Association of LY9 in UK and Canadian SLE Families

ORIGINAL ARTICLE Association of LY9 in UK and Canadian SLE families

DS Cunninghame Graham1, TJ Vyse1,11, PR Fortin2,11, A Montpetit3, Y-c Cai4, S Lim5, T McKenzie4, L Farwell6, B Rhodes1, L Chad3, TJ Hudson3, A Sharpe7, C Terhorst8, CMT Greenwood5, J Wither9,12, JD Rioux6,10,12 and CaNIOS GenES Investigators13 1Section of Molecular Genetics and Rheumatology, Imperial College Faculty of Medicine, Hammersmith Hospital, London, UK; 2University of Toronto Lupus Clinic, Centre for Prognosis Studies in the Rheumatic Diseases, Toronto Western Hospital, University Health Network; Department of Medicine, University of Toronto, Toronto, Ontario, Canada; 3McGill University and Genome Quebec Innovation Centre, Montreal, Quebec, Canada; Ontario Institute for Cancer Research, Toronto, Ontario, Canada; 4Toronto Western Hospital Research Institute, University Health Network, Toronto, Ontario, Canada; 5Program in Genetics and Genome Biology, The Hospital for Sick Children Research Institute, Toronto, Ontario, Canada; 6The Broad Institute of MIT and Harvard, 7 Cambridge Center, Cambridge, MA, USA; 7Harvard Medical School, Pathology, NRB-837, Boston, MA, USA; 8Division of Immunology, BIDMC, Harvard Institute of Medicine, Boston, MA, USA; 9Arthritis Centre of Excellence; Division of Genetics and Development, Toronto Western Hospital Research Institute, University Health Network; Departments of Medicine and Immunology, University of Toronto, Toronto, Ontario, Canada and 10University of Montreal and the Montreal Heart Institute, Research Center, Montreal, Quebec, Canada

Systemic lupus erythematosus (SLE) is a complex disease trait of unknown aetiology. Genome-wide linkage studies in human SLE identified several linkage regions, including one at 1q23, which contains multiple susceptibility genes, including the members of the signalling lymphocyte activation molecule (SLAM) locus. In mice there is a syntenic linkage region, Sle1. The SLAM genes are functionally related cell-surface receptors, which regulate signal transduction of cells in the immune system. Family-based association study in UK and Canadian SLE families identified variants in the promoter and coding region of SLAMF7 and LY9 contributing to SLE disease susceptibility. The strongest association was from rs509749, in exon 8 of LY9 (P ¼ 0.00209). rs509749 encodes a Val/Met nonsynonymous change in amino acid 602 in the cytoplasmic domain of LY9. In the parents and affected individuals from the Canadian SLE families, the risk allele of rs509049 skews the T-cell population by increasing the number of CD8 þ memory T cells, while decreasing the proportion of CD4 þ naı¨ve T cells and activated T cells. Since rs509749 lies within the consensus binding site for SAP/SH2D1a, which influences downstream signalling events from LY9, the mechanism for increased CD8 þ memory T cells may include differential binding SAP/SH2D1a to the cytoplasmic domain of LY9. Genes and Immunity (2008) 9, 93–102; doi:10.1038/sj.gene.6364453; published online 24 January 2008

