1997 Oxford University Press Human Molecular Genetics, 1997, Vol. 6, No

Transcription

1 1997 Oxford University Press Human Molecular Genetics, 1997, Vol. 6, No Identification of a major susceptibility locus on chromosome 6p and evidence for further disease loci revealed by a two stage genome-wide search in psoriasis Richard C. Trembath 1, *, R. Lee Clough 1, Jane L. Rosbotham 1,2, Andrew B. Jones 2, Richard D. R. Camp 1, Angela Frodsham 1, Julie Browne 1, Ruth Barber 1, Joseph Terwilliger 3,4, G. Mark Lathrop 3 and Jonathan N. W. N. Barker 2 1 Departments of Genetics and Medicine and Therapeutics, University of Leicester, Leicester LE1 7RH, UK, 2 St. John s Institute of Dermatology, UMDS St. Thomas Hospital, London SE1 7EH, UK, 3 Wellcome Trust Centre for Human Genetics, University of Oxford, Windmill Road, Oxford OX3 7BN, UK and 4 Columbia University Department of Genetics and Development and Department of Psychiatry, 722 West 168th Street, New York, NY 10032, USA Received January 2, 1997; Revised and Accepted February 19, 1997 Psoriasis is a common chronic inflammatory disorder of the skin. To further understand the pathogenesis of psoriasis we have chosen to investigate the molecular genetic basis of the disorder. We have used a twostage approach to search the human genome for the location of genes conferring susceptibility to psoriasis, using a total of 106 affected sibling pairs identified from 68 independent families. As over a third of the extended kindreds included affected relatives besides siblings, in addition to an analysis of allele sharing between affected sibling pairs, a novel linkage strategy was applied that extracts full non-parametric information. Four principal regions of possible linkage were identified on chromosomes 2, 8, 20 (p <0.005) and markers from the MHC region at 6p21 (p < ) for which significant evidence of linkage disequilibrium was also observed (p < ). Whilst data from limited case control associations exist to implicate the MHC, the results of this genome wide analysis demonstrate that, at least in the population studied, a gene or genes located within the MHC and close to the class 1 HLA loci, represent the major determinant of the genetic basis of psoriasis. INTRODUCTION Psoriasis vulgaris is characterised by the appearance of multiple red, scaly, indurated plaques. The disease is common affecting 2% of the UK population and is found in all racial groups (1). Although the disease is rarely fatal, objective measurements indicate that with more severe disease, psoriasis has a significant adverse effect on the quality of life (2). The impact for health care of psoriasis is considerable. In the USA the annual expenditure has been estimated to exceed $1.6 billion per year. All topical and systemic treatments for psoriasis are symptomatic, none leads to a cure and all are associated with potentially serious side effects. The causes of psoriasis are unknown (3,4). Epidemiological studies, including concordance rates in twin pairs and siblings, strongly implicate genetic factors in the pathogenesis of psoriasis (5). The risk to siblings of an affected individual of developing psoriasis is four-fold that observed in the general population (1). Occasionally multigeneration kindreds with psoriasis are observed. In one such large extended white American family anonymous microsatellite markers at 17q demonstrated linkage to psoriasis under an assumed autosomal dominant model of inheritance (6). Studies in extended kindreds from Northern Europe have failed to confirm linkage to the long arm of chromosome 17 and suggest that a susceptibility locus in this region is unlikely to account for a significant proportion of the genetic contribution to psoriasis (Burden et al., manuscript in preparation) (7,8). Recently a further locus at 4q (D4S1535, Z max = 3.03, θ = 0.08) has been proposed through the analysis of extended families from Ireland and the UK (9). The authors of both these reports suggest caution in the assessment of linkage to psoriasis susceptibility loci as a number of factors complicate the analyses. These include incomplete penetrance, phenocopies, misdiagnosis and the lack of a robust genetic model that accurately accounts for the observed familial aggregation, which all serve to compound parametric linkage studies of common disorders with non-mendelian inheritance such as psoriasis. Several investigators have conducted case control association studies in psoriasis typically concentrating upon genes involved in the regulation of the immune response (10,11) and particularly the major histocompatibility complex [human leucocyte antigen (HLA) region] on the short arm of chromosome 6 (12,13). Of the *To whom correspondence should be addressed. Tel: ; Fax: ; rtrembat@hgmp.mrc.ac.uk

