and Immunity (2008) 9, 249–258 & 2008 Nature Publishing Group All rights reserved 1466-4879/08 $30.00 www.nature.com/gene

ORIGINAL ARTICLE Combination of KIR and HLA variants augments the risk of developing birdshot chorioretinopathy in HLA-A*29-positive individuals

RD Levinson1,6,ZDu2,6, L Luo2, D Monnet3, T Tabary4, AP Brezin3, L Zhao5, DW Gjertson2,5, GN Holland1, EF Reed2, JHM Cohen4 and R Rajalingam2 1Ocular Inflammatory Disease Center, Jules Stein Eye Institute, David Geffen School of Medicine at UCLA, University of California at Los Angeles, Los Angeles, CA, USA; 2Department of Pathology and Laboratory Medicine, UCLA Immunogenetics Center, David Geffen School of Medicine at UCLA, University of California at Los Angeles, Los Angeles, CA, USA; 3Service d0ophtalmologie, Hoˆpital Cochin Assistance Publique, Universite´ Paris Descartes, Hoˆpitaux de Paris, France; 4Laboratoire d’Immunologie EA 3798, Institut Fe´de´ratif de Recherche, 53 Universite´ de Reims Champagne Ardenne, CHU Reims, France and 5Department of Biostatistics, School of Public Health, University of California at Los Angeles, Los Angeles, CA, USA

Birdshot chorioretinopathy (BCR), a chronic ocular inflammatory disease with characteristic choroidal lymphocytic infiltrates, has been strongly associated with human leukocyte antigen (HLA)-A29. Although HLA-A29 occurs frequently in all populations, BCR affects only a small percentage of HLA-A29-positive Caucasians, indicating additional susceptibility factors for BCR. Discovery of HLA class I-specific killer cell immunoglobulin-like receptors (KIR) led to a series of epidemiological studies implicating KIR–HLA gene combinations in disease. Here, we characterized KIR–HLA pairs in BCR patients and controls carrying HLA-A*29 as well as controls lacking HLA-A*29. KIR–HLA pairs implicated for weak inhibition (KIR2DL2/3 þ HLA-C1 and KIR3DL1 þ HLA-Bw4T80) in combination with activating KIR genes associated with autoimmunity (KIR2DS2, 2DS3 and 2DS4) augment the risk of developing BCR in HLA-A*29-positive individuals. The reciprocal association of strong inhibitory pairs (KIR3DL1 þ HLA-Bw4I80 and KIR2DL1 þ HLA-C2) in combination with those implicated in protection from infection (KIR3DS1 þ HLA-Bw4I80 and KIR2DS1 þ HLA-C2) was observed in HLA-A*29-negative controls. These results suggest that a profound effect of KIR2DS2/S3/S4 in the absence of strong inhibition may enhance the activation of natural killer cells and T-cell subsets against intraocular self-antigens, thereby contributing to pathogenesis of BCR. Genes and Immunity (2008) 9, 249–258; doi:10.1038/gene.2008.13; published online 13 March 2008

Keywords: natural killer cells; autoimmunity; MHC; birdshot chorioretinopathy; killer cell immunoglobulin-like receptors; disease association

Introduction in Caucasians, 3.5% in Africans and 1.7% in Asians,3 BCR is almost exclusively found in Caucasians and is rare Despite its immunologically privileged property, the eye even in HLA-A*29-positive Caucasians, suggesting that can develop inflammation and immunological disease.1 host genetic factors besides HLA-A*29 play a role in BCR Birdshot chorioretinopathy (BCR) is a rare chronic disease. autoimmune uveitis with a characteristic focal lympho- The recent discovery of killer cell immunoglobulin-like cytic infiltrate in the choroid-affecting function.2 BCR is receptors (KIR) led a series of epidemiological studies strongly associated with human leukocyte antigen implicating KIRs in antiviral immunity,4–6 autoimmune (HLA)-A29, one of the 21 serologically defined variants diseases7–10 and cancer progression.11,12 KIRs are key of HLA-A gene.2 The relative risk of developing BCR receptors of human natural killer (NK) cells, a subset of among HLA-A*29-positive individuals has been esti- lymphocytes that trigger early immune response against mated to be 50–224. Although HLA-A*29 occurs com- infection and tumors, while maintaining tolerance to monly in all ethnic populations with a frequency of 4.5% healthy self.13,14 Further, CD8 þ cytotoxic T lymphocytes (CTL) with memory phenotype and CD4 þ CD28null Correspondence: Dr R Rajalingam, Department of Pathology T cells also express KIRs indicating KIRs’ regulatory role and Laboratory Medicine, UCLA Immunogenetics Center, David in antigen-specific adaptive immunity.15,16 Given the Geffen School of Medicine at UCLA, University of California at importance of KIR receptors in innate and adaptive Los Angeles, 1000 Veteran Avenue, Room No. 1-536, Box 951652, immunity and their genomic associations with various Los Angeles, CA 90095-1652, USA. diseases, we examined whether BCR could also be E-mail: [email protected] 6These authors contributed equally to this work. associated with some KIR genes in a Caucasian popula- Received 17 January 2008; revised 12 February 2008; accepted 15 tion. Although it was never studied whether KIR- February 2008; published online 13 March 2008 expressing NK cells or T cells were crucial players in KIR–HLA association with birdshot chorioretinopathy RD Levinson et al 250 BCR pathogenesis, the aggregations of lymphocytes in tion of these unlinked gene loci produce diversity in the the deep choroid and in the optic nerve head observed number and type of KIR þ HLA gene combinations by a histopathological analysis of a BCR patient17 inherited in individuals.47 To examine whether indicates a possible role of these lymphocyte subsets. KIR–HLA combinations confer risk for developing BCR, Fourteen KIR receptors triggering either inhibition we typed KIR and HLA class I genes in patients and (3DL1–3, 2DL1–3, 2DL5) or activation (3DS1, 2DS1–5) or controls carrying HLA-A*29 as well as controls lacking both (2DL4) have been identified.18 Inhibitory KIR HLA-A*29 from a French Caucasian population. This receptors (iKIR) recognize distinct motifs of the poly- provides an appropriate system to assess the sole effect morphic HLA class I molecules and trigger signals that of KIR genes in conferring risk of BCR since the impact of stop effector function of NK cells. KIR2DL2 and 2DL3 HLA-A*29, the allele known to confer risk for disease, bind a subset of HLA-C allotypes containing an its linked HLA-B and HLA-C loci3 and their force on asparagine at amino-acid position 80 of the heavy chain coevolving KIR genes48 are likely identical in patients (HLA-C1 alleles) and KIR2DL1 binds the remaining and controls carrying HLA-A*29. HLA-C allotypes with lysine 80 (HLA-C2 alleles).19–21 KIR3DL1 binds 40% of the HLA-B allotypes containing a serologically defined Bw4 epitope22,23 and KIR3DL2 Results binds HLA-A3/A11.24,25 The strength of these inter- actions is highly sensitive to polymorphism of the KIR Differential distribution of KIR genes in patients and controls and HLA genes, as well as the HLA-bound peptide carrying HLA-A*29 sequence.26–32 KIR2DL4 binds trophoblast-specific To test the possibility that KIR genes are involved in risk HLA-G and regulates vascular growth in the decidua of developing BCR, we first compared the KIR gene that facilitates successful pregnancy.33,34 Ligands for profiles among the HLA-A*29-positive BCR patients, PC KIR2DL5, KIR3DL3 or any activating KIR (aKIR) have HLA-A*29-positive controls (termed A*29 ) and HLA- NC not been identified. A*29-negative controls (termed A*29 ). Based on the Based on genetic epidemiological studies,4 in vitro presence and absence of known KIR genes, 32 distinct activation35,36 and weak binding of aKIR-Fc proteins37 KIR genotypes were identified from all individuals and tetramers,38 three aKIRs are thought to bind the analyzed in the present study (Figure 1). The patients same HLA class I ligands as their homologous inhibitory displayed only 13 of 32 genotypes. The overall distribu- counterparts at low affinity. KIR3DS1 that shares 97% tion of KIR genotypes between patients and either sequence similarity with KIR3DL1 in the extracellular control group was significantly different (Po0.05). immunoglobulin (Ig)-domains is believed to bind HLA- Particularly, the KIR genotypes with C2-T1 motifs Bw4 ligand.4 Similarly, KIR2DS1 (homologue of 2DL1) carrying 2DS2, 2DL2, 2DL1, 3DL1 and 2DS4 genes were and 2DS2 (homologue of 2DL2) are considered to bind found more commonly in patients (57.3%) than HLA- PC weakly to HLA-C2 and HLA-C1, respectively.37,38 The A*29 (40%), P ¼ 0.017, odds ratios (OR) ¼ 2.0, 95% functional analogues of aKIR receptors in mice, LY49D confidence interval (CI) ¼ 1.15–3.51 (Figure 1). and LY49H were shown to bind directly to virus-encoded All patients carried KIR3DL1 and KIR2DS4, while B expressed by cells infected with mouse cytome- 7% of controls lacked them (Figure 1). Genotypes galovirus and initiate the activation cascade leading to deleted for both KIR3DL1 and 2DS4 were reported to cytolysis of infected cells.39,40 A prediction from these occur 5–7% in Caucasian populations including 49 50 studies was that human aKIR may also bind pathogen- those studied from southeast and western France. encoded, or pathogen-induced determinants expressed The ubiquitous distribution of 3DL1 and 2DS4 was on the surface of infected or stressed cells and trigger previously reported only in populations of African 46 NK lysis. ancestry. Subtyping analyses revealed that the majority KIR receptors are encoded by a family of tightly of patients, but not controls, carried functional 3DL1 clustered genes on leukocyte complex at and 2DS4 variants (Figure 2). Further, more patients PC NC 19q13.4.41 The number and type of KIR (20.6%) than HLA-A*29 (10.6%) or HLA-A*29 (7.8%, genes vary substantially between haplotypes41,42 and P ¼ 0.0098, OR ¼ 3.08, 95% CI ¼ 1.3–7.28) carried two display sequence polymorphism.43,44 Patterns of linkage alleles of 3DL1 that encode strong inhibitory receptors disequilibrium divide the KIR haplotype between 3DP1 (Table 1). Conversely, homozygous carriers of activating and 2DL4 into two halves.44,45 For genes within each half KIR3DS1 that shares allelic relationship with inhibitory NC there is significant linkage disequilibrium, but it is much KIR3DL1 were more frequent in A*29 (17.6%) com- less for genes in the two different halves. Based on the pared to patients (5.6%, P ¼ 0.0086, OR ¼ 0.28, 95% PC gene content, the centromeric (C) and telomeric (T) CI ¼ 0.11–0.74) and A*29 (9.6%). A trend toward higher halves can be subdivided into small motif groups such as frequencies of KIR2DS2, 2DS3 and 2DS4 and lower C1, C2, C3, Cx, T1, T2 and Tx. Combinations of these frequencies of KIR3DS1, 2DS1 and 2DS5 was observed in C and T motif groups were suggested to generate patients compared to controls (Figure 2). exponential KIR haplotypic diversity.46 The independent segregation of KIR haplotypes with different gene Distribution of KIR-binding HLA motifs is comparable content yields diversity in the number and type of KIR between patients and controls carrying HLA-A*29 genes inherited in individuals, which consequently The distribution of HLA class I alleles and KIR-binding produce a population diversity in KIR gene frequencies motifs in patients was comparable to HLA-29PC, and genotype profiles.46 but differed substantially from HLA-A*29NC (Table 2; Since KIR genes at chromosome 19q13.4 and HLA Supplementary Table 1). This was primarily due to genes at chromosome 6p21.3 are polymorphic and the known positive linkage disequilibrium between display significant variations, the independent segrega- HLA-A*2902 and B*4403-Cw*1601 alleles, resulting in