Keywords: SLE; LY9; SLAMF7; genetic association and T cells

Introduction of studies have been published, twin concordance rates provide additional evidence for a genetic contribution in Epidemiological studies reveal a significant genetic SLE; the concordance rate is significantly greater in contribution to the pathogenesis of systemic lupus monozygotic than dizygotic twins, 24 vs 2% in one study erythematosus (SLE). The relative risk to siblings of and 57 vs 5% in a smaller study.2,3 Although small, the affected individuals (ls) is estimated to be 15- to 20-fold twin studies indicate a 10-fold increase in disease over that of the unrelated population.1 This increase in concordance in monozygotic twins compared with risk is similar in magnitude to that seen in several other nonidentical twins. autoimmune diseases, such as multiple sclerosis and Over the last several years, a number of groups have type 1 diabetes mellitus. Although only limited numbers published genome-wide and targeted linkage analyses in SLE4–7 There is considerable heterogeneity within these linkage studies, no doubt reflecting small study sizes, Correspondence: Dr DS Cunninghame Graham, Section of Mole- cular Genetics and Rheumatology, Imperial College Faculty of genetic heterogeneity and clinical heterogeneity in SLE. Medicine, Hammersmith Hospital, Hammersmith Campus, First The most consistently mapped lupus susceptibility loci Floor J Block, Room 118/119, Du Cane Road, London W12 ONN, reside in the following regions: 1q23, 1q25–31, 1q41–42, UK. 2q35–37, 4p16–15.2, 6p11–21 (MHC) and 16q12, all of E-mail: [email protected] 11 which have been corroborated at least once in an This is a co-second author. independent cohort.8,9 Of these loci, the evidence for 12This is a co-last author. 13The complete list is shown in the Appendix. the 1q23 locus is therefore striking since (1) it has been Received 17 September 2007; revised 23 November 2007; accepted identified in multiple genome-wide linkage scans in 26 November 2007; published online 24 January 2008 humans, (2) it has been replicated in subsequent linkage Association of LY9 in SLE families DS Cunninghame Graham et al 94 studies that have targeted this region and (3) the syntenic susceptibility, two family collections, one from the region in mice has also been linked in more than one United Kingdom and the other from Canada, were mouse model of spontaneous lupus. Specifically, in independently typed for markers across this locus. Both humans this region has been identified in three genome research groups selected single nucleotide polymorph- wide scans and in two replication studies. Furthermore, isms (SNPs) across the region from the SLAMF7 gene at the evidence of linkage in these studies was found in the telomeric end of this 1p locus all the way to the populations of European and African derived popula- intelectin genes at the centromeric end. The genomic tions.4,5,10–14 Finer localization is not possible with these organization of the SLAM locus, from SLAMF7 through studies since the sample sizes and linkage approach limit to ITNL1, is shown in Figure 1. In both populations, the the power of resolution. Nonetheless, the studies are composition of which is shown in Table 1, the initial very informative in the sense that they consistently genotyping showed evidence of association for multiple provide evidence for linkage to this region. SNPs around SLAMF7/LY9 (Table 3). To maximize the In mice, genome-wide linkage studies have implicated power of this candidate region association approach both the region that is syntenic to the chromosome 1 region groups were interested in combining the two datasets, identified in human studies, in three different models of and so a second round of genotyping was performed in spontaneous lupus: the F1 intercross (New Zealand both populations to increase the density of markers in Black Â New Zealand White; NZB/W), New Zealand this region and to ensure that associated SNPs from Mixed/Aeg2410 New Zealand mice (NZM) and BXSB either population were typed in both set of samples. This mouse strains.15–17 The phenotype of these mice is very meant that the final list of 40 markers genotyped across similar to that in SLE patients, with the production of the locus included a total of 31 SNPs genotyped in the autoantibodies as well as multi-organ involvement, 461 UK SLE nuclear families and an overlapping set of 32 including severe nephritis. Interestingly, it has been markers typed in the 271 Canadian families (Table 2). shown that this locus is necessary for the production of The genotyping in both populations was subject to nephrophilic autoantibodies and clinical glomerulone- quality control, as described in Materials and methods, phritis.18 Fine mapping of this locus in one mouse strain to determine the final list of markers suitable for the has demonstrated the presence of a cluster of genes that combined analysis. Since the list of markers typed in contribute to disease susceptibility.19 This latter observa- both populations was overlapping, there was a common tion is very consistent with the human linkage data core of 23 variants genotyped in both populations. The insofar as a cluster of disease susceptibility genes is likely identities of all analysable markers, together with those to lead to a consistently observed linkage region as is the removed following quality control are listed in Table 2 case for 1q23 in human SLE. and the chromosomal location of these markers is Within the 1q23 linkage region, the cluster of genes in illustrated in Figure 1. Using this common core of 22 the signalling lymphocyte activation molecule (SLAM) variants, a cross population comparison of parental locus constitutes a strong set of candidate genes. This minor allele frequencies was used to demonstrate that locus is a cluster of functionally related genes that belong there were no significant differences in parental allele to the CD2 superfamily of molecules that regulate the frequency between the two populations for any of the responses of cells of the immune system. These mole- genotyped markers across the SLAM locus, therefore cules are expressed in T cells, B cells, NK cells, myeloid supporting a combined analysis of these two populations and plasmacytoid dendritic cells, macrophages and (Table 2). monocytes and are involved in the regulation of immune responses in a wide variety of cell types. Six molecules in TDT analysis for single SNPs this cluster (SLAMF6 (Ly108), CD84, SLAMF1 (SLAM), The results of the transmission disequilibrium test (TDT) SLAMF7 (CS1, CRACC), LY9 (CD229) and CD244) recruit analysis using TRANSMIT, in both complete trios and active src kinases to their cytoplasmic tails using the single parent families, which have one or more un- adaptors of SAP (SLAM-associated protein), in T cells, affected siblings, are presented in Table 3. These data NK cells and thrombocytes and EAT-2, in antigen- demonstrate that there were two variants in the LY9- presenting cells (APCs) and B cells,20–22 while the lipid- CD244 region of the SLAM cluster showing significant linked CD48 interacts with src kinases in lipid rafts association (Po0.05) in the UK SLE families: rs509749 in present in the plasma membrane of T cells and APCs. LY9 (P ¼ 0.0149) and rs480104 in CD244 (P ¼ 0.0145). Recently, genetic polymorphisms in this cluster in There were also multiple associated variants in the different mouse strains have been found to be associated region around the intelectin genes, ITNL1 and ITNL2, the with production of anti-nuclear autoantibodies, a hall- strongest of which was a variant in the ITLN1–ITLN2 mark of SLE, as well as functional alterations of T and B intergenic region, rs4656959 (P ¼ 0.00823). In the Cana- cells.23,24 Thus, emerging data provide strong arguments dian samples, there was significant association from a that these genes play an important role in the key second variant within LY9, rs3817407 (P ¼ 1.17 Â 10À3), processes believed to lead to SLE and therefore represent but no significant evidence of association in the ITLN1– strong functional candidates. ITLN2 intergenic region. The association of these variants from LY9 in two independent populations suggest that LY9 may contribute to disease susceptibility, but there Results was no significant evidence for association of other genes in these populations. Two independent association screens of the SLAM locus Before carrying out TDT analysis on the combined Given the strong prior probability, based upon human UK/Canadian dataset, we demonstrated homogeneity in and mouse linkage and functional results, for the the proportions of transmitted to untransmitted alleles molecules in the SLAM locus having a role in SLE between the UK and Canadian populations, using

Genes and Immunity Association of LY9 in SLE families DS Cunninghame Graham et al 95

Figure 1 Genomic organisation and single nucleotide polymorphisms (SNPs) genotyped in the UK and Canadian families. The location of the signalling lymphocyte activation molecule (SLAM) locus, at 1q23, showing the order of the genes along chromosome 1 illustrated according to the direction of translation, with the positions according to the scale indicated. The SNPs genotyped are numbered according the SNPs identified in Table 2. The associated SNPs are marked with an asterisk and the P-values quoted in the joint UK/Canadian dataset from Table 3.

Table 1 Composition of the UK and Canadian SLE study cohorts associated variants that were genotyped across the locus. Therefore, given the overall similarity between the UK Collection type Sample collection Numbers of familiesa and Canadian samples, to maximize the information from both populations, it was deemed appropriate to Parent-affected trios UK SLE families 461 undertake TRANSMIT-TDT analysis on a single com- Canadian SLE families 271 bined dataset, by merging all the samples from the UK Single-parent families UK families 199 Canadian families NA and Canadian families into a single dataset (Table 3). The strongest association was found for the coding SNP rs509749, located in exon 8 of LY9. There was under- Abbreviation: NA, not available; SLE, systemic lupus erythemato- transmission of the rare G allele of rs509749 (P ¼ 0.00209). sus. This SNP causes a nonsynonymous change of residue Details of the sample collections included in this study: UK SLE families and Canadian SLE families. 602 (Met-602 (A) to Val-602 (G)) in the cytoplasmic aThe total numbers of UK families was 521. This was made up from domain of LY9. There is a second variant, rs1333065, 461 trios of which 322 had siblings and 199 single parent families. In located 6.1 kb upstream of LY9 which also shows under- the Canadian families, 103 of the 271 families had siblings. transmission of the rare G allele (P ¼ 0.025). The association in rs509749 (P ¼ 0.0292) was also confirmed in complete trios from the joint UK/Canadian dataset (Supplementary Table 1). In the joint UK/Canadian Pearson’s w2 test (Supplementary Table 2) and homo- dataset, there is an overtransmitted haplotype which is geneity in odds ratios by the Breslow–Day test (P40.05) tagged by the overtransmitted A alleles of both rs509749 in both populations for all markers tested, except for and rs1333065 (P ¼ 0.0177). This haplotype, which rs493646, in the upstream region of SLAMF7. rs493646 stretches 34 kb across LY9 and into the first intron of does not impact on the observed pattern of association CD244, carries the A alleles of rs1333065, rs3817407, because it shows an r2 of less than 0.12 with the other rs509749 and the G allele of rs480104. When the