2 814 Human Molecular Genetics, 1997, Vol. 6, No. 5 many case control studies performed, associations with psoriasis have been identified for multiple class I and class II MHC antigens including HLA- B13, B17, B39, B57, Cw6, Cw7, DR4 and DR7 and are likely to reflect extended linkage disequilibrium across the region (14). The strongest association appears to be with the class I antigen HLA-Cw6 which, in some series, is seen in up to 80% of patients compared to 10% in the general population. There are, however, several factors which make interpretation of these data problematic including patient ascertainment bias, control population characteristics, small sample size and reliance on serologically defined class I antigens. Furthermore, studies performed in small numbers of families have failed to demonstrate linkage to the MHC (15,16). Hence the molecular genetic basis for these associations remains unclear. As the number of genes involved in the pathogenesis of psoriasis and their chromosomal location are presently unknown, we have performed two stage (17) systematic genome wide genotyping in a large cohort of nuclear psoriatic families and sibling pairs and analysed the data by including a novel method designed to extract full non-parametric data from all affected relatives. RESULTS Linkage studies in 41 multiplex families The primary study group consisted of a total of 254 individuals recruited in Britain. The mapping panel consisted of 66 pairs of affected siblings from 41 families, with extended relationships as described in Table 1. In 94% of the affected cohort, chronic plaque type psoriasis was the presenting form. The other forms observed within the extended families were guttate psoriasis (3%) and palmoplantar psoriasis (3%). The group was genotyped at a total of 260 autosomal microsatellite markers. For each marker, a non-parametric statistical estimate of linkage was calculated for affected siblings with psoriasis (Fig. 1). Nominal evidence for linkage was defined at the level of p Using these criteria, linkage was identified at 10 markers (7.8%), distributed over eight chromosomes; on chromosome 2 (D2S134), chromosome 6 (three adjacent markers, D6S276, D6S273 and TNFα, which span an interval of 1.5 centimorgans and include the major histocompatibility region), chromosome 8 (D8S284), chromosome 11 (D11S910), chromosome 12 (D12S99), chromosome 14 (D14S50), chromosome 16 (D16S422) and chromosome 20 (D20S186). For three distinct chromosomal regions two adjacent markers from the mapping panel reached a p value of p 0.1; D4S418 and D4S405 separated by 10 cm on chromosome 4p, D14S72 and D14S50 with a marker interval of 5 cm and D16S289 which lies within 9 cm from the marker D16S422 at chromosome 16q. We found no evidence for distortion of allele sharing for either markers D17S784 (p = 0.5) or D4S408, both of which map within the linkage intervals previously reported for psoriasis susceptibility loci (6,9). Table 1. Details of the two psoriasis family panels used for linkage analysis. Number of affected relations and families MULTIPLEX ASP1 2 sibs sibs sibs 2 5 sibs 1 Total affected sib pairs Total affected relative pairs (non-sib) 15 3 Total independent families In the MULTIPLEX panel, 21 affected sibships had both parents, eight had one parent and two had neither parent available for genotyping. In the ASP1 panel, the corresponding number of parents available for genotyping were 21, 4 and 3. In the statistical analyses dependence of multiple affected sibling pairs was taken into account as previously described (22,23). In 13 and one of the MULTIPLEX and ASP1 families respectively, affected relatives existed in two or more generations. Table 2. Results of full non-parametric (association and affected relative) analysis performed on the combined family cohorts (MULTIPLEX and ASP1) Marker Heterogeneity Within sibpairs Between sibpairs All linkages All linkages Combined linkage p-value p-value p-value LOD score and association a D6S D6S D6S D6S HLA-C TNFα n.d. D6S D6S p-value a Data presented for markers with evidence of allelic association only. n.d., Allelic association at TNFα not assessed as fluorescence based typing did not allow unambiguous calling of alleles.