Genes and Immunity 3DL1 C2 ttsial significant, statistically nte without another controls; stenme fidvdashvn h eoyedvddb h ubro niiul tde ( absence studied and individuals box) of number (shaded the presence by the divided by genotype other the each having differed individuals KIR which ( of receptors observed, number immunoglobulin-like were the cell as genotypes killer distinct 14 of 32 box) individuals, (white 307 of group study n otos h rqece of frequencies The controls. and iue2 Figure 1 Figure rsn nalo otindividuals. most or all in present 3DS1 ¼ eecnet h etoei C n eoei T avsaesbiie notefloigmtfgroups: motif following the into subdivided are halves (T) telomeric and (C) centromeric the content, gene 2DS2 * þ 049N 2DS1 A ifrnildsrbto fkle elimngoui-iercpos( receptors immunoglobulin-like cell killer of distribution Differential þ eoye containing Genotypes * 29 or 2DL2 þ NC 2DS3 or , HLA-A þ HLA-A 3DS1 2DL1 * 003N þ P * þ 100

% Frequency * 29 2DS1 25 32 30 22 20 18 12 11 10 17 16 15 31 29 28 27 26 24 23 21 19 14 13 ¼ 29 20 40 60 80 1 9 8 7 6 5 4 3 2 , eecniee sngtv o h ie oi aafor Data loci. given the for negative as considered were 0 KIR Genotypes .1,OR 0.017, ngtv controls. -negative ( C3 A * À * D12L D32L D52S D32S D12S 2DS5 2DS1 3DS1 2DS4 2DS3 2DS2 2DL5 2DL2 2DL3 2DL1 3DL1 3DL3 ¼ 29 h ifrnei h itiuinof distribution the in difference The .

KIR 2DS2 2DS2 NC 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ KIR2DS2-2DL2-2DL1-3DL1-2DS4 3 2 2 3 3 2 3 3 3 2 2 3 3 2 2 2 2 Centromeric half 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 .Teidvdashmzgu o h nxrse ainsof variants unexpressed the for homozygous individuals The ). 2DL2/3 ee nBRptet r oprdwt hs ftocnrlgop,oewith one groups, control two of those with compared are patients BCR in genes ¼ þ

.,9%CI 95% 2.0, 2DL5 2DL2 2DS3 2DP1 Number ofgenotypes þ

2DL1 2DL1

3DP1 KIR Genecontent C2 C2 C2 C2 C2 C2 C2 C2 C2 C2 C2 C2 C2 C2 C C C C C C C C C C C C C C C C C C xC Cx ¼ 1 1 3 2 1 2 2 1 3 1 1 1 1 3 3 3 1 1 À *** BCRvs. .535.Tefaeokgnsaedpce ndr boxes. dark in depicted are genes framework The 1.15–3.51. ** BCRvs. , * BCRvs. C C C C C C C C C C C C C C C C C C C C C C C C C C C Cx Cx Cx motif Cx 3 1 1 1 1 1 3 1 1 1 3 3 1 3 1 1 1 2 2 2 1 1 1 1 1 1 1 1 ¼ KIR oprsnpvleO 95%CI OR p-value Comparison 2DL4 2DL2 /S1 L1/S1 L1/S1 L1/S1 L1/S1 L1/S1 L1/S1 L1/S1 L1/S1 L1/S1 L1/S1 A*29 S1 S1 S1 S1 S1 S1 S1 S1 A*29 L1 L1 L1 L1 L1 L1 L1 L1 L1 L1 L1 L1 L1 ee.Feunyo ahgntp sepesda ecnaeaddefined and percentage a as expressed is genotype each of Frequency genes. ) A*29 3DL1/S1 Telomeric half þ PC NC PC 2DS2 r rdmnn nbrso hroeioah BR ains nour In patients. (BCR) chorioretinopathy birdshot in predominant are cn .2 .81.09-3.98 2.08 0.027 con. cn .0340 1.65-9.95 4.05 0.0013 con. con. 2DS5 C2-T1

2DS1 Levinson RD KIR À r2DS2 or – .1 .71.20-3.92 2.17 0.013 2DS4 3DL2 HLA oisbtenptet 5.% and (57.3%) patients between motifs T1 T1 T1 T1 T1 T1 T1 T1 T1 T1 T1 T1 T1 T1 T T T T T T T T T T T T T T T T T T T KIR soito ihbrso chorioretinopathy birdshot with association 2 1 1 1 1 1 1 1 1 1 1 1 ** 2 2 2 2 2 2 2 tal et 2DL4 T T T T T T T T T T T T T T T T T T T T T T T Tx Tx Tx Tx Tx Tx Tx Tx ee nptet ihbrso hroeioah (BCR) chorioretinopathy birdshot with patients in genes )