Genes and Immunity 96 ee n Immunity and Genes

Table 2 Selection of markers across the SLAM locus

Marker SNPs Genomic Location Gene Function Population For Exclusion Population Alleles Parental MAF comparison identified (UCSC March 2006) initial genotyping criterion for SNP analysis UK Canada Cross population P-value

rs352685 1 158948460 CD48 50 near gene UK/Can HWE GoC 0.43 0.36 0.311 rs352692 2 158953118 CD48-SLAMF7 UK/Can UK/Can GoA 0.39 0.37 0.771 rs493646 3 158960499 CD48-SLAMF7 UK/Can UK/Can GoC 0.35 0.39 0.561 rs503832 4 158962928 CD48-SLAMF7 Can Can ToC 0.33 rs576627 5 158969099 CD48-SLAMF7 UK ME CoT 0.15 rs579272 6 158969415 CD48-SLAMF7 UK/Can UK/Can GoA 0.34 0.37 0.657 rs983494 7 158970589 CD48-SLAMF7 UK/Can UK/Can GoA 0.23 0.21 0.732 rs489286 8 158989174 SLAMF7 I6 UK/Can UK/Can GoA 0.35 0.33 0.765 rs526030 9 158990929 SLAMF7 30 UTR UK/Can UK/Can ToC 0.03 0.03 1.000 rs522606 10 159004192 SLAMF7-LY9 UK HWE CoT 0.33 rs564883 11 158999408 SLAMF7-LY9 UK/Can UK/Can ToC 0.35 0.29 0.363 of Association rs1333065 12 159026143 SLAMF7-LY9 UK/Can UK/Can GoA 0.34 0.33 0.882 Graham Cunninghame DS rs540224 13 159029994 SLAMF7-LY9 UK ME AoG 0.29 rs579589 14 159054349 LY9 I5 Can Can GoA 0.31 rs3817407 15 159054755 LY9 I6 UK/Can UK/Can GoA 0.24 0.23 0.867 rs2027015 16 159058402 LY9 I7 UK ME GoA 0.41 LY9 a rs509749 17 159060184 LY9 E8 UK/Can UK/Can GoA 0.42 0.45 0.669 families SLE in rs480104 18 159067670 CD244 30 UTR UK/Can UK/Can CoG 0.22 0.22 1.000 rs1319651 19 159075100 CD244 I4 UK ME GoC 0.49 rs2990703 20 159078454 CD244 I1 Can Can CoT 0.46 rs1855188 21 159096329 CD244 I1 Can Can CoT 0.22 al et rs955370 22 159101975 CD244-ITLN1 UK/Can UK/Can ToG 0.20 0.22 0.728 rs955371 23 159102207 CD244-ITLN1 Can Can ToC 0.22 rs869167 24 159105953 CD244-ITLN1 UK/Can UK/Can ToC 0.31 0.31 1.000 rs952804 25 159111369 CD244-ITLN1 UK HWE ToC 0.41 rs4656955 26 159114034 ITLN1 I7 UK/Can UK/Can GoA 0.33 0.32 0.879 rs3766356 27 159115561 ITLN1 I7 UK/Can UK/Can ToC 0.34 0.32 0.764 rs2236515 28 159117560 ITLN1 I5 UK/Can UK/Can ToC 0.32 0.31 0.879 rs2274908 29 159118518 ITLN1 E3b UK ME GoA 0.31 rs2274910 30 159118670 ITLN1 I3 UK ME ToC 0.38 rs2039415 31 159121069 ITLN1 I2 UK/Can UK/Can ToC 0.31 0.30 0.877 rs4656959 32 159125585 ITLN1-ITLN2 UK/Can UK/Can GoA 0.35 0.32 0.653 rs2225591 33 159142380 ITLN1-ITLN2 UK/Can UK/Can GoA 0.33 0.32 0.879 rs2184064 34 159154099 ITLN1-ITLN2 UK/Can UK/Can AoC 0.30 0.31 0.877 rs7516660 35 159164281 ITLN1-ITLN2 UK/Can UK/Can GoT 0.32 0.31 0.879 rs2774277 36 159166105 ITLN1-ITLN2 UK/Can UK/Can AoG 0.09 0.09 1.000 rs6693472 37 159187673 ITLN2 E4c Can Can GoA 0.02 rs2255842 38 159193450 ITLN2-F11R Can Can GoA 0.30 rs1556260 39 159281070 USF1 I1 Can Can AoG 0.19 rs1556259 40 159281273 USF1 I1 Can Can CoT 0.19

Abbreviations: Can, Canada; SLAM, signalling lymphocyte activation molecule; SNP, single nucleotide polymorphism. The SNPs across the SLAM locus were numbered from 1 to 40 and the chromosomal position listed on the UCSC March 2006 genome assembly. The location of each SNP in relation to SLAM genes was shown and where the variant is located within a gene, the intron/exon was stated. Of a total of 40 markers identified, there are 23 in the UK population, 8 of which are unique to the UK samples. An overlapping 32 markers were typed in the Canadian samples, 9 of which are unique to the Canadian dataset. The population used for initial genotyping is listed, together with the reasons for exclusion of particular variants in the individual populations. There were a total of 23 SNPs analysed in both the Canadian and UK datasets, but one further SNP (rs352685) was removed from the analysis due because it had a Hardy–Weinberg (HWE) P-value of less than 0.05. SNPs marked ME were removed from the analysis because more than 5% families showed Mendelian errors. The alleles of each SNP are shown as minor allele ominor allele. The cross-population differences in parental minor allele frequency were quantified by w2 analysis. a rs509749 is a non-synonymous change (Val/Met) in exon 8 of LY9. b rs2274908 is a synonymous (His/His) change in E3 of ITLN1. c rs6693472 is a synonymous change (Gly/Gly) in exon 4 of ITNL2. Association of LY9 in SLE families DS Cunninghame Graham et al 97 Table 3 Analysis of single SNPs across the SLAM locus by TRANSMIT

