Different HLA Class IA Region Complotypes for HLA-A29.2 and -A29.1 Antigens, Identical in Birdshot Retinochoroidopathy Patients or Healthy Individuals

Abstract

purpose. Birdshot retinochoroidopathy (BSCR) is a rare posterior uveitis characterized by distinctive, multiple, hypopigmented choroidal and retinal lesions. At least 96% of patients, if not all, share the major histocompatibility antigen HLA-A29. Although it was hypothesized earlier that more frequently the A*2902 subtype was closely associated with BSCR, new patients were found to share the A*2901 subtype and were further investigated. The present study was designated to check patients’ HLA-A*2901 subtyping and the polymorphisms available in the HLA region in patients and control subjects sharing the A*2901 and A*2902 subtypes.

methods. HLA-A29 was assessed and subtyped by molecular biology. cDNA from one patient (HLA-A*2901) was sequenced. A29.1 antigenic expression on peripheral blood lymphocytes was checked by microlymphocytotoxicity (MLCT). Four homozygous A29.2 and 4 heterozygous A29.2 patients, 3 homozygous A29.2 healthy subjects, 3 heterozygous A29.1 patients, and 11 heterozygous A29.1 healthy subjects were tested for the microsatellite alleles MOGa, -b, -c, and e (of the myelin oligodendrocyte glycoprotein [MOG]gene), D6S265, D6S510, RF, C5_4_5, D6S105, and D6S276 and the mutation H63D of the familial hemochromatosis gene (HFE).

results. The patients’ cDNA sequences and MLCT reactivities of HLA-A29.1 subtypes were found to be identical with published data from healthy individuals. Surprisingly, though A*2901 and A*2902 differed only by a single mutation (G376C/ D102H) two strong A*2901 and A*2902 complotypes were observed in patients and control subjects, the polymorphisms being identical at all loci near HLA-A, whereas more distant loci exhibited some diversity.

conclusions. Susceptibility to BSCR thus appeared to be located between the left and right remote markers C5_4_5 and D6S276, if not relying on the HLA-A29 molecule itself.

Birdshot retinochoroidopathy (BSCR) is a rare posterior uveitis characterized by distinctive, multiple, hypopigmented choroidal and retinal lesions, first identified by Ryan and Maumenee. ¹ Blurred vision and floaters are the most prevalent visual symptoms. Patients may also report dyschromatopsia and poor contrast sensitivity. Macular edema or atrophy is the most common cause of decline in visual acuity and ∼10% of patients are legally blind. 

In the review by Shah et al., ² 99.5% of patients in reported cases were white, and 95.7% of 467 patients were HLA-A29 positive. Furthermore, for research purposes, most investigators recognize the presence of the HLA-A29 allele as a necessary criterion for BSCR diagnosis. The relative risk of BSCR among HLA-A29-positive individuals has been estimated to be 50 to 224. HLA-A29 is present in as many as 7% of whites and is subdivided into 11 subtypes—mostly A*2902 in whites and A*2901 in Asians. Other subtypes are anecdotal descriptions of very rare alleles. Even Asians who live in the same country as whites in Europe or in the United States appear to be exempt from BSCR. We found that gene sequences from all patients and healthy individuals sharing the A*2902 subtype are identical. ^{3 4} Then, three white patients exhibiting a clinical pattern compatible with BSCR were found to share the A*2901 subtype. 

The first part of this study was conducted to show identity between the A*2901 subtype sequence in patients and control subjects. The second part was to check polymorphisms in the HLA-A region in patients and control subjects, to localize and map the susceptibility locus for BSCR. Four homozygous A29.2 patients, 4 heterozygous A29.2 patients, 3 homozygous A29.2 healthy subjects, 3 heterozygous A29.1 patients, and 11 heterozygous A29.1 healthy subjects were tested for microsatellite alleles RF (GAA repeats, 250-kb telomeric side), MOGa and MOGb (CA repeats in intron 2 of the myelin oligodendrocyte glycoprotein [MOG] gene, located 400-kb telomeric side of the HLA-A locus), MOGc (CA repeats upstream of the MOG gene), MOGe (TAAA repeats in exon 8 of the MOG gene), D6S510 (CA-GA repeats, 27-kb centromeric side), and D6S265 (CA repeats, 100–70-kb centromeric side), as well as for the more remote microsatellite alleles C5_4_5 (TTTA repeats, 200-kb centromeric side), D6S105 (CA repeats, 1500–2500-kb, telomeric side), mutation H63D of the familial hemochromatosis (HFE gene) and D6S276 (CA repeats, 6000-kb telomeric side). 

