNP6 and SNP11 truly increase suscepti- could be a false positive finding, although this cannot bility to BPAD in the UK population, perhaps in com- be concluded with certainty. bination with other (unknown) susceptibility loci In the UK sample significant association to BPAD found more commonly in the UK population than in was found with SNP6 (allele 1) (P = 0.003) and with the Danish population. Alternatively, strong linkage SNP11 (allele 2) (P = 0.004). Combining genotypes from disequilibrium may exist between SSTR5 and one or these two SNPs into haplotypes, the significance several susceptibility alleles in a gene close to SSTR5. increased to P = 0.0007. This haplotype association These susceptibility alleles for BPAD could be located was not replicated in the Danish case-control study. on different SSTR5 haplotypes. Within the two Danish families transmitting BPAD, In view of the prior evidence from linkage analysis, allele 1 from SNP6 and allele 2 from SNP11 were altered somatostatin level in bipolar affected patients, present on haplotype H7. H7 was seen with a fre- physical interaction between the SST5 receptor and D2 quency of 19% in affecteds and 17% in the background receptor and a cluster of markers in the SSTR5 gene population. It is dubious if the Danish haplotype study with P-values nearing significance level, we suggest supports the British association. that SSTR5 may be a susceptibility gene for BPAD. We are aware though, that the association between Further studies with additional samples and inclusion BPAD and SSTR5 may be a false-positive result. As of DRD2 polymorphisms will clarify this. Finally, the with many association studies, multiple testing is a possible significance of SSTR5 gene variants in other major issue. The number of tests carried out, combined neuropsychiatric diseases should be investigated. with the low prior probability of pointing out the right gene from a number of potential candidates, increases Acknowledgements the risk of getting a false-positive result. Using the Bonferroni correction for multiple testing The SSTR5 clone used in this study was kindly pro- = − − n Pcorr 1 (1 P) , where n is the number of inde- vided by Dr Graeme I Bell, Howard Hughes Medical pendent tests carried out, a P value of 0.003 increases Institute. We thank Drs Robert Plomin (SGDP Centre) = to Pcorr 0.06 in 21 tests (11 tests in Danish case-con- and Simon Lovestone (Old Age Psychiatry) for access trols, 10 tests in British case-controls). Thus, none of to the British control samples. This work was sup- the single SNP P-value remains significant after correc- ported by the Danish Medical Research Council tion. This is, however, a very conservative correction (9303757, 9602007, 9902685, 9902769), Fonden til Lae- as some of the SNPs are in linkage disequilibrium and gevidenskabens Fremme, Fonden til Psykiatriens therefore the tests are not independent. Currently there Fremme, The Psychiatric Research Foundation, The Eli is no good method for adjusting P-values for multiple and Egon Larsen Foundation, The Geert-Jørgensen testing when loci involved are in strong linkage dis- Foundation, The Axel Thomsen Foundation, The equilibrium. Trier-Hansen Foundation, and The Jacob Madsen A major problem is that we do not know how much Foundation. signal to expect, when analyzing potential susceptibility genes for BPAD and other complex disorders, References where a substantial degree of heterogeneity is expected to exist. Power calculations show that detecting associ- 1 Ewald H, Mors O, Flint T, Koed K, Eiberg H, Kruse TA. A possible locus for manic depressive illness on chromosome 16p13. Psychi- ations with alleles having frequencies under 5% will atr Genet 1995; 5:71–81. typically require thousands of patients (except in rare 2 Dib C, Faure S, Fizames C, Samson D, Drouot N, Vignal A et al.A cases in which the attributable risk is large).41 Long and comprehensive genetic map of the human genome based on 5264 Langley (1999) indicate that there is sufficient power microsatellites. Nature 1996; 380:152–154. to detect the presence of causative polymorphisms of 3 Edenberg HJ, Foroud T, Conneally PM, Sorbel JJ, Carr K, Crose C et al. Initial genomic scan of the NIMH genetics initiative bipolar small effects if on the order of 500 individuals are pedigrees: chromosomes 3, 5, 15, 16, 17, and 22. Am J Med Genet 42 sampled. Compared to the number of individuals 1997; 74:238–246. mentioned above our samples are relatively small. 