rs number SNP Gene UK Canadian UK–Canadian analysed Fam Obs Exp w2 P-value Fam Obs Exp w2 P-value Fam Obs Exp w2 P-value rs352692 2 CD48-SLAMF7 25 123 94 95 0.0218 0.883 276 196 201 0.465 0.496 399 290 295 0.350 0.554 rs493646 3 CD48-SLAMF7 26 416 282 269 2.22 0.13 267 195 201 0.650 0.420 683 477 470 0.355 0.552 rs503832 4 CD48-SLAMF7 27 276 188 179 1.32 0.251 rs579272 6 CD48-SLAMF7 28 479 257 271 2.38 0.123 271 186 200 3.60 0.0578 750 443 472 5.93 0.0149 rs983494 7 CD48-SLAMF7 29 406 175 181 0.611 0.434 278 117 114 0.185 0.668 684 292 295 0.0801 0.777 rs489286 8 SLAMF7 30 367 254 238 3.39 0.0657 268 186 175 2.10 0.148 635 440 413 5.51 0.0189 rs526030 9 SLAMF7 31 367 21 17 2.25 0.134 272 14 18 2.30 0.129 639 35 35 0.00302 0.956 rs564883 11 SLAMF7-LY9 32 366 264 262 0.0385 0.845 268 164 155 1.73 0.189 634 428 417 0.888 0.346 rs1333065 12 SLAMF7-LY9 33 471 193 204 1.84 0.175 271 166 179 3.36 0.0700 742 359 383 4.97 0.0258 rs579589 14 LY9 35 275 176 177 0.0255 0.873 rs3817407 15 LY9 36 367 163 163 0.00193 0.965 270 103 125 10.5 1.17x10À3 637 266 287 4.33 0.0374 rs509749 17 LY9 37 510 377 403 5.92 0.0149 270 237 251 3.18 0.0744 780 614 655 9.47 0.00209 rs480104 18 CD244 38 437 157 177 5.98 0.0145 268 123 121 0.0905 0.764 705 280 298 3.02 0.0821 rs2990703 20 CD244 39 274 264 253 1.68 0.195 rs1855188 21 CD244 40 275 129 127 0.0815 0.775 rs955370 22 CD244-ITLN1 41 417 130 140 1.86 0.172 269 121 120 0.0509 0.822 686 251 260 0.831 0.362 rs955371 23 CD244-ITLN1 42 274 126 126 0.00113 0.973 rs869167 24 CD244-ITLN1 43 438 239 253 2.23 0.135 269 171 168 0.141 0.707 707 410 421 0.878 0.349 rs952804 25 CD244-ITLN1 44 502 269 251 5.09 0.0241 rs4656955 26 ITLN1 45 442 237 260 6.58 0.0103 268 175 174 0.0100 0.920 710 412 435 3.77 0.0520 rs3766356 27 ITLN1 46 439 242 258 2.88 0.0897 264 172 171 0.0112 0.916 703 414 429 1.60 0.206 rs2236515 28 ITLN1 47 431 226 244 4.34 0.0372 266 169 167 0.0457 0.831 697 395 412 2.17 0.141 rs2039415 31 ITLN1 48 441 233 254 5.50 0.0190 267 165 165 0.00150 0.969 708 398 420 3.44 0.0638 rs4656959 32 ITLN1–ITLN2 49 440 253 277 6.98 0.00823 250 170 167 0.166 0.684 690 423 444 3.32 0.0683 rs2225591 33 ITLN1–ITLN2 50 439 244 266 5.88 0.0154 268 175 176 0.0244 0.876 707 419 443 3.91 0.0479 rs2184064 34 ITLN1–ITLN2 51 426 198 219 6.53 0.0106 265 168 167 0.0176 0.894 691 366 386 3.31 0.0690 rs7516660 35 ITLN1–ITLN2 52 408 215 227 2.20 0.138 268 167 167 0.00197 0.965 676 382 395 1.28 0.257 rs2774277 36 ITLN1–ITLN2 53 442 70 81 3.46 0.0630 270 41 46 1.16 0.281 712 111 127 4.50 0.0339 rs6693472 37 ITLN2 54 271 2 3 0.920 0.338 rs2255842 38 ITLN2-F11R 55 268 160 160 0.00222 0.962 rs1556260 39 USF1 56 270 106 108 0.110 0.740 rs1556259 40 USF1 57 270 106 107 0.0777 0.780

Abbreviations: SLAM, signalling lymphocyte activation molecule; SNP, single nucleotide polymorphism; TDT, transmission disequilibrium test. TDT analysis by TRANSMIT for each SNP tested in the UK SLE families or in the Canadian dataset. A separate analysis was performed for the UK and Canadian families and also for joint UK/Canadian samples. The column marked ‘SNP analysed’, gives the order of analysed SNPs. For each variant the P-value is quoted with one degree of freedom. The column marked ‘Fam’ shows the number of families included in the analysis. For the minor allele of each variant, the number of observed transmissions (‘Obs’) and the number of expected transmissions (‘Obs’) is stated.

association of the SNPs tagging this risk haplotype were a single association signal from the two regions and not modelled in the joint UK/Canadian dataset, a dominant two independent signals. The results of these conditional model for association best fitted the pattern of associa- analyses showed that there was only a single association tion, with the strongest effect being for rs509749 signal from the two regions. This is because the effect of (P ¼ 0.0328). the SLAMF7-LY9 haplotype disappeared, when it was In an attempt to dissect out the causative alleles in LY9, estimated conditional on the variant showing the we used conditional logistic analysis to determine which strongest association in the ITLN1–ITLN2 region variants made the strongest contribution to the associa- (P ¼ 0.373) and the effect from the ITLN1–ITLN2 haplo- tion, using SNPs typed in both the UK and Canadian type disappeared when it was estimated conditional on families within LY9 and rs1333065 which is located 6.4 kb the synonymous coding variant in LY9, rs509749 immediately upstream of the gene. When the effects of (P ¼ 0.376) (Supplementary Table 4). each SNP were estimated conditional on the haplotype background across LY9, the nonsynonymous SNP Effect of LY9 polymorphism on T-cell subsets rs509749 showed the strongest reduction in association Since LY9 is expressed on peripheral T cells, we wanted (P ¼ 0.7) (Supplementary Table 3), which supports the to see whether LY9 polymorphisms affected the size of hypothesis that rs509749 makes a major contribution to individual T-cell populations in peripheral blood. Re- the association in LY9. In addition, because we found gression analysis in Canadian parental samples estab- associated variants in both the SLAMF7-LY9 region and lished that the genotype of the LY9 variant, rs509749, was in the intelectin genes ITLN1 and ITLN2 for the UK correlated to the relative numbers of the major T cell samples (Table 3), we used further conditional logistic subsets, since there was a significant association between regression analysis to investigate whether there was only the A allele of rs509749 and a decrease in the proportion

