Distribution of HLA antigens in Spanish Gypsies: A comparative study Rosario de Pablo, Carlos Vilches, Maria €.Moreno, M. Carmen Rementeria, Rosario Solis and Miguel Kreisler
R. de Pablo, C. Vilches, M. E. Moreno, M. C. Rementeria, R. Solis, ' M. Kreisler. Distribution of HLA antigens in Spanish Gypsies: A comparative study. 1 Tissue Antigens 1992: 40: 187-196.
Abstract: We have studied the HLA-class I and class I1 antigen distribution in a sample of 75 Spanish Gypsies and 74 Spanish non-Gypsies by serology, restriction fragment length polymorphism, and protein chain reaction and hybridization with allele-specific oligonucleotide probes. When both population samples are compared, we find that Gypsies have a statistically significantly higher frequency of Al, A l l , B61, Cw6, DQ5 and haplotypes DRi6 DQ5 Dw21 and DR14 DQ5 Dw9 DR52b. Frequency of A3, A29, B44, DR4, DQ2, DQ8 and haplotypes DRI DQ5 and DR7 DQ2 DB17 DK53 are significantly lower in this ethnic group. The analysis of the serological data in the two populations demonstrates that Cw6 can be split into long Cw6 (Cw6.1) and short Cw6 (Cw6.2). Haplotype Al-Cw6-B61-DR14-DQ5is the most characteristic in Gypsies, with a frequency of 13%. Estimation of the genetic distances shows that Spanish Gypsies are closer to Indian Caucasoid populations than to the Spanish non-Gypsy population. HLA data support the proposed historical origin
Introduction
Gypsies are an ethnic group that is historically considered to be of northern Indian origin. They moved out of that area probably between AD 800 and AD 950 and, migrating westwards, they arrived in Europe around AD 1100. Linguistic evidence ( 1 ) suggests that the ancestors of all Gypsy populations are the Domba or Dombari (2), who constituted a single race speaking the same language, which subsequently diverged into two linguistic branches: the Nawar, Kurbat, Karaci and Helebi now found throughout Egypt and the Middle East on the one hand, and the Bosa in Armenia and eastern Turkey, and the Rom or Roma in Europe, on the other. However, on the basis of more recent scholarship, there is some reason to believe that the three populations usually thought to comprise the descendants of the Domba may in fact have left India at different times and under different circumstances (2). The three branches exhibit considerable lexical adoption from Persia, for example, but there are no items shared by all three branches. The same is true for the Armenian items in Central and Western Gypsies. Conservative Romani dialects remain two-thirds or more Indian in their basic lexicon and grammar,
Servicio de Inmunologia, Hospital Puerta de Hierro, Madrid, Spain
-
Key words: HLA-A,B,C HLA-class II allogenotypes - Spanish Gypsies - SpMi:;h non-Gypsies HLA Class II RFLP - HLA Class II PCR-SSOP
I
-
Received 14 April, revised, accepted for publication 1 JUIY 1992
retaining features which have become lost in their neo-Indic cognates. The nomadic lives and cultural peculiarities of Gypsies made them an ethnic group that seldom integrated with other societies, endogamy being almost complete in the gypsy community. Since they arrived in Spain, they have continued to be genetically isolated up to the present day. To characterize human populations and their genetic relationships, anthropologists and geneticists have taken advantage of the high polymorphism of the major histocompatibility complex (MHC). HLA products are essential to the immune response because they represent the molecular context for antigen presentation to CD8+ and CD4+ T cells. Moreover, the requirements for HLA matching in transplantation and the association of several diseases to certain HLA class I and I1 antigens underline its medical importance. All these reasons induced us to analyze HLA distribution in the second most populous ethnic group of Spaniards. We have studied the HLA-A, B, C, DP, DQ and D R antigenic distribution in Spanish Gypsies and non-Gypsies. We have compared Spanish Gypsies with Hungarian Gypsies (3), with four different populations from India, with eight European populations and with the Armenians (4). 187
de Pablo et a]. Material and methods Population studied
We have studied a sample of unrelated Spanish Gypsies (75 individuals by serology, 76 by RFLP and 77 PCR) and Spanish non-Gypsies (74 individuals by serology, 96 by RFLP), all of them living in Madrid but native to different regions of Spain. One Gypsy family was studied for segregation analysis. Serology
HLA-A,B,C and HLA-DR, HLA-DQ typing were performed on purified T- and B-lymphocyte suspensions, respectively, using the standard NIH microlymphocytotoxicity technique. For this purpose, we used the set of sera of the anthropological component of the 1 1th International Histocompatibility Workshop (1 1th IHW) and our local sera. DNA samples
Genomic DNA was obtained from peripheral blood leukocytes using the salting-out method (5). Southern blot analysis
Aliquots of 15 pg DNA were digested with TuqI and MspI according to the manufacturer's recommendations. Agarose gel electrophoresis, vacuum blotting of size-separated DNA fragments onto nylon membranes, labeling of purified probe inserts with random primers, prehybridization, hybridization, high-stringency washes and autoradiography were performed according to standard techniques with minor modifications (as described in (6)).
PCR amplification of genomic DNA
The primers used in this study were provided by the 11th International Histocompatibility Workshop. The amplification was performed following the protocols of this Workshop (12). The primers were those provided by the organizers of the 11thIHW (13). Dot blot and SSOP hybridization
Dot blotting .and hybridization were performed as described in the 11 thIHW protocols (12), using the sequence-specific oligonucleotide probes (SSOP) provided by the organizers (1 3). The washing procedure after hybridization was that described by Wood et al. (14). Statistical analysis
HLA-A,B,C,DR and DQ gene frequencies were derived from serology results (15, 16). Genetic distances were calculated as previously described (17). RFLP genotypic relative frequencies were calculated with respect to the total number of haplotypes studied in each sample. Phenotypic frequencies of HLA-DRBI, DRB3, DQA1, DQBI, DPAl and DPBl were calculated as a percentage. Haplotype comparisons were made using the reference tables of the 1 1th International Histocompatibility Workshop (4,18). Data were analyzed by the chi-square test with the Yates' correction. Probability (p) values were always corrected with a factor dependent on the number of alleles or DR-DQ haplotypes studied (p,= 1 -(1 -p)", with n being the number of antigens in that allelic series). DRBI-DQAl -DQB 1 associations were deduced from RFLP and PCR-SSOP results Results
RFLP probes
The DNA probes used were: DRB, a 785 bp Sac11 Hind11 fragment of the cDNA clone pII-B-4 complementary to the DRBI gene (7); DQA, a 725 bp ApaI fragment of the cDNA clone p-II-a-5 complementary to the DQAZ gene (8); and DQB, a 625 bp h a 1 fragment of the cDNA clone pII-B-1 complementary to the DQBl gene (9). Assignment of RFLP haplotypes
Tag1 DRB, DQA and DQB, and MspI DQA restriction fragment patterns were analyzed as previously described (10, 11). 188
The analysis of the frequencies of the samples compared here demonstrate that both populations are under Hardy-Weinberg equilibrium. Serological' results
The gene frequency of the HLA-A allelic series of the two population samples studied are shown in Table 1. The elevated frequency of A1 and A l l and the lower frequency of A3 and A29 are the main features of Gypsies when compared with nonGypsies. When the gene frequencies of the alleles of the HLA-B series of the two population samples are compared (Table 2), the main differences are found
HLA in Spanish Gypsies Table 1. Gene frequencies of the HLA-A allelic series of a sample of Spanish Gypsies and non Gypsies
HLA-A
Spanish Gypsies N375 ~~
A1 A2 A3 A1 1 A23 A24 A25 A26 A28 A29 A30 A31 A32 A33 A34 ABL
Spanish nowGypsies N=74
Xu2
Pc
18.755 7.718 10.167 19.042 0.549 0.587 0.706 0.3W 0.000 12.297 0.549
0.005
~
0.283 0.119
0.058 0.258 0.020
0.063 0.013 0.047 0.01 3 0.01 3 0.020 0.000 0.073 0.007 0.000 0.01 3
0.077 0.264 0.1 69 0.070 0.034 0.077 0.027 0.027 0.01 3 0.130 0.034 0.041 0.01 3 0.020 0.006 0.000
6.055 6.056 1.015 0.830
-
N.S.
0.05 0.005 N.S. N.S. N.S. N.S. N.S.
0.05 N.S. N.S. N.S. N.S. N.S.
-
in B44 and B61. While B44 is the allele with the highest frequency in the Spanish non-Gypsy population, B61 is the most characteristic among Gypsies. The high frequency of B27 among Gypsies,
Table 2. Gene frequencies of the HLA-8 allelic series of a sample of Spanish Gypsies and non Gypsies
HLA-8 87
88 814 818 827 835 837 838 839 841 B44 845 849 850 851 852 855 656 857 858 860 861 862 863 870 BBL
Spanish Gypsies N=75
Spanish nowGypsies N=74
xvz
0.047 0.01 3 0.01 3 0.087
0.122 0.048 0.055 0.041 0.01 3 0.1 07
4.923 2.880 3.706 2.766 6.903 0.299
0.000
6.055
0.048 0.006 0.01 3 0.21 1
6.898 3.455 2.002 13.792
0.006
0.000
0.048 0.020 0.092 0.01 3 0.034 0.006 0.01 3 0.006 0.027 0.020 0.020 0.020 0.006 0.005
0.1 00 3.1 33 0.066 1.975 0.549 0.830 1.975 0.954 4.1 83 16.915 0.217 3.133 0.000
0.080 0.1 27 0.040 0.000 0.040 0.000
0.053 0.007 0.040 0.000 0.093 0.040 0.020 0.000 0.047 0.020 0.000 0.21 3 0.01 3 0.000 0.007 0.000
-
while not statistically significant with respect to non-Gypsies, is, however, of interest given the described association of this antigen with ankylosing spondilitis. Among the alleles of the HLA-C seriles, Cw6 is highly represented in the Gypsy population (Table 3). When we look at the reaction patterns of the Cw6-specific sera of the 1 1thIHW (1 9) - SVEll4 [Workshop Number 4321, DDC103 FV.N. 4331, ROO103 [W.N. 4341 and AST109 [W.N. 4351 -with the cells of the population samples under study, two patterns emerge that we call long-Cw6 and short-Cw6 (Table 4). Long Cw6-bearing cells show a positive reaction with SVEll4, DDC103, ROO103 and AST109, while short Cw6 cells react only with ROO103 and AST109. The: so-called short pattern can only be defined in the absence of Cw4 because AST109 recognized all Cw6 cells and 36% of Cw4. All except one of the clells (KRE 1047) showing the short-Cw6 reaction paittern were of Gypsy origin. Our finding of two clear-cut reaction patterns in such a great proportion of cells points to the existence of two Cw6 splits, Cw6.1 (long) and Cw6.2 (short). The segregation of the short pattern in a Gypsy family (Table 4) confirms the split. When the serological data on the IDR allelic series (Table 5) are compared in the two populations, the main differences are the higher frequency of DR16 and DR14, and the lower frequency of DR4 and DR7 in Gypsies. In the DQ series (Table 5), we find DQ1 frequency increased in Gypsies, while DQ;! is significantly lower among them.