Materials and Methods

Patients and Healthy Subjects

We are the French reference laboratory in BSCR studies (http://www.orphanet.net). Among ∼70 patients with BSCR analyzed, two women and six men were chosen as either A*2902 homozygous or as bearers of other HLA-A alleles, facilitating the study. The three A*2901 patients were women. All patients were white. They met internationally defined criteria for the diagnosis of BSCR, ² and their conditions were diagnosed independently by two ophthalmologists. Among healthy subjects, two women and seven men ranging in age from 30 to 55 years were blood donors or were employees at the laboratory. They were white except one, who was of mixed white and Asian origin. In addition, five A*2901 DNA samples from Asian subjects, analyzed in the HLA-Diversity/Anthropology Workshop (http://www.ncbi.nlm.gov/mhc/ihwg.fcgi/ National Center for Biotechnology Information, National Institutes of Health, Bethesda, MD) were studied. Three samples were from Vietnamese (kind gift of Chen Chung Chu, Immunohematology Reference Laboratory, Mackay Memorial Hospital, Taipei, Taiwan) and two were from Singapore Chinese (kind gift of Fionnuala Williams, Northern Ireland Regional Histocompatibility and Immunogenetics Laboratory, Belfast, Northern Ireland). HLA-A*29 genotyping given in Table 1was performed in compliance with French law and in accordance with the Declaration of Helsinki for research involving human subjects. 

HLA Typing

HLA class I (HLA-A, -B, and -C) typing was performed by polymerase chain reaction with sequence-specific primers (PCR-SSP) with a commercial kit (MicroSSP 1L; One Lambda, Canoga Park, CA) on genomic DNA extracted from the peripheral blood lymphocytes (PBLs). The amplified DNA fragments were separated by agarose gel electrophoresis, visualized by staining with ethidium bromide, and exposed to ultraviolet light. HLA-A29 subtyping was assessed by amplification refractory mutation system/polymerase chain reaction (ARMS/PCR) as described by Krausa et al. ⁵ with AL#2, AL#F and AL#6 and A*2901 primer combinations or the PCR-SSP method with another commercial kit (SSP HLA-A*29; Olerup, Saltsjöbaden, Sweden). The antigenic reactivity of the patients’ HLA-A29.1 subtype was checked by serologically typing for HLA class I antigens, by using monoclonal antibody trays (Special HLA Class I Tissue Typing Tray; One Lambda) and the standard lymphocytotoxicity method (MLCT). ⁶  

cDNA Study

First-strand cDNA was synthetized by using oligo(dT) and avian myeloblastosis virus reverse transcriptase from mRNA extracted from peripheral blood mononuclear cells of a patient sharing the HLA-A*2901 subtype. cDNA was amplified by PCR with priming oligonucleotides, as described by Ennis et al. ⁷ (forward primer: 5′-GGGCGTCGACGGACTCAGAATCTCCCCAGACGCCGAG-3′, reverse primer: 5′-GCCCAAGCTTTCTCAGTCCCTCACAAGGCAGCTGTC-3′) and Taq polymerase (ampliTaq DNA polymerase; Applied Biosystems, Foster City, CA). The product of amplification was then cloned by using DNA topoisomerase I from a commercial source (Topo TA Cloning kit; Invitrogen, Carlsbad, CA). A*2901 clones were screened by specific A29 ARMS/PCR and then sequenced (ESGS [Euro Séquences Gènes Services], Evry, France). Sequence alignment was performed by using the Infobiogen Web site (http://www.infobiogen.fr/ provided in the public domain by the Infobiogen Group, Villejuif, France) with HLA-A*29010101 sequencing obtained from the Anthony Nolan Research Institute Web site (http://www.anthonynolan.org.uk/ provided in the public domain by the Anthony Nolan Research Institute, London, UK). 