Genes and Immunity Association of LY9 in SLE families DS Cunninghame Graham et al 98 Table 4 Association between rs509749 genotype and cellular phenotype in Canadian parental samples

Cell type Cellular phenotype b-Coefficient s.e. t-statistic P-value

CD4+ memory/effector cells CD4+ CD45RO+ 1.872 1.062 1.763 0.079 CD4+ naı¨ve T cells CD4+ CD45RA+ À2.057 0.861 À2.390 0.017 Activated CD4+ T cells CD4+ HLADR+ À1.534 0.682 À2.248 0.025 CD8+ memory/effector cells CD8+ CD45RO+ 2.011 0.813 2.473 0.014 CD8+ naı¨ve T cells CD8+ CD45RA+ À1.732 1.104 À1.569 0.117 Recently activated CD4+ T cells log-CD4+ CD69+ 0.054 0.036 1.478 0.140 Invariant NKT cells log-CD3+ Va24+ Vb11+ 0.012 0.041 0.299 0.765

Regression analysis for cellular phenotypes in Canadian SLE parental samples. The values for the b-coefficient are calculated by SPSS from the percentages of each cell type in the raw FACS data.

of both CD4 þ naı¨ve T cells (P ¼ 0.017) and activated T However, when SAP is bound to LY9, the downstream cells (P ¼ 0.025) but an increase in the number of CD8 þ signalling molecule SH-2 cannot be activated and memory T cells (P ¼ 0.014) (Table 4). Similar trends in remains in its inactive state.28 Inactive SHP-2 promotes T-cell populations were observed in SLE affected apoptosis, but active SHP-2 (bound to the phosphory-

individuals (data not shown). lated pY603 in LY9), will promote the production of inflammatory cytokines,28 as illustrated in Figure 2. Interestingly, the SNP showing the strongest associa- Discussion tion in LY9, rs509749, is a nonsynonymous variant lying within the consensus binding site for SAP/SH2D1a

Seven members of the SLAM family are located within (T601-(I/V)602-pY603-N604-N605-(I/V)606; Figure 2). The var- the SLE linkage region on human chromosome 1 (1q23), iant alleles of rs509749 variant are located in the first and the syntenic linkage region in mouse (Sle1b). Linkage position of the codon for (I/V602). Since SAP/SH2D1a is studies in humans and in animal models of lupus have an adaptor molecule that directly interacts with LY9 and suggested an important role in SLE. This has been plays an important role in downstream signalling events, further supported by functional studies in mice and it is possible that the Met602Val change and other humans that have demonstrated that the SLAM and variation within the binding site may influence the SLAM-related molecules are cell-surface receptors ex- interaction between these molecules as well as in their pressed on multiple haematopoietic cells, where they signalling cascade. We, therefore, hypothesize that the mediate adhesion between T cells and APCs in the rs509479 mutation will affect the binding affinity of immune synapse and hence regulation of signal trans- SAP/SH2D1a for LY9 and thereby determine the direc- duction in B cells, T cells and APCs.25 The current family- tion of these downstream signalling events, as shown in based association testing of this genomic region in two Figure 2. If the A allele, coding for Met602, reduces the independent populations has provided significant evi- affinity of SAP/SH21Da for LY9, the phosphorylated dence for association to SLE. Specifically, multiple SNPs pT603 will be exposed to activate SHP-2, with consequen- within the SLAMF7/LY9 region were found to be tial increased production of inflammatory cytokines and associated in each of the studies and a combined analysis enhanced immune response. identified that the strongest association within this With this in mind, we used regression analysis in region was to rs509749, a nonsynonymous SNP in exon Canadian samples to demonstrate that the overtrans- 8ofLY9. We used conditional logistic regression to show mitted A allele of rs509749 affected proportions of that rs509749 was the variant making the strongest different T-cell sub-populations, because the risk allele contribution to the association observed from the of this variant was associated with decreased numbers of SLAMF7-LY9 region. CD4 þ naı¨ve T cells and activated T cells and with LY9 is expressed on the surface of monocytes and increased numbers of CD8 þ memory T cells (Table 4). present at low levels on NKT cells. However, there is This skewing in the T-cell populations may indicate a little information regarding the exact function of LY9. state of chronic T-cell activation. In Ly9À/À knockout mice there is a defect in T-cell Although we have presented evidence that rs509479 is proliferation and Th2 cytokine production, indicating a the SNP showing the strongest association in the UK and positive role for LY9 in T-cell function.26 Conversely, in Canadian SLE families, that it is located within the human T cells, co-ligation of LY9 with the T-cell receptor consensus binding site for SAP/SH2D1a, and that it is partially reduces ERK activation and production of the correlated with differences in specific T-cell populations, Th1 cytokine IFN-g, which may indicate a negative role it is formally possible that other variants in LD with for the protein in T-cell function.25 This apparent rs509479 could actually be the causal variants and/or discrepancy may be partly due to different down-stream that variants in addition to rs509479 have a role to play in signalling events following interaction of LY9 with one of determining an individual’s risk to developing SLE. its adaptor molecules, SAP/SH2D1a. The Src 2 homology In summary, therefore, the data presented in this paper domain of SAP/SH2D1a binds with high affinity to two represent an initial screen across the SLAM locus, and immunoreceptor tyrosine based switch motifs (ISTM) (T- the identification of novel associations within the LY9 (I/V)-pY-N-N-(I/V)) in the cytoplasmic tail of LY927 and gene. The variant showing the strongest association is is required for maximal phosphorylation of LY9.25 correlated to skewed populations of T cells in peripheral

Genes and Immunity Association of LY9 in SLE families DS Cunninghame Graham et al 99 CD4+ memory/effector cells CD8+ memory/effector cells CD8+ naive T cells activated CD4+ T cells CD4+ naive T cells apoptosis of activated cells inflammatory signals SAP

inactive (closed) SHP-2

activated (open) SHP-2

SAP SAP

pY603 pY603 LY9 ATG GTG Met 602 Val602 Risk Allele Protective Allele

LY9 SAP/SH2D1a T601 M602 Y603 A604 Q605 V606 Rsk allele A (Met ) 602

SAP/SH2D1a consensusT601 I/V602 Y603 N604 N605 I/V 606 Protective allele G (Val ) 602 genomic sequence ACC A/GTG TAT GCA CAA GTG

