Molecular characterization of HLA-C incompatibilities in HLA-ABDR-matched unrelated bone marrow donor-recipient pairs. Sequence of two new Cw alleles (Cw02023 and Cw0707) and recognition by cytotoxic T lymphocytes

(1)

Article

Reference

Molecular characterization of HLA-C incompatibilities in

HLA-ABDR-matched unrelated bone marrow donor-recipient pairs.

Sequence of two new Cw alleles (Cw02023 and Cw0707) and recognition by cytotoxic T lymphocytes

GRUNDSCHOBER, C, et al.

Abstract

While the influence of HLA-AB and -DRB1 matching on the outcome of bone marrow transplantation (BMT) with unrelated donors is clear, the evaluation of HLA-C has been hampered by its poor serological definition. Because the low resolution of standard HLA-C typing could explain the significant number of positive cytotoxic T lymphocyte precursor frequency (CTLpf) tests found among HLA-AB-subtype, DRB1/B3/B5-subtype matched patient/donor pairs, we have identified by sequencing the incompatibilities recognized by CD8+ CTL clones obtained from such positive CTLpf tests. In most cases the target molecules were HLA-C antigens that had escaped detection by serology (e.g. Cw*1601, 1502 or 0702). Direct recognition of HLA-C by a CTL clone was demonstrated by lysis of the HLA class I-negative 721.221 cell line transfected with Cw*1601 cDNA. Because of the functional importance of Cw polymorphism, a PCR-SSO oligotyping procedure was set up allowing the resolution of 29 Cw alleles. Oligotyping of a panel of 382 individuals (including 101 patients and their 272 potential unrelated donors, 5 related donors and 4 platelet donors) allowed [...]

GRUNDSCHOBER, C, et al. Molecular characterization of HLA-C incompatibilities in

HLA-ABDR-matched unrelated bone marrow donor-recipient pairs. Sequence of two new Cw alleles (Cw*02023 and Cw*0707) and recognition by cytotoxic T lymphocytes. Tissue Antigens, 1997, vol. 49, no. 6, p. 612-23

DOI : 10.1111/j.1399-0039.1997.tb02809.x PMID : 9234483

Available at:

http://archive-ouverte.unige.ch/unige:16902

Disclaimer: layout of this document may differ from the published version.

(2)

T I S S U E A N T I G E N S ISS N 000 1-28 IS

Molecular characterization of HLA-C

incomDatibihties in HLA-ABDR-matched unrelGed bone marrow donor-recipient pairs

Sequence of two new Cw alleles (Cw02023 and Cw0707) and recognition by cytotoxic T lymphocytes

C. Grundschober? N. Rufer, A. Sanchez-Mazas, A. Madrigal, M. Jeannet, E. Roosnek, J.-M. Tiercy. Molecular characterization of HLA-C

incompatibilities in HLA-ABDR-matched unrelated bone marrow donor-recipient pairs. Sequence of two new Cw alleles (Cw*02023 and Cw*0707) and recognition by cytotoxic T lymphocytes.

Tissue Antigens 1997: 49: 612-623. 0 Munksgaard, 1997

While the influence of HLA-AB and -DRBI matching on the outcome of bone marrow transplantation (BMT) with unrelated donors is clear, the evaluation of HLA-C has been hampered by its poor serological definition.

Because the low resolution of standard HLA-C typing could explain the significant number of positive cytotoxic T lymphocyte precursor frequency (CTLpf) tests found among HLA-AB-subtype, DRB I/B3/BS-subtype matched patient/donor pairs, we have identified by sequencing the incompatibilities recognized by CD8+ CTL clones obtained from such positive CTLpf tests. In most cases the target molecules were HLA-C antigens that had escaped detection by serology (e.g. Cw*1601, 1502 or 0702). Direct recognition of HLA-C by a CTL clone was demonstrated by lysis of the HLA class I-negative 721.22 1 cell line transfected with Cw* 160 1 cDNA.

Because of the functional importance of Cw polymorphism, a PCR-SSO oligotyping procedure was set up allowing the resolution of 29 Cw alleles.

Oligotyping of a panel of 382 individuals (including 101 patients and their 272 potential unrelated donors, 5 related donors and 4 platelet donors) allowed to determine HLA-C and HLA A-B-Cw-DRB I allelic frequencies, as well as a number of A-Cw, B-Cw, and DRB 1-Cw associations. Two new HLA-Cw alleles (Cw*02023 and Cw*0707) were identified by DNA sequencing of PCR-amplified exon 2-intron 2-exon 3 amplicons. Further- more, we determined the degree of HLA-C compatibility in 287 matched pairs that could be formed from 73 patients and their 184 potential unrelated donors compatible for HLA-AB by serology and for HLA-DRBI/

B3/B5 by oligotyping. Cw mismatches were identified in 42.1% of these pairs, and AB-subtype oligotyping showed that 3oy0 of these Cw-incampat- ible pairs were also mismatched for A or B-locus subtype. The degree of HLA-C incompatibility was strongly influenced by the linkage with B al- leles and by the ABDR haplotypes. Cw alkles linked with B*4403, B*5101, B i 8, and B62 haplotypes were frequently mismatched. Apparently high resolution DNA typing for HLA-AB does not result in full matching at

locus C. Since HLA-C polymorpliism is recognized by alloreactive CTLs, such incompatibilities might be as relevant as AB-subtype mismatches in clinical transplantation.