Microsatellite Markers and the H63D Mutation of the HFE Gene

Names, distances from the HLA-A gene, and localization of microsatellite markers studied are depicted in Figure 1and Table 2 , along with the repetition, size range, allele number, and heterozygosity estimated by Foissac et al. ⁸ on 56 chromosomes for the markers reported by Généthon (http://www.genethon.fr/; provided in the public domain by the French Association against Myopathies, Evry, France). PCR primers and conditions are listed in Table 3 . To determine the allele size, we amplified genomic DNA from 11 patients and 14 healthy subjects by PCR with Taq DNA polymerase (ampliTaq DNA polymerase; Applied Biosystems). Amplified products were denatured for 7 minutes at 96°C and chilled on ice before being electrophoresed on 6% acrylamide gels with urea at 7 M followed by silver staining (Promega, Madison, WI). A 10-bp DNA ladder from 30 to 330 bp and a 25-bp DNA ladders from 25 to 500 bp (Invitrogen) were used to estimate alleles sizes. The H63D mutation was assessed by the PCR-specific sequence oligonucleotide (PCR-SSO) method with a commercial kit (ViennaLab, Vienna, Austria). 

Results

Comparison of Patients’ HLA-A29.1 Subtype Molecule with Healthy Subjects’ HLA-A29.1 Subtype Molecule

The PCR-SSP methodology is based on the principle that completely matched oligonucleotide primers are more efficiently used when amplifying a target by recombinant Taq polymerase than is a mismatched oligonucleotide primer. Primer pairs are designated to be perfectly matched only with a single allele or group of alleles. Under strictly controlled PCR conditions, perfectly matched primer pairs result in the amplification of target sequences (i.e., positive result), whereas mismatched primer pairs do not result in amplification (i.e., negative result). ARMS/PCR is based on the same principle: a mismatch at the 3′ residue of the primer inhibits nonspecific amplification. In our study, three patients with BSCR were subtyped A*2901. 

Total RNA from one patient having the A29.1 subtype was used to prepare single-stranded cDNA, to provide the substrate for specific amplification of the HLA-A, -B, and -C sequences by PCR. The oligonucleotides used to prime specific amplification derive from relatively conserved sequences in the 5′ and 3′ untranslated regions of the HLA-A, -B, and -C genes. In our work, amplification resulted in the expected product of approximately 1.1 kb, predicted to contain a mixture of HLA-A, -B, and -C sequences. The product was directly cloned using DNA topoisomerase I, which functions both as a restriction enzyme and as a ligase in the sequencing vector containing the T7 and T3 priming sites. Clones were screened by ARMS/PCR. Three clones were sequenced and found to be identical with the HLA-A*29010101 sequence, by alignment. 

Viable peripheral blood lymphocytes obtained by magnetic anti-CD8 beads separation from an HLA-A*2901-bearing patient were incubated with complement-binding monoclonal antibodies and then with rabbit complement. As antigen–antibody complexes initiate the complement cascade that leads to cell lysis, in a negative reaction the lymphocytes are alive and in a positive reaction the lymphocytes are dead. In our study, the microlymphocytotoxic (MLCT) reactions were fully positive, with the two anti-HLA-A19-specific monoclonal antibodies (X4034 and Z5911; A19 is the HLA-A29, -A30, -A31, -A32, and -A33 antigenic group) and with the two anti-HLA-A29-specific monoclonal antibodies (X5518 and Z1064) from the typing tray. 

Microsatellite Markers and the H63D Mutation of the HFE Gene

Figure 2shows an example of silver-stain polyacrylamide gel electrophoresis of the microsatellite alleles C5_4_5. Allele size depends on the number of TTTA repeats. Molecular weight markers allow the estimation of allele sizes. Alleles in each microsatellite marker are usually named on the basis of the amplified fragment length. Table 4shows the allele sizes of the different microsatellite markers as well as the percentage of D63 mutation of the HFE gene. For D6S510 and MOGb all alleles are the same, whatever the A29 subtype and healthy or BSCR status. For D6S265, MOGa, MOGc, and MOGe all alleles are the same for the 12 haplotypes from A*2902 patients and 6 haplotypes from A*2902 healthy subjects. The alleles of these microsatellite markers are also the same for the 3 haplotypes from A*2901 patients and 11 haplotypes from A*2901 healthy subjects, but distinct from alleles linked to A*2902. Locus RF and the more distant loci C5_4_5, D6S105, H63D, and D6S276 exhibit some diversity. All studied A*2901 haplotypes shared the same 170-bp allele for the RF locus. Only one A*2902 haplotype (17%) possessed the 327-bp allele among healthy subjects and 50% among patients with BSCR. For the C5_4_5 locus the 14 A*2901 haplotypes shared the same 307-bp allele, but all A*2902 haplotypes from patients with BSCR and half of the A*2902 haplotypes from healthy subjects shared the same 299-bp allele, whereas the other half from A*2902 healthy subjects shared a 303-bp allele. For D6S105 all the A*2901 haplotypes studied shared the same 124-bp allele, as did 6 of the 12 A*2902 haplotypes from patients with BSCR and 3 of 4 A*2902 haplotypes from the healthy subjects. The H63D mutation was absent in the three A*2901-bearing patients. Among the A*2901 healthy subjects, one was homozygous for D63, three were heterozygous for H63D, and seven were homozygous for H63. Among the four homozygous A*2902 patients with BSCR studied two were homozygous for D63, one was heterozygous for H63D, and one was homozygous for H63. One A*2902 healthy subject was homozygous for D63, and two others were homozygous for H63. For D6S276, a 224-bp allele was found in two of the three A*2901 patients, in two of five of the A*2901 healthy subjects, in four of seven A*2902 patients, and in none of two homozygous A*2902 healthy subjects. 