PC
N.S. N.S.
N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. 0.05 N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. 0.01 N.S.
N.S. N.S.
-
RFLP results
The frequencies of the HLA-DR-DQ a.llogenotypes determined by RFLP are shown in Table 6. The most striking differences are the fo bowing: Table 3. Gene frequencies of the HLA-C allelic series of a sample of Spanish Gypsies and non Gypsies
HLA-C Cwl cw2 cw4 cw5 Cw6 cw7 cw9 CWlO
CBL
Spanish Gypsies N-75
Spanish non-Gypsies N=74
0.000 0.1 16 0.098 0.041 0.293 0.1 27 0.027 0.000 0.298
0.020
2.002
N.S.
0.048 0.137 0.1 22 0.055 0.237
3.887
N.S. N.S. N.S. 0.0005 N.S. N.S. N.S.
0.048
0.041 0.292
xu2 -o.tjao 6.055 25.626 4.51 9 1.723
6.055
-
Pc
-
Cw8 was tested but it was absent in both population samples.
189
de Pablo et al. Table 4. Cw6 reaction patterns
Antisera
cell
A
6
BW
cw
KRE2Oll KRE2023 KRE2024 KRE2058 KRE1050 KRE2063 KRE2038 KRE2074 KRElO57 KRE1067 KRE1069 KRE1072 KRE2006 KRE2008 KRE2037 KRE2066 KRE2078 KRE1012 KRE2003 KRE2052 KRE2059 KRE2067 KRE2056 KRE2012
A11A23 A 1 A 1 A l,A 2 A31,A33 A 1,A23 A11,A 1 A26,A32 A 2,A26 AJ,A 2 A 2.A 2 A 1,A24 A26A25 AllA 1 A24,A26 A11,A32 A 1 A 2,A23 A11,A26 A 1,A 3 A 1,A32 A 1,A 2 A 1,A29 A1
844,661 844,657 835,661 635,857 8 7,635 ' 844,661 835,861 627,837 650,814 651,657 849,650 6 8.657 862,637 839.637 637,655 645,627 637,661 844,650 837.639 635,657 627,657 6 7,657 644,657 661
6~416 Bw4 Bw6 Bw416 Bw6 Bw416 Bw6 Bw4 Bw6 Bw4 Bw416 6~416 Bw416 6~416 6~416 Bw416 Bw416 Bw416 6~416 6~416 Bw4 6~416 Bw4 Bw6
Cw6,Cw4 Cw6,Cw4 Bw6,Cw4 Cw6,Cw4 Cw6,Cw4 Cw6,Cw4 Cw6,Cw4 Cw6,Cw2 Cw6 Cw6,Cw2 Cw6 Cw6 Cw6,Cw9 Cw6 Cw6,Cw9 Cw6,Cw2 Cw6 Cw6,Cw5 Cw6 Cw6 Cw6,Cw2 Cw6,Cw7 Cw6 Cw6
KR E2042 KRE2018 KRE2004 KRE2014 KRE2Ol9 KRE2070 KRE2073 KRE2131 KRElO47 KRE2007 KRE2Ol6 KRE2033 KRE2041 KRE2048 KRE2053 KRE2071 KRE2064
A11,A 1 A 3,All A1 1,A32 A1 A1 1 A11 A 1,All A l,A 2 A 1,A24 A 1,All A 1.A68 A 1,All A1 A l.A 2 A 1,All A 1,All A11,A32
6 7,661 6 7,661
627,861 661 661 661 661 618,661 651,662 639,661 651,661 649,661 661 651.661 658,861 662,861 651,627
Bw6 Bw6 8~416 Bw6 Bw6 Bw6 Bw6 Bw6 Bw416 Bw6 Bw416 6~416 Bw6 Bw416 Bw416 Bw6 Bw4
Cw6,Cw2 Cw6 Cw6,Cw2 Cw6 Cw6 Cw6 Cw6 Cw6,Cw5 Cw6 Cw6 Cw6 Cw6 Cw6 Cw6 Cw6,Cw7 Cw6 Cw6,Cw2
KRE2062' KRE2073t KRE20755
A26,A32 A 1,All A11,A32
827,652 661 627,661
Bw4 Bw6 6~416
cw2 Cw6.2 Cw2,Cw6.2
c
c
w 6
w 4
+
+
+
+ +
+
+ + + +
+ +
+ + + + + + +
+ + + + + + + +
+ + + + + +
D D C 1 0 3
S V E 1 1 4
++ ++ ++
++ ++ ++ ++
++ ++ ++ ++ ++ ++ ++ ++ ++ ++
+
+
+ +
+ +
+
+
+ +
++ ++ ++ ++ ++ ++ ++ -k ++
++ ++ +
+ + +
+ -
-
-
-
Y E T 2 0 3
+ +
++ ++
++ ++ ++ ++
+
K W C 2 2 8
++ ++ ++ ++ ++ ++ ++ ++ ++
++ +
+ + + + + + + +
R O O 1 0 3
++
+ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ + ++ ++ + ++ ++ + ++ ++ + ++ ++ + ++ ++ - + ++
+
+ -
A S T 1 0 9
-
+
+
++ ++ +
++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ -
-
++
-
-
++
-
-
++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++
++
-
-
-
-
-
-
-
-
++ - + - -
+ + - + + - -
~
* Father. t Mother.
5
Daughter.
++ Strong positive reaction. + Positive reaction. - Negative reaction.
- D Rl is significantly lower in Gypsies than in non-Gypsies. - When the frequency distributions of the DR2 splits in the population samples are compared (Table 6), it is clear that haplotype DR15 DQ6 Dw2 has a lower frequency in Spanish Gypsies while, in contrast, haplotype DR15 DQ6 Dw12 has '
a higher frequency in this group, but neither of these differences is statistically significant. DR16 DQ5 Dw21 is significantly overrepresented in Gypsies. This contrasts with the DR2 haplotype distribution in Spanish non-Gypsies where DR15 is more frequent than DR16. Comparing the two ethnic groups, the overall frequency of DR2 is
HLA in Spanish Gypsies Table 5. Gene frequencies of the HLA-DR and DQ allelic series of a sample of Spanish Gypsies and non Gypsies ~
DR1 OR15 DR16 DR3 OR4 DR11 DR12 DR13 OR14 DR7 OR8 OR9 DRBL DQ1 DQ2 DQ3 DQ4 DQBL
Spanish Gypsies N-75
Spanish nonGypsies N=-74
0.034 0.083 0.1 35 0.083 0.040 0.113 0.013 0.062 0.288 0.076 0.006
0.1 22 0.092 0.01 3 0.099 0.203 0.092 0.007 0.070 0.01 3 0.21 1 0.041 0.020 0.01 7 0.332 0.283 0.302 0.055
0.000 0.067 0.600 0.1 05 0.1 91 0.01 3 0.091
XVZ
Pc
7.309 0.066 12.799 0.221 15.552 0.335 0.303 0.077 27.976 9.87 3.584 3.1 33
N.S. N.S. 0.01 N.S. 0.005 N.S. N.S. N.S. 0.0005 0.05 N.S. N.S.
13.661 13.013 4.1 92 3.706
0.005 0.005 N.S. N.S.
-
-
0.028
-
-
similar because the two haplotype frequencies compensate for one another. - DR4 frequency is significantly diminished in Spanish Gypsies with respect to non-Gypsies. This Table 6. Relative haplotypic frequencies in % of DR-DQ-Dw haplotypes in Spanish Gypsies and non Gypsies Samples
DRw,DQw,Dw Taql-DRE allogenotypes DR1IBr,DQ5,Dwl ,Dw20 DR15,DQ6,Dw2 DR15,DQ6,Dwl2 DR15,DQ5 DR16,DQ5,Dw21 DR17,DQ2,Dw3,DR52(Dw24) DR17,W2,Dw3,DR52(Dw25) DR4,DQ7,Dw4/13,DR53
3.2 0.7 7.2 0.7 14.5 2 7.9 2 DR4,DQ8',Dw4/10114115,DR53 2.6 DR11,W6,DwFS,DR52(Dw25) 0 DR11.DQ7,Dw5/DwJVM,DR52(Dw25) 11.2 OR12,DQ7,DwDB6,DR52(0w25) 0.7 DR13,DQ6,Dwl8,DR52(Dw25) 0.7 DR13a,DQ6,Dwl8,DR52(Dw25) 1.3 D R ~ ~ ~ . D Q ~ , D w I ~ , D R ~ ~ ( D w ~ ~ )5.2 DR13b,DQ7,DwHAG,DR52(0w24) 0 DRl4,WS,Dw9,DR52(Dw25) 30.9 DR7,DQ2,Dwl7,DR53 2.6 DR7,DQ2,DwDBI,DR53 0 DR7,DQS,Dwll 5.2 DR8,DQ8.1/8.2 1.3 DR9,DQ9,Dw23,DR53 0 DR1O,DQ5 0 *: See reference 20.
Studied
GYPSY Non GYPSY % (N=76) % (N=96)
14 6.3 1.6 2.1 1.6 5.2 7.8 4.2 12 0.5 8.3 1 4.7 1 1 0.5 1.6 17.7 1.6 1 3.6 1.6 0
is mainly due to the low frequency of haplotype DR4 DQ4/8 (Table 6). When DR4 haplotype frequencies are compared, no significant differences are found. - DR7 is poorly represented in Gypsies, especially due to the low frequency of allogenotype DR7 DQ2 Dw17 DR53 with respect to the Spanish non-Gypsy population. Haplotype DR7, DQ9, Dwll is more frequent among Gypsies than in Spanish non-Gypsies, but this does not compensate for the already mentioned DR7-DQ2 underrepresesntation. - Considering the aforesaid and given that DR9 is absent in the sample of Gypsies studied, DR53 frequency is significantly low in this population. The frequency of serologically-detected DR53 is even lower (data not shown) because haplotype DR7, DQ9, Dwll harbors a pseudogene in the DRB4 locus. - DR6 frequency is considerabty increased among Gypsies (p, < 0.000001), especially due to the overrepresentation of DR14 DQ5 Dw9 (30.9% in Gypsies vs. 1.6% in non-Gypsies), the frequency of DR13 being similar in the two population samples. The above-mentioned haplotypic differences imply a significantly high frequency of DR52b (Dw25), Dw9 and Dw21 among Gypsies, while the frequencies of DR53 and Dw17 are significantly low in this group. As can be seen in Table 7, DQ2, DQ5 and DQ8 show significant differences when the two ethnic groups are compared. Considering that DQ4 is seldom associated with DR.4 in individuals of Caucasoid origin. DQ4/8 can be taken as DQS (* in Table 6) (20).
pc
0.05 N.S. N.S. N.S. 0.01 N.S. N.S. N.S. N.S. N.S. N.S.
N.S. N.S. N.S. N.S. N.S. 10-5 0.05 N.S. N.S. N.S. N.S.