4 McInnes LA, Escamilla MA, Service SK, Reus VI, Leon P, Silva S Martin and coworkers have carried out as SNP analy- et al. A complete genome screen for genes predisposing to severe bipolar disorder in two Costa Rican pedigrees. Proc Natl Acad Sci sis of the APOE in Alzheimer’s disease, to investigate USA 1996; 93: 13060–13065. whether it was possible to detect association near a 5 Levitas AS, Reid CS. Rubinstein–Taybi syndrome and psychiatric complex disease gene using SNPs.43 They used 60 disorders. J Intellect Disabil Res 1998; 42:284–292. SNPs spanning 40 kb on either side of APOE in 220 6 Adams LJ, Salmon JA, Kwok JB, Vivero C, Donald JA, Mitchell PB cases and 220 controls. They found that looking for et al. Exclusion of linkage between bipolar affective disorder and chromosome 16 in 12 Australian pedigrees. Am J Med Genet 1997; regions of markers with P values nearing significance 74:304–310. for association (significance level 0.05 without correc- 7 Blackwood DH, He L, Morris SW, McLean A, Whitton C, Thomson tion for multiple testing) is an important indicator for M et al. A locus for bipolar affective disorder on chromosome 4p. the presence of a disease gene. Nat Genet 1996; 12:427–430. 8 Coon H, Jensen S, Hoff M, Holik J, Plaetke R, Reimherr F et al.A Considering the new results revealing that SSTR5 genome-wide search for genes predisposing to manic-depression, and DRD2 form functional heterodimers, it is not assuming autosomal dominant inheritance. Am J Hum Genet 1993; unlikely that the impact of an SSTR5 mutation is 52: 1234–1249.

Molecular Psychiatry SSTR5 polymorphisms and bipolar disorder M Nyegaard et al 754 9 Detera-Wadleigh SD, Badner JA, Berrettini WH, Yoshikawa T, Gol- terminating inhibitors. Proc Natl Acad Sci USA 1977; 74: 5463– din LR, Turner G et al. A high-density genome scan detects evi- 5467. dence for a bipolar-disorder susceptibility locus on 13q32 and 26 Lathrop GM, Lalouel JM, Julier C, Ott J. Strategies for multilocus other potential loci on 1q32 and 18p11.2. Proc Natl Acad Sci USA linkage analysis in humans. Proc Natl Acad Sci USA 1984; 81: 1999; 96: 5604–5609. 3443–3446. 10 Friddle C, Koskela R, Ranade K, Hebert J, Cargill M, Clark CD et al. 27 Terwilliger JD, Ott J. Handbook of Human Genetic Linkage. The Full-genome scan for linkage in 50 families segregating the bipolar John Hopkins University Press: Baltimore, 1994. affective disease phenotype. Am J Hum Genet 2000; 66:205–215. 28 Sham PC, Curtis D. Monte Carlo tests for associations between dis- 11 Morissette J, Villeneuve A, Bordeleau L, Rochette D, Laberge C, ease and alleles at highly polymorphic loci. Ann Hum Genet 1995; Gagne B et al. Genome-wide search for linkage of bipolar affective 59:97–105. disorders in a very large pedigree derived from a homogeneous 29 Lewontin RC. The interaction of selection and linkage. I. General population in Quebec points to a locus of major effect on chromo- consideration; heterotic models. Genetics 1964; 49:49–67. some 12q23–q24. Am J Med Genet 1999; 88: 567–587. 30 Greenwood MT, Panetta R, Robertson LA, Liu JL, Patel YC. 12 Murphy VE, Mynett-Johnson LA, Claffey E, Bergin P, McAuliffe M, Sequence analysis of the 5Ј-flanking promoter region of the human Kealey C et al. Search for bipolar disorder susceptibility loci: the somatostatin receptor 5. Biochem Biophys Res Commun 1994; 205: application of a modified genome scan concentrating on gene-rich 1883–1890. regions. Am J Med Genet 2000; 96: 728–732. 31 Quandt K, Frech K, Karas H, Wingender E, Werner T. MatInd and 13 Ewald H, Kruse TA. Bipolar affective disorder, chromosome MatInspector: new fast and versatile tools for detection of consen- 16p13.3, and recessive disease genes [letter]. Am J Med Genet 1997; sus matches in nucleotide sequence data. Nucleic Acids Res 1995; 74:549–450. 