Discussion

BSCR shares homology with experimental autoimmune uveitis. Response to immunosuppressive treatments and strong association with the HLA-A29 antigen make BSCR likely to be an autoimmune illness. At least 96% of patients, if not all, are A29 positive. HLA-A29 is subdivided into 11 subtypes, mostly A*2901 and A*2902. BSCR is not restricted to the A*2902 subtype and is also found in A*2901, as previously described in one ⁹ and in four cases, ¹⁰ as well as in three patients reported herein. It is noteworthy that cDNA A29 subtypes are strictly identical between patients and healthy subjects. ^{3 4} Antigenic reactivity by lymphocytotoxicity and previous experiments ¹¹ with HLA-A29 gene transfections are unlikely to generate a defect in the gene expression process as alternative splicing for example. Moreover, transgenic mice expressing HLA-A29 cDNA exhibit spontaneous retinopathy showing a striking resemblance to BSCR ¹² but originating from a single founder. 

Since BSCR seemed to be an autoimmune illness, because familial cases other than twins ¹³ have appeared between brothers and sisters (Brézin A, Levinson R, Monnet D, personal observations, November 2002), we investigated the genetic polymorphisms of microsatellite markers located near the locus HLA-A. We studied microsatellite markers of DNA directly extracted from PBLs and not from cell lines, to avoid the in vitro mutations known to give rise to artifactual polymorphisms during cell culture. ¹⁴ Table 4clearly indicates the lack of disequilibrium linkage between any polymorphism of the microsatellite markers and BSCR. We can thus define two different complotypes: A*2901 and A*2902. The former is more extended than the latter. Surprisingly, although A*2902 and A*2901 differ only by a single mutation (G376C/D102H), their surrounding microsatellite polymorphisms appear to be fully different. HLA-A*2902, -B44, and -Cw16 is an ancestral haplotype probably fixed for some 10,000 years. Some microsatellites appear to have been fixed in frozen haplotype blocks, but others have mutated independently and will be informative if the same mutation is directly responsible for the disease phenotype ¹⁵ or reflects the short physical distance between the expressed gene implicated in the physiopathological process. For example, C5_4_5 does not exhibit Hardy-Weinberg equilibrium in a sample population examined, ¹⁶ because of an unknown molecular mechanism. More distant loci D6S105 and D6S276 exhibit diversity but without any link between BSCR or healthy status. A significant disequilibrium between the H63D mutation and HLA-A29-containing haplotypes in a Portuguese study has been reported. ¹⁷ In the absence of the study of family segregation in our work, we can only refer to assessments of A29 subjects homozygous for H63D mutations or homozygous for D63 or for wild-type H63 subjects for heterozygous A29. Of eight haplotypes from healthy subjects, only one A*2901 haplotype was D63. The three A*2901 patients with BSCR were wild-type H63. Fifty percent of haplotypes from homozygous A*2902 patients with BSCR and healthy subjects studied are D63—so much so that there is a significant difference between D63 mutated or H63 wild-type frequencies in A*2901 and A*2902 subjects (P = 0.03 using the Fisher exact probability test). In the Portuguese study, ¹⁷ 43.5% of the 62 haplotypes containing HLA-A29 had the H63D mutation. Assuming that this HLA-A29 population is essentially A*2902, as Europeans tend to be (94% in the HLA-Diversity/Anthropology workshop (http://www.ncbi.nlm.gov/mhc/ihwg.fcgi)), there is no significant difference between this population and the A*2902 population presently studied (P = 0.20 using the Fisher exact probability test). The H63D mutation is thought to be old, according to the high allele frequencies reported in a wide geographical distribution. Strong linkage disequilibrium at so large a distance could be explained rather by a coselection of this combination of alleles imposed by some biological advantage, such as the CD8⁺ lymphocyte acting as an iron storage compartment, than by a founder effect. Finally, this linkage disequilibrium between HLA-A29 and H63D involves the A*2902 allele but not the A*2901 allele. 