-
PCR SSOP data
The distribution of HLA-DRBI and DRB3 alleles among Gypsies is shown in Table 8. These data confirm the existence of a characteri:stically high frequency of DRBl*1401/ 1404 and an increase in Table 7. Relative haplotypic frequencies in % of DQ antigens in Spanish Gypsies and non Gypsies, defined by RFLP
DQ5 DQ6 DQ2 DQ7 DQ8' DQ9 DQ4
GYPSY % (N=76)
Non-Gypsy % (N-96)
49.3 15.1 12.5 13.9 2.6 5.2 1.3
19.3 15 32.3 14
12 3.0 3.6
XZ
Pc
32.250 0.021 17.337 0.01 6 10.766 0.521 1.009
< 0.0001 N.S.
< 0.005 N.S.
< 0.05 N.S. N.S.
In Caucasians, DR4 was always found associated to Dlaw8 (20).
191
de Pablo et al. Table 8. Frequency in % of DRB1, DRB3, DQB1, DQAl, DPB1 and DPA1 in Spanish Gypsies determined by PCR-SOP
ORB1 (N=77)
FF
DQB1 (N=77)
0101 0102 1501 1502 1601 1602 0301 0401 0402 0404 1101 1102 1103 1104 1201 1301 1302 1303 1401+140w) 07 0801 0901
5.20 1.30 2.60 15.6 22.1 1.30 14.3 1.30 1.30 5.20 11.7 1.30 2.60 5.20 1.30
0501 0502 0503 0504 0601 0602
DRB3 (N=47)
0101 0201 0202 0301
FF
6.60 25.0 51.3 0000 15.8 1.30 3.90 10.5 0000 22.3 26.3 5.20 10.5 0000 2.60
0603 0604
0605 0201 0301 0302 0303 0401 0402
6.50 7.80 DOAl (N=77) 1.30 0101 50.6 0102 10.3 0103 2.60 0201 2.60 0301 0401 FF 0501 3.90 0601 0000 93.5 7.80
FF 57.9 38.1 18.4 15.8 9.20 2.60 39.5 0000
DPBl (N-73) 0101 0201 0202 0301 0401 0402 0501 0601 0801 0901 1001 1101 1301 1401 1501 1601 1701 1801 1901 DPAl (N=73)
01 02
FF 0000
35.6 4.1 0 9.60 53.4 24.6 1.30
5.50 0000 0000 0000 0000 10.9 0000
0000 0000 0000 0000 2.70
FF 89.3
38.6
FF: Phenotypic frequency. * In presence of DR3, DRwll and DRwl3, DRB1.1404 can not be distinguishedfrom DRB1'1401 by using the Ubrkshop oligonucleotides. DRBl'1404 phenotypic frequency is 28.57%. DRBl'14OlI 1404 phenotypic frequency is 22%.
DRB1*1601, while DR1, DR4, DR7 and DR9 alleles are poorly represented. In the presence of DR3, D R l l and DR13, allele DRB 1* 1404 is not distinguishable from
DRB 1 * 1401 by using the Workshop reagents. 28.57% of the Spanish Gypsies are DRB1*1404, while 22% of them are either DRB1*1401 or DRB 1* 1404. As expected from serology and RFLP, the results of DQAl and DQB1 PCR-SSOP typing (Table 8) indicate a high frequency of DQA1*0102, DQB1*0502 and especially of DQA1*0101 and DQB1*0503, and a poor representation of DQA1*0201, *0301 and *0401, and of DQB 1*0301, *0302, *0303 and *0402. The frequencies of DPAl and DPBl alleles are shown in Table 8. Special mention should be made of the absence of DPB1*0101 and the high frequency of DPBl * 1301. Characteristic haplotypes
The most characteristic haplotypes found in the populations under study (4) are shown in Table 9. Those that were not found in any of the other ethnic groups dealt with during the IlthIHW are labelled with an asterisk. Special mention should be made of haplotype A1 Cw6 B61 DR14 DQ1 among Gypsies which is the most frequently found in this population. When DRBl-DPBI haplotypes are considered (4), the Gypsy group has a characteristic association not found in any other group studied during the 1 IthIHW; that is, DRBl*1401/ 1404-DPB1*0401 (H.F.: 4.20/, L.D.: 0.4). However, the most frequent association among Gypsies is DRB 1* 1401 / 1404DPB1*0201 (H.F.: 10%, L.D.: 4.1), which is also found in Buyi (a Mongoloid group) (4). Genetic distance
Genetic distances among several Caucasoid groups (3,4) are illustrated in Table 10 (and Fig. 1). As can
Table 9. HLA-A, B. C. DR and DQ haplotypes Spanish non Gypsies
Spanish Gypsies
A
C
B
LD
DQ
DR
A
C
~
A1 A32 A1 A11 A3 A26 A2 A11 A11 A30 A24
Cw6 cw2 CBL cw7 CBL Cw6 CBL CBL Cw6 cw5 cw9
661 827 635 849 B18 837 851 639 661 E l8 855
DR14 DR16 OR14 DR11 DR15 DR11
*
DQ1 DP1 DQ1'
Dar
DR16
OR1* DQ7' DP1
DR14 OR1 4 OR3 DR14
DQ1 DQ2 DQ1
ow
13.0 5.7 4.7 4.0 3.1 2.7 2.7 2.7 2.3 2.0 2.0
HF: Haplotype frequency. LO Linkage desequilibriumX 1O-*.
192
12.7 5.7 4.5 4.0 3.1 2.7 2.7 2.7 2.3 2.0 2.0
B
DR
DQ
HF (%)
LO
844 88
DR7 0r3 DR3 DR15 DR7 OR13 DR1 DR11 0r1 0r7 DR4
DO2 DQ2 DQ2
4.7 2.3 1.8 1.5 1.4 1.4 1.3 1.2 1.1 1.1 1.1
4.7 2.3 1.8 1.5 1.4 I .4 1.3 1.2
~
A29 A1 A30 A2 A30 A2 All A2 A33 A2 A2
CBL CW7 Cw5 Cw7 Cw6 Cw5 Cw4 CBL CBL Cw5 Cw5
818
87 813 844 835 B51 814 844 844
DQ1
DQ~ DQl DQl
DQ7
om
DQ2 DQ7
1.1
1.1 1.1
HLA in Spanish Gypsies Table 10. HLA-A. E and DR gene frequncies were used to calculate the genetic distances among several Caucasoid groups, following Nei et al. (17)
EH IY TR IN SG HG S F GE AU RU YG CZ HU AR
BH
IY
TR
IN
000 182 235 236 432 307 203 154 190 208 240 205 185 181 245
000 224 135 462 461 237 205 237 252 276 259 237 242 226
000 162 412 229 237 171 200 211 136 168 175 167 218
000 171 351 229 148 205 190 157 194 177 199 227
1
SG
HG
S
F
GE
AU
RU
YG
CZ
HU
AR
000 382 434 291 331 307 275 296 341 313 416
000 339 278 276 293 234 258 269 221 401
000 044 074 084 125 106 077 085 190
000 032 040 070 054 039 039 133
000 027 076 053 030 039 106
000 066 059 039 054 145
000 058 048 077 118
000 062 046 128
000 056 117
000 134
000
be seen, of the different groups considered, Spanish Gypsies are closer to Indians than to Hungarian Gypsies. Spanish non-Gypsies are closer to the European groups considered here than to Gypsies. Fig. 1 represents a dendrogram showing the relationships among 15 Caucasoid groups, constructed by the Neighbor-Joining method, and based on the allele frequencies of HLA-A, B and DR. Discussion
Among all the ethnic groups studied so far, Spanish Gypsies present a unique HLA class I and class I1
-
HG SG
Figure 1. Dendrogram showing the relationship among 15 Caucasoid groups. S: Spanish, F: French, CZ: Czech, AU: Austrian, GE: German, HL': Hungarian, YG: Yugoslavian, RU: Rumanian, AR: Armenian, BH: Bhargavas, IY: Iyers, TR: Trival, HG: Hungarian Gypsies, S G Spanish Gypsies and IN: Indian.
antigen distribution. Of those groups dealt with during the 1 IthIHW, they present the highest frequency of A l , Cw6, DRB1*1401/1404, DRB3*0202 and DQB1*0503, with A11 and B61 also being well-represented among them. The overrepresentation of Al, A1 1 and B61 matches Spanish Gypsies with Hungarian Gypsies (3) and Indians (4). although the corresponding allele frequencies in the latter two groups are at a lower level than in the former. The proximity of Gypsies to some Caucasoid groups of the Indian subcontinent is also suggested by the low frequency of A2 in all of them, which is exceptional among Caucasian populations. Our study of the Cw6 reactivity patterns in Gypsies with the 1 1thIHW sera was pivotal to split this antigen into a short and a long Cw61.As will be shown in the Joint Report of the Antigen Society No 114 of the 1 IthIHW, the short Cw61pattern was evidenced in 25% of Spanish Gypsies ( 18 cells) and in a few cells of Armenian, British and Asian Indian origin (8, 6 and 2 cells, respectively). The Spanish Gypsy community shows a frequency distribution of HLA-DR, DQ allogenotypes different from that found among Spanish nonGypsies. Nevertheless, new DR-DQ associations were not found in any of them. The lower representation of DR15, DQ6, Dw2 in Gypsies, when compared to non-Gypsies, is of interest for two reasons. First, this haplotype has a lower frequency in Asian Indians (21) and this probably represents a genetic trait conipatible with the Indian origin of the Gypsy population. And, second, in a recent publication Tackics et al. (22) correlated the low incidence of multiple sclerosis 193
de Pablo et al.