C. Grunds~hober~

,

^N.Ruferl, A. Sanchex-Mazas2, A. Madrigal3, M. Jeannet', E. Roosnek' and J.-M. Tiercyl

'Transplantation Immunology Unit, Division of Immunology and Allergology, Hdpitaux Universitaires de GenBve, 'Department of Anthropology and Ecology, University of Geneva, Switzerland, 3Anthony Nolan Research Centre, Royal Free Hospital, London, United Kingdom

Key words: bone marrow transplantation -

CTL - HLA-C - oligotyping - incompatibility - H t A haplotype frequency -

Cw'1601-transfectant - unrelated donor Received 23 September 1996, revised, accepted for publication 6 February 1997

(3)

HLA-C incompatibility in unrelated BNIT The increased incidence of post-transplant compli-

cations after transplantation with bone marrow from an unrelated donor reflects the high frequency of HLA incompatibilities not resolved by standard tissue typing. Clinical studies have shown the influence of HLA-AB (serology) and -DRBl (DNA typing) matching on the outcome of BMT with unrelated donors. However, a thorough evaluation of HLA-C mismatches has not been possible yet because of the poor serological definition of HbA-C antigens, possibly due to the low level of cell surface expression (1, 2). HLA-C antigens are functional class I molecules that like HLA-A and B antigens are recognized by NK cells (3, 4) as

w0lI as by CD8-positive T cells ( 5 ) . Despite recent developments in PCR-based approaches for the analysis of HLA-C (69), very little information is currently available on the functional role of the diversity of locus C in BMT.

I n our center, selection of unrelated bone mar- raw donors is based on AB serology, DRBl/

DRB3/DRBS oligotyping analysis (10, 11) and on a negative CTLpf assay (12). We have studied in detail the mismatches recognized by the CTLs in order to develop comprehensive molecuiar typing methods that could eventually lead to a replace- ment of the CTLpf test.

In a previous analysis (13), high resolution typing of HLA-ABDR ^{I -}¹⁴matched donor/recipient pairs showed that only 28% of the donors were compatible on the basis of AB-subtype, DRB i/B3/

85-oligo matching and CTLp-negativi ty. Common HLA class I subtype mismatches within A2, B35 or B44 (14, IS) serotypes were shown to cause the positive CTLpf tests (1 3). Furthermore, CTL clones derived from these combinations recognized the mismatched antigens, even when the sequence differences between the two alleles consisted of one or two amino acids onIy (16, 17).

In this paper we present a sequencing anaIysis of HLA-C incompatibilities of three donor/recipient combinations with a positive CTLpf test that could not be accounted for by an A or B-subtype differ- ence. In all three cases, HLA-C incompatibilities could be identified. Furthermore, direct recognition by the CTL clones of C w mismatches was demonstrated by transfection of Cw* 160 1 cDNA into a class I-negative human cell line. In addition

wc have developed a PCR-SSO oligotyping protocol in order to: a) determine the frequency of HLA-C alleles in our patient popuiation, b) to in- vestigate the overall rate of HLA-C incompatibilities in AB-sero-DRB l/B3/BS-oligo matched pairs, and c) to analyze the co-occurrence of A- Cw, B-Cw, and DRB1-Cw antigens in the sampled phenotypes in order to propose ABDR haplotypes

Lit risk to be mismatched for HLA-C.

Since HLA-C incompatibilities are recognized by CTLs, it is important to determine the extent of HLA-C incompatibility in the unrelated bone marrow transplantation setting. Knowledge of preferential HLA class I - class I1 associations, including linkage with HLA-Cw alleles, should therefore be helpful in focusing the pre-transplant molecular biology analyses on those haplotypes with highest probability to be incompatible.

Material and methods Patients and donors

One hundred and one non-selected patients with blood disorders and their related (n=5) and unrelated (n=272) potential donors, as well as 4 platelet donors were typed by serology for HLA-AB (382 individuals in total) and by PCR/SSO-oligotyping for class 11. Subtypes of HLA-A2, 1335, B44, B7, B51 were determined by oligotyping as described elsewhere (14, 15, 18). HLA-C polymorphism was determined by oligotyping as described below. DRBl/B3/B5 oligotyping was done by the microtiter plate oligotyping assay (bioMtr- ieux) detecting 60 alleles (1 9), and by dot blot SSO (20) and/or PCR-SSP (2 1) for high resolution D R B 1 typing. Calculation of HLA-C allelic and haplotype frequencies was done on the 101 patients and 4 platelet donors, and calculation of

A/

C, B/C, C/DR associations was done on 382 individuals. For the evaluation of HLA matching we excluded 4 patients with a related donor ( n = 5 ) and the 4 platelet donors. From the remaining 97 patients and their 272 unrelated donors selected from international registries (369 individuals), 73 pa- tients and their 184 donors were compatible for HLA-AB by serology and for DRB1/83/B5 by oli-.

gotyping. This group formed 287 matched pairs that were used to calculate the rate of HLA-A/B and/or -Cw mismatching.

HLA-C amplification

Primers 5CZnl-61 and 3BCIn3-12 (22) were used to amplify the complete exon 2, intron 2 and exon 3 of HLA-C genes. Genomic DNA was prepared from granulocytes as previously described (14).

PCR reactions were performed in a total volume of 50 pl containing l x P C R buffer (Tris pH 8.8 10 mM, KCl 50 mM, MgCll 1.25 mM, BSA 0.2 cis/

ml), 0.5-1 pg of genomic DNA, 0.2 mM dNTPs, 0.25 pM each of the two primers, 1U Taq poly- merase (Gibco BRL). After an initial denaturation step of 4 min at 94°C the cyciing conditions were as follow: 5 cycles with denaturation at 94°C for 30 s. annealing at 65°C for 50 s and extension at 72°C for 50 ^sfollowed by 25 cycles with denatura-

(4)

Crundschober et al.

tion at 94°C for 30 s, annealing at 63°C for 50 s and extension at 72°C for 50 s. Amplification ef- ficiency was confirmed by agarose gel electro- phoresis of 1/25 of the PCR product.

Hybridization with seq uence-specif ic oligonucleotide (SSO) probes

For high-resolution HLA-C oligotyping a total of 27 SSO probes were designed. The sequence, codon positions, and washing temperatures of all probes are given in Table 1. Probes El63 (#9), G170 (#lo), A152-Ql56 (#14) and Pl05 (#22) are from Kennedy et a1 (9). The following HLA-C alleles were recently reported as being sequencing ar- tifacts: Cw*0101, 1201, 1501 (23). Therefore the probes S178 (AGGAGAECTGCAG), V116 (CCAGGTCGCCTACG) and H 129 (TGAAC- GAGCACCTG) aimed at detecting these alleles and which always gave a negative signal, were dropped. Probe A152-Ql56 is only used to dis- criminate Cw*0704 from Cw*0701 in the presence of Cw"0602.