The pathophysiological mechanism of BSCR is still unknown. The inner retina should be affected first, but the lesions are believed to be located in the choroid. ² The S-antigen and interphotoreceptor retinoid-binding protein are located in the outer retina. The immune response to retinal antigens may play a role in the pathophysiology of disease and HLA-A29 may play a role in triggering the autoimmune response. Portions of the S-antigen were reported to bind to the HLA molecule in vitro. ¹¹ Why BSCR is so uncommon when as many as 7% of the white population is HLA-A29 positive remains unexplained. In any event, HLA-A29 seems to be necessary for the development of BSCR, whatever the subtype, A*2901 or A*2902, and finding patients of the two subtypes should rule out a different presentation of a putative uveitogenic peptide by each subtype. 

We cannot explain the selection pressure leading to the similar alleles A*2901 and A*2902 in such different haplotypes. Most whites express the A*2902 subtype, whereas Asians express the A*2901 subtype. ¹⁸ In this study, HLA-A*2901 haplotypes from white patients and Asian healthy subjects seemed identical, indicating that the susceptibility factor for BSCR is HLA-A29 itself or a very close gene product. Even Asians living in the same country as whites in Europe or the United States seem to be exempt from BSCR. Of interest, the Asian A*2901 complotype is identical with that of the non-Asian carrier of A*2901, and both the A*2901 and A*2902 complotypes seem to be identical in patients with BSCR or in healthy individuals. The occurrence of BSCR in rare A*2901 individuals, all of them without recent Asian ancestry, leads to a new hypothesis: although the HLA-A29 molecule seems to be mandatory for susceptibility to BSCR, another factor, probably not linked to the major histocompatibility complex, is either protective in Asians or conversely triggers an autoimmune reactivity that is possibly present in whites and absent in Asians. 

 Submitted for publication July 22, 2004; revised December 13, 2004; accepted January 17, 2005.

 Disclosure: B. Donvito, None; D. Monnet, None; T. Tabary, None; E. Delair, None; M. Vittier, None; B. Réveil, None; R. Sivaradjam, None; C. Edelson, None; F. Philbert, None; A.P. Brézin, None; J.H.M. Cohen, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

HLA-A*29 Genotypes of Subjects Analyzed

	Number of Subjects	A*29 Heterozygous	A*29 Homozygous
A*2902 healthy	3	0	3
A*2902 BSCR	8	4	4
A*2901 healthy	11	11	0
A*2901 BSCR	3	3	0

Figure 1.

 

The position of the microsatellite markers on the major histocompatibility complex map. In parentheses: the distance in megabases from the extremity of the short arm of chromosome 6, which is 170,914,576 bases long, to the gene or marker considered. Data were collected from the Sanger Institute (Wellcome Trust Genome Campus, Hinxton, UK; http://www.ensembl.org/). Marker C5_4_5 and RF are not mapped to the assembly in the current Ensembl database.

Table 2.

 

Microsatellite Markers and H63D Mutation of the HFE Gene Located around the HLA-A Locus^*

Name	Localization/HLA-A	Repeat	Number of Alleles	Size Range (bp)	% Heterozygous
C5_4_5	Centr 200 kb	(TTTA)9	7	295–313	74
D6S265	Centr 100–70 kb	(CA)12	8–12	118–140	79
D6S510	Centr 27 kb	(CA-GA)n	8	178–196	74
RF	Tel 250 kb	(GAA)n	19	168–340	95
MOGa	tel 400 kb	(CA)13	6	124–134	59
MOGb	Tel 400 kb	(CA)11	10	160–190	51
MOGc	Upstream MOG gene	(CA)15	11	122–150	81
MOGe	Exon 8 of MOG gene	(TAAA)10	7	206–230	75
D6S105	Tel [phone] kb	(CA)n	10–12	116–138	87
H63D	HFE gene tel 4000 kb	—	—	—	26
D6S276	Tel 6000 kb	(CA)12	10	198–230	84

*  Adapted from Foissac et al. ⁸

 Centr, centromere; Tel, telomere.