(MS) in Hungarian Gypsies with the lack of DR15, DQ6, Dw2 in this ethnic group. As in the Hungarian Gypsy population, MS seems to be a rare disease among the Spanish Gypsies (23), corroborating the possible clinical implications of the low frequency of this haplotype. The reported association of DR4 and MS (reviewed in 24) in Arabian (25) and Sardinian (26) patients, and the lower frequency of this antigen among Spanish Gypsies in comparison with nonGypsies, may also correlate with the aforesaid low incidence of the disease in this ethnic group. The reported frequency of this antigen in Czech Gypsies (27) (frequency: 26.5!/0) is higher than in Spanish Gypsies (PCR-SSOP frequency: 7.8%). Certain DR2 haplotypes provide resistance to insulin-dependent diabetes mellitus (IDDM) (28), but haplotype DR16,'DQ5, Dw21 (AZH or MN2 type) is specifically found more frequently in patients than in controls. This is the only DR2 haplotype that has a DQBI B chain allele that encodes a non-charged amino acid at position 57. At present, we have no data about the incidence of IDDM in Spanish Gypsies, but the higher frequency of DR16 that we are describing with respect to Spanish non-Gypsies may indicate a lower resistance to the disease among them. Further studies are in progress to clarify this aspect. When the distributions of subtypes of DR2, DQ1 haplotypes of Spanish Gypsies and Spanish non-Gypsies are compared, it is evident that DR15, DQ6, Dw12 is elevated in the former, this difference not being significant. A report (21) comparing the distribution of DR2 haplotypes in Asian Indians and Canadian Caucasians revealed that haplotype DR15, DQ6, Dw12 is overrepresented in the first group, which may be coincidental due to the geographical origin of Gypsies. Moreover, a low frequency of DRB1*1501 in Czech Gypsies has been reported (27), which points to a common origin of both Gypsy communities. By contrast, the frequency of DR16, DQ6, Dw21 in Indians is 0% (21), while the frequency of this haplotype among the DR2 Spanish Gypsies is 62.8%. Recent data report a DRB 1*1601 phenotypic frequency of 2.9% in Czech Gypsies (27) which contrasts with the frequency of this allele in Spanish Gypsies (F.F.: 22.1%) and is a point of coincidence of this group with Indians. Another point of contrast between these two Gypsy groups is the high frequency of DRB1*1602 in Czech Gypsies while in Spanish Gypsies it is quite low. A high frequency of haplotype DR14, DQ5, Dw9 has been reported in non-Caucasian populations of Asian origin (29). It is peculiar that Gypsies, considered to be anthropologically distant 194
from these populations, also show an increment in this haplotype. The higher frequency of DR14(DR6) in Spanish Gypsies with respect to Caucasian populations is in agreement with the reported frequency of this allele in Czech Gypsies (27). The underrepresentation of DR6 among Hungarian Gypsies might be explained by the difficulty encountered in defining this antigen serologically when this group was studied, when neither RFLP or PCR-AS0 were available to perform class I1 typing at the DNA level. As DR7 is one of the most frequent antigens in the Spanish non-Gypsy population, its lower frequency in Gypsies is very striking. The distribution of DR7 haplotypes is different in the two population groups compared here, DR7, DQ9, Dwll being that most frequent in Gypsies. The underrepresentation of DR4, DR7 and DR9 determines a low frequency of DR53 and is compensated by the already mentioned increments. PCR-SSOP results confirm, as expected, the data previously obtained by RFLP. It was of interest to note the higher frequency of DRBl* 1404 (28.5%) in Spanish Gypsies which is quite similar to that found in Czech Gypsies (29.4%). However, given that, in the presence of DR3, DR11 and DRI 3 we can not dstinguish DRB1*1401 from DRB1*1404, it is probable that many individuals cqnsidered to be DRB1*1401/1404 would be 1404. New DRBlDQB1 or DQA1-DQB1 associations were not found. When DRB 1-DPB 1 associations are considered, the Gypsy group shows a characteristic one, that is DRB 1 * 1401-DPB1*0401. The haplotypes found in Spanish Gypsies (Table 9) confirm the uniqueness of this group, as the majority of them are found only in this ethnic group. However, A2-CBL-BS1-DR 16-DQI is also found in Rumanians, while A30-Cw5-B18-DR3DQ2 is found in Spanish non-Gypsies, Basques and Sardinians. The association A 1-B61 has been reported in Hungarian Gypsies; however, Gyody et al. report B61 to be linked to DR8 instead of to DR6. It must be kept in mind that a clear-cut serological definition of these two alleles is difficult unless DNA typing techniques are used. Analysis of genetic distance data shows that Spanish Gypsies are closer to Indians than to Hungarian Gypsies. These results confirm the linguistic data that point to the Indian origin of this ethnic group. As the two Gypsy groups share a common language, it seems probable that the differences between them could be due to genetic drift. When we analyze the frequency distribution of A l , A l l , B61, DR14 in all the ethnic groups studied during the 1IthIHW, we find that many Oriental groups that are not related to Gypsies
HLA in Spanish Gypsies from an anthropological point of view also show very high frequencies of these antigens. We cannot convincingly explain how it is that ethnic groups in close proximity when conventional genetic distance is considered do not share extended haplotypes (as is the case of Spanish Gypsies and Indians). At the same time, it is also difficult to explain how it is that groups which are not very close one another in terms of allelic frequencies share haplotypes, as is the case of Hungarian and Spanish Gypsies. In conclusion, the Spanish Gypsy population has a unique HLA class I and class I1 antigenic distribution, different from that found in all the other ethnic groups studied so far. They are close to Indians in terms of genetic distance, and show a considerable similarity to Hungarian and Czech Gypsies. Acknowledgments
We want to express our thanks for their collabor-' ation to Alfonso Ram6n from the Centro de Promoci6n de Salud (C.P.S.) of Montesa and to the staff of the following C.P.S. of the Comunidad de Madrid: Hortaleza, Latina, Tetuan, Vallecas Puente. Vallecas Villa, Vicalvaro, Villaverde, and the following Centros de Atencibn Primaria: Potes and Gomez Acebo. We thank Alicia Brea, Consuelo Pascau, Isabel Perez, M. Luisa Crespo and Rosario Martin for their skillful technical assistance. cDNA probes were kindly provided by Ann-Kristin Jonsson, Swedish University of Agricultural Sciences, Uppsala Biomedical Center, Uppsala, Sweden. This work was partially supported by FISS (INSALUD) grant 90/200.
References 1. Sampson J. On the origin and early migration of the Gypsies. J G.vpsy Lore Society 1923: 3rd ser., 2(4): 156-69.
2. Hancock I. The Pariah Syndrome. Ann Arbor: Karoma Publishers, Inc. 1987. 3. Gyody E, Tauszik T, Petranyi Gy, et al. The HLA antigen distribution in the gypsy population in Hungary. Tissue Antigens 1981: 18: 1-12. 4. Imanishi T, Akaza T, Kimura A, Tokunaga K, Gojohori T. Reference Tables, Chapter 15. In Tsuji K, ed. HLA 1991, vol. 1. Oxford: Oxford University Press (in press). 5. Miller SA, Dykes DD, Polesky HF. A simple salting-out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res 1988: 1215. 6. Carlsson B, Wallin J, Bohme J, Moller G. HLA-DR-DQ haplotypes defined by restriction fragment analysis. Correlation to serology. Hum Irnmunol 1987: 20: 95-1 13. 7. Gustaffson K, Wiman K, Emmoth E, et al. Mutations and selection in the generation of class I1 histocompatibility antigen polymorphism. EMBO J 1984: 3: 1655-61. 8. Scheming L, Larhammar D, Bill P, et at. Both A and
B chains of HLA-DC class I1 histocompattibility antigens display extensive polymorphism in their aminotenninal domains. EMBO J 1984 3: 447-525. 9. Larhammar D, Scheming L, Gustafsson K, et al. Complete amino acid sequence of an HLA-DR antigen-like B chain as predicted from the nucleotide sequence: Similarities with immunoglobulins and HLA-A, -B and -C antigens. Proc Natl Acad Sci USA 1982: 79: 3687-91. 10. Martell M, Marcadet A, Cohen D. HLA Class I1 Typing by the Workshop Defined RFLP Specificities in Unrelated Individuals. In: Dupont B, ed. Immunobiology of HLA, vol. I. Berlin-Heidelberg-New York Springer Verlag, 1989. 947-52. 11. Bidwell J. DNA-RFLP analysis and genotyping of HLADR and DQ antigens. Immunology Today 1988: 9: 18-23. 12. Kimura A, Sasazuki T. Workshop Reference Protocol of the HLA-DNA Typing Technique. In: Tsuji K, ed. HLA 1991, vol. 1. Oxford: Oxford University Press (in press). 13. Kimura A, Dong R-P, Harada H, Sasazuki T. DNA Typing of HLA class I1 genes in B-lymphoblastoid. cell lines homozygous for HLA. Tissue Antigens 1992: 401: 5 . 14. Wood W, Gitschier J, Larsky LA, et al. Base compositionindependent hybridization in tetramethylainmonium chloride: A method for oligonucleotide screening of highly complex gene libraries. Proc Natl Acad Sci USA 1985: 82: 1585-8. 15. Pickbourne P, Piazza A, Bodmer WF. Poplulation Analysis. In: Bodmer WF, Batchelor JR, Bodmer T.J, Festenstein H, Morns PJ, eds. Histocompatibility Testing 1977. Copenhagen: Munksgaard, 1978: 259-78. 16. Baur MP, Danilovs JA. Population analysis of HLA-A. B, C, DR, and other genetic markers. In: Terasaki PI, ed. Histocompatibility Testing 1980. Los Angeles: UCLA Tissue Typing Laboratory, 1980: 955-8. 17. Nei M. Estimation of average heterozygosity and genetic distance from a small number of individua'ls. Generics 1978: 89: 583-90. 18. Imanishi 1.Akaza T. Kimura A, Tokunaya K, Gojobori T. Estimation of allele and haplotype frequencies for HLA and complement loci. In: Tsuji K, ed. H.LA 1991, vol. I , chapter 1. Oxford: Oxford University Press (in press). 19. 1 1th International Histocompatibility Workshop. Data Analysis Book 11. Central Analysis Committee, 1 1 th International Histocompability Workshop, Yokohama. Japan 1991. 20. Tait BD, Park MS, Barbetti A, Langley J, Terasaki PI, Aizawa M. Antigen Society #28. Report (.DQw3).In: Dupont B, ed. Immunobiology of HLA. Vol. 1, p. 269. New York, Berlin, Heidelberg, London, Paris, Tokyo, Hong Kong: Springer Verlag, 1987. 21. Singal DP, D'Souza M, Sood SK. Polymclrphism of HLADR2, DQwl haplotypes in Asian Indians. Tissue Antigens 1990: 36: 164-70. 22. Takacs K, Kalmhn B, Gyodi E, et al. Association between the lack of DQw6 and the low incidence of multiple sclerosis in Hungarian Gypsies. Zmmunogenetics 1990: 31: 383-5. 23. Trueba JL, Garcia Merino JA. Personal communication. 24. Olerup 0, Hillert J. HLA class II-associateid genetic susceptibility in multiple sclerosis: A critical evaluation. Tissue Anrigens 1991: 38: 1-15. 25. Kurdi A, Ayesh I, Abdallat A, et al. Different B lymphocyte alloantigens associated with multiple sclerosis in Arabs and North Europeans. Lancer 1977: i: 1123-5. 26. Marrosu MG, Muntoni F, Murru MR, et al. Sardian multiple sclerosis is associated with HLA-DR4. Neurology 1988: 38: 1749-53. 27. Cerna M, Fernandez-Viiia M, Ivaskova E., Stastny P. Comparison of HLA class I1 alleles in Gypsy and Czech populations by DNA typing with oligonucleotide probes. Tissue Antigens 1992: 39: 1 11-6.