Two p1 of each PCR amplification was dot-blot- ted on a nylon membrane (Nytran, Schleicher

& Schull) by a robotic workstation Biornek 1000.

Table 1.

Sequence-specific exon 2 and 3 oligonucleotide probes far HLA-C typing

The DNA was denatured by i 0 rnin in NaOW 0.4 M, neutralized by 10 min in NH4-acetate 1 M and cross-linked (optional) to the membrane by UV irradiation (120 mJ) with a Stratalinker (Stratagene). After blocking the membrane by 30 min incubation at 42°C with blocking solu- tion (4xSSPE, 1"/0 blocking reagent (Boehringer Mannheim), 0.1% lauroylsarcosine) in a hybridization oven, the membranes were hybridized with 6 pmol of digoxigenin-labeled probes at 42°C for 2 h in TMAC buffer (3 M tetramethyl ammonium chloride (Fluka), 50 mM Tris-HC1, pH=8.0, 2 m M EDTA, 0.1% SDS). Washing was for 10 min in 2xSSPE - 0.1% SDS followed by 15 min of strin- gent wash in TMAC buffer (Table 1). Detection of the DIG-labeIed probes by cherniluminescence, using CSPD as substrate, was done following the indications of the manufacturer (Boehringer- Mannheim).

Cloning and sequencing of exons 2 and 3 of HLA-C After HLA-C specific amplification of exon 2, intron 2 and exon 3 with primers 5CInl-61 and 3BCIn3-12 (22), the PCR product was electro- phoresed on a 1 % ~ agarose gel, the band excised

Sequence 5'4' Exon: (specific nucleotides

Probe nucleotides underlined) Specificity Washing "C

s77 N77 c99 A73 T134' Y99' W156 El63 G170 L147 s99 A1 52 A1 52-01 56 Q156 P47*

L163 R91 PI 05' Q35 w D 1 1 4-51 1 6

W97-F99 HI134116 T152 E l 5 5 4 1 56 H21 w T73 195 w

21222-235 21222-235

3:31-18 2121 1-224 311 24-137 3 ; 20-33 31187-205 3:209-224 3230-245 311 63-1 78

3: 19-32 311 76-1 90 31183-198 3: 186203 2: 1 3 6 1 50 3:210-223 2:263-intron 3170-83 335-49 2:95-108 3:15-23 3:71-84 311 77-1 91 3:187-200

2:54-67 2:21&223

314-1 7

CGAGTGACCCTGCG CGAGTGMCCTGCG GCAGCCACACATCC CACAGGCTG ACCGA CCTGGACIGCCGCG G ATGTAGGG CTG CG

G CCTCGTCECTC CGG AT GGAGGGCGAGTGCGTG GCTCCGCGGATACCTG

GCAAG~EGAGGCG

GGATGTCTGGCTGC GGCCCGTGCGGCGGA GCGGCGGAGCAG CAG A GGAGCAGC4GAGAGCC AGCCCCGGGCGCCGT GAG G GCCJG TG C GT CG AGGCCAGTGAGT ACC A G T ~ ~ G C C T A C CCTGGGGCCGGACGG GTTCGTGCAGTTCG CAGIGGATGQTGG CCAGTTAGCCTACG GCCCGTgGGCGGAG CGGAGEAGCGGAGA G AGCCCCACllCAT GCACAGACTGACCG C t C ACAlJATCCAG

01,03,07 (except 0707),08,12,13,14.1601 02,04,0501,0602,15,1602,1701 01

04,0602,07,12,13,1503, 170f 1203

02 (except 02023),13,1202 02,0602,12,13

02,1701 1701

0701,0702,0704,0707 0702,0703 07,16

16,07 weak (but not 0704) 1601, 1602

0602,0704,1301 03

0303

02,0302,0602,0701/2/3/7,12,13,14,16 04

0501,08 14 1502,1503 0402

02,03,1301 ,t403,15

01,02,03,0501,08,14,16,1502/4/5 0303,0J,15

080i,oao3

45 45 45 45 45 45 55 55 55 45 45 45 50 45 45 45 45 45 45 45 45 45 45 45 45 45 45

* silent mutation w w a a k probe

(5)

HLA-C incompatibility in unrelated BMT and purified with the Nucleotrdp kit (Macherey-

Nagel) and ligated to the TA cloning vector ( h i - trogen). After transformation the bacterial colo- nies were screened by PCR followed by SSO hybridization for the HLA-C allele of interest. Two ciones from two independent PGR were sequenced in both directions using Sequenase 2.0 (Amer- sham),

Cloning and stable transfection of HLA-CwV601

The mRNA was extracted from 1.106 cells (Fast Track, Pharmacia) and reverse transcribed by Su- perscript enzyme (Gibco BRL). 1/20 of the cDNA reaction was amplified with primers 5p2 and 3pC (24) in 1 XPCR buffer containing 1.5 mM MgC12.

The initial denaturation step of 4 min at 94°C was fallawed by 30 cycles with denaturation at 94°C for 60 s, annealing at 60°C for 60 s and extension at 72°C for 90 s. After gel purification and diges- tion with XhoVHindIII, the PCR product was sub- cloned in the pBJI-Neo vector (25). The identity

oif Cw*1601 was confirmed by sequencing. 20 pg of the plasmid was linearized by Xmnl and elec- troporated in the class I-negative cell line 721.22 1 (26). After selection with G418 a stable transfectant was isolated and the expression of Cw*1601 tested by FACS analysis using the anti-class I anti- body W6/32. The transfected cell line was used in a conventional cytotoxici ty chromium release assay with the CTL clone as described (1 6, 17).