Figure 2 Hypothesis to explain the functional mechanism for the G allele of rs509479 in the immune response. (a) In the absence of SAP/

SH2D1a (SAP), SHP-2 can bind to pY603 in LY9 and become activated. Activated SHP-2 promotes the production of inflammatory signals. In the presence of SAP/SH2D1a (SAP), SHP-2 cannot bind to LY9 and behaves as an adaptor protein to stimulate activation of caspases and promote apoptosis. (b) Alignment of the consensus sequence for the LY9 SAP/SH2D1a sequence, the SAP/SH2D1a consensus sequence and the genomic sequence coding for the SAP/SH2D1a binding sites in LY9, showing both alleles of rs509749. blood. However, there are compelling arguments to chloroform extraction. For Canadian samples, the DNA support the other members of the SLAM family as SLE was extracted from whole blood using the Gentra susceptibility genes, and so future work will involve Autopure LS isolation system (Gentra Systems Inc., high-density mapping of the entire SLAM locus to Minneapolis, MN, USA) according to the manufacturer’s discover novel associations in these other genes and by instructions. the generation of a detailed haplotype structure in the region, to determine the relationship between the Selection of markers existing associations and unknown causal alleles. In this study a total of 33 markers were selected across the SLAM locus, which were identified from the public databases, dbSNP (http://www.ncbi.nlm.nih.gov/SNP/ Materials and methods index.html) and from HAPMAP (http://www.hapmap. Details of study cohorts org/). Following preliminary genotyping in 461 UK trios, The collection of the UK Caucasian SLE families markers were excluded from the analysis if they showed consisted of a total of 461 complete trios and 199 a genotyping success of less than 85%, had more than 5% independent single-parent families (Table 2). There were of families with Mendelian errors as identified by also 271 independent complete trios collected from PEDCHECK and/or had Hardy–Weinberg (HWE) several Canadian cities, 103 of which also had siblings. P-value in the parental samples of less than P ¼ 0.05 For risk alleles having minor allele frequencies 40–60%, (Table 2). Six markers from the list of SNPs typed in the the 732 families used in this collection have a power of UK samples were removed from the analysis, because 40–42% to detect an association of P ¼ 0.05 with an a they had more than 5% families showing Mendelian score of 0.1 (http://pngu.mgh.harvard.edu/~purcell/ errors, and two further markers were removed because gpc/dtdt.html). All probands conformed to the ACR they had a HWE P-value of less than 0.05 in the founder criteria for SLE29 with a diagnosis of SLE being chromosomes. A further marker genotyped in both the established by telephonic interview, health questionnaire UK and Canadian populations was removed from the and details from clinical notes. Written consent was analysis due to possessing a HWE P-value of less than obtained from all participants, including relatives. In the 0.05. However, none of the 32 markers typed in the United Kingdom, ethical approval was obtained from Canadian samples failed quality control. Multi-Centre Research Ethics Committee (MREC) and in Canada, the study was approved by the Research Ethics Genotyping methodology Board of the University Health Network and each The genotyping in the UK samples was performed using participating recruitment centre. MALDI–TOF mass spectrometry (Sequenom, San Diego, CA, USA)30 and analysis of the raw genotype data Preparation of genomic DNA carried out using the MassArray Typer v3.4 software Genomic DNA from the UK samples was isolated from (Sequenom). For the Canadian samples, genotyping was anti-coagulated whole blood by a standard phenol– performed by the SNPstream UHT platform (Beckman

Genes and Immunity Association of LY9 in SLE families DS Cunninghame Graham et al 100 Coulter, Mississauga, Ontario, Canada), with genotypes iant rs509749 in LY9 was used to condition (control for) being called by the UHTGetGenos software (Beckman the association from the variants in the ITLN1–ITLN2 Coulter). After visual inspection of the clusters, manual genes. adjustments were made for some of the assays, or by the The association between genotype at rs509749 and Sequenom iPlex system. cellular phenotype was evaluated in a mixed regression model using SPSS v.14 in Canadian parental samples. All Cellular phenotyping and serologic testing cellular data showed a normal distribution, with the Peripheral blood mononuclear cells (PBMC) were iso- exception of that for invariant NK T cells (CD3 þ lated from heparinized blood by Ficoll density gradient Va24 þ Vb11 þ ) and recently activated CD4 þ T cells centrifugation. Cells were isolated within 16–20 h of the (CD4 þ CD69 þ ). The data for both cell types showed blood drawing. Samples were transported, processed marked positive skews and required log-transformation and analysed by flow cytometry previously described by prior to analysis. Alleles at rs509749 were assumed to act Wither et al. (manuscript submitted to Arthritis in an additive (co-dominant) manner. and Rheumatism, 2007). The cellular phenotypes mea- sured were the percentage of invariant NKT cells (CD3 þ Va24 þ Vb11 þ ), CD4 þ memory/effector T cells (CD4 þ CD45RO þ CD45RA), CD4 þ naı¨ve T cells Acknowledgements (CD4 þ CD45ROÀ CD45RA þ ), recently activated CD4 þ T cells (CD4 þ CD69 þ ), CD8 þ memory/effector This work was funded by a grant from the National T cells (CD8 þ CD45RO þ CD45RAÀ) and CD8 þ naı¨ve Institutes of Allergy and Infectious Diseases (AI065687) T cells (CD8 þ CD45ROÀ CD45RA þ ). to Timothy J Vyse, John D Rioux and Cox Terhorst and Arlene Sharpe, through a Senior Fellowship Award from Statistical analysis the Wellcome Trust to Timothy J Vyse and by a grant All sample genotype and phenotype data were managed from the Canadian Institutes of Health Research (no. by, and analysis files generated with, BC/GENE and 62840) to Joan Wither and Paul Fortin. We acknowledge BC/CLIN software (Biocomputing Platforms Ltd, Espoo, the work of Christine Stevens and Angela Richardson for Finland). Alleles were counted in parental samples, as help with the genotyping of the UK samples, Andrew population controls, for each variant analysed across the Wong and Paul Spencer in recruiting patients and SLAM locus. These allele frequencies were calculated for families into the study and we would like to thank our each allele as a fraction of the total alleles for each SNP. A clinical colleagues for helping us recruit study partici- comparison of the parental allele frequencies between pants. Our thanks and appreciation is extended to all the the UK and Canadian collections was made from w2 patients and their relatives for generously donating analysis using a 2 Â 2 contingency table, with the level blood samples and all the general practitioners and of significance being presented as a P-value with practice nurses for collecting them. We appreciate the 1 d.f. Pearson’s test was used to compare the ratio of contribution to the genotyping from the Broad Institute transmitted to untransmitted (T:U) families for each Center for Genotyping and Analysis, which is supported marker and the Breslow-Day test in PLINK was used to by grant U54 RR020278-01 from the National Center for test homogeneity of odds ratios between the two Research Resources. We appreciate the contribution to populations. CaNIOS studies made by the Arthritis and Autoimmune Association of alleles to SLE for both individual and Research Centre Foundation and by Lupus Canada. Dr multiple SNPs was tested by TDT, which compares the Fortin’s salary is supported in part by a Distinguished observed and expected transmission of alleles from Senior Research Investigator Award from The Arthritis heterozygous parents to affected offspring. This analysis Society (Canada) and the Arthritis Centre of Excellence, was carried out using GENEHUNTER (complete University of Toronto. We thank Glinda Cooper, a GenES trios)31,32 or TRANSMIT (trios and single parent fa- co-investigator, for helpful discussions and Jaime O milies). Claudio, the National Scientific and Development To determine the variants on a haplotype making the Coordinator for CaNIOS for assistance in preparation strongest contribution to the haplotype association, of the article and Jiandong Su, CaNIOS Database conditional logistic regression, using WHAP (http:// Administrator for assistance with management of the pngu.mgh.harvard.edu/purcell//whap/) was per- GenES database. We also acknowledge the investigators formed in samples from the UK and Canadian paren- who contributed patients to the study (see Appendix). tal-affected trios. This analysis tested the individual contribution to the association in LY9 from SNPs located in the gene or immediately upstream, by conditioning on each constituent variant in turn. To determine whether References there were independent association signals in the SLAMF9-LY9 region and the ITLN1–ITLN2 regions, we 1 Rus V, Hochberg MC. The epidemiology of systemic lupus performed conditional logistic regression for each region. erythematosus. In: Wallace DJ, Hahn BH (eds). Dubois’ Lupus To investigate whether there was an independent signal Erythematosus. Lippincott, Williams and Wilkins: Baltimore, from the SLAMF7-LY9 region, the SNP showing the 2002, pp 65–83. 2 Deapen D, Escalante A, Weinrib L, Horwitz D, Bachman B, strongest association in the ITLN1–ITLN2 region, Roy-Burman P et al. A revised estimate of twin concordance in rs4656959, was used to condition the association from systemic lupus erythematosus. Arthritis Rheum 1992; 35: the variants in the SLAMF7-LY9 region. To find out 311–318. whether there was an independent signal from the 3 Block SR, Winfield JB, Lockshin MD, D’Angelo WA, Christian ITLN1–ITLN2 region, the nonsynonymous coding var- CL. Studies of twins with systemic lupus erythematosus. A