195
de Pablo et a]. 28. Todd JA. Genetic control of autoimmunity in type 1 diabetes. Immunology Today 1990: 11: 122-9. 29. Kohonen-Codsh MRJ, Dunckley H, Serjeantson Sw. HLA-DR and -DQ DNA genotyping in seven populations of Asia-Oceania and Australia. Tissue Antigens 1988: 32: 3240.
196
Address: Rosario de Pablo Servicio de Inmunologa Hospital Puerta de Hierro ClSan Martin de Porres 4 28035 Madrid Spain
R. de Pablo, C. Vilches, M. E. Moreno, M. C. Rementeria, R. Solis, ' M. Kreisler. Distribution of HLA antigens in Spanish Gypsies: A comparative study. 1 Tissue Antigens 1992: 40: 187-196.
Abstract: We have studied the HLA-class I and class I1 antigen distribution in a sample of 75 Spanish Gypsies and 74 Spanish non-Gypsies by serology, restriction fragment length polymorphism, and protein chain reaction and hybridization with allele-specific oligonucleotide probes. When both population samples are compared, we find that Gypsies have a statistically significantly higher frequency of Al, A l l , B61, Cw6, DQ5 and haplotypes DRi6 DQ5 Dw21 and DR14 DQ5 Dw9 DR52b. Frequency of A3, A29, B44, DR4, DQ2, DQ8 and haplotypes DRI DQ5 and DR7 DQ2 DB17 DK53 are significantly lower in this ethnic group. The analysis of the serological data in the two populations demonstrates that Cw6 can be split into long Cw6 (Cw6.1) and short Cw6 (Cw6.2). Haplotype Al-Cw6-B61-DR14-DQ5is the most characteristic in Gypsies, with a frequency of 13%. Estimation of the genetic distances shows that Spanish Gypsies are closer to Indian Caucasoid populations than to the Spanish non-Gypsy population. HLA data support the proposed historical origin
Introduction
Gypsies are an ethnic group that is historically considered to be of northern Indian origin. They moved out of that area probably between AD 800 and AD 950 and, migrating westwards, they arrived in Europe around AD 1100. Linguistic evidence ( 1 ) suggests that the ancestors of all Gypsy populations are the Domba or Dombari (2), who constituted a single race speaking the same language, which subsequently diverged into two linguistic branches: the Nawar, Kurbat, Karaci and Helebi now found throughout Egypt and the Middle East on the one hand, and the Bosa in Armenia and eastern Turkey, and the Rom or Roma in Europe, on the other. However, on the basis of more recent scholarship, there is some reason to believe that the three populations usually thought to comprise the descendants of the Domba may in fact have left India at different times and under different circumstances (2). The three branches exhibit considerable lexical adoption from Persia, for example, but there are no items shared by all three branches. The same is true for the Armenian items in Central and Western Gypsies. Conservative Romani dialects remain two-thirds or more Indian in their basic lexicon and grammar,
Servicio de Inmunologia, Hospital Puerta de Hierro, Madrid, Spain
-
Key words: HLA-A,B,C HLA-class II allogenotypes - Spanish Gypsies - SpMi:;h non-Gypsies HLA Class II RFLP - HLA Class II PCR-SSOP
I
-
Received 14 April, revised, accepted for publication 1 JUIY 1992
retaining features which have become lost in their neo-Indic cognates. The nomadic lives and cultural peculiarities of Gypsies made them an ethnic group that seldom integrated with other societies, endogamy being almost complete in the gypsy community. Since they arrived in Spain, they have continued to be genetically isolated up to the present day. To characterize human populations and their genetic relationships, anthropologists and geneticists have taken advantage of the high polymorphism of the major histocompatibility complex (MHC). HLA products are essential to the immune response because they represent the molecular context for antigen presentation to CD8+ and CD4+ T cells. Moreover, the requirements for HLA matching in transplantation and the association of several diseases to certain HLA class I and I1 antigens underline its medical importance. All these reasons induced us to analyze HLA distribution in the second most populous ethnic group of Spaniards. We have studied the HLA-A, B, C, DP, DQ and D R antigenic distribution in Spanish Gypsies and non-Gypsies. We have compared Spanish Gypsies with Hungarian Gypsies (3), with four different populations from India, with eight European populations and with the Armenians (4). 187
de Pablo et a]. Material and methods Population studied
We have studied a sample of unrelated Spanish Gypsies (75 individuals by serology, 76 by RFLP and 77 PCR) and Spanish non-Gypsies (74 individuals by serology, 96 by RFLP), all of them living in Madrid but native to different regions of Spain. One Gypsy family was studied for segregation analysis. Serology
HLA-A,B,C and HLA-DR, HLA-DQ typing were performed on purified T- and B-lymphocyte suspensions, respectively, using the standard NIH microlymphocytotoxicity technique. For this purpose, we used the set of sera of the anthropological component of the 1 1th International Histocompatibility Workshop (1 1th IHW) and our local sera. DNA samples
Genomic DNA was obtained from peripheral blood leukocytes using the salting-out method (5). Southern blot analysis
Aliquots of 15 pg DNA were digested with TuqI and MspI according to the manufacturer's recommendations. Agarose gel electrophoresis, vacuum blotting of size-separated DNA fragments onto nylon membranes, labeling of purified probe inserts with random primers, prehybridization, hybridization, high-stringency washes and autoradiography were performed according to standard techniques with minor modifications (as described in (6)).
PCR amplification of genomic DNA
The primers used in this study were provided by the 11th International Histocompatibility Workshop. The amplification was performed following the protocols of this Workshop (12). The primers were those provided by the organizers of the 11thIHW (13). Dot blot and SSOP hybridization
Dot blotting .and hybridization were performed as described in the 11 thIHW protocols (12), using the sequence-specific oligonucleotide probes (SSOP) provided by the organizers (1 3). The washing procedure after hybridization was that described by Wood et al. (14). Statistical analysis
HLA-A,B,C,DR and DQ gene frequencies were derived from serology results (15, 16). Genetic distances were calculated as previously described (17). RFLP genotypic relative frequencies were calculated with respect to the total number of haplotypes studied in each sample. Phenotypic frequencies of HLA-DRBI, DRB3, DQA1, DQBI, DPAl and DPBl were calculated as a percentage. Haplotype comparisons were made using the reference tables of the 1 1th International Histocompatibility Workshop (4,18). Data were analyzed by the chi-square test with the Yates' correction. Probability (p) values were always corrected with a factor dependent on the number of alleles or DR-DQ haplotypes studied (p,= 1 -(1 -p)", with n being the number of antigens in that allelic series). DRBI-DQAl -DQB 1 associations were deduced from RFLP and PCR-SSOP results Results
RFLP probes
The DNA probes used were: DRB, a 785 bp Sac11 Hind11 fragment of the cDNA clone pII-B-4 complementary to the DRBI gene (7); DQA, a 725 bp ApaI fragment of the cDNA clone p-II-a-5 complementary to the DQAZ gene (8); and DQB, a 625 bp h a 1 fragment of the cDNA clone pII-B-1 complementary to the DQBl gene (9). Assignment of RFLP haplotypes
Tag1 DRB, DQA and DQB, and MspI DQA restriction fragment patterns were analyzed as previously described (10, 11). 188
The analysis of the frequencies of the samples compared here demonstrate that both populations are under Hardy-Weinberg equilibrium. Serological' results
The gene frequency of the HLA-A allelic series of the two population samples studied are shown in Table 1. The elevated frequency of A1 and A l l and the lower frequency of A3 and A29 are the main features of Gypsies when compared with nonGypsies. When the gene frequencies of the alleles of the HLA-B series of the two population samples are compared (Table 2), the main differences are found
HLA in Spanish Gypsies Table 1. Gene frequencies of the HLA-A allelic series of a sample of Spanish Gypsies and non Gypsies
HLA-A
Spanish Gypsies N375 ~~
A1 A2 A3 A1 1 A23 A24 A25 A26 A28 A29 A30 A31 A32 A33 A34 ABL
Spanish nowGypsies N=74
Xu2
Pc
18.755 7.718 10.167 19.042 0.549 0.587 0.706 0.3W 0.000 12.297 0.549
0.005
~
0.283 0.119
0.058 0.258 0.020
0.063 0.013 0.047 0.01 3 0.01 3 0.020 0.000 0.073 0.007 0.000 0.01 3
0.077 0.264 0.1 69 0.070 0.034 0.077 0.027 0.027 0.01 3 0.130 0.034 0.041 0.01 3 0.020 0.006 0.000
6.055 6.056 1.015 0.830
-
N.S.
0.05 0.005 N.S. N.S. N.S. N.S. N.S.
0.05 N.S. N.S. N.S. N.S. N.S.
-
in B44 and B61. While B44 is the allele with the highest frequency in the Spanish non-Gypsy population, B61 is the most characteristic among Gypsies. The high frequency of B27 among Gypsies,
Table 2. Gene frequencies of the HLA-8 allelic series of a sample of Spanish Gypsies and non Gypsies
HLA-8 87
88 814 818 827 835 837 838 839 841 B44 845 849 850 851 852 855 656 857 858 860 861 862 863 870 BBL
Spanish Gypsies N=75
Spanish nowGypsies N=74
xvz
0.047 0.01 3 0.01 3 0.087
0.122 0.048 0.055 0.041 0.01 3 0.1 07
4.923 2.880 3.706 2.766 6.903 0.299
0.000
6.055
0.048 0.006 0.01 3 0.21 1
6.898 3.455 2.002 13.792
0.006
0.000
0.048 0.020 0.092 0.01 3 0.034 0.006 0.01 3 0.006 0.027 0.020 0.020 0.020 0.006 0.005
0.1 00 3.1 33 0.066 1.975 0.549 0.830 1.975 0.954 4.1 83 16.915 0.217 3.133 0.000
0.080 0.1 27 0.040 0.000 0.040 0.000
0.053 0.007 0.040 0.000 0.093 0.040 0.020 0.000 0.047 0.020 0.000 0.21 3 0.01 3 0.000 0.007 0.000
-
while not statistically significant with respect to non-Gypsies, is, however, of interest given the described association of this antigen with ankylosing spondilitis. Among the alleles of the HLA-C seriles, Cw6 is highly represented in the Gypsy population (Table 3). When we look at the reaction patterns of the Cw6-specific sera of the 1 1thIHW (1 9) - SVEll4 [Workshop Number 4321, DDC103 FV.N. 4331, ROO103 [W.N. 4341 and AST109 [W.N. 4351 -with the cells of the population samples under study, two patterns emerge that we call long-Cw6 and short-Cw6 (Table 4). Long Cw6-bearing cells show a positive reaction with SVEll4, DDC103, ROO103 and AST109, while short Cw6 cells react only with ROO103 and AST109. The: so-called short pattern can only be defined in the absence of Cw4 because AST109 recognized all Cw6 cells and 36% of Cw4. All except one of the clells (KRE 1047) showing the short-Cw6 reaction paittern were of Gypsy origin. Our finding of two clear-cut reaction patterns in such a great proportion of cells points to the existence of two Cw6 splits, Cw6.1 (long) and Cw6.2 (short). The segregation of the short pattern in a Gypsy family (Table 4) confirms the split. When the serological data on the IDR allelic series (Table 5) are compared in the two populations, the main differences are the higher frequency of DR16 and DR14, and the lower frequency of DR4 and DR7 in Gypsies. In the DQ series (Table 5), we find DQ1 frequency increased in Gypsies, while DQ;! is significantly lower among them.