þ motif 2 2 1 2 1 1 1 2 2 2 1 2 2 2 1 1 2 2 2 2 2 2 2 2DL2 n10 n9)(n=102) (n=95) (n=110) , *** BCR 3DL2 641. 13.7 24.5 13.7 1.0 30.5 0.91 1.0 16.4 25.5 1.1 8.8 3.9 0.91 4.9 0.91 8.4 9.8 0.91 1.1 0.91 13.7 5.3 2.0 3.64 4.2 5.45 16.8 7.27 1.1 10.9 19.1 7.27 À 32 22 20 13 , T1 , 3DL3 ¼ A*29 3DL1 n 2.1 1.1 1.0 1.1 1.0 4.2 2.1 2.9 1.1 1.1 1.1 2.9 1.1 1.1 2.1 ntegvnsuygop ae nthe on Based group. study given the in ) 3DL1 , 2DP1 PC þ 2DS4 * 004 A*29 A*29 BCR A*29 and 1.0 1.0 1.0 1.0 1.0 1.0 2.0 , 1 þ NC PC 2DL2 , 3DP1 NC A T2 * 29 * HLA-A HLA-A ¼ 004 PC r o hw sthey as shown not are 3DS1 , C1 , HLA-A 2DS4 * ¼ * 29 þ 29 2DL3 2DS1 PC * ( 003/004/006 A * 29 4% was (40%) ee n Immunity and Genes * 29 þ -positive þ 2DL1 PC , and ) Tx ¼ þ , , 251 KIR–HLA association with birdshot chorioretinopathy RD Levinson et al 252 Table 1 Frequency of KIR3DL1/S1 alleles and phenotypes in patients with BCR and controls

Functiona BCR (n ¼ 106) HLA-A*29PC (n ¼ 94) HLA-A*29NC (n ¼ 102) BCR vs A*29NC

P-value OR (95% CI)

%F (N+) %F (N+) %F (N+) Allelesb 3DL1*00101c Strong inhibition 15.1 (32) 16.0 (30) 11.8 (24) 3DL1*002 Strong inhibition 14.6 (31) 9.6 (18) 8.8 (18) 3DL1*01502 Strong inhibition 7.5 (16) 3.2 (6) 4.9 (10) 3DL1*008 Strong inhibition 5.2 (11) 4.8 (9) 4.9 (10) 3DL1*020 Strong inhibition 1.4 (3) 0 (0) 1.5 (3) 3DL1*009 Strong inhibition 3.3 (7) 1.1 (2) 2.0 (4) 3DL1*00501 Weak inhibition 10.4 (22) 14.9 (28) 11.3 (23) 3DL1*007 Weak inhibition 3.8 (8) 1.6 (3) 1.0 (2) 3DL1*019 Weak inhibition 1.4 (3) 0 (0) 0.5 (1) 3DS1*01301c Activating 17.4 (37) 19.1 (36) 21.1 (43) 3DL1*004d Unknown 14.6 (31) 17.0 (32) 16.2 (33) 3DS1*49N No function 0 (0) 1.1 (2) 1.5 (3) 3DL1*-new Unknown 1.9 (4) 2.7 (5) 1.5 (3) 3DL1/S1-blank 3.3 (7) 9.0 (17) 13.2 (27)

Phenotypese SI/SI Strong inhibition 20.6 (22) 10.6 (10) 7.8 (8) 0.0098 3.08 (1.30–7.28) SI/WI Strong inhibition 13.1 (14) 11.7 (11) 7.8 (8) SI/no Weak inhibition 18.7 (20) 18.1 (17) 25.5 (26) WI/WI Weak inhibition 1.9 (2) 0 (0) 0 (0) WI/no Weak inhibition 9.4 (10) 16.0 (15) 12.7 (13) A/SI Unknown 21.5 (23) 19.2 (18) 18.6 (19) A/WI Unknown 7.5 (8) 9.6 (9) 5.9 (6) A/no Activation 5.6 (6) 9.6 (9) 17.6 (18) 0.0086 0.28 (0.11–0.74) No/no No function 1.0 (1) 5.3 (5) 3.9 (4)

Abbreviations: A, activation; BCR, birdshot chorioretinopathy; HLA, human leukocyte antigen; no, no function; OR, odds ratio; SI, strong inhibition; WI, weak inhibition. aThe function of KIR3DL1*00101, *002, *01502, *00501, *007, *004 and 3DS1*01301 alleles was assigned on the basis experimental data by Yawata et al.51 For the remainders, the function was predicted on the basis of their highest sequence homologies with either strongly inhibiting 3DL1*015 allele (3DL1*008, *009 and *020) or weakly inhibiting allele 3DL1*005 (3DL1*019) per Norman et al. study.52 KIR3DL1*003, *006, *01501, *016, *017 and *018, and 3DS1*010, *011, *012 and *014 alleles were not encountered in this study. KIR3DL1/S1-blank indicates individuals carrying a single 3DL1/3DS1 allele, and these individuals may carry two identical alleles of 3DL1/3DS1. bAllele frequency (%F) is expressed as a percentage and defined as number of times the allele is present (N+) divided by the total number of haplotypes in the given study group (2n, where n is the number of individuals in the panel). co7% of these assignments consisted with one of the following ambiguities: 3DL1*00101/016 and 3DL1*01301/010. d7 BCR patients (6.6%), 13 (13.8%) HLA-A*29-positive controls (A*29PC) and 22 (21.6%) HLA-A*29-negative controls (A*29NC) were either homozygous for 3DL1*004/*004 (unexpressed allele) or homozygous for 3DS1/3DS1 (activating alleles) or heterozygous for 3DL1*004/3DS1, and thus do not express a functional KIR3DL1 receptor (BCR vs A*29PC: P ¼ NS; BCR vs A*29NC: P ¼ 0.0023, OR ¼ 0.26, 95% CI ¼ 0.1–0.63). ePhenotype frequency (%F) is expressed as a percentage and defined as number of individuals predicted to be expressing the phenotype (N+) divided by the total number of individuals studied in the panel (n).

high frequencies of HLA-Bw4 and HLA-C1 ligands HLA-A23/A24/A25/A32 carry the Bw4I80 epitope, compared to HLA-A*29NC (Table 2). There was a trend and frequency of these combined epitopes from HLA-B toward higher frequencies of homozygotes for C1/C1 and and HLA-A allotypes (termed Bw4I80 þ ) was higher in Bw4/Bw4 among BCR patients. In contrast, the reciprocal HLA-A*29NC compared to A*29-positive subjects (Table 2). association displaying increased frequencies of C2/C2 and Bw6/Bw6 homozygotes was observed in HLA-A*29- negative controls. The homozygous carriers of A*29/A*29 iKIR þ HLA combinations implicated for weak inhibition were and/or B*44/B*44 were observed more frequently among frequent in BCR patients patients than HLA-A*29PC (A*29/A*29: 21.8 vs 1.1%, Since KIRs on chromosome 19q13.4 and HLA on P ¼ 0.0034, OR ¼ 11.51, 95% CI ¼ 1.47–90.3; B*44/B*44: chromosome 6p21.3 are substantially polymorphic, their 19.1 vs 7%, P ¼ 0.023, OR ¼ 2.97, 95% CI ¼ 1.2–7.3; independent segregation produces diverse KIR þ HLA Supplementary Table 1). combinations among individuals.47 To investigate Over 75% of the patients and HLA-A*29PC carried the whether KIR–HLA interactions confer risk for BCR, we subset of Bw4 allotypes containing threonine 80 (Bw4T80), assessed four well-defined iKIR þ HLA pairs in each which provides weaker inhibition than Bw4 containing subject. Patients displayed increased frequencies of isoleucine at this position (Bw4I80)23,53 (Table 2). In 3DL1 þ Bw4 and 2DL2/3 þ C1, and decreased frequencies contrast, most Bw4-positive individuals in the HLA- of 2DL1 þ C2 and 3DL2 þ A3/11, compared to controls A*29NC group carried the strongly inhibitory Bw4I80 (Table 2). Since KIR2DL2 binds HLA-C1 with greater subset. In addition to one-third of HLA-B allotypes, affinity than its allelic variant 2DL3,27,41 we assessed

Genes and Immunity KIR–HLA association with birdshot chorioretinopathy RD Levinson et al 253 Table 2 Frequency of KIR-binding HLA motifs and KIR–HLA combinations in patients with BCR and controls