Gytotoxicity assays

CTLpf tests and specificity assays with cytotoxic T cell clones were performed as described ( 12, f 3, 16, 17), using autoIogous (donor) cells and third party cells that do not express HLA antigens in common as negative controls.

Frequency and linkage disequilibrium of HLA-C with other HLA loci

HLA-C and DRBl allelic frequencies were estimated in 101 patients and 4 platelet donors of European origin living in Switzerland (total: 105 individuals) M aximum-likelihood allele and haplotype frequencies and their standard deviations were estimated by using the program package ARLE- QUIN (Schoeider S., Kueffer J.-M., Roessli D., and Excoffier L., personal communication). Gene frequencies were obtained starting with 10 initial values, and the iterative process was stopped when the epsilon value between the frequencies of two consecutive iterations was as low as The standard deviations were computed by a bootstrap procedure ( 100 bootstraps). Pairwise allelic associ-

Table 2.

HLA class I PCR/SSO-oligotyping and sequencing analysis of three patient/

donor pairs with a positive CTLpf test. In alf three pairs a single Cw mismatch was detected by DNA sequencing. In pairs 1 and 2 the incompatible Cw alleles had been typed as serologicat blanks. P: patient, 0: donor.

CTLp

Pair A 6 cw frequency

1 P 1/0301 0702/4403 0702/1601 911 o8

D 1/0301 07W4403 070U0401

~~ ~~~

2 P 0201/24 510114403 040111502 911 O*

D 0201124 510114403 0401/1601

3 P 0301/24 18/3501 0401/0702 6/1 P D 0301/24 18/3501 040110501

-

ations between different HLA loci were assessed by classical contingency tables using Yates' correction when necessary.

Results

Identification of HLA-C incompatibilities by DNA

sequencing

Three HLA-ABDRB l/B3/B5 matched donorhe- cipient pairs with a positive CTLp test (Table 2) were analyzed by SSO-oligotyping for HLA-AB and by sequencing for HLA-C in order to identify class I incompatibilities undisclosed by serology.

Table 2 shows that in all 3 cases the class I incompatibility mapped to locus HLA-C. In pairs 1 and 2, mismatched Cw alleles were Cw*160l ^(us0401) and 1502 (vs 1601), respectively, both of which had been typed as serological blanks. In pair 3, the incompatible allele was a subtype of Cw7 (Cw*0702

vs 0501).

Analysis of the CTL specificity on a panel of HLA-A,B,C,DR well characterized cells showed that these HLA-C incompatibilities were indeed the targets recognized (data not shown). Two CTL clones established in pair I recognized 16 of 27 B*4403-Cw serologically blank panel cells so that it became likely that these CTL were directed against the C w blank antigen (16). After identification of the Cw*1601 allele in the patient of this pair 1, we typed the 27 cells from the panel used for the CTL specificity test by hybridization with probe Q 156 which is Cw* 160 1/1602-specific. We found that all 16 target cells lysed by the CTLs were Ql56-positive, whereas 10 of 1 I cells that were not lysed were Q156-negative. To study this in more detail, we sequenced the HLA-C of the Q 156-positive cell (AY3 B5 1/7 Cw7/blank) that was not Iysed by the CTL clones and found that it corresponded to Cw* 1602. allele (28) that differs from 1601 at 2 residues (positions 77 and 80).

Similarly, the specificity of a CTL clone derived

(6)

Grundschober et al.

from pair 2 was tested on a panel of I3 cells: oligotyping showed a perfect correlation between the presence of the Cw*1502 allele on target cells and lysis by CTLs. The Cw*0702 allele could be desig- nated as the target recognized by showing that CTLs from pair 3 lysed all 7 targets tested that expressed this allele. Thus CTL allorecognition of Cw antigens is highly specific, comparable to the one observed for HLA-A and B subtypes (16, 17, 30).

Direct allorecognition of the mismatched Cw'l601 antigen Since all target cells lysed by the clones isolated from pair 1 (Table 2) typed B*4403-Cw*l601, we considered the possibility of an indirect allorecognition (3 1) of either a B*4403-peptide presented by

Cw* 160 1 or a Cw* 160 1 -peptide presented by 8*4403. Therefore we cloned Cw"1601 cDNA by RTPCR into the expression vector PBJI-neo. Fol- lowing stable transfection of the HLA class I-negative mutant cell line 721.221 ( 2 6 ) , the recognition of the transfected HLA-Cw'1601 by one of the two CTL clones was tested. Because the transfectant was efficiently lysed by the CTLs (Figure 1 ) we concluded that the Cw*1601 antigen was recognized directly.

HLA-C oligotyping analysis of patients an# their unrelated bone marrow donors: HLA-C allelic ftequencies and linkage disequilibrium with other HLA loci

Because of the functional importance of the HLA- C incompatibilities, we set up a PCR-SSO oligotyping procedure with locus C-specific primers (22) and a total of 27 SSO probes (Table 1) essentially based on the protocol in ref. 9. This system allows the discrimination of 29 HLA-C alleles. A total of 22 C w alleles were observed in our panel of 382 individuals. In this group, 90 serological blanks (50%), out of 180 serological typings, turned out to be defined alleles. The Cw*0402, 0801, 0803, 1301,

- transfected not trafl'sfected control + control -

F ~ ~ I I I ' L J f , Specifcity of anti-Cw*1601 CTL clones derived rrom the Cw-mismatched pair I of Table I demonstrated by lysis of the class I-negative 721.221 cell line transfected with Cw*1601 cDNA clorird from the patient. Positive control: Al/3 B*4JO3/

7 Cw7/1601 NegLitive control: A113 l3*4403/7 Cw7/3

Table 3.