PC
N.S. N.S.
N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. 0.05 N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. 0.01 N.S.
N.S. N.S.
-
RFLP results
The frequencies of the HLA-DR-DQ a.llogenotypes determined by RFLP are shown in Table 6. The most striking differences are the fo bowing: Table 3. Gene frequencies of the HLA-C allelic series of a sample of Spanish Gypsies and non Gypsies
HLA-C Cwl cw2 cw4 cw5 Cw6 cw7 cw9 CWlO
CBL
Spanish Gypsies N-75
Spanish non-Gypsies N=74
0.000 0.1 16 0.098 0.041 0.293 0.1 27 0.027 0.000 0.298
0.020
2.002
N.S.
0.048 0.137 0.1 22 0.055 0.237
3.887
N.S. N.S. N.S. 0.0005 N.S. N.S. N.S.
0.048
0.041 0.292
xu2 -o.tjao 6.055 25.626 4.51 9 1.723
6.055
-
Pc
-
Cw8 was tested but it was absent in both population samples.
189
de Pablo et al. Table 4. Cw6 reaction patterns
Antisera
cell
A
6
BW
cw
KRE2Oll KRE2023 KRE2024 KRE2058 KRE1050 KRE2063 KRE2038 KRE2074 KRElO57 KRE1067 KRE1069 KRE1072 KRE2006 KRE2008 KRE2037 KRE2066 KRE2078 KRE1012 KRE2003 KRE2052 KRE2059 KRE2067 KRE2056 KRE2012
A11A23 A 1 A 1 A l,A 2 A31,A33 A 1,A23 A11,A 1 A26,A32 A 2,A26 AJ,A 2 A 2.A 2 A 1,A24 A26A25 AllA 1 A24,A26 A11,A32 A 1 A 2,A23 A11,A26 A 1,A 3 A 1,A32 A 1,A 2 A 1,A29 A1
844,661 844,657 835,661 635,857 8 7,635 ' 844,661 835,861 627,837 650,814 651,657 849,650 6 8.657 862,637 839.637 637,655 645,627 637,661 844,650 837.639 635,657 627,657 6 7,657 644,657 661
6~416 Bw4 Bw6 Bw416 Bw6 Bw416 Bw6 Bw4 Bw6 Bw4 Bw416 6~416 Bw416 6~416 6~416 Bw416 Bw416 Bw416 6~416 6~416 Bw4 6~416 Bw4 Bw6
Cw6,Cw4 Cw6,Cw4 Bw6,Cw4 Cw6,Cw4 Cw6,Cw4 Cw6,Cw4 Cw6,Cw4 Cw6,Cw2 Cw6 Cw6,Cw2 Cw6 Cw6 Cw6,Cw9 Cw6 Cw6,Cw9 Cw6,Cw2 Cw6 Cw6,Cw5 Cw6 Cw6 Cw6,Cw2 Cw6,Cw7 Cw6 Cw6
KR E2042 KRE2018 KRE2004 KRE2014 KRE2Ol9 KRE2070 KRE2073 KRE2131 KRElO47 KRE2007 KRE2Ol6 KRE2033 KRE2041 KRE2048 KRE2053 KRE2071 KRE2064
A11,A 1 A 3,All A1 1,A32 A1 A1 1 A11 A 1,All A l,A 2 A 1,A24 A 1,All A 1.A68 A 1,All A1 A l.A 2 A 1,All A 1,All A11,A32
6 7,661 6 7,661
627,861 661 661 661 661 618,661 651,662 639,661 651,661 649,661 661 651.661 658,861 662,861 651,627
Bw6 Bw6 8~416 Bw6 Bw6 Bw6 Bw6 Bw6 Bw416 Bw6 Bw416 6~416 Bw6 Bw416 Bw416 Bw6 Bw4
Cw6,Cw2 Cw6 Cw6,Cw2 Cw6 Cw6 Cw6 Cw6 Cw6,Cw5 Cw6 Cw6 Cw6 Cw6 Cw6 Cw6 Cw6,Cw7 Cw6 Cw6,Cw2
KRE2062' KRE2073t KRE20755
A26,A32 A 1,All A11,A32
827,652 661 627,661
Bw4 Bw6 6~416
cw2 Cw6.2 Cw2,Cw6.2
c
c
w 6
w 4
+
+
+
+ +
+
+ + + +
+ +
+ + + + + + +
+ + + + + + + +
+ + + + + +
D D C 1 0 3
S V E 1 1 4
++ ++ ++
++ ++ ++ ++
++ ++ ++ ++ ++ ++ ++ ++ ++ ++
+
+
+ +
+ +
+
+
+ +
++ ++ ++ ++ ++ ++ ++ -k ++
++ ++ +
+ + +
+ -
-
-
-
Y E T 2 0 3
+ +
++ ++
++ ++ ++ ++
+
K W C 2 2 8
++ ++ ++ ++ ++ ++ ++ ++ ++
++ +
+ + + + + + + +
R O O 1 0 3
++
+ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ + ++ ++ + ++ ++ + ++ ++ + ++ ++ + ++ ++ - + ++
+
+ -
A S T 1 0 9
-
+
+
++ ++ +
++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ -
-
++
-
-
++
-
-
++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++
++
-
-
-
-
-
-
-
-
++ - + - -
+ + - + + - -
~
* Father. t Mother.
5
Daughter.
++ Strong positive reaction. + Positive reaction. - Negative reaction.
- D Rl is significantly lower in Gypsies than in non-Gypsies. - When the frequency distributions of the DR2 splits in the population samples are compared (Table 6), it is clear that haplotype DR15 DQ6 Dw2 has a lower frequency in Spanish Gypsies while, in contrast, haplotype DR15 DQ6 Dw12 has '
a higher frequency in this group, but neither of these differences is statistically significant. DR16 DQ5 Dw21 is significantly overrepresented in Gypsies. This contrasts with the DR2 haplotype distribution in Spanish non-Gypsies where DR15 is more frequent than DR16. Comparing the two ethnic groups, the overall frequency of DR2 is
HLA in Spanish Gypsies Table 5. Gene frequencies of the HLA-DR and DQ allelic series of a sample of Spanish Gypsies and non Gypsies ~
DR1 OR15 DR16 DR3 OR4 DR11 DR12 DR13 OR14 DR7 OR8 OR9 DRBL DQ1 DQ2 DQ3 DQ4 DQBL
Spanish Gypsies N-75
Spanish nonGypsies N=-74
0.034 0.083 0.1 35 0.083 0.040 0.113 0.013 0.062 0.288 0.076 0.006
0.1 22 0.092 0.01 3 0.099 0.203 0.092 0.007 0.070 0.01 3 0.21 1 0.041 0.020 0.01 7 0.332 0.283 0.302 0.055
0.000 0.067 0.600 0.1 05 0.1 91 0.01 3 0.091
XVZ
Pc
7.309 0.066 12.799 0.221 15.552 0.335 0.303 0.077 27.976 9.87 3.584 3.1 33
N.S. N.S. 0.01 N.S. 0.005 N.S. N.S. N.S. 0.0005 0.05 N.S. N.S.
13.661 13.013 4.1 92 3.706
0.005 0.005 N.S. N.S.
-
-
0.028
-
-
similar because the two haplotype frequencies compensate for one another. - DR4 frequency is significantly diminished in Spanish Gypsies with respect to non-Gypsies. This Table 6. Relative haplotypic frequencies in % of DR-DQ-Dw haplotypes in Spanish Gypsies and non Gypsies Samples
DRw,DQw,Dw Taql-DRE allogenotypes DR1IBr,DQ5,Dwl ,Dw20 DR15,DQ6,Dw2 DR15,DQ6,Dwl2 DR15,DQ5 DR16,DQ5,Dw21 DR17,DQ2,Dw3,DR52(Dw24) DR17,W2,Dw3,DR52(Dw25) DR4,DQ7,Dw4/13,DR53
3.2 0.7 7.2 0.7 14.5 2 7.9 2 DR4,DQ8',Dw4/10114115,DR53 2.6 DR11,W6,DwFS,DR52(Dw25) 0 DR11.DQ7,Dw5/DwJVM,DR52(Dw25) 11.2 OR12,DQ7,DwDB6,DR52(0w25) 0.7 DR13,DQ6,Dwl8,DR52(Dw25) 0.7 DR13a,DQ6,Dwl8,DR52(Dw25) 1.3 D R ~ ~ ~ . D Q ~ , D w I ~ , D R ~ ~ ( D w ~ ~ )5.2 DR13b,DQ7,DwHAG,DR52(0w24) 0 DRl4,WS,Dw9,DR52(Dw25) 30.9 DR7,DQ2,Dwl7,DR53 2.6 DR7,DQ2,DwDBI,DR53 0 DR7,DQS,Dwll 5.2 DR8,DQ8.1/8.2 1.3 DR9,DQ9,Dw23,DR53 0 DR1O,DQ5 0 *: See reference 20.