BCR (n ¼ 110) A*29PC (n ¼ 95) A*29NC (n ¼ 102) BCR vs A*29PC BCR vs A*29NC

P-value OR (95% CI) P-value OR (95% CI)

%F (N+) %F (N+) %F (N+) KIR-binding motif: HLA-C2 43.6 (48) 54.7 (52) 69.6 (71) NS 0.00018 0.34 (0.19-0.6) HLA-C1 92.7 (102) 88.4 (84) 78.4 (80) NS 0.0031 3.51 (1.48–8.29) C1/C1 56.4 (62) 45.3 (43) 30.4 (31) NS 0.00018 2.96 (1.68–5.21) C1/C2 36.4 (40) 43.2 (41) 48.0 (49) NS NS C2/C2 7.3 (8) 11.6 (11) 21.6 (22) NS 0.0031 0.29 (0.12–0.67) HLA-Bw4 82.7 (91) 84.2 (80) 56.9 (58) NS 0.000044 3.63 (1.93–6.83) Bw4I80 10.9 (12) 15.8 (15) 37.2 (38) NS 0.0000083 0.21 (0.10–0.42) Bw4T80 78.2 (86) 74.7 (71) 28.4 (29) NS 2 Â 10À12 9.02 (4.83–16.84) Bw4/Bw4 34.5 (38) 22.1 (21) 14.7 (15) NS 0.0014 3.06 (1.56–6.0) Bw4/Bw6 48.2 (53) 62.1 (59) 42.2 (43) 0.05 0.57 (0.32–1.0) NS Bw6/Bw6 17.3 (19) 15.8 (15) 44.1 (45) NS 0.000024 0.26 (0.14–0.5) HLA-Bw4+A23/24/25/32 84.5 (93) 89.5 (85) 68.6 (70) NS 0.0087 2.5 (1.29–4.86) Bw4I80 +A23/24/25/32 (Bw4I80+) 22.7 (25) 36.8 (35) 56.9 (58) 0.031 0.5 (0.27–0.93) 3.6 Â 10À6 0.22 (0.12–0.40) HLA-A3/11 16.4 (18) 29.5 (28) 39.2 (40) 0.029 0.47 (0.24–0.92) 0.00021 0.3 (0.16–0.58)

KIR–HLA combination: 2DL1+C2 42.7 (47) 52.6 (50) 68.6 (70) NS 0.00018 0.34 (0.19–0.60) 2DS1+C2 11.8 (13) 23.2 (22) 36.3 (37) 0.04 0.44 (0.21–0.94) 0.000038 0.24 (0.12–0.48) 3DL2+A3/11 16.4 (18) 29.5 (28) 39.2 (40) 0.03 0.47 (0.24–0.92) 0.00021 0.3 (0.16–0.58) 3DL1+Bw4 79.1 (87) 70.5 (67) 46.1 (47) NS 9.4 Â 10À7 4.43 (2.42–8.1) 3DL1+Bw4I80 10.9 (12) 15.8 (15) 36.3 (37) NS 0.000016 0.22 (0.10–0.44) 3DL1+Bw4T80 78.2 (86) 68.4 (65) 25.5 (26) NS 8.6 Â 10À15 10.47 (5.55–19.76) 3DL1*001/2/8/9/15/20+Bw4I80 9.1 (10) 8.4 (8) 22.6 (23) NS 0.008 0.34 (0.15–0.76) 3DL1*001/2/8/9/15/20+Bw4T80 56.4 (62) 43.2 (41) 15.7 (16) NS 5.1 Â10À10 6.94 (3.61–13.34) 3DS1+Bw4 27.3 (30) 30.5 (29) 26.5 (27) NS NS NS 3DS1+Bw4I80 1.8 (2) 7.4 (7) 17.6 (18) NS 0.000073 0.086 (0.02–0.38) 3DS1+Bw4T80 26.4 (29) 26.3 (25) 13.7 (14) NS 0.026 2.25 (1.11–4.56) 3DS1+Bw4I80+ 6.4 (7) 13.7 (13) 22.5 (23) NS 0.00076 0.23 (0.09–0.57) 3DL1+A23/24/25/32 11.8 (13) 14.7 (14) 35.3 (36) NS 0.000071 0.25 (0.12–0.50) 3DL1+Bw4/A23/24/25/32 79.1 (87) 74.7 (71) 58.8 (60) NS 0.0017 2.65 (1.45–4.85) 2DL2/3+C1 92.7 (102) 88.4 (84) 78.4 (80) NS 0.0031 3.51 (1.48–8.30) 2DL2+C1 55.5 (61) 43.2 (41) 45.1 (46) NS NS NS 2DL3+C1 82.7 (91) 83.2 (79) 70.6 (72) NS 0.05 2.0 (1.04–3.83) 2DS2+C1 54.5 (60) 41.0 (39) 45.1 (46) NS NS NS

Abbreviations: A*29PC, HLA-A*29-positive controls; A*29NC, HLA-A*29-negative controls; BCR, birdshot chorioretinopathy; CI, confidence interval; KIR, killer-cell immunoglobulin-like receptors; HLA, human leukocyte antigen; OR, odds ratio; NS, not significant. Frequency (%F) of each phenotype is expressed as a percentage and defined as the number of individuals having the genotype (N+) divided by the number of individuals studied (n) in the study group.

2DL2 þ C1 and 2DL3 þ C1 combinations, and found both Bw4T80 and decreased frequency of KIR3DS1 þ Bw4I80 increased in patients. compared to HLA-A*29NC. Sequence polymorphisms of 3DL1 and Bw4 affect their expression, binding and inhibitory function.23,51 The Combination of certain inhibitory iKIR þ HLA pairs and weakly inhibitory 3DL1 þ Bw4T80 combination was more aKIRs enhances the risk of developing BCR in HLA-A*29- frequent in patients (78.2%) compared to HLA-A*29PC positive individuals (68.4%) and HLA-A*29NC (25.5%, P ¼ 8.6 Â 10À15, To determine the additive effect of four iKIR þ HLA pairs, OR ¼ 10.45, 95% CI ¼ 5.55–19.76) (Table 2). In contrast, we compared the frequency of individuals carrying the strongly inhibitory 3DL1 þ Bw4I80 pair was most combinations of these receptor-ligand pairs (Figure 3). common among HLA-A*29NC (36.3%) compared to The strongest inhibitory genotype, consisting all four HLA-A*29PC (15.8%) and patients (10.9%, P ¼ 0.000016, iKIR þ HLA pairs, was observed more frequently in HLA- OR ¼ 0.22, 95% CI ¼ 0.1–0.44). The combinations of A*29PC than patients. More patients (45%) carried 3DL1*001/*002/*008/*009/*015/*020 and Bw4I80 that have 3DL1 þ Bw4 and 2DL2/3 þ C1 pairs, while lacking other been implicated as having the strongest inhibition23,51 two (2DL1 þ C2 and 3DL2 þ A3/11), than HLA-A*29PC and peptide-dependent KIR3DL1 þ HLA-A23/24/25/32 (26%, P ¼ 0.0057, OR ¼ 2.3, 95% CI ¼ 1.29–4.21) or HLA- interactions26 were found to be significantly decreased A*29NC (6%, P ¼ 2 Â 10À10,OR¼ 13.3, 95% CI ¼ 5.39–33). in patients compared to HLA-A*29NC (Table 2). Among In contrast, the presence of 3DL2 þ A3/11 with the three predicted aKIR þ HLA combinations, 3DL1 þ Bw4 and 2DL2/3 þ C1 (with or without KIR3DS1 þ Bw4 and KIR2DS2 þ HLA-C1 distributed 2DL1 þ C2) was less frequent in patients (9.1%) compared equally in all three groups, while KIR2DS1 þ HLA-C2 to HLA-A*29PC (24.2%, P ¼ 0.0041, OR ¼ 0.31, 95% CI ¼ was underrepresented in patients (Table 2). Nevertheless, 0.14–0.7). In general, genotypes positive for 3DL2 þ A3/11 patients displayed an increased frequency of KIR3DS1 þ or lacking either 3DL1 þ Bw4 or 2DL2/3 þ C1, or both,

Genes and Immunity KIR–HLA association with birdshot chorioretinopathy RD Levinson et al 254 50