3 815 Human Nucleic Molecular Acids Research, Genetics, 1994, 1997, Vol. Vol. 22, No. 6, No Table 3. Proportion of zero, one or two alleles shared common by descent by affected siblings for the three chromosome 6p21 markers showing strongest evidence for linkage in the combined datasets of MULTIPLEX and ASP1 Marker HLA-C TNFα D6S A major psoriasis susceptibility locus on chromosome 6 Analysis of the eight chromosomal regions suggestive of linkage was extended in a second independent panel of 40 siblings (ASP1) affected with psoriasis (all with chronic plaque type psoriasis) recruited within the UK. From the marker mapping set, two adjacent markers located at chromosome 6p21 each showed evidence of linkage with psoriasis in the second sample of siblings: D6S273 (p <0.03) and TNFα (p <0.001). Additional markers adjacent to and within the intervening linkage interval and including the MHC locus HLA-C were then studied in the combined cohort of affected siblings and multiplex families. As 15 (33%) of the families in MULTIPLEX had either multiple affected generations and/or multiple affected relative pairs other than sibs we analysed the data by a novel approach to extract all non-parametric linkage information, to include a search for allelic association (see Materials and Methods). Significant evidence of linkage was observed in the extended data set for the consecutive markers D6S273 (p = ) and TNFα (p = ). The gene encoding the class I molecule HLA-C which lies within 2 Mb of the markers displaying linkage, was genotyped using molecular methods and also gave support for linkage at p = Evidence for allelic association was observed for alleles at two of the three linked loci; allele 5 (136 mobility units) at D6S273, p = and a total of 62 HLA-Cw6 alleles were transmitted from parents to an affected offspring compared to 22 not transmitted (p = ). The results of the combined analysis of all independent nonparametric data for the chromosome 6p21 markers are presented in Table 2 and the proportion of alleles shared by common descent for the three most closely linked markers in Table 3. Suggestive linkage to non-mhc loci in psoriasis The most compelling evidence in favour of linkage to non-mhc loci was identified at three chromosomal regions: chromosome 2 (D2S134; p = , LOD score = 1.27), chromosome 8 (D8S284; p = , LOD score = 2.60 ) and chromosome 20 (D20S186; p = , LOD score = 2.01). Results are presented in Table 4 for these markers in the individual data sets (MULTIPLEX and ASP1) and the combined sets. No allele at any of these three loci showed evidence of allelic association. Support for linkage was not observed for any flanking marker from the mapping panel at these three loci, although the intermarker distances were large; D2S cm-d2s cm-d2s139, D8S cm-d8s cm-d8s272, tel-d20s186-9 cm- D20S112-8 cm-d20s200. The remaining four markers, D11S910, D12S99, D14S50 and D16S422 all showed reduced evidence for linkage in the second cohort of affected siblings and consequently in the combined data set. Table 4. Significance of linkage tests (shown as p-values) for markers characterised in the two family panels, MULTIPLEX and ASP1 Marker MULTIPLEX ASP1 Combined p-value p-value p-value D2S D8S D20S Mean IBD (identity by descent) estimates on the combined datasets of 106 sibling pairs with psoriasis are given for the markers with the strongest evidence for linkage. DISCUSSION These results provide significant evidence for the presence of a psoriasis susceptibility locus within the MHC region on chromosome 6p21. The contribution of a single locus to the familial clustering of a disease, termed λs, may be estimated by calculating the ratio of the expected and observed proportions of affected sibling pairs who share zero alleles identical by descent (IBD) (18). An overall estimate of sibling recurrence risk for psoriasis is 4.0 (19) and using estimates for affected IBD allele sharing from the combined analysis for the class 1 molecule HLA C, λ s 1 = 1.4 indicating that the contribution of this locus to the relative risk of developing psoriasis is 35%. MHC Class 1 genes have previously been implicated in disease susceptibility in psoriasis using case-control disease association studies (20). However a number of limitations apply to many of these reports and include small numbers of subjects studied, difficulties of rigid genetic population stratification and the use of insensitive methods to assign