Genes and Immunity Association of LY9 in SLE families DS Cunninghame Graham et al 101 review of the literature and presentation of 12 additional sets. 17 Kono DH, Burlingame RW, Owens DG, Kuramochi A, Am J Med 1975; 59: 533–552. Balderas RS, Balomenos D et al. Lupus susceptibility loci in 4 Moser KL, Neas BR, Salmon JE, Yu H, Gray-McGuire C, New Zealand mice. Proc Natl Acad Sci USA 1994; 91: Asundi N et al. Genome scan of human systemic lupus 10168–10172. erythematosus: evidence for linkage on chromosome 1q in 18 Mohan C, Morel L, Yang P, Watanabe H, Croker B, Gilkeson G African-American pedigrees. Proc Natl Acad Sci USA 1998; 95: et al. Genetic dissection of lupus pathogenesis: a recipe for 14869–14874. nephrophilic autoantibodies. J Clin Invest 1999; 103: 1685–1695. 5 Shai R, Quismorio Jr FP, Li L, Kwon OJ, Morrison J, Wallace DJ 19 Morel L, Blenman KR, Croker BP, Wakeland EK. The major et al. Genome-wide screen for systemic lupus erythematosus murine systemic lupus erythematosus susceptibility locus, susceptibility genes in multiplex families. Hum Mol Genet Sle1, is a cluster of functionally related genes. Proc Natl Acad 1999; 8: 639–644. Sci USA 2001; 98: 1787–1792. 6 Gaffney PM, Ortmann WA, Selby SA, Shark KB, Ockenden 20 Tangye SG, Cherwinski H, Lanier LL, Phillips JH. 2B4- TC, Rohlf KE et al. Genome screening in human systemic mediated activation of human natural killer cells. Mol Immunol lupus erythematosus: results from a second Minnesota cohort 2000; 37: 493–501. and combined analyses of 187 sib-pair families. Am J Hum 21 Tangye SG, Phillips JH, Lanier LL. The CD2-subset of the Ig Genet 2000; 66: 547–556. superfamily of cell surface molecules: receptor-ligand pairs 7 Lindqvist AK, Steinsson K, Johanneson B, Kristjansdottir H, expressed by NK cells and other immune cells. Semin Immunol Arnasson A, Grondal G et al. A susceptibility locus for human 2000; 12: 149–157. systemic lupus erythematosus (hSLE1) on chromosome 2q. 22 Sidorenko SP, Clark EA. The dual-function CD150 receptor J Autoimmun 2000; 14: 169–178. subfamily: the viral attraction. Nat Immunol 2003; 4: 19–24. 8 Wakeland EK, Liu K, Graham RR, Behrens TW. Delineating 23 Tsao BP, Grossman JM, Riemekasten G, Strong N, Kalsi J, the genetic basis of systemic lupus erythematosus. Immunity Wallace DJ et al. Familiality and co-occurrence of clinical 2001; 15: 397–408. features of systemic lupus erythematosus. Arthritis Rheum 9 Tsao BP. The genetics of human systemic lupus erythemato- 2002; 46: 2678–2685. sus. Trends Immunol 2003; 24: 595–602. 24 Rao S, Olson JM, Moser KL, Gray-McGuire C, Bruner GR, 10 Tsao BP, Cantor RM, Kalunian KC, Chen CJ, Badsha H, Kelly J et al. Linkage analysis of human systemic lupus Singh R et al. Evidence for linkage of a candidate chromosome erythematosus-related traits: a principal component approach. 1 region to human systemic lupus erythematosus. J Clin Invest Arthritis Rheum 2001; 44: 2807–2818. 1997; 99: 725–731. 25 Ma CS, Nichols KE, Tangye SG. Regulation of cellular and 11 Johanneson B, Steinsson K, Lindqvist AK, Kristjansdottir H, humoral immune responses by the SLAM and SAP families of Grondal G, Sandino S et al. A comparison of genome-scans molecules. Annu Rev Immunol 2007; 25: 337–379. performed in multicase families with systemic lupus erythe- 26 Ostrakhovitch EA, Li SS. The role of SLAM family receptors in matosus from different population groups. J Autoimmun 1999; immune cell signaling. Biochem Cell Biol 2006; 84: 832–843. 13: 137–141. 27 Graham DB, Bell MP, McCausland MM, Huntoon CJ, van DJ, 12 Tsao BP, Cantor RM, Grossman JM, Kim SK, Strong N, Lau CS Faubion WA et al. Ly9 (CD229)-deficient mice exhibit T cell et al. Linkage and interaction of loci on 1q23 and 16q12 may defects yet do not share several phenotypic characteristics contribute to susceptibility to systemic lupus erythematosus. associated with SLAM- and SAP-deficient mice. J Immunol Arthritis Rheum 2002; 46: 2928–2936. 2006; 176: 291–300. 13 Gray-McGuire C, Moser KL, Gaffney PM, Kelly J, Yu H, 28 Hoff H, Brunner-Weinzierl MC. The tyrosine phos- Olson JM et al. Genome scan of human systemic lupus phatase SHP-2 regulates differentiation and apoptosis of erythematosus by regression modeling: evidence of linkage individual primary T lymphocytes. Eur J Immunol 2007; 37: and epistasis at 4p16–15.2. Am J Hum Genet 2000; 67: 1460–1469. 1072–1086. 14 Johanneson B, Lima G, von SJ, Alarcon-Segovia D, Alarcon- 29 Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, Rothfield Riquelme ME. A major susceptibility locus for systemic lupus NF et al. The 1982 revised criteria for the classification erythemathosus maps to chromosome 1q31. Am J Hum Genet of systemic lupus erythematosus. Arthritis Rheum 1982; 25: 2002; 71: 1060–1071. 1271–1277. 15 Rozzo SJ, Vyse TJ, Drake CG, Kotzin BL. Effect of genetic 30 Storm N, rnhofer-Patel B, van den BD, Rodi CP. MALDI-TOF background on the contribution of New Zealand black loci to mass spectrometry-based SNP genotyping. Methods Mol Biol autoimmune lupus nephritis. Proc Natl Acad Sci USA 1996; 93: 2003; 212: 241–262. 15164–15168. 31 Kruglyak L, Daly MJ, Reeve-Daly MP, Lander ES. Parametric 16 Hogarth MB, Slingsby JH, Allen PJ, Thompson EM, and nonparametric linkage analysis: a unified multipoint Chandler P, Davies KA et al. Multiple lupus susceptibility approach. Am J Hum Genet 1996; 58: 1347–1363. loci map to chromosome 1 in BXSB mice. J Immunol 1998; 161: 32 Kruglyak L, Lander ES. Faster multipoint linkage analysis 2753–2761. using fourier transforms. J Comput Biol 1998; 5: 1–7.