Table 3.

 

PCR Characteristics of the Microsatellite Markers

Name	Forward Primer Reverse Primer	MgCl2 (mM)	Cycles	Denaturation Time (s)	Hybridization Temperature (°C)	Hybridization Time (s)	Elongation Time (s)
C5_4_5	AGCATCAAAGTCCAGGCTGG	2.5	35	60	60	60	60
	TCTTGCCTTCTCCCCGCTAC
D6S265	ACGTTCGTACCCATTAACCT	2.5	35	60	60	60	60
	ATCGAGGTAAACAGCAGAAA
D6S510	AATGGGCTACTACTTCACACC	3.0	35	60	60	60	30
	CAACACACTGATTTCCATAGC
RF	CTGTCCTATTTCATATGCTCAGG	6.0	35	60	61	60	60
	ATGAACTTGTCCTGAGAATGAAG
MOGa	ACCCTGTATTTGTGAGCGC	3.0	35	45	55	45	45
	GTACAGCCAAAAGGTGACATC
MOGb	TTCACTGCCACATCCTCA	3.0	35	45	55	45	45
	ATGGTTTTATCTTTCTCTTAG
MOGc	GAAATGTGAGAATAAAGGAGA	3.0	35	45	55	45	45
	GATAAGGGGAACTACTACA
MOGe	Mog50:ATCTTTCCTTCCTCTATCC	4.0	20	30	58	30	30
	Mog51:GGTGGAGTAGAGGGAG
	Mog52:CCAGGAGGCAGAGGTTG
D6S105	TAACCAGTCTCCACAATTGCA	3.0	35	45	55	45	45
	CTGGAAATGCTGGTTTGCCAG
D6S276	TCAATCAAATCATCCCCAGAAG	3.0	35	30	55	60	60
	GGGTGCAACTTGTTCCTCCT

 MOGe was assessed by nested PCR (Mog50–51 then 50–52).

Figure 2.

 

Microsatellite alleles C5_4_5. Alleles were amplified from homozygous A*2902 healthy subjects (lanes 4, 5, 6, 7, 13 and 14), homozygous A*2902 patients with BSCR (lanes 8 and 12), and heterozygous A*2901 healthy subjects (lanes 9, 10, and 11), by PCR, and analyzed by silver-stain polyacrylamide gel electrophoresis. Lanes 1, 2, 3, 15, and 16: molecular weight markers. Lanes 4, 5, 7, 8, 12, and 13: homozygous 299 alleles; lanes 6 and 14: heterozygous 299/303 alleles; lanes 9 and 10: heterozygous 303/307 alleles; lane 11: heterozygous 307/311 alleles. pb, base pairs.

Table 4.

 

Allele Sizes of the Different Microsatellite Markers and the Percentage of D63 Mutation of HFE Gene

	C5_4_5	D6S265	D6S510	RF	MOGa	MOGb	MOGc	MOGe	D6S105	H63D	D6S276
A* 2902 healthy	303 (50%) 299 (50%)	126	190	327 (17%)	126	160	122	222	124 (75%)†	33 %	NC†
A* 2902 BSCR	299	126	190	327 (50%)	126	160	122	222	124 (50%)	62† %	NC†
A* 2901 healthy	307	140	190	170	128	160	132	218	124	25† %	NC†
A* 2901 BSCR	307	140	190	170	128	160	132	218	124	0 %	NC†

 The number of studied haplotypes was 6 in A*2902 healthy subjects, 12 in A*2902 BSCR patients, 11 in A*2901 healthy subjects, 3 in A*2902 BSCR patients, except for cases indicated in the text and noted by †. Alleles sizes are in base pairs. In parentheses percentage of allele studied when different from 100%. H63D: percentage having the D63 mutation of HFE gene. NC, no consensus (i.e. more than four alleles).

The authors thank Ralph D. Levinson, Ocular Inflammatory Disease Center and Department of Ophthalmology (Jules Stein Eye Institute, UCLA, Los Angeles, CA) for kindly providing biological material from one HLA-A*2901 patient with BSCR; Marie Lin, Immunohematology Reference Laboratory (Mackay Memorial Hospital 92 Taipei, Taiwan) and Derek Middleton, Northern Ireland Regional Histocompatibility and Immunogenetics Laboratory (Blood Transfusion Building, City Hospital, Belfast, Ireland) for the kind gift of Asian HLA-A*2901 DNA; and Jocelyne Wuibout for help in writing the article.