class 1 MHC genotypes. We used a molecular method, sequence specific oligonucleotide priming (SSOP), to detect HLA-C alleles (21) as a significant number are not detected (null alleles) by widely used serological assays. Linkage was detected using robust non-parametric methods in a cohort of over 100 affected sibling pairs with additional linkage information obtained from other affected relatives. Allelic association was sought by a genotype-based haplotype relative risk approach that compares genotypes in an affected individual with the two parental alleles that were not transmitted to the affected offspring. An increasing number of genome wide scans searching for susceptibility loci in common diseases with complex inheritance characteristics are being reported (22,23). Most studies have chosen to utilise affected sibling pairs and to seek evidence of distortion of the frequencies of allele sharing from expected values if random segregation existed. This approach has the benefit of making no prior assumptions about parameters required to perform a recombination based analysis; however the power to detect linkage by ASP analysis is dependent upon the mode of inheritance of the disease allele. Furthermore, in many diseases multigeneration families can be identified with greater power characteristics for detecting linkage than simple sibling pairs. The size of the patient resource required to efficiently identify the genetic contribution to psoriasis is difficult to estimate and prior segregation analysis does not accurately indicate the number of susceptibility genes for a given disorder. We performed linkage simulations (SLINK) using the 61 nuclear and multiplex pedigrees ascertained. The results indicate a high probability of detecting linkage using this data set even under

4 816 Human Molecular Genetics, 1997, Vol. 6, No. 5 inheritance models extending to 50% heterogeneity. Under autosomal recessive models expected maximum LOD (EMLOD) scores ranged from at θ = 0, in the absence of heterogeneity to 3.12 at θ = 0.05 with α = 0.5 (proportion of families linked = 50%) and a 0.5% phenocopy frequency. The comparable EMLOD for autosomal dominant models were and 3.03 respectively. Hence we developed a novel strategy to maximise the power to detect linkage under full non-parametric assumptions and demonstrate the utility of this approach when evidence of both linkage and association exists, as seen for markers within the MHC at 6p21. We did not find support for linkage using markers previously genotyped in large extended families [D17S548 (6) and D4S408 (9)]. Psoriasis is likely to be a genetically heterogeneous disorder and our data does not exclude the localisation of psoriasis susceptibility genes to those regions but would suggest that neither locus accounts substantially for the genetic contribution to psoriasis in the population studied. However, our studies did identify support for linkage at several loci outside the MHC region. Confirmation of the significance of these non-mhc loci and analysis of their potential role will require studies of further family panels. It remains of interest that the non-mhc marker that shows strongest evidence for linkage was D8S284 (p = ) which maps within 10 cm of a gene associated with the skeletal disorder, hereditary multiple exostosis (HME). We have previously reported a unique family within which HME and psoriasis appeared to cosegregate in three generations (24). HME at 8q24.1 results from mutation of a putative tumour suppressor gene (EXT1) (25) and studies to identify a possible contiguous gene defect in this family are now underway. These results have important implications for future studies of the molecular genetics of psoriasis. Identification of linkage for

5 817 Human Nucleic Molecular Acids Research, Genetics, 1994, 1997, Vol. Vol. 22, No. 6, No Figure 1. Non-parametric probability estimates of linkage at all loci analysed in affected sibling pairs with psoriasis from the MULTIPLEX cohort. Filled blocks represent those loci with an estimate of linkage at a nominal value of p Shaded blocks identify loci from the genome mapping set that lie within linkage intervals previously reported for psoriasis (6,9). Full details of loci genotyped are available upon request (see Materials and Methods). the MHC region on chromosome 6p21 support the validity of whole genome scanning in family collections as a priori data from limited association