*# iKIR+HLA only iKIR+HLA+aKIR BCR Comparison p-value OR 95% CI p-value OR 95% CI A*29PC PC 40 * BCR vs. A*29 con. 0.0057 2.3 1.29-4.21 0.0071 3.2 1.37-7.49 A*29NC -10 # BCR vs. A*29 NCcon. 2x10 13.3 5.39-33 0.000084 7.2 2.41-21.53 ** BCR vs. A*29 PC con. 0.011 0.24 0.07-0.76 NC 30 *** BCR vs. A*29 con. 0.0038 0.2 0.06-0.63 0.015 0.1 0.01-0.88 **** BCR vs. A*29 NC con. 0.004 0.08 0.01-0.67

20 % Frequency

10 ** *** *** * 0 3DL2+A3/11 --+++---++-+ 2DL1+C2 -+-+++ ++-+-+ 2DL2/3+C1 ++ ++--+-+++- 3DL1+Bw4 ++ ++-+---+-- Figure 3 The combination of four inhibitory killer-cell immunoglobulin-like receptors (iKIR) þ human leukocyte antigen (HLA) pairs in patients with birdshot chorioretinopathy (BCR) and control groups. The square symbols indicate the frequency of individuals within each iKIR þ HLA combination group that carry two or three activating KIR (aKIR) genes among 2DS2, 2DS3 and 2DS4. The triangle symbols indicate the frequency of individuals within each iKIR þ HLA combination group that carry two or three aKIR genes among 3DS1, 2DS1 and 2DS5. A*29PC, HLA-A*29-positive controls; A*29NC, HLA-A*29-negative controls.

were at lower frequency in patients compared to controls 70 BCR (n=110) (Figure 3). Taken together, 67.3% of the patients, but only A*29 PC (n=95) 46.3% of HLA-A*29PC, carried genotypes containing HLA- 60 A*29, 3DL1 þ Bw4 and 2DL2/3 þ C1, but not 3DL2 þ A3/11 50 (P ¼ 0.0029, OR ¼ 2.38, 95% CI ¼ 1.35–4.2) (Figure 4). Within these groups, individuals carrying two or three 40 aKIR genes among 2DS2, 2DS3 and 2DS4 were more frequent in patients (32.5%) than HLA-A*29PC (7.7%, 30 À6 P ¼ 1.4 Â 10 ,OR¼ 6.91, 95% CI ¼ 2.91–16.4). 20 % Frequency 10 Discussion 0 2 Strong association of HLA-A*29, lymphocytic infiltration . in the choroid17 as well as the beneficial effects of -neg. -neg, -pos, 54 -pos, -pos, -pos, -pos, -pos, immunosuppressive agents such as cyclosporine in -pos, -pos/neg, -pos/neg, BCR clearly demonstrate the central role for an HLA- 2DS2/3/4 A29-restricted CTL response. The molecular or antigenic HLA-A*29 2DL1+C2 HLA-A*29 2DL2/3+C1 3DL1+Bw4 3DL2+A3/11 2DL2/3+C1 2 or 3 of 3DL2+A3/11 3DL1+Bw4 mimicry between the ‘external’ and ‘self’ peptide ligands 2DL1+C2 of HLA-A29 may result in some activated CTLs having p-value 0.0029 0.0000014 a response against self, which in turn may lead to OR 2.38 6.91 autoimmunity. Supporting this model was the elution of 95% CI 1.35-4.2 2.91-16.4 retinal autoantigens from HLA-A29 molecules.55 Predo- minance of KIR2DS2, 2DS3 and 2DS4, the genes Figure 4 Compound genotype of certain killer cell immunoglobulin- 56–58 like receptors (KIR) and human leukocyte antigen (HLA)loci implicated in autoimmune diseases in BCR patients confers risk of birdshot chorioretinopathy (BCR). Combination of compared to HLA-A*29-matched controls, suggests that KIR3DL1 þ Bw4, KIR2DL2/3 þ HLA-C1 and KIR2DS2/S3/S4 in the these aKIR receptors may enhance the T-cell autoreactiv- absence of KIR3DL2 þ A3/11 is predominantly represented in HLA- ity. Although the direct role of NK cells in BCR A*29-positive BCR patients as compared to HLA-A*29-positive controls pathogenesis is not clear, the genetic data presented (A*29PC). herein indicate that KIR–HLA combinations likely influ- ence both CTL and NK cell responses toward tolerance in studied here can generate highly competent NK cells as controls and autoimmunity in BCR patients (Figure 5). they express iKIRs for both HLA-B and HLA-Cw In addition to providing self-tolerance, the interaction molecules. Relatively weaker inhibitory signals triggered of iKIRs with self-HLA class I molecules confers NK cell by the predominant interactions of KIR2DL2/3 þ HLA- functional competence, a process called ‘licensing’,59 C1 and KIR3DL1 þ HLA-Bw4T80 in patients may be suf- ‘arming’60 or ‘education’.61 Expression of progressively ficient to establish NK cell tolerance to self in the resting higher numbers of iKIRs for self-HLA-B and HLA-Cw state,27 but may be easily overcome by activating signals molecules has been correlated with an increased effector triggered in response to stress such as infection, leading capacity.62 Accordingly, most patients and controls to disease. The preponderance of autoimmunity-

Genes and Immunity KIR–HLA association with birdshot chorioretinopathy RD Levinson et al 255 Innate Immune Response Adaptive Immune Response

Infection Target NK cell Memory CTL Target Infection

HLA-C1 KIR2DL2/3 Weak inhibition - Weak inhibition - KIR2DL2/3 HLA-C1 HLA-Bw4T80 KIR3DL1 activation unable to suppress KIR3DL1 HLA-Bw4T80 can override TCR activation

Activation Activation ?? KIR2DS2/3/4 KIR2DS2/3/4 ?? against self against self

Activation against self due to TCR HLA-A29

HLA-A*29 positive BCR patient mimicry with pathogen

HLA-C1 KIR2DL2/3 KIR2DL2/3 HLA-C1 Strong inhibition - HLA-Bw4T80 HLA-Bw4T80 KIR3DL1 Strong suppress KIR3DL1 inhibition HLA-A3/11 HLA-A3/11 KIR3DL2 TCR activation KIR3DL2

Pathogen- Pathogen- Specific KIR2DS1 HLA-C2 KIR2DS1 Specific HLA-C2 activation activation

HLA-A*29 positive control TCR HLA-A29 Self-tolerance BCR Pathogenesis

HLA-C2 KIR2DL1 HLA-C2 KIR2DL1 Strong inhibition - I80 Strong I80 HLA-Bw4 KIR3DL1 suppress KIR3DL1 HLA-Bw4 inhibition TCR activation HLA-A3/11 KIR3DL2 KIR3DL2 HLA-A3/11 HLA-Bw4I80 HLA-Bw4I80 KIR3DS1 Pathogen- Pathogen- KIR3DS1 HLA-C2 KIR2DS1 Specific Specific KIR2DS1 HLA-C2 activation activation

Self-tolerance

HLA-A*29 negative control & Pathogen- TCR HLA (other specific activation than A29)