23: 4878–4884. Ј 14 Takeda J, Fernald AA, Yamagata K, Le Beau MM, Bell GI. Localiz- 32 Sasi R, Puebla L, Khare S, Patel YC. Polymorphism in the 5 flank- ation of human somatostatin receptor 5 gene (SSTR5) to chromo- ing region of the human somatostatin receptor subtype 5. Gene some band 16p13.3 by fluorescence in situ hybridization. Genomics 1998; 214:45–49. 1995; 26:638–639. 33 Thompson EA, Deeb S, Walker D, Motulsky AG. The detection of 15 Deloukas P, Schuler GD, Gyapay G, Beasley EM, Soderlund C, Rod- linkage disequilibrium between closely linked markers: RFLPs at riguez-Tome P et al. A physical map of 30 000 human genes. the AI-CIII apolipoprotein genes. Am J Hum Genet 1988; 42: Science 1998; 282: 744–746. 113–124. 16 Brazeau P, Vale W, Burgus R, Ling N, Butcher M, Rivier J et al. 34 Przeworski M, Hudson RR, Di Rienzo A. Adjusting the focus on Hypothalamic polypeptide that inhibits the secretion of immuno- human variation. Trends Genet 2000; 16:296–302. reactive pituitary growth hormone. Science 1973; 179:77–79. 35 Cruts M, van Duijn CM, Backhovens H, Van den Broeck M, Wehnert A, Serneels S et al. Estimation of the genetic contribution 17 Patel YC, Greenwood MT, Panetta R, Demchyshyn L, Niznik H, of presenilin-1 and -2 mutations in a population-based study of Srikant CB. The somatostatin receptor family. Life Sci 1995; 57: presenile Alzheimer disease. Hum Mol Genet 1998; 7:43–51. 1249–1265. 36 Patel YC, Greenwood M, Panetta R, Hukovic N, Grigorakis S, Rob- 18 Rubinow DR, Davis CL, Post RM. Somatostatin in neuropsychiatric ertson LA et al. Molecular biology of somatostatin receptor sub- disorders. Prog Neuropsychopharmacol Biol Psychiatry 1988; 12: types. Metabolism 1996; 45:31–38. 137–155. 37 Hukovic N, Panetta R, Kumar U, Rocheville M, Patel YC. The cyto- 19 Bissette G, Myers B. Somatostatin in Alzheimer’s disease and plasmic tail of the human somatostatin receptor type 5 is crucial depression. Life Sci 1992; 51: 1389–1410. for interaction with adenylyl cyclase and in mediating desensitiz- 20 Patel YC. Somatostatin and its receptor family. Front Neuroendo- ation and internalization. J Biol Chem 1998; 273: 21416–21422. crinol 1999; 20:157–198. 38 Knowles JA, Fyer AJ, Vieland VJ, Weissman MM, Hodge SE, Hei- 21 Panetta R, Greenwood MT, Warszynska A, Demchyshyn LL, Day man GA et al. Results of a genome-wide genetic screen for panic R, Niznik HB et al. Molecular cloning, functional characterization, disorder. Am J Med Genet 1998; 81:139–147. and chromosomal localization of a human somatostatin receptor 39 International Molecular Genetic Study of Autism Consortium. A (somatostatin receptor type 5) with preferential affinity for somato- full genome screen for autism with evidence for linkage to a region statin-28. Mol Pharmacol 1994; 45:417–427. on chromosome 7q. Hum Mol Genet 1998; 7: 571–578. 22 Yamada Y, Kagimoto S, Kubota A, Yasuda K, Masuda K, Someya 40 Ewald H, Mors O, Eiberg H. Linkage analysis between manic- Y et al. Cloning, functional expression and pharmacological depressive illness and 35 classical markers. Am J Med Genet 1994; characterization of a fourth (hSSTR4) and a fifth (hSSTR5) human 54: 144–148. somatostatin receptor subtype. Biochem Biophys Res Commun 41 Risch N, Merikangas K. The future of genetic studies of complex 1993; 195:844–852. human diseases. Science 1996; 273: 1516–1517. 23 Rocheville M, Lange DC, Kumar U, Patel SC, Patel RC, Patel YC. 42 Long AD, Langley CH. The power of association studies to detect Receptors for dopamine and somatostatin: formation of hetero-olig- the contribution of candidate genetic loci to variation in complex omers with enhanced functional activity. Science 2000; 288: traits. Genome Res 1999; 9: 720–731. 154–157. 43 Martin ER, Lai EH, Gilbert JR, Rogala AR, Afshari AJ, Riley J et 24 American Psychiatric Association. Diagnostic and Statistical Man- al. SNPing away at complex diseases: analysis of single-nucleotide ual of Mental Disorder, 4th edn. APA: Washington DC, 1994. polymorphisms around APOE in Alzheimer disease. Am J Hum 25 Sanger F, Nicklen S, Coulson AR. DNA sequencing with chain- Genet 2000; 67:383–394.

Molecular Psychiatry