studies exists to implicate the MHC in psoriasis disease determination and suggests that assessment of other linkages in additional datasets should include analysis of interaction (epistasis) with the HLA complex. Furthermore, although there is consensus that T lymphocytes play a critical role in the pathogenesis of psoriasis, controversy exists as to whether the T cells mediating the disease are CD8 (cytotoxic/suppressor) or CD4 (helper) T cells. Since CD8 T cells recognise antigen in the context of class I MHC, our linkage disequilibrium data would favour CD8 as the primary disease-associated sub-type of T cell. Further characterisation of this linkage region will be required to confirm this suggestion. MATERIALS AND METHODS Family ascertainment A total of 61 European Caucasoid families with a minimum of two affected members with psoriasis were initially identified by co-operating dermatologists (60%), through the Psoriasis Association of Great Britain (15%) and national press advertisement (25%). All affected individuals and their relatives were assessed

6 818 Human Molecular Genetics, 1997, Vol. 6, No. 5 by a single dermatologist (JLR). A diagnosis of psoriasis was made using standard clinical criteria as described elsewhere (1). In this and the ASP1 panel, the diagnosis of psoriasis was verified in a sample of affected individuals (15%) by an experienced dermatologist (JNWNB). Concordance of the diagnosis was 100%. All probands had chronic plaque type psoriasis. In a total of 26 (12.3%) individuals with suggestive skin lesions, including skin disease confined to flexural sites, the diagnosis of psoriasis could not be confirmed. These individuals were therefore considered as of unknown phenotype and either genotyped to aid likelihood estimates for allele sharing when parents were missing, or if only singleton affecteds remained, the family was not included in the study. A primary panel of 41 multiplex families (MULTIPLEX) comprising a total of 262 individuals with 66 affected sibling pairs were used for the whole genome analysis. The mean age of onset in this group of 135 affected individuals [60 male (44.4%) and 75 female (55.6%)] was 16.8 (range 1 59) years. An independent panel (ASP1) was recruited to allow further evaluation of the results of the primary genome mapping data and ascertained from similar sources using identical clinical criteria (1). These kindreds were assessed by a single research nurse with 4 years experience in dermatology (AJ), and comprised 28 families with a total of 40 affected siblings [25 male (41%) and 36 female (59%)] all present in single generations. The mean age of onset of psoriasis in the ASP1 panel was aged 14.6 (range 3 45) years. DNA was extracted from peripheral venous blood from all affected family members and willing first degree relatives and stored at 70 C until genotyped. Approval of this study was provided by the Guy s and St. Thomas Hospitals Ethics Committees. Genotyping Genotyping was performed for 260 highly polymorphic microsatellite markers spanning the 22 autosomes as previous described (27). For further details of markers used in the study, rtrembat@hgmp.mrc.ac.uk. Using data from the Genethon human linkage map (28), the spacing between adjacent markers was estimated to be 14.1 cm. For a total of 66 markers the interval was greater than 14 cm with the greatest being 38 cm between D1S107 and D1S252, hence a total of 97.1% of the genome lies within 20 cm of a marker and the mean heterozygosity of markers used was 78.9%. Genotypes at each marker were determined using fluorescencelabelled primers and semi-automated techniques, as described previously (22). Briefly, conditions for PCR were optimised for each primer pair over a range of annealing temperatures (53 61 C) and magnesium concentrations ( mm MgCl 2 ). Amplifications were performed in 96-well microtitre plates (Costar) on MJ Research machines (PTC 225). Each 15 µl PCR mixture contained 50 ng DNA, 0.2 U Taq polymerase, 40 ng of each primer, 1.5 µl 10 potassium chloride buffer and magnesium chloride in distilled water. Pooled amplified DNA was electrophoresed on 6% acrylamide gels, which were run for 4.5 h at 900V using model 373A automated sequencer (Applied Biosystems). Semi-automated DNA fragment sizing was performed using GENESCAN TM 672 (version 1.1) software, and genotyping carried out using GENOTYPER TM version 1.1 software (Applied Biosystems). HLA-C genotypes were