Figure 5 Model of killer cell immunoglobulin-like receptors (KIR) þ human leukocyte antigen (HLA) interactions associated with the risk of developing birdshot chorioretinopathy (BCR). The genes encoding weak inhibitory KIR2DL2/3 þ HLA-C1 and KIR3DL1 þ HLA-Bw4T80 interactions in combination with the activating receptors (KIR2DS2/S3/S4 and self-reactive T-cell antigen receptor, TCR) may drive the immune response toward breaking the self-tolerance and cause autoimmune condition in BCR. In contrast, the genes encoding strongly inhibitory KIR3DL1 þ HLA-Bw4I80 and KIR2DL1 þ HLA-C2 interactions in combination with the activating interactions (KIR2DS1 þ HLA-C2, KIR3DS1 þ HLA-Bw4I80 and non-HLA-A*29-restricted TCR) may drive the immune response that maintains self-tolerance but mounts early immune response against infection. Symbols ‘-’ indicate activation and ‘>’ indicate inhibition. associated activating receptors KIR2DS2, 2DS3 and HLA-B*2705 molecules,64 which in turn may further 2DS456–58 in patients suggests that the interaction reduce inhibition and contribute to the increased NK and between these aKIRs and determinants expressed on CTL activation. the surface of stressed intraocular tissue trigger signals Strong inhibitory combinations KIR3DL1 þ HLA-Bw4I80 that may overcome the weaker inhibition, favoring and KIR2DL1 þ HLA-C2 in HLA-A*29-negative controls localized NK cell hyperresponsiveness against self. Of may render protection against BCR. Predominance of note, individuals carrying the compound genotype KIR2DS1 þ HLA-C2 and KIR3DS1 þ HLA-Bw4I80 combi- (KIR2DL2/3 þ HLA-C1, KIR3DL1 þ HLA-Bw4T80 and nations in controls raises the possibility that these weak KIR2DS2/S3/S4) are extremely rare in black and Asian interactions may be established only upon infection4,6,65 populations,47 which is consistent with the epidemiology and trigger NK response at the initial stage of infection, of BCR.2 Alternatively, neighboring loci linked to 2DS2, which perhaps avoids subsequent immune responses 2DS3 and 2DS4 genes may also be involved in BCR that may trigger inflammation and autoimmunity. pathogenesis. In conclusion, this study shows a strong association HLA-A29 and its genetically linked HLA-B*4403 between certain allelic combination of 7 KIR and 3 HLA molecule carry a leucine residue at position 156 in the loci and risk of developing BCR. These data are a2 helix and share high similarity in peptide-binding consistent with loss of tolerance in BCR, perhaps in the cleft (E pocket), which was implicated in loading unique context of infection, being in part mediated by cytotoxic peptides.63 Since the HLA-B44 is the primary allotype lymphocytes expressing KIRs. A better understanding of contributing Bw4 ligand in HLA-A*29-positive subjects, the pathophysiology of disease provides the basis for the altered antigen-binding cleft may avoid its binding to designing strategies for screening and suggests that inhibitory receptor KIR3DL1 as demonstrated with novel therapeutic approaches may be possible.

Genes and Immunity KIR–HLA association with birdshot chorioretinopathy RD Levinson et al 256 Materials and methods characterize the allelic variants of KIR3DL1/S1. The strategy coupled gene-specific PCR amplification and Study subjects and samples direct sequencing of the amplified products after enzy- Genomic DNA samples from 110 unrelated patients with matic purification, an approach we have successfully BCR and 197 unrelated healthy volunteer from a used for sequencing the KIR2DS3 and KIR3DS1 genes.67 Caucasian population of France were included in this The 3DL1/S1-specific amplicon of B4.3 kb consists exons study. The patients were recruited at Hoˆpital Cochin, 3–5, which encodes the ligand-binding extracellular Ig- Paris, France. All patients met criteria for diagnosis of like domains D0, D1 and D2. The oligonucleotide BCR that had been defined previously by an interna- 66 primers used in this direct DNA sequencing analysis tional group of investigators. Briefly, the criteria are listed in Supplementary Table 2. include bilateral involvement; no more than mild (1 þ ) Expand Long Template (LT) PCR System consisting anterior uveitis or moderate (2 þ ) or less vitreous Tgo DNA polymerase, a thermostable DNA polymerase inflammatory reaction; and at least three hypopigmented with proofreading activity (Roche Applied Science, choroidal lesions inferior or nasal to the optic disk. Indianapolis, IN, USA) was used to avoid miss-incor- Although the presence of the HLA-A*29 allele was not poration during the amplification of 3DL1/S1 gene. The required for the diagnosis of BCR, all patients included PCR reaction (20 ml volume) consisted a final concentra- in this study had the HLA-A*29 allele. tion of 1 Â LT buffer II, 500 mM of each deoxyribonucleo- The control DNA samples were collected at the CHU tide triphosphates (dNTPs), 0.3 mM of each forward and de Reims Champagne Ardenne, Reims, France. Controls reverse primers, 1.5 U of LT Tgo DNA polymerase and were two groups: one group included samples collected 100 ng genomic DNA. The thermal cycling conditions randomly and another group included samples collected used for these PCR amplifications were 95 1C for 3 min on the basis of HLA-A*29 positivity (n ¼ 95). We excluded 1 1 1 B followed by 10 cycles of 94 C for 15 s, 64 C for 30 s, 72 C the A*29-positive individuals ( 5%) from the randomly for 4 min and 22 cycles of 94 1C for 15 s, 61 1C for 30 s, collected control group, so that 102 individuals negative 72 1C for 4 min completed with 10 min extension at 72 1C. for HLA-A*29 were available in this group. The study The PCR amplicons were purified from unincorporated adhered to the tenets of the Declaration of Helsinki for primers and dNTPs by digesting with ExoSAP-IT research involving human subjects. Fully informed exonuclease-I (USB Corporation, Cleveland, OH, USA) consent was obtained from all subjects included in this according to the manufacturer’s protocol, and were used study. The regional Comite´ Consultatif pour la Protection as template in the sequencing reactions. Exons 3, 4 and 5 des Personnes se Preˆtant a` la Recherche Biome´dicale and were sequenced in both directions using the BigDye the UCLA Human Subject Protection Committee ap- terminator V1.1 Cycle Sequencing Kit (Applied Biosys- proved this study. The DNA samples were isolated from tems, Foster City, CA, USA). The heterozygous DNA peripheral blood samples using a standard salting out sequence analysis was performed using Assign SBT method or QIAamp Blood Kit (Qiagen, Chatsworth, CA, v3.5.1 software (Conexio Genomics, Applecross, Australia). USA). Quality and quantity of DNA was determined by The program combined both forward and reverse UV spectrophotometry and the concentration was sequences, and generates a contiguous sequence from m adjusted to 100 ng/ l. multiple exons. The Assign software compares the test sequence with a library of known KIR3DL1/S1 sequences KIR and HLA genotyping and assigns the alleles. DNA samples were typed for the presence and absence The alleles identical in the region that we sequenced of 16 KIR genes using a gene-specific PCR typing system but differ in exon 2 and 9 were resolved by additional as described recently.47 Using similar typing systems, PCR-SSP typing. The typing system distinguished the the unexpressed alleles 2DS4*003/004/006, 2DS3*003N, following alleles and allele groups: 3DL1*007, 3DL1*016 3DS1*049N, 3DL1*004 and 2DL2*004 were distinguished and 3DS1*013, 3DL1*001/004/005/009 and 3DS1*010/011/ from their expressed counterparts,47,67 and only the 012. The primers used for these typing are listed in the expressed functional alleles were included for receptor- Supplementary Table 2. The PCR reagents used for these ligand analysis. HLA-A,-B and -C typing was performed assays are same as those used recently for KIR genotyp- by either sequence-specific primer-directed PCR ampli- ing.47 The thermal cycling conditions used for this fication (PCR-SSP) or sequence-specific oligonucleotide amplifications were 95 1C for 2 min followed by 5 cycles hybridization methods using the commercial kits per of 94 1C for 20 s, 69 1C for 20 s, 72 1C for 1 min and 30 manufacturer’s instructions (One Lambda, Canoga Park, cycles of 94 1C for 20 s, 65 1C for 30 s, 72 1C for 1 min CA, USA). The KIR-binding HLA class I epitopes were completed with 10 min extension at 72 1C. PCR products predicted from the HLA typing results. If the HLA typing were electrophoresized on 2% agarose gels in the results were ambiguous, we SSP typed the KIR-binding presence of ethidium bromide and visualized using a HLA motifs as described earlier.47 The presence of UV illuminator. The PCR product sizes for the positive the following four iKIR þ HLA pairs was analyzed in reactions are provided in the Supplementary Table 2. each individual: 2DL1 þ HLA-C2, 2DL2/3 þ HLA-C1, 3DL1 þ HLA-Bw4 and 3DL2 þ HLA-A3/11. KIR3DL3 and Data analysis and statistical methods 2DL5 were excluded from the study because their Statistical analysis was performed using Stata 9 software. expression, ligand and function are not known.18 Differences between controls and patients in the dis- tribution of KIR genes and their profiles were tested by Allele-level typing of KIR3DL1/S1 gene two-tailed Fisher’s exact probabilities (P), and Po0.05 The function of different KIR3DL1 alleles was assigned was considered to be statistically significant. ORs with on the basis functional studies and sequence homo- 95% CI were calculated for the comparisons showing logies.51,52 Direct sequencing method was developed to significant difference between patient and control