HLA-C allelic and haplotype frequencies in the sample of 105 European individuals

Standard Frequency deviation H LA- Cw a 1 I e I e s

Cw'0401 Cw '0702 Cw * 070 1 Cw*0501 Cw'0602 Cw "0304 Cw'1502 Cw ' 150U03 Cw'1203 Cw"0303 Cw '1 601 Cw'0802 cw '02022 CW'1202 cw*0102 Cw'0704 Cw'1602 Cw'l701 Cw'1402 Cw*02023 null

0.1621 0.1402 0.1242 0.0999 0.0685 0.0476 0.0048 0.0476 0.0345 0.0307 0.0238 0.01 90 0.0190 0.0143 0.0099 0.0048 0.0048 0.0048 0.051 7 0.0589

0.0286

0.0242 0.0233 0.0202 0.0202 0.01 55 0.01 39 0.01 53 0.0046 0.01 49 0.01 50 0.01 03 0.01 09 0.01 03 0.0099 0.0069 0.0069 0.0049 0.0045 0.0049 0.01 97 0.0085

Haplotypes

A ~ - B ~ - C W ' O ~ O ~ - O R B ~ '1 SO1 Al-B8-C~'0701 -DRB1'0301 A2-8'3501-C~'0401-DR81 '0101 A2-8"4402-C~ "0501 -ORB1 '1 301 A2-8'4402-CW '0501-DR81'0401 A23-8'4403-CW '04Ol-DRB1'0701 A29-B'4403-C~'1601 -ORB1 '0701 A1 -H57-C~'0602-DR81'0701 A2-B13-C~'0602-DR81'0701 A2-8'4403-C~'0401 -DAB1 '0701 A2-B18-C~'0701 -DRBI '1104

0.061 1 0.0476 0.0238 0.0229 0.0143 0.0143 0.0143 0.01 43 0.01 43 0.01 43 0.0143

0.01 63 0.01 63 0.01 05 0.0092 0.0086 0.0061 0.0086 0.0093 0,0094 0.0083 o+ooa8

1403, I504 and 1505 alleles were not detected in our panei.

The frequencies of the 21 HLA-C alleles and 11 most frequent A-B-Cw-DRB 1 haplotypes found in the patient population (n= 105) are shown in Table 3. As the A, B, and DRBl frequencies in the patients' group are close to the frequencies previously determined in a local panel of 151 platelet donors (271, the HLA-C allelic frequencies in the patients' group should also be representative of the Swiss population. No bias seems to be introduced, either by the diseases, or by the fact that onty patients for whom serologically identica1 donors could be selected from the bone marrow donor registries were analyzed. Cw*0401, 0702, 0701, and 0501 are the most frequent alleles, reflecting the high frequency of the (linked) B35, 97, B8, and B44 antigens in Etiropeans. The heterozygosity level of locus C is 0.909. which is very close to the heterozygosity levels previously crilcuiated for HLA-B

(7)

HLA-C incompatibility in unrelated BMT (0.926) and DRBl (0.923) loci (27). This result

contrasts with the very low level of Cw diversity previously detected by serology, and suggests that this locus may be subject to heterozygous advan- tage, as it is often shown for other HLA loci. How- ever, the homozygosity observed in the Swiss sample (Fob,=O.09l) here falls within the limits of a 95% confidence interval for the homozygosity ex- pected under neutrality (F,,,=O. 1402, confidence interval=(0.0841; 0.2521), as given by an Ewens- Watterson test). Therefore, the neutral hypothesis cannot be rejected.

Preferential co-occurrences of A and Cw, B and Cw, as well as DRBl and Cw antigens observed in the HLA phenotypes of 382 sampled individuals are shown in Table 4. These results suggest strong as- sociations between Cw and some HLA-B alleles, confirming previous observations made by serology (78) and by DNA typing (6,7). In particular the fol- lowing associations were observed: B62-Cw*O303, B*3501 -Cw*O40 1, A'020 l-B*4402-C~*050 1,

B*

I3-C~*0602, B*57-C~*0602, B*52-Cw* 1202, A1-B8-C~*0701-DRB1*0301, A*0301-B*0702- C~*0702-DRB1*1501, A33-B14-C~*0802, A1 1- B52-Cw*1202-DRBl* 1502, and B*4101-Cw*f701.

Sequence of two new HLA-C alleles

upon analysis of the HLA-C mismatch in pair 3 (Table 2), we sequenced the Cw7 allele of the pa-

tient and found that the sequence differed from that of the initially reported Cw*0702 allele (32) by 2 bp substitutions in exon 3 (data not shown), The exon 2 + 3 sequence from this patient (accession number: 249 1 12) corresponds to the Cw*0702 allele that has been recently corrected (33). In the further process of SSO oligotyping of our patienudonor group we found two unusual hybridization patterns with the Cw probes. After cloning and sequencing the (exon 2-in tron 2-exon 3) segment we identified two new Cw alleles, HLA- Cw"O2023 was sequenced from genomic DNA of a leukemic patient of Eastern European origin (AI, A2; B18, B27; DRB1*1104/1301; DRB3*02).

This allele (Figure 2) differs from Cw"O2022 by a single silent substitution at codon 99 (TAT+TAC) that results in the loss of hybridization signal with probe Y99 (probe C7 of ref. 9). The second new Cw allele, now referred to as Cw*0707, was sequenced from genornic DNA of a Swiss European Caucasoid blood donor (A 1 -B8-Cw*0707- This allele (Figure 3) differs from Cw*0701 at 3 non-synonymous nucleotide positions (229, 239 and 268) in exon 2, resulting in a Ser to Asn, Asn to Lys, and Asp to Ala amino acid exchange at positions 77, 80, and 90 respectively. This is the single Cw7 allelic subtype that presents the N77- K80 motif. The presence of this allele on the A l - B8-DR3 haplotype which is in high linkage dis- DRB1*0301 / A l l

-

B 4 4 - C ~ * 0 5 0 1 -DRB1*0101).

Qble 4.