determined by locus-specific amplification of the polymorphic sequences that encode the peptide-binding α1 and α2 domains of the class 1 protein (28). Alleles were detected using 26 sequence specific oligonucleotide probes as these DNA based methods have been shown to detect all common heterozygous individuals (21) detecting alleles beyond the resolution of presently available serological methods. Statistical analysis Full nonparametric analysis. The data were analyzed with a novel nonparametric analysis method which partitions the overall evidence of linkage into three components: (a) linkage within sibships; (b) linkage between sibships; (c) association between pedigrees (i.e. linkage within populations between pedigrees). (a) Linkage within sibships Initially, extended pedigrees were divided into component nuclear pedigrees and affected sib-pair analysis was undertaken with the method described by Satsangi et al. (22) which is essentially equivalent to the LOD score calculated under the assumption of a simple recessive disease model with phase-unknown matings. Briefly, the likelihood contribution for meioses from a heterozygous parent with n affected offspring, of which m inherited one marker allele and n m the other, is [ρ m (1 ρ) n m + ρ n m (1 ρ) m ], and for the whole family the contribution is the product of the two parental contributions. If parental contributions are missing, the likelihood is a sum of terms corresponding to each of the possible parental genotype combinations, weighted by the genotype frequencies calculated under the assumption of Hardy-Weinburg equilibrium and incorporating information on all offspring (assuming Mendelian segregation for unaffected offspring). The likelihood ratio test statistic, 2ln[L(ρ)/L(ρ = 0.5)] has been shown in sharp contrast to traditional sibpair methods, to have a null hypothesis distribution that is almost perfectly a mixture of a point mass at 0, and a χ 2 distribution (22,23). The likelihood computed for each value of ρ [denoted L WS (ρ)] comprises the first element of the combined test statistic. (b) Linkage between sibships within pedigrees In contrast to the situation for sibpairs, in the absence of consanguinity more distantly related individuals can share at most one allele IBD at the trait locus from a single common ancestor. Linkage analysis of such data should be nearly equivalent to considering the pedigree likelihood under a dominant mode of inheritance. To implement a non-parametric statistic, we first trimmed the extended pedigrees of all unaffected persons with no affected descendants (as they contain no information about IBD among the affected individuals in the pedigree). Then for any siblings left, they are further truncated such that one and only one affected sibling per sibship is included in the analysis; the affected sib is selected at random from the sibship. If there are multiple sibs each of whom has an affected descendant, then they are all included (of necessity) in this information twice in the joint analysis. What is left is a minimal set of affected, connecting and married-in individuals. The analysis is undertaken assuming a two-allele trait locus with genotypes 1/2 for all affected and connecting individuals, 1/1 for all married-in, unaffected individuals and unknown phenotype who could have transmitted the disease. The frequency of this 2 allele is set infinitesimally small such that only one of these parents is likely to carry the trait allele. Similarly, all individuals who have parents in the pedigree, and affected descendants are

7 819 Human Nucleic Molecular Acids Research, Genetics, 1994, 1997, Vol. Vol. 22, No. 6, No given trait locus genotype 1/2, since if the affecteds received the trait allele IBD this is necessarily the genotype of each intermediate individual. Analogous with the above, the likelihood, taking account of the marker locus is written as a function of µ where µ is equivalent to the recombination fraction under the simple dominant model. The likelihood ratios [denoted L BS (µ)] comprises the second element of our combined test statistic. The distribution of the likelihood ratio statistic itself, 2ln [L BS (µ)/l BS (µ= 0.5)] is also a mixture of point mass at 0 and χ 2. (c) Allelic association between pedigrees The final component of the test statistic measures the association of the marker and the trait loci which could either be due to linkage disequilibrium or a direct