Genes and Immunity KIR–HLA association with birdshot chorioretinopathy RD Levinson et al 257 groups. Given the rarity of BCR disease the OR was over activating signals. Cancer Immunol Immunother 2005; 54: interpreted as an approximation for relative risk. 172–178. 13 Biron CA, Brossay L. NK cells and NKT cells in innate defense against viral infections. Curr Opin Immunol 2001; 13: 458–464. 14 Diefenbach A, Raulet DH. The innate immune response to Acknowledgements tumors and its role in the induction of T-cell immunity. Immunol Rev 2002; 188: 9–21. This work was supported by start-up funds from the 15 Phillips JH, Gumperz JE, Parham P, Lanier LL. Superantigen- Department of Pathology and Laboratory Medicine dependent, cell-mediated cytotoxicity inhibited by MHC class (to RR), Stein Oppenheimer Endowment Award (to RR I receptors on T lymphocytes. Science 1995; 268: 403–405. and RL) and the MacDonald Family Foundation (to RL). 16 Namekawa T, Snyder MR, Yen JH, Goehring BE, Leibson PJ, We thank Damian Goodridge, Conexio Genomics, Weyand CM et al. Killer cell activating receptors function Western Australia for his help in building and optimiz- as costimulatory molecules on CD4+CD28null T cells ing KIR libraries for the use in Assign software. clonally expanded in rheumatoid arthritis. J Immunol 2000; 165: 1138–1145. 17 Gaudio PA, Kaye DB, Crawford JB. Histopathology of bird- shot retinochoroidopathy. Br J Ophthalmol 2002; 86: 1439–1441. 18 Vilches C, Parham P. KIR: diverse, rapidly evolving receptors Conflict of interest of innate and adaptive immunity. Annu Rev Immunol 2002; 20: The authors declare no financial or commercial conflict 217–251. of interest. 19 Colonna M, Borsellino G, Falco M, Ferrara GB, Strominger JL. HLA-C is the inhibitory ligand that determines dominant resistance to lysis by NK1- and NK2-specific natural killer cells. Proc Natl Acad Sci USA 1993; 90: 12000–12004. References 20 Wagtmann N, Biassoni R, Cantoni C, Verdiani S, Malnati MS, Vitale M et al. Molecular clones of the p58 NK cell receptor 1 de Smet MD, Chan CC. Regulation of ocular inflammation— reveal immunoglobulin-related molecules with diversity in what experimental and human studies have taught us. Prog both the extra- and intracellular domains. Immunity 1995; 2: Retin Eye Res 2001; 20: 761–797. 439–449. 2 Shah KH, Levinson RD, Yu F, Goldhardt R, Gordon LK, 21 Winter CC, Long EO. A single amino acid in the p58 killer cell Gonzales CR et al. Birdshot chorioretinopathy. Surv Ophthalmol inhibitory receptor controls the ability of natural killer cells to 2005; 50: 519–541. discriminate between the two groups of HLA-C allotypes. 3 Maiers M, Gragert L, Klitz W. High-resolution HLA alleles J Immunol 1997; 158: 4026–4028. and haplotypes in the United States population. Hum Immunol 22 Gumperz JE, Litwin V, Phillips JH, Lanier LL, Parham P. The 2007; 68: 779–788. Bw4 public epitope of HLA-B molecules confers reactivity 4 Martin MP, Gao X, Lee JH, Nelson GW, Detels R, Goedert JJ with natural killer cell clones that express NKB1, a putative et al. Epistatic interaction between KIR3DS1 and HLA-B HLA receptor. J Exp Med 1995; 181: 1133–1144. delays the progression to AIDS. Nat Genet 2002; 31: 429–434. 23 Cella M, Longo A, Ferrara GB, Strominger JL, Colonna M. 5 Khakoo SI, Thio CL, Martin MP, Brooks CR, Gao X, NK3-specific natural killer cells are selectively inhibited by Astemborski J et al. HLA and NK cell inhibitory receptor Bw4-positive HLA alleles with isoleucine 80. J Exp Med 1994; genes in resolving hepatitis C virus infection. Science 2004; 305: 180: 1235–1242. 872–874. 24 Pende D, Biassoni R, Cantoni C, Verdiani S, Falco M, di 6 Qi Y, Martin MP, Gao X, Jacobson L, Goedert JJ, Buchbinder S Donato C et al. The natural killer cell receptor specific for et al. KIR/HLA pleiotropism: protection against both HIV and HLA-A allotypes: a novel member of the p58/p70 family of opportunistic infections. PLoS Pathog 2006; 2: e79. inhibitory receptors that is characterized by three immuno- 7 van der Slik AR, Koeleman BP, Verduijn W, Bruining GJ, Roep globulin-like domains and is expressed as a 140-kD dis- BO, Giphart MJ. KIR in type 1 diabetes: disparate distribution ulphide-linked dimer. J Exp Med 1996; 184: 505–518. of activating and inhibitory natural killer cell receptors in 25 Dohring C, Scheidegger D, Samaridis J, Cella M, Colonna M. patients versus HLA-matched control subjects. Diabetes 2003; A human killer inhibitory receptor specific for HLA-A1, 2. 52: 2639–2642. J Immunol 1996; 156: 3098–3101. 8 Nelson GW, Martin MP, Gladman D, Wade J, Trowsdale J, 26 Thananchai H, Gillespie G, Martin MP, Bashirova A, Yawata Carrington M. Cutting edge: heterozygote advantage in N, Yawata M et al. Cutting edge: allele-specific and peptide- autoimmune disease: hierarchy of protection/susceptibility dependent interactions between KIR3DL1 and HLA-A and conferred by HLA and killer Ig-like receptor combinations in HLA-B. J Immunol 2007; 178: 33–37. psoriatic arthritis. J Immunol 2004; 173: 4273–4276. 27 Winter CC, Gumperz JE, Parham P, Long EO, Wagtmann N. 9 Nikitina-Zake L, Rajalingam R, Rumba I, Sanjeevi CB. Killer Direct binding and functional transfer of NK cell inhibitory cell immunoglobulin-like receptor genes in Latvian patients receptors reveal novel patterns of HLA-C allotype recognition. with type 1 diabetes mellitus and healthy controls. Ann NY J Immunol 1998; 161: 571–577. Acad Sci 2004; 1037: 161–169. 28 Peruzzi M, Parker KC, Long EO, Malnati MS. Peptide 10 Yen JH, Moore BE, Nakajima T, Scholl D, Schaid DJ, Weyand sequence requirements for the recognition of HLA-B*2705 CM et al. Major histocompatibility complex class I-recognizing by specific natural killer cells. J Immunol 1996; 157: receptors are disease risk genes in rheumatoid arthritis. J Exp 3350–3356. Med 2001; 193: 1159–1167. 29 Rajagopalan S, Long EO. The direct binding of a p58 killer cell 11 Carrington M, Wang S, Martin MP, Gao X, Schiffman M, inhibitory receptor to human histocompatibility leukocyte Cheng J et al. Hierarchy of resistance to cervical neoplasia antigen (HLA)-Cw4 exhibits peptide selectivity. J Exp Med mediated by combinations of killer immunoglobulin-like 1997; 185: 1523–1528. receptor and human leukocyte antigen loci. J Exp Med 2005; 30 Maenaka K, Juji T, Nakayama T, Wyer JR, Gao GF, Maenaka T 201: 1069–1075. et al. Killer cell immunoglobulin receptors and T cell receptors 12 Naumova E, Mihaylova A, Stoitchkov K, Ivanova M, Quin L, bind peptide-major histocompatibility complex class I with Toneva M. Genetic polymorphism of NK receptors and their distinct thermodynamic and kinetic properties. J Biol Chem ligands in melanoma patients: prevalence of inhibitory 1999; 274: 28329–28334.