Studied
GYPSY Non GYPSY % (N=76) % (N=96)
14 6.3 1.6 2.1 1.6 5.2 7.8 4.2 12 0.5 8.3 1 4.7 1 1 0.5 1.6 17.7 1.6 1 3.6 1.6 0
is mainly due to the low frequency of haplotype DR4 DQ4/8 (Table 6). When DR4 haplotype frequencies are compared, no significant differences are found. - DR7 is poorly represented in Gypsies, especially due to the low frequency of allogenotype DR7 DQ2 Dw17 DR53 with respect to the Spanish non-Gypsy population. Haplotype DR7, DQ9, Dwll is more frequent among Gypsies than in Spanish non-Gypsies, but this does not compensate for the already mentioned DR7-DQ2 underrepresesntation. - Considering the aforesaid and given that DR9 is absent in the sample of Gypsies studied, DR53 frequency is significantly low in this population. The frequency of serologically-detected DR53 is even lower (data not shown) because haplotype DR7, DQ9, Dwll harbors a pseudogene in the DRB4 locus. - DR6 frequency is considerabty increased among Gypsies (p, < 0.000001), especially due to the overrepresentation of DR14 DQ5 Dw9 (30.9% in Gypsies vs. 1.6% in non-Gypsies), the frequency of DR13 being similar in the two population samples. The above-mentioned haplotypic differences imply a significantly high frequency of DR52b (Dw25), Dw9 and Dw21 among Gypsies, while the frequencies of DR53 and Dw17 are significantly low in this group. As can be seen in Table 7, DQ2, DQ5 and DQ8 show significant differences when the two ethnic groups are compared. Considering that DQ4 is seldom associated with DR.4 in individuals of Caucasoid origin. DQ4/8 can be taken as DQS (* in Table 6) (20).
pc
0.05 N.S. N.S. N.S. 0.01 N.S. N.S. N.S. N.S. N.S. N.S.
N.S. N.S. N.S. N.S. N.S. 10-5 0.05 N.S. N.S. N.S. N.S.
-
PCR SSOP data
The distribution of HLA-DRBI and DRB3 alleles among Gypsies is shown in Table 8. These data confirm the existence of a characteri:stically high frequency of DRBl*1401/ 1404 and an increase in Table 7. Relative haplotypic frequencies in % of DQ antigens in Spanish Gypsies and non Gypsies, defined by RFLP
DQ5 DQ6 DQ2 DQ7 DQ8' DQ9 DQ4
GYPSY % (N=76)
Non-Gypsy % (N-96)
49.3 15.1 12.5 13.9 2.6 5.2 1.3
19.3 15 32.3 14
12 3.0 3.6
XZ
Pc
32.250 0.021 17.337 0.01 6 10.766 0.521 1.009
< 0.0001 N.S.
< 0.005 N.S.
< 0.05 N.S. N.S.
In Caucasians, DR4 was always found associated to Dlaw8 (20).
191
de Pablo et al. Table 8. Frequency in % of DRB1, DRB3, DQB1, DQAl, DPB1 and DPA1 in Spanish Gypsies determined by PCR-SOP
ORB1 (N=77)
FF
DQB1 (N=77)
0101 0102 1501 1502 1601 1602 0301 0401 0402 0404 1101 1102 1103 1104 1201 1301 1302 1303 1401+140w) 07 0801 0901
5.20 1.30 2.60 15.6 22.1 1.30 14.3 1.30 1.30 5.20 11.7 1.30 2.60 5.20 1.30
0501 0502 0503 0504 0601 0602
DRB3 (N=47)
0101 0201 0202 0301
FF
6.60 25.0 51.3 0000 15.8 1.30 3.90 10.5 0000 22.3 26.3 5.20 10.5 0000 2.60
0603 0604
0605 0201 0301 0302 0303 0401 0402
6.50 7.80 DOAl (N=77) 1.30 0101 50.6 0102 10.3 0103 2.60 0201 2.60 0301 0401 FF 0501 3.90 0601 0000 93.5 7.80
FF 57.9 38.1 18.4 15.8 9.20 2.60 39.5 0000
DPBl (N-73) 0101 0201 0202 0301 0401 0402 0501 0601 0801 0901 1001 1101 1301 1401 1501 1601 1701 1801 1901 DPAl (N=73)
01 02
FF 0000
35.6 4.1 0 9.60 53.4 24.6 1.30
5.50 0000 0000 0000 0000 10.9 0000
0000 0000 0000 0000 2.70
FF 89.3
38.6
FF: Phenotypic frequency. * In presence of DR3, DRwll and DRwl3, DRB1.1404 can not be distinguishedfrom DRB1'1401 by using the Ubrkshop oligonucleotides. DRBl'1404 phenotypic frequency is 28.57%. DRBl'14OlI 1404 phenotypic frequency is 22%.
DRB1*1601, while DR1, DR4, DR7 and DR9 alleles are poorly represented. In the presence of DR3, D R l l and DR13, allele DRB 1* 1404 is not distinguishable from
DRB 1 * 1401 by using the Workshop reagents. 28.57% of the Spanish Gypsies are DRB1*1404, while 22% of them are either DRB1*1401 or DRB 1* 1404. As expected from serology and RFLP, the results of DQAl and DQB1 PCR-SSOP typing (Table 8) indicate a high frequency of DQA1*0102, DQB1*0502 and especially of DQA1*0101 and DQB1*0503, and a poor representation of DQA1*0201, *0301 and *0401, and of DQB 1*0301, *0302, *0303 and *0402. The frequencies of DPAl and DPBl alleles are shown in Table 8. Special mention should be made of the absence of DPB1*0101 and the high frequency of DPBl * 1301. Characteristic haplotypes
The most characteristic haplotypes found in the populations under study (4) are shown in Table 9. Those that were not found in any of the other ethnic groups dealt with during the IlthIHW are labelled with an asterisk. Special mention should be made of haplotype A1 Cw6 B61 DR14 DQ1 among Gypsies which is the most frequently found in this population. When DRBl-DPBI haplotypes are considered (4), the Gypsy group has a characteristic association not found in any other group studied during the 1 IthIHW; that is, DRBl*1401/ 1404-DPB1*0401 (H.F.: 4.20/, L.D.: 0.4). However, the most frequent association among Gypsies is DRB 1* 1401 / 1404DPB1*0201 (H.F.: 10%, L.D.: 4.1), which is also found in Buyi (a Mongoloid group) (4). Genetic distance
Genetic distances among several Caucasoid groups (3,4) are illustrated in Table 10 (and Fig. 1). As can
Table 9. HLA-A, B. C. DR and DQ haplotypes Spanish non Gypsies
Spanish Gypsies
A
C
B
LD
DQ
DR
A
C
~
A1 A32 A1 A11 A3 A26 A2 A11 A11 A30 A24
Cw6 cw2 CBL cw7 CBL Cw6 CBL CBL Cw6 cw5 cw9
661 827 635 849 B18 837 851 639 661 E l8 855
DR14 DR16 OR14 DR11 DR15 DR11
*
DQ1 DP1 DQ1'
Dar
DR16
OR1* DQ7' DP1
DR14 OR1 4 OR3 DR14
DQ1 DQ2 DQ1
ow
13.0 5.7 4.7 4.0 3.1 2.7 2.7 2.7 2.3 2.0 2.0
HF: Haplotype frequency. LO Linkage desequilibriumX 1O-*.
192
12.7 5.7 4.5 4.0 3.1 2.7 2.7 2.7 2.3 2.0 2.0
B
DR
DQ
HF (%)
LO
844 88
DR7 0r3 DR3 DR15 DR7 OR13 DR1 DR11 0r1 0r7 DR4
DO2 DQ2 DQ2
4.7 2.3 1.8 1.5 1.4 1.4 1.3 1.2 1.1 1.1 1.1
4.7 2.3 1.8 1.5 1.4 I .4 1.3 1.2
~
A29 A1 A30 A2 A30 A2 All A2 A33 A2 A2
CBL CW7 Cw5 Cw7 Cw6 Cw5 Cw4 CBL CBL Cw5 Cw5
818
87 813 844 835 B51 814 844 844
DQ1
DQ~ DQl DQl
DQ7
om
DQ2 DQ7
1.1
1.1 1.1
HLA in Spanish Gypsies Table 10. HLA-A. E and DR gene frequncies were used to calculate the genetic distances among several Caucasoid groups, following Nei et al. (17)
EH IY TR IN SG HG S F GE AU RU YG CZ HU AR
BH
IY
TR
IN
000 182 235 236 432 307 203 154 190 208 240 205 185 181 245
000 224 135 462 461 237 205 237 252 276 259 237 242 226
000 162 412 229 237 171 200 211 136 168 175 167 218
000 171 351 229 148 205 190 157 194 177 199 227
1
SG
HG
S
F
GE
AU
RU
YG
CZ
HU
AR
000 382 434 291 331 307 275 296 341 313 416
000 339 278 276 293 234 258 269 221 401
000 044 074 084 125 106 077 085 190
000 032 040 070 054 039 039 133
000 027 076 053 030 039 106
000 066 059 039 054 145
000 058 048 077 118
000 062 046 128
000 056 117
000 134
000
be seen, of the different groups considered, Spanish Gypsies are closer to Indians than to Hungarian Gypsies. Spanish non-Gypsies are closer to the European groups considered here than to Gypsies. Fig. 1 represents a dendrogram showing the relationships among 15 Caucasoid groups, constructed by the Neighbor-Joining method, and based on the allele frequencies of HLA-A, B and DR. Discussion
Among all the ethnic groups studied so far, Spanish Gypsies present a unique HLA class I and class I1
-
HG SG
Figure 1. Dendrogram showing the relationship among 15 Caucasoid groups. S: Spanish, F: French, CZ: Czech, AU: Austrian, GE: German, HL': Hungarian, YG: Yugoslavian, RU: Rumanian, AR: Armenian, BH: Bhargavas, IY: Iyers, TR: Trival, HG: Hungarian Gypsies, S G Spanish Gypsies and IN: Indian.
antigen distribution. Of those groups dealt with during the 1 IthIHW, they present the highest frequency of A l , Cw6, DRB1*1401/1404, DRB3*0202 and DQB1*0503, with A11 and B61 also being well-represented among them. The overrepresentation of Al, A1 1 and B61 matches Spanish Gypsies with Hungarian Gypsies (3) and Indians (4). although the corresponding allele frequencies in the latter two groups are at a lower level than in the former. The proximity of Gypsies to some Caucasoid groups of the Indian subcontinent is also suggested by the low frequency of A2 in all of them, which is exceptional among Caucasian populations. Our study of the Cw6 reactivity patterns in Gypsies with the 1 1thIHW sera was pivotal to split this antigen into a short and a long Cw61.As will be shown in the Joint Report of the Antigen Society No 114 of the 1 IthIHW, the short Cw61pattern was evidenced in 25% of Spanish Gypsies ( 18 cells) and in a few cells of Armenian, British and Asian Indian origin (8, 6 and 2 cells, respectively). The Spanish Gypsy community shows a frequency distribution of HLA-DR, DQ allogenotypes different from that found among Spanish nonGypsies. Nevertheless, new DR-DQ associations were not found in any of them. The lower representation of DR15, DQ6, Dw2 in Gypsies, when compared to non-Gypsies, is of interest for two reasons. First, this haplotype has a lower frequency in Asian Indians (21) and this probably represents a genetic trait conipatible with the Indian origin of the Gypsy population. And, second, in a recent publication Tackics et al. (22) correlated the low incidence of multiple sclerosis 193
de Pablo et al.