aetiological relationship between the marker and disease. A haplotype relative risk approach has been adopted in which one affected individual per pedigree is ascertained to determine case/control alleles. The selection criteria are: (i) first look for trios of an affected offspring with both parents all typed at the marker locus. If within a single pedigree a set of such trios is found, one is randomly chosen for inclusion in the analysis, the genotype of the affected offspring is used as the case sample genotype from this pedigree, and the control sample genotype is obtained from the two alleles which were not transmitted to this child from his parents. (Note that heterozygous and homozygous parents are both used, as this is more powerful in an association test than discarding the homozygotes as in the TDT linkage test). If no such trio is found in a pedigree, an affected offspring who is genotyped and has at least one parent genotyped is selected; in this case there would be two transmitted alleles identified and only one non-transmitted allele in the study. As a last resort, an affected individual with no genotyped parents is selected to contribute two alleles to the case sample in lieu of discarding all information in the pedigree. The likelihood is computed following the method of Terwilliger (1995) (30) as a function of λ, the percentage of the total possible association which is observed in this sample (i.e. it is on a scale between 0 and 1 with 0 meaning no association and 1 meaning complete association). The likelihood thus computed is then denoted L LD(λ). The distribution of the statistic, 2ln [L LD(λ)/L LD (λ=0)] is unfortunately very complicated, but assuming a mixture of χ 2 and point mass at 0 is very conservative and yet very powerful. In reality the point mass at zero can be as large as 80 90% for some markers under the null hypothesis (31). (d) Overall evidence of linkage and association To ensure a conservative overall test statistic, each of the three likelihoods was maximised independently over their respective parameters. The overall likelihood test is then L Linkage = [maxl WS (θ)] maxl BS(θ)/L WS (θ = 0.5) L BS (θ =0.5), and the overall likelihood test for linkage and association together is just L complete = [L Linkage)(maxL LD(λ)]L LD(λ = 0). The distribution of these likelihood statistics is derived from the independent distributions of their independent components, and p-values are thus derived. Note that this is less than optimally powerful, since if there is linkage there is some relationship between the θs from the within and between sibpairs analyses, and taking advantage of this relationship may be more powerful. This has not been exploited in this study as even this more conservative approach was very significant in this set of data for psoriasis. ACKNOWLEDGEMENTS G. Mark Lathrop is a Wellcome Trust Principal Fellow. Joseph D. Terwilliger is supported by a Hitchings-Elion Fellowship from the Burroughs Wellcome Fund and by a NIH grant (HG000008). We thank Susan Daniels for help with computer programmes and Dr K. Welsh for advice on HLA C typing. The financial support of the Psoriasis Association of Great Britain and Henry Smith Charity (RCT, RDC and JNWNB) the Dunhill Medical Trust (JNWNB) and the Wellcome Trust (RCT and JNWNB) are most gratefully acknowledged. We wish to record our sincere appreciation for the support of the many dermatologists who have notified us of families with psoriasis and to the patients and their families for their forbearance in participating in these studies. REFERENCES 1. Camp, R.D.R. (1992) Psoriasis. In: Textbook of Dermatology, Champion, R.H., Burton, J.L. and Ebling, F.J.G., eds. Oxford, Blackwell, pp Jobling, R.G. (1976) Psoriasis a preliminary questionnaire study of sufferers subjective experience. Clin. Exp. Dermatol., 1, Baadsgaard, O., Fisher, G., Voorhees, J.J. and Cooper, K.D. (1990) The role of the immune response system in the pathogenesis of psoriasis. J. Invest. Dermatol., 95, Barker, J.N.W.N. (1991) Pathophysiology of psoriasis. Lancet, 338, Elder, J.T., Nair, R.P. and Voorhees, J.J. (1994) Epidemiology and the genetics of psoriasis. J. Invest. Dermatol., 102, 24S 27S. 6. Tomfohrde, J., Silverman, A., Barnes, R., Fernandez-Vina, M.A., Young, M., Lory, D., Morris, L., Wuepper, K.D., Stastny, P., Menter, A. and Bowcock, A. (1994) Gene for familial psoriasis susceptibility mapped to the distal end of human