Supplementary Information accompanies the paper on Genes and Immunity website (http://www.nature.com/gene)

Appendix Scotia, Canada; C Douglas Smith: Division of Rheuma- tology, Ottawa Hospital, Ottawa, Ontario, Canada; Ann CaNIOS GenES Investigators are as follows. Janet Pope: Clarke, Sasha Bernatsky and Christian Pineau: Division Division of Rheumatology, St Joseph’s Health Centre, of Rheumatology, McGill University Health Center, London, Ontario, Canada; Dafna Gladman and Murray Montreal, Quebec, Canada; Christine Peschken and Urowitz: University of Toronto Lupus Clinic, Centre for Carol Hitchon: Winnipeg Health Science Center, Winni- Prognosis Studies in the Rheumatic Diseases, Toronto peg, Manitoba, Canada; Michel Zummer: Department of Western Hospital, University Health Network; Depart- Rheumatology, Maisonneuve-Rosemont Hospital, Mon- ment of Medicine, University of Toronto, Toronto, treal, Quebec, Canada; Susan Barr: Division of Rheuma- Ontario, Canada; John Hanly: Division of Rheumatology, tology, Department of Medicine, University of Calgary, Department of Medicine, Queen Elizabeth II Health Calgary, Alberta, Canada; Gilles Boire: Division of Sciences Centre and Dalhousie University, Halifax, Nova Rheumatology, Department of Medicine, Faculty of

Genes and Immunity Association of LY9 in SLE families DS Cunninghame Graham et al 102 Medicine and Health Sciences, Universite´ de Sherbrooke, University, Halifax, Nova Scotia, Canada; Kathryn Sherbrooke, Quebec, Canada; Eric Rich, Jean-Luc Drouin: Division of Rheumatology, Ottawa Hospital, Senecal: Division of Rheumatology, Centre Hospitalier Ottawa, Ontario, Canada; Nancy Branco and Elizabeth de l’Universite´ de Montre´al, Department of Medicine, Piniero: Division of Clinical Epidemiology, Montreal University of Montreal School of Medicine, Montreal, General Hospital, and McGill University, Montreal, Quebec, Canada; Simon Carette and Robert Inman: Quebec, Canada; Andrea Craig, Diane Ferland, and Toronto Western Hospital, University Health Network, Donna Hart: Winnipeg Health Science Center, Winnipeg, Toronto, Ontario, Canada; and the CaNIOS research Manitoba, Canada Winnipeg; Diane Ferland: Depart- assistants/coordinators that recruited the patients: Sara ment of Rheumatology, Maisonneuve-Rosemont Hospi- Hewitt and Janine Ouimet: Division of Rheumatology, tal, Montreal, Quebec, Canada; Whitney Steber and St Joseph’s Health Centre, London, Ontario, Canada; Patrice Nedinis: Calgary Health Sciences Centre, Tamara McKenzie, Diona Dobaille, Menisha Hodge, University of Calgary, Calgary, Alberta, Canada; Celine Tammy Koonthanan, Kiran Pabla and Yang Zhou: Boulet and Isabelle Gagnon: Department of Medicine, Division of Rheumatology, Toronto Western Hospital, Division of Rheumatology, University of Sherbrooke, University Health Network; Tina Linehan: Division Sherbrooke, Quebec, Canada and Diane Therrien: of Rheumatology, Department of Medicine, Queen Division of Rheumatology, Hoˆpital Notre-Dame, Elizabeth II Health Sciences Centre and Dalhousie Montreal, Quebec, Canada.

Genes and Immunity