Qo-occurrence of A-Cw, B-Cw and DRB1 -Cw antigens in the sample of 382 individualsn (best associations)

Cw allele A allele f l ~ yo A-CW' B allele flB % 8-CW' DRBl allele ORB^ % DRB1-Cw' Cw'O102

Cw'0303 Cw'0304 Cw'0401 Cw'0501 Cw'0602

Cw'O701 Cw"0702 Cw'0704 Cw'0802 c w 1202 Cw* 1203 Cw* 1402 Cw'1502 Cw * 1601 Cw' 1701

A25 A'0201 A31 A'0201 A1 A30 A1 A'0301 A33 A10 A1 1 A23

A3 1 A29

4 99 20 99 110 18 110 43

7 5 43 26

20 16

50 13 80 34 31 77 49 44 86 80 12 23

45 81

862 B60 8'3501 B"4402 81 3 857 837 88 81 8 6*0702 81 4 852 838 627 8'5102 8'5101 6'4403 8'4101

34 17 57 65 17 15 22 57 48 51 19 8 20 19' 1 2 43 39 4

68 100 100 80 100 100 91 8 9 58 100 95 100 95 74 42 42 54 75

DRBl"1601 DRH1'1501 DRBl 1501

OR81 '0701 ORB1 "1001 ORB1 '0301 DRB1'1501 DRBl*0408 DRBl'0102 DRB1*0403 DR61'1301 DRB1'0801 ORB1 '1201 DRB1'0701

6 87 87

58 14 69 87 '

3 8 10 71 23 15 58

33 16 21

36 57 60 63 33 87 20 24 13 40 24

The table gives the proportion (YO) of individuals carrying a given HLA-Cw allele (either at the homozygous or the heterozygous state), among those (nA, nE, noRer) carrying the corresponding A, B or ORB1 allele. For example, 50% of all A25 antigens co-occur with Cw'0303.The table lists only the cases where a high proportion of A, 6 or ORB1 antigens segregate with Cw.

(8)

Grundschober et al.

Exon 2 ( 1 - 2 7 0 )

c w * 0 2 0 2 1 c w * 0 2 0 2 2 Cw* 02 0 2 3

Consensus cw* 02 02 1 Cwf 02 02 2 Cw"O2023

Consensus C w ' 0 2 0 2 1 c w * 0 2 0 2 2 Cw* 02 023

C o n s e n s u s cw* 02 02 1 c w * 02 022 Cw* 02 023

7 1 1 4 0

AGTGGGCTAC GTGGACGACA C G " C G T GCGGTTCGAC.AGCGACGCCG CGAGTCCAAG AGGGGAGCCG

2 1 1 2 7 0

CACAGACTGA CCGAGTGAAC CTGCGGAJWc TGCGCGGCTA CTACAACCAG AGCEAGGCCG

Intron 2 ( 2 7 1 - 5 1 5 )

C w * 0 2 0 2 3 GTGAGTGACC CCGGCCCGGG GCGCAGGTCA CGACCCCTCC CCATCCCCCA CGGACGGCCC GGGTCGCCCC Cwf02023 GAGTCTCCGG GTCTGAGATC CACCCCGAGG CTGCGGAACC CGCCCAGACC CTCGACCGGA GAGAGCCCCA Cw*02023 GTCACCTTTA CCCGGTTTCA TTTTCAGTTT AGGCCAAAAT CCCCGCGGGT TGGTCGGGGC TGGGGCGGGG C w * 0 2 0 2 3 CTCGGGGGAC GGGCTGACCA CWGGGCGGG GCCAG

7 2 6 7 9 1

Consensus AGGGCGAGTG CGTGGAGTGG CTCCGCAGAT ACCTGGAGAA CGGGAAGGAG ACGCTGCAGC GCGCCG (-"*02021 _ _ _ _ _ _ _ _ _ _ ^_^_^_^_^_^-^d^-^-^-+ . - - - _ _ - - - - * - - - _ _ _ _ _ I _ _ + - I-----

c w * 0 2 0 2 2 - - - - - _ _ - * - _ - - - - - - - - --- _ _ _ _ _ _ _ _ _ _ - - - cw*02023 _ _ _ _ I _ _ _ - - - - _ _ _ _ _ - I - I - - - --- - - I - - -

_ _ _ _ _ _ _ _ _ _

^--_--...

F i g m 2. Full exon 3, intron 2, and exon 3 nucleotide sequence of Cw*02023 (EMBL accession number 272007) from cell KACD.

For comparison the exons 2 + 3 sequences of Cw*OZOZ1 and 02022 are indicated (34). The name Cw*02033 hus been officidty assigned by the WHO Nomenclature Committee in May 1996. This follows the iigreed policy that, subject to the conditions stated in the most recent nomenclature report (35). names will be ussignrd to new sequences ;ts they are identified. Lists of such new names will be published in the Vollowing WHO nomenc1;rtiire report.

(9)

HLA-C incompatibility in unrelated BMT Exon 2 ( 1 - 2 7 0 )

Cw*0707

Cwf07 07

Cw*0707

1 74

GCTCCCACTCCATGA~TATTTCGACACCGACACCGCCG~TCCCGGCCCGGCCGCGGAGAGCCCCGC~CATCTCAG~

C S f l S M R Y F D T A V S R P G R G E P R F I S V -

1 10 2 0 2s

75 149

GGCTACGTGGACGACACGCAGTTCGTGCGGTTCGACAGCGACCCCGCGAGTCCGA~AGGGGAGCCGCGGGCGCCG

G Y V D D T Q F V R F D S D A A S P R G E P R A P

3 0 4 0 5 0

1 5 0 2 2 4

T G G C T G G A G C A G G A G G G G C C ~ A G T A ~ A C C G G G A G A C A C A G ~ C T A C ~ G C G C C A ~ C A C A ~ C T G A C C G A

W V E Q E G P E Y W D R E T Q N Y K R Q A Q A D R

60 7 0

22 5 27 0

GTGA&ZCTGCGGAA&2TGCCCGGCTACTACA.&CCAGAGCGAGGCCG

V N L R K L R G Y Y N Q S E A

80 90

J n t r o n 2 (271-520)