chromosome 17q. Science, 264, Matthews, D., Fry, L., Powles, A., Weissenbach, J. and Williamson, R. (1995) Confirmation of genetic heterogeneity in familial psoriasis. J. Med. Genet., 32, Nair, R.P., Guo, S.W., Jenisch, S., Henseler, T., Lange, E.M., Terhune, M., Westphal, E., Christophers, E., Voorhees, J.J. and Elder, J.T. (1995) Scanning chromosome 17 for psoriasis susceptibility: lack of evidence for a distal 17q locus. Hum. Hered., 45, Matthews, D., Fry, L., Powles, A., Weber, J., McCarthy, M., Fisher, E., Davies, K. and Williamson, R. (1996) Evidence that a locus for familial psoriasis maps to chromosome 4q. Nature Genet., 14, Rosbotham, J., Barker, J.N.W.N. and Trembath, R.C. (1995) Exclusion of interleukin-1 receptor antagonist as a primary determinant for psoriasis. J. Invest. Dermatol., 104, Wyatt, R.J., Wang, C., Hudson, E.C., Jones, R.M., Noah, P.W. and Rosenberg, E.W. (1989) Complement phenotypes in patients with psoriasis. Hum. Hered., 39, Russel, T.J., Schultes, L.M. and Kuban, D.J. (1972) Histocompatibility (HLA) antigens associated with psoriasis. N. Engl. J. Med., 287, White, S.H., Newcomer, V.D., Mickey, M.R. and Terasaki, P.I. (1972) Disturbances of HLA antigen frequency in psoriasis. N. Engl. J. Med., 287, Gottlieb, A.B. and Krueger, J.G. (1990) HLA region genes and immune activation in the pathogenesis of psoriasis. Arch. Dermatol., 126, Nair, R.P., Henseler, T. and Lunetta, K. (1993) Exclusion of tight linkage between familial psoriasis and HLA genes under a single gene autosomal dominant model. J. Invest. Derm., 100, 539(abstr). 16. Elder, J.T., Nair, R.P., Guo, S., Henseler, T., Christophers, E. and Voorhees, J.J. (1994) The genetics of psoriasis. Arch. Dermatol., 130, Lander, E. and Kruglyak, L. (1995) Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nature Genet., 11, Risch, N. (1987) Assessing the role of HLA-linked and unlinked determinants of disease. Am. J. Hum. Genet., 40, Lomholt, G. (1993) Psoriasis, prevalence, spontaneous course and genetics. (G.E.C Gad), Copenhagen. 20. Suarez-Almazor, M. and Russel, A. (1990) The genetics of psoriasis. Haplotype sharing in siblings with the disease. Arch. Dermatol., 126,

8 820 Human Molecular Genetics, 1997, Vol. 6, No Kennedy, L.J., Poulton, K.V., Dyer, P.A., Ollier, W.E.R., and Thompson, W. (1995) Definition of HLA-C alleles using sequence specific oligonucleotide probes (PCR-SSOP). Tissue Ant. 46, Satsangi, J., Parkes, M., Louis, E., Hashimoto, L., Kato, N., Welsh, K., Terwilliger, J.D., Lathrop, G.M., Bell, J. and Jewell, D.P. (1996) Two-stage genome wide search in inflammatory bowel disease: evidence for susceptibility loci on chromosomes 3,7,and 12. Nature Genet., 14, Kuokkanen,S., Sundvall,M., Terwilliger, JD., Tienari P. J., Wikstrom,J., Holmdahl,R., Pettersson,U and Peltonen,L. (1996) A putative vulnerability locus to multiple sclerosis maps to 5p14-p12 in a region syntenic to the murine locus Eae2. Nature Genet., 13, Zhong, F., McCombs, C.C., Olson, J.M., Elston, R.C., Stevens, F.M., McCarthy, C.F. and Michalski, J.P. (1996) An autosomal screen for genes that predispose to celiac disease in western counties of Ireland. Nature Genet., 14, Rosbotham, J., Trembath, R.C., Glover, M., Leigh, I.M. and Barker, J.N.W.N. (1994) An association of psoriasis and hereditary multiple exostosis. A clue for the mapping of psoriasis susceptibility genes? Br. J. Dermatol., 130, Ahn, J., Ludecke, H.J., Lindow, S., Horton, W.A., Lee, B., Wagner, M.J., Horsthemke, B. and Wells, D.E. (1995) Cloning of the putative tumour suppressor gene for hereditary multiple exostoses (EXT1). Nature Genet., 11, Reed, P.W. (1994) Chromosome-specific microsatellite sets for fluorescencebased semi-automated genome mapping. Nature Genet., 7, Dib, C., Faure, S., Fizames, C., Samson, D., Drouot, N., Vignal, A., Millasseau, P., Marc, S., Hazan, J., Seboun, E., Lathrop, M., Gyapay, G., Morissette, J. and Weissenbach, J. (1996) A comprehensive genetic map of the human genome based on 5,264 microsatellites. Nature, 380, Zemmour, J. and Parham, P. (1993) HLA class 1 nucleotide sequences. Immunobiology, 187, Terwilliger, J.D. (1995) A powerful likelihood method for the analysis of linkage disequilibrium between trait loci and one or more polymorphic marker loci. Am. J. Hum. Genet., 56, Sham, P.C., Curtis, D. and MacLean, C.J. (1996) Likelihood ratio tests for linkage and linkage disequilibrium: asymptotic distribution and power. Am. J. Hum. Genet., 58,