Genes and Immunity KIR–HLA association with birdshot chorioretinopathy RD Levinson et al 258 31 Zappacosta F, Borrego F, Brooks AG, Parker KC, Coligan JE. 50 Denis L, Sivula J, Gourraud PA, Kerdudou N, Chout R, Ricard Peptides isolated from HLA-Cw*0304 confer different degrees C et al. Genetic diversity of KIR natural killer cell markers in of protection from natural killer cell-mediated lysis. Proc Natl populations from France, Guadeloupe, Finland, Senegal and Acad Sci USA 1997; 94: 6313–6318. Reunion. Tissue Antigens 2005; 66: 267–276. 32 Boyington JC, Motyka SA, Schuck P, Brooks AG, Sun PD. 51 Yawata M, Yawata N, Draghi M, Little AM, Partheniou F, Crystal structure of an NK cell immunoglobulin-like receptor Parham P. Roles for HLA and KIR polymorphisms in natural in complex with its class I MHC ligand. Nature 2000; 405: killer cell repertoire selection and modulation of effector 537–543. function. J Exp Med 2006; 203: 633–645. 33 Faure M, Long EO. KIR2DL4 (CD158d), an NK cell-activating 52 Norman PJ, Abi-Rached L, Gendzekhadze K, Korbel D, receptor with inhibitory potential. J Immunol 2002; 168: Gleimer M, Rowley D et al. Unusual selection on the 6208–6214. KIR3DL1/S1 natural killer cell receptor in Africans. Nat Genet 34 Ponte M, Cantoni C, Biassoni R, Tradori-Cappai A, Bentivo- 2007; 39: 1092–1099. glio G, Vitale C et al. Inhibitory receptors sensing HLA-G1 53 Rojo S, Wagtmann N, Long EO. Binding of a soluble p70 killer molecules in pregnancy: decidua-associated natural killer cell inhibitory receptor to HLA-B*5101: requirement for all cells express LIR-1 and CD94/NKG2A and acquire p49, an three p70 immunoglobulin domains. Eur J Immunol 1997; 27: HLA-G1-specific receptor. Proc Natl Acad Sci USA 1999; 96: 568–571. 5674–5679. 54 Nussenblatt RB, Palestine AG, Chan CC, Stevens Jr G, Mellow 35ChewningJH,GudmeCN,HsuKC,SelvakumarA,DupontB. SD, Green SB. Randomized, double-masked study of cyclos- KIR2DS1-positive NK cells mediate alloresponse against the C2 porine compared to prednisolone in the treatment of HLA-KIR ligand group in vitro. J Immunol 2007; 179: 854–868. endogenous uveitis. Am J Ophthalmol 1991; 112: 138–146. 36 Alter G, Martin MP, Teigen N, Carr WH, Suscovich TJ, 55 Boisgerault F, Khalil I, Tieng V, Connan F, Tabary T, Cohen JH Schneidewind A et al. Differential natural killer cell mediated et al. Definition of the HLA-A29 peptide ligand motif allows inhibition of HIV-1 replication based on distinct KIR/HLA prediction of potential T-cell epitopes from the retinal soluble subtypes. J Exp Med 2007; 204: 3027–3036. antigen, a candidate autoantigen in birdshot retinopathy. 37 Biassoni R, Pessino A, Malaspina A, Cantoni C, Bottino C, Proc Natl Acad Sci USA 1996; 93: 3466–3470. Sivori S et al. Role of amino acid position 70 in the binding 56 Khakoo SI, Carrington M. KIR and disease: a model system or affinity of p50.1 and p58.1 receptors for HLA-Cw4 molecules. system of models? Immunol Rev 2006; 214: 186–201. Eur J Immunol 1997; 27: 3095–3099. 57 Rodacki M, Svoren B, Butty V, Besse W, Laffel L, Benoist C 38 Stewart CA, Laugier-Anfossi F, Vely F, Saulquin X, Riedmuller et al. Altered natural killer cells in type 1 diabetic patients. J, Tisserant A et al. Recognition of peptide-MHC class I Diabetes 2007; 56: 177–185. complexes by activating killer immunoglobulin-like receptors. 58 Yen JH, Lin CH, Tsai WC, Wu CC, Ou TT, Hu CJ et al. Killer Proc Natl Acad Sci USA 2005; 102: 13224–13229. cell immunoglobulin-like receptor gene’s repertoire in rheu- 39 Arase H, Mocarski ES, Campbell AE, Hill AB, Lanier LL. matoid arthritis. Scand J Rheumatol 2006; 35: 124–127. Direct recognition of cytomegalovirus by activating and 59 Kim S, Poursine-Laurent J, Truscott SM, Lybarger L, Song YJ, inhibitory NK cell receptors. Science 2002; 296: 1323–1326. Yang L et al. Licensing of natural killer cells by host major 40 Smith HR, Heusel JW, Mehta IK, Kim S, Dorner BG, Naidenko histocompatibility complex class I molecules. Nature 2005; 436: OV et al. Recognition of a virus-encoded ligand by a natural 709–713. killer cell activation receptor. Proc Natl Acad Sci USA 2002; 99: 60 Raulet DH, Vance RE. Self-tolerance of natural killer cells. 8826–8831. Nat Rev Immunol 2006; 6: 520–531. 41 Wilson MJ, Torkar M, Haude A, Milne S, Jones T, Sheer D et al. 61 Anfossi N, Andre P, Guia S, Falk CS, Roetynck S, Stewart CA Plasticity in the organization and sequences of human KIR/ et al. Human NK cell education by inhibitory receptors for ILT gene families. Proc Natl Acad Sci USA 2000; 97: 4778–4783. MHC class I. Immunity 2006. 42 Martin AM, Kulski JK, Gaudieri S, Witt CS, Freitas EM, 62 Yu J, Heller G, Chewning J, Kim S, Yokoyama WM, Hsu KC. Trowsdale J et al. Comparative genomic analysis, diversity Hierarchy of the human natural killer cell response is and evolution of two KIR haplotypes A and B. Gene 2004; 335: determined by class and quantity of inhibitory receptors 121–131. for self-HLA-B and HLA-C ligands. J Immunol 2007; 179: 43 Garcia CA, Robinson J, Guethlein LA, Parham P, Madrigal JA, 5977–5989. Marsh SG. Human KIR sequences 2003. Immunogenetics 2003; 63 Macdonald WA, Purcell AW, Mifsud NA, Ely LK, 55: 227–239. Williams DS, Chang L et al. A naturally selected dimorphism 44 Shilling HG, Guethlein LA, Cheng NW, Gardiner CM, within the HLA-B44 supertype alters class I structure, Rodriguez R, Tyan D et al. Allelic polymorphism synergizes peptide repertoire, and T cell recognition. J Exp Med 2003; with variable gene content to individualize human KIR 198: 679–691. genotype. J Immunol 2002; 168: 2307–2315. 64 Stewart-Jones GB, di Gleria K, Kollnberger S, McMichael AJ, 45 Middleton D, Meenagh A, Gourraud PA. KIR haplotype Jones EY, Bowness P. Crystal structures and KIR3DL1 content at the allele level in 77 Northern Irish families. recognition of three immunodominant viral peptides com- Immunogenetics 2007; 59: 145–158. plexed to HLA-B*2705. Eur J Immunol 2005; 35: 341–351. 46 Yawata M, Yawata N, Abi-Rached L, Parham P. Variation 65 Zhi-ming L, Yu-lian J, Zhao-lei F, Chun-xiao W, Zhen-fang D, within the human killer cell immunoglobulin-like receptor Bing-chang Z et al. Polymorphisms of killer cell immuno- (KIR) gene family. Crit Rev Immunol 2002; 22: 463–482. globulin-like receptor gene: possible association with 47 Du Z, Gjertson DW, Reed EF, Rajalingam R. Receptor-ligand susceptibility to or clearance of hepatitis B virus infection in analyses define minimal killer cell Ig-like receptor (KIR) in Chinese Han population. Croat Med J 2007; 48: 800–806. humans. Immunogenetics 2007; 59: 1–15. 66 Levinson RD, Brezin A, Rothova A, Accorinti M, Holland GN. 48 Single RM, Martin MP, Gao X, Meyer D, Yeager M, Kidd JR Research criteria for the diagnosis of birdshot chorioretino- et al. Global diversity and evidence for coevolution of KIR and pathy: results of an international consensus conference. HLA. Nat Genet 2007; 39: 1114–1119. Am J Ophthalmol 2006; 141: 185–187. 49 Frassati C, Touinssi M, Picard C, Segura M, Galicher V, Papa K 67 Luo L, Du Z, Sharma SK, Cullen R, Spellman S, Reed EF et al. et al. Distribution of killer-cell immunoglobulin-like receptor Chain-terminating natural mutations affect the function of (KIR) in Comoros and Southeast France. Tissue Antigens 2006; activating KIR receptors 3DS1 and 2DS3. Immunogenetics 2007; 67: 356–367. 10: 779–792.

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

Genes and Immunity