Cw*0707 GTGAGTGACC CCGGCCCGGG GCGCAGGTCA CGACCCCTCC CCATCCCCCA CGGACGGCCC GGGTCGCCCC Cw*0707 GAGTCTCCCC GTCTGAGATC CACCCCAAGG TGGATCTGCG GAACCCGCCC AGACCCTCGA CCGGAGAGAG C w f 0 7 0 7 CCCCAGTCGC CTTTACCCGG TTTCATTTTC GGTTTAGGCC AAAATCCCCG CGGGTTGGTC GGGGCGGGGC Cw*0707 GGGGCTCGGG GGACTGGGCT GACCGCGGGG GCGGGGCCAG

on 3 (521-796)

Cw*0707

Cw* 07 07

Cw*0707

5 2 1 5 9 4

GGTCTCACACCCTCCAGAGGATGTATGTATGGC~CGACCTGGGGCCCGAC~GCGCCTCCTCC~C~~TA~ACCAG

A S H T L Q R M Y G C D L C ~ D G R L L R G Y D Q

9 1 10 0 110

5 9 5 6 6 9

TCCGCCTACGACGGCElAGGATTACATCGCCc~~CGAGGACCTGCGCTCCTGGACCGCCGCGG~CACCGCGGCT

S A Y D G K D Y I A L N E D L R S W T A A D T A A

120 13 0 140

67 0 744

CAGATCACCCAGCGCAAGTTGGAGGCGGCCCGTGCGGCGGAGCAGCTGAGAGCCTACCTGGA~GCACGTGCG~

Q I T Q R K L E A A R A A E Q L R A Y L E G T C V

150 160

745 796

GAG’IY3GCTCCGCAGATACCTGGAGAACGGGAAGGAGAAGGAGACGCTGCAGCGCGCAG E W L R R Y L E N G K E T L O R A

17 0 180

Figwrre 3. Full exon 2, intron 2, and exon 3 nucleotide sequence of Cw*0707 (EMBL accession number 279751) from cell HAHR.

The nucleotide positions that differ from the Cw*0701 sequence (34) are underlined. The intron 2 sequences of ^Cw*0707and CW*O701(36) are identical. The deduced Cw*0707 amino acid sequence differs from that of Cw*0701 at codons 77 (Ser ws Asn), 80 (Asn vs Lys) and 90 (Asp vs Ala). The name Cw*0707 has been officially assigned by the WHO Nomenclature Committee in Ocilober 1996, This follows the agreed policy that subject to the conditions stated in the most recent nomenclature report ( 3 9 , natrics will be assigned to new sequences as they are identified. A list of such new iiarnes will be published in the following WHO norwmhiture report.

equilibrium with Cw*0701, as well as the intron 2 sequence comparison (Figure 3) (36), support the hypothesis that Cw*0707 resulted from a gene con- version-like event that involved transfer of ⁱⁱshort

segment of the alpha I domain comprising at least codons 77-90 in a Cw*0701 recipient allele. The boundaries of this putative recombination would be situated within codons 44-76 at the 5’-end de-

(10)

Grundschober et al.

pending on the donor allele, and between nucleotides 1-80 of intron 2 at the 3'-end. Interestingly, the CTLpf test performed in the intra-Familial combination that differed only at the locus HLA- C (Cw*0707 versus Cw*0701) was positive (3 pre- cursors/106). This implies that the motif S77-N80- D90 on the a-helix is important for CTL allorec- ogni tion.

Frequency of mismatched HLA-C alleles in AB-sero- and DRB1/B3/BS-oligo-compatible pairs

We have previously shown that high-resolution DRB 1/B3/B5 matching significantly affects the rate of HLA-A and -B subtype compatibility (1 5).

Linkage disequilibria will also determine the rate of HLA-C incompatibilities in combinations matched for ABDR antigens. We assessed the chance of being mismatched for HLA-C by study- ing the 287 matched pairs that could be formed from 73 patients and their 184 potential unrelated donors compatible for HLA-AB by serology and for DRBl/B3/BS by oligotyping. In all of these pairs AB-subtype and HLA-C polymorphism were determined by oligotyping and/or sequencing. As shown in Figure 4, the overall degree of ABC match was 51.3%. That is, even after high resolution DRB matching, about half of the potential donors will still be incompatible for at least one CIass I antigen. 24.6% of the pairs were incompatible for HLA-C only, whereas 12.50/0 were mismatched for HLA-A/B and HLA-C and 6.6% for HLA-A/B only.

It should be noted that of the 44 patient/donor combinations that were incompatible for HLA-C, 41 were CTLp-positive. All of the 17 pairs that had an additional A or €3 mismatch as determined by oligotyping were in the group with a positive CTLp test. In the 27 pairs, in which the absence of

4

A+E+Cw match

A or B mismatch

A+Cw or B+Cw mismatch

F ~ ~ I I I Y J 4. Schematic representation ut' the occurrence of HLA class I incompatibilities in 287 pairwise conipxisons derived t'roin 73 leukeiiiic puticnts and their IYJ HLA-AB-sero. -DRB I / BYBj-oligo matched potential unreluted dunurs.

Table 5.

Co-occurrence of B and Cw antigens in the sample of 382 individuals: the table tists only the cases where a high heterogeneity of Cw co-occurs with each E antigen'

HLA-8 alleles 4 HLA-Cw allele % 6-Cw 6 ' 4403 39 1601 54

0401 44

0701 10

0702 8

0501 5

0602 5

6*5101 43 1502 42

01 02 21

1402 12

0702 9

0701 5

1602 5

1503 2

E l 8 48 0701 58

. 1203 17

0702 15

0602 6

03 2

862 34 0303 67

15 3

0401 3

050 1 3

a The table gives the proportion (%) of individuals carrying a given HLA-Cw allele (either at the harnozygous or the heterozygous state), among those (nB) carrying the corresponding 6 allele.

additional A/B-subtype mismatches was confirmed by oligo typing or sequencing, CTLp frequencies were in the range of 2-45/106 with a mean of 10/

lo6. Although these vahes are somewhat lower that the ones found in 25 CTLp-tests with an HLA-A or B ( ~ C W ) mismatch (3-91/106, mean of 29/106), it is clear that HLA-C incompatibilities are efficiently recognized by CTLs.

The rate of HLA-C incompatibilities in various ABDR haplotypes

The results shown in Tables 3 and 4 suggest that cer- tain ABDR haplotypes are at relatively tow risk to be mismatched for HLA-C. The common haplo- types A1 -B8-Cw*070 1-DRB 1 *0301 -DRB3*0101, A*030l -B*0702-Cw*0702-DRBl* 1501, or Ax- B*4402-Cw*O501-DRBl*04/1301/1501 can thus be considered as "low risk haplotypes" with respect to Cw incompatibility. In contrast, Cw mismatches occur frequently in haplotypes containing the B*4403, B*5101, B18, or 862, alleles (Table 5 ) . For instance only 54'M) of the B'4403-positive donors typed Cw*1601 and only 42%) of B*5101 haplotypes were Cw* 1502. Clearly high resolution DNA typing for HLA-A and -B m d DRBl does not guarantee fill1

(11)

HLA-C incompatibility in unrelated BMT matching at locus C. Interestingly, haplotypes

matched for HLA- C and DRBl by high resolution oligotyping could still be mismatched for HLA-A/

B subtypes: such mismatches were identified for the subtypes of B35 where B*3501,3503 or 3508 could all be associated with Cw*0401.

Discussion

The data presented here illustrate the relevance of HLA-C incompatibilities for the matching of unrelated donorlrecipient combinations. Firstly, they show that CTLs recognize HLA-C incompati- biIities almost as efficiently as they recognize HLA- A or HLA-B incompatibilities. In all but 3 of the 27 tested combinations with a Cw mismatch only, the Cw incompatibility resulted in a positive CTLp test. Of the three HLA-C incompatible pairs with a negative CTLp test, in one combination the mis- rrbatched alleles were Cw*0303 versus Cw*0304, which differ for a single amino acid (position 91) outside the peptide binding groove. One might ar- gue that such differences are immunologically ir- relevant (37) and therefore, could be considered as an acceptable mismatch. The other two cornbi- n,a tions however, were common HLA-mismatches (Cw* 1502 versus 170 1, Cw*0304 versus 050 1) which should have induced an ailogenic reaction.

We are reluctant to speculate on these two cases, in particular because we could not repeat the CTLp tests due to lack of viable patient cells.

The second finding of this report is that, even in well matched donodrecipient combinations HLA- C mismatches are rather frequent. A large scale analysis of donodrecipient pairs was done with an aligotyping protocol based on hybridization with 27 SSO probes following one locus C-specific ex- clns 2 + 3 PCR. This oligotyping procedure covered practically all of the sequenced C w alleles and has qow been introduced in our routine typing labora- tory. It is highly efficient to pick-up new alleles by identifying new hybridization patterns, which allowed us to detect 2 new Cw alleles, Cw*02023 and Cw*0707. The analysis of 287 pairs of HLA AB-serology, DRB lIB3IBS-oligo matched patients and unrelated donors revealed an overall HLA-C qisrnatching rate of 42.1%. In 30% of these cases

the Cw incompatibility was accompanied by a subtype-mismatch at either one or both HLA-A and E$ loci, as determined by SSO-oligatyping for A?,

~ 4 3 ~ B7, B41, B44, or B5 1, or by sequencing analy- stis. This rate of HLA-C mismatch is comparable ( 3 8 4 0 ) or higher (7) to those observed previously cbn a significantly lower number of pairs. The identification of “high-risk haplotypes” (e.g. B44,

85

1 ) with respect to Cw incompatibility iinderline

t h e idvantage of H LA-Cw DNA typing for

the optimal selection of HLA class I-matched donors.

A strong correlation between high CTLp frequencies and incompatibilities at locus HLA-C has recently been reported (39). Because of the variable linkage disequilibria between

B

and C alleles, the effect of individual C mismatches on cellular reac- tivity is difficult to evaluate without high resolution A/B typing. It is interesting to note that 8 of the 9 Cw-mismatched combinations in this study (39) are linked to either B14, B44 or BYB51, three antigens known to be split into subtypes. The high CTLpf values reported by these authors in 8 of the 9 combinations could therefore also be accounted for by a B+Cw mismatch (eg. typically B*4402- Cw*0501 versus B*4403-Cw*I601).

Differences in the clinical outcome for patients with serologically distinguishable (“major”) HLA- AB mismatches or with (“minor”) mismatches of HLA-AB antigens that share serologically defined epitopes (41) could be partially explained by the presence of additional Cw incompatibilities that are frequently associated with the major serological AB incompatibilities. As a consequence, sero- -logically A13 mismatched donodrecipien t pairs will

more often be mismatched for a second (HLA-C) antigen, possibly increasing the risk of post transplant complications. On the other hand, depending of the HLA-B-subtype mismatch, additional Cw incompatibility may also occur more frequently, as was observed in the case of B”4402 versus B*4403.

Such pairs, that would have been classified in a group with “minor” HLA incompatibilities, are in fact mismatched at both HLA-B and -C loci.

The impact on patient’s survival of HLA class I mismatches identified by serology (42-44), by the CTLpf test (12, 45), or by a Combination of serology/DNA typing and CTLp analysis (46) has been reported. in the present study we have shown that, although the frequencies in the CTLpf test might be lower, HLA-C incompatibilities are recognized by alloreactive CTLs. Clinical consequences of a single HLA-C incompatibility can now be analyzed. Our clinical data are still preliminary, because unti! now we could study only ten patients transplanted with an HLA-AB subtype, DRB 1/B3/

BS-oligo-matched but Cw-mismatched unrelated donor. Of these patients 40% are alive (follow-tip:

0.5 to 6.5 years) which is in between the 3-year survival probability of our group of patients with a HLA-ABCDRB l/B3/BS/DQB 1-matched- CTLpf-negative unrelated donor and that of a mismatched group (46). The predictive value of a positive CTLp test for increased risk of GvHD and decreased survival ( 1 2, 45) suggests that, whenever possible, we should consider Cw incompatibilities in the donor selection procedure. Such incompati-