(MS) in Hungarian Gypsies with the lack of DR15, DQ6, Dw2 in this ethnic group. As in the Hungarian Gypsy population, MS seems to be a rare disease among the Spanish Gypsies (23), corroborating the possible clinical implications of the low frequency of this haplotype. The reported association of DR4 and MS (reviewed in 24) in Arabian (25) and Sardinian (26) patients, and the lower frequency of this antigen among Spanish Gypsies in comparison with nonGypsies, may also correlate with the aforesaid low incidence of the disease in this ethnic group. The reported frequency of this antigen in Czech Gypsies (27) (frequency: 26.5!/0) is higher than in Spanish Gypsies (PCR-SSOP frequency: 7.8%). Certain DR2 haplotypes provide resistance to insulin-dependent diabetes mellitus (IDDM) (28), but haplotype DR16,'DQ5, Dw21 (AZH or MN2 type) is specifically found more frequently in patients than in controls. This is the only DR2 haplotype that has a DQBI B chain allele that encodes a non-charged amino acid at position 57. At present, we have no data about the incidence of IDDM in Spanish Gypsies, but the higher frequency of DR16 that we are describing with respect to Spanish non-Gypsies may indicate a lower resistance to the disease among them. Further studies are in progress to clarify this aspect. When the distributions of subtypes of DR2, DQ1 haplotypes of Spanish Gypsies and Spanish non-Gypsies are compared, it is evident that DR15, DQ6, Dw12 is elevated in the former, this difference not being significant. A report (21) comparing the distribution of DR2 haplotypes in Asian Indians and Canadian Caucasians revealed that haplotype DR15, DQ6, Dw12 is overrepresented in the first group, which may be coincidental due to the geographical origin of Gypsies. Moreover, a low frequency of DRB1*1501 in Czech Gypsies has been reported (27), which points to a common origin of both Gypsy communities. By contrast, the frequency of DR16, DQ6, Dw21 in Indians is 0% (21), while the frequency of this haplotype among the DR2 Spanish Gypsies is 62.8%. Recent data report a DRB 1*1601 phenotypic frequency of 2.9% in Czech Gypsies (27) which contrasts with the frequency of this allele in Spanish Gypsies (F.F.: 22.1%) and is a point of coincidence of this group with Indians. Another point of contrast between these two Gypsy groups is the high frequency of DRB1*1602 in Czech Gypsies while in Spanish Gypsies it is quite low. A high frequency of haplotype DR14, DQ5, Dw9 has been reported in non-Caucasian populations of Asian origin (29). It is peculiar that Gypsies, considered to be anthropologically distant 194
from these populations, also show an increment in this haplotype. The higher frequency of DR14(DR6) in Spanish Gypsies with respect to Caucasian populations is in agreement with the reported frequency of this allele in Czech Gypsies (27). The underrepresentation of DR6 among Hungarian Gypsies might be explained by the difficulty encountered in defining this antigen serologically when this group was studied, when neither RFLP or PCR-AS0 were available to perform class I1 typing at the DNA level. As DR7 is one of the most frequent antigens in the Spanish non-Gypsy population, its lower frequency in Gypsies is very striking. The distribution of DR7 haplotypes is different in the two population groups compared here, DR7, DQ9, Dwll being that most frequent in Gypsies. The underrepresentation of DR4, DR7 and DR9 determines a low frequency of DR53 and is compensated by the already mentioned increments. PCR-SSOP results confirm, as expected, the data previously obtained by RFLP. It was of interest to note the higher frequency of DRBl* 1404 (28.5%) in Spanish Gypsies which is quite similar to that found in Czech Gypsies (29.4%). However, given that, in the presence of DR3, DR11 and DRI 3 we can not dstinguish DRB1*1401 from DRB1*1404, it is probable that many individuals cqnsidered to be DRB1*1401/1404 would be 1404. New DRBlDQB1 or DQA1-DQB1 associations were not found. When DRB 1-DPB 1 associations are considered, the Gypsy group shows a characteristic one, that is DRB 1 * 1401-DPB1*0401. The haplotypes found in Spanish Gypsies (Table 9) confirm the uniqueness of this group, as the majority of them are found only in this ethnic group. However, A2-CBL-BS1-DR 16-DQI is also found in Rumanians, while A30-Cw5-B18-DR3DQ2 is found in Spanish non-Gypsies, Basques and Sardinians. The association A 1-B61 has been reported in Hungarian Gypsies; however, Gyody et al. report B61 to be linked to DR8 instead of to DR6. It must be kept in mind that a clear-cut serological definition of these two alleles is difficult unless DNA typing techniques are used. Analysis of genetic distance data shows that Spanish Gypsies are closer to Indians than to Hungarian Gypsies. These results confirm the linguistic data that point to the Indian origin of this ethnic group. As the two Gypsy groups share a common language, it seems probable that the differences between them could be due to genetic drift. When we analyze the frequency distribution of A l , A l l , B61, DR14 in all the ethnic groups studied during the 1IthIHW, we find that many Oriental groups that are not related to Gypsies
HLA in Spanish Gypsies from an anthropological point of view also show very high frequencies of these antigens. We cannot convincingly explain how it is that ethnic groups in close proximity when conventional genetic distance is considered do not share extended haplotypes (as is the case of Spanish Gypsies and Indians). At the same time, it is also difficult to explain how it is that groups which are not very close one another in terms of allelic frequencies share haplotypes, as is the case of Hungarian and Spanish Gypsies. In conclusion, the Spanish Gypsy population has a unique HLA class I and class I1 antigenic distribution, different from that found in all the other ethnic groups studied so far. They are close to Indians in terms of genetic distance, and show a considerable similarity to Hungarian and Czech Gypsies. Acknowledgments
We want to express our thanks for their collabor-' ation to Alfonso Ram6n from the Centro de Promoci6n de Salud (C.P.S.) of Montesa and to the staff of the following C.P.S. of the Comunidad de Madrid: Hortaleza, Latina, Tetuan, Vallecas Puente. Vallecas Villa, Vicalvaro, Villaverde, and the following Centros de Atencibn Primaria: Potes and Gomez Acebo. We thank Alicia Brea, Consuelo Pascau, Isabel Perez, M. Luisa Crespo and Rosario Martin for their skillful technical assistance. cDNA probes were kindly provided by Ann-Kristin Jonsson, Swedish University of Agricultural Sciences, Uppsala Biomedical Center, Uppsala, Sweden. This work was partially supported by FISS (INSALUD) grant 90/200.
References 1. Sampson J. On the origin and early migration of the Gypsies. J G.vpsy Lore Society 1923: 3rd ser., 2(4): 156-69.
2. Hancock I. The Pariah Syndrome. Ann Arbor: Karoma Publishers, Inc. 1987. 3. Gyody E, Tauszik T, Petranyi Gy, et al. The HLA antigen distribution in the gypsy population in Hungary. Tissue Antigens 1981: 18: 1-12. 4. Imanishi T, Akaza T, Kimura A, Tokunaga K, Gojohori T. Reference Tables, Chapter 15. In Tsuji K, ed. HLA 1991, vol. 1. Oxford: Oxford University Press (in press). 5. Miller SA, Dykes DD, Polesky HF. A simple salting-out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res 1988: 1215. 6. Carlsson B, Wallin J, Bohme J, Moller G. HLA-DR-DQ haplotypes defined by restriction fragment analysis. Correlation to serology. Hum Irnmunol 1987: 20: 95-1 13. 7. Gustaffson K, Wiman K, Emmoth E, et al. Mutations and selection in the generation of class I1 histocompatibility antigen polymorphism. EMBO J 1984: 3: 1655-61. 8. Scheming L, Larhammar D, Bill P, et at. Both A and
B chains of HLA-DC class I1 histocompattibility antigens display extensive polymorphism in their aminotenninal domains. EMBO J 1984 3: 447-525. 9. Larhammar D, Scheming L, Gustafsson K, et al. Complete amino acid sequence of an HLA-DR antigen-like B chain as predicted from the nucleotide sequence: Similarities with immunoglobulins and HLA-A, -B and -C antigens. Proc Natl Acad Sci USA 1982: 79: 3687-91. 10. Martell M, Marcadet A, Cohen D. HLA Class I1 Typing by the Workshop Defined RFLP Specificities in Unrelated Individuals. In: Dupont B, ed. Immunobiology of HLA, vol. I. Berlin-Heidelberg-New York Springer Verlag, 1989. 947-52. 11. Bidwell J. DNA-RFLP analysis and genotyping of HLADR and DQ antigens. Immunology Today 1988: 9: 18-23. 12. Kimura A, Sasazuki T. Workshop Reference Protocol of the HLA-DNA Typing Technique. In: Tsuji K, ed. HLA 1991, vol. 1. Oxford: Oxford University Press (in press). 13. Kimura A, Dong R-P, Harada H, Sasazuki T. DNA Typing of HLA class I1 genes in B-lymphoblastoid. cell lines homozygous for HLA. Tissue Antigens 1992: 401: 5 . 14. Wood W, Gitschier J, Larsky LA, et al. Base compositionindependent hybridization in tetramethylainmonium chloride: A method for oligonucleotide screening of highly complex gene libraries. Proc Natl Acad Sci USA 1985: 82: 1585-8. 15. Pickbourne P, Piazza A, Bodmer WF. Poplulation Analysis. In: Bodmer WF, Batchelor JR, Bodmer T.J, Festenstein H, Morns PJ, eds. Histocompatibility Testing 1977. Copenhagen: Munksgaard, 1978: 259-78. 16. Baur MP, Danilovs JA. Population analysis of HLA-A. B, C, DR, and other genetic markers. In: Terasaki PI, ed. Histocompatibility Testing 1980. Los Angeles: UCLA Tissue Typing Laboratory, 1980: 955-8. 17. Nei M. Estimation of average heterozygosity and genetic distance from a small number of individua'ls. Generics 1978: 89: 583-90. 18. Imanishi 1.Akaza T. Kimura A, Tokunaya K, Gojobori T. Estimation of allele and haplotype frequencies for HLA and complement loci. In: Tsuji K, ed. H.LA 1991, vol. I , chapter 1. Oxford: Oxford University Press (in press). 19. 1 1th International Histocompatibility Workshop. Data Analysis Book 11. Central Analysis Committee, 1 1 th International Histocompability Workshop, Yokohama. Japan 1991. 20. Tait BD, Park MS, Barbetti A, Langley J, Terasaki PI, Aizawa M. Antigen Society #28. Report (.DQw3).In: Dupont B, ed. Immunobiology of HLA. Vol. 1, p. 269. New York, Berlin, Heidelberg, London, Paris, Tokyo, Hong Kong: Springer Verlag, 1987. 21. Singal DP, D'Souza M, Sood SK. Polymclrphism of HLADR2, DQwl haplotypes in Asian Indians. Tissue Antigens 1990: 36: 164-70. 22. Takacs K, Kalmhn B, Gyodi E, et al. Association between the lack of DQw6 and the low incidence of multiple sclerosis in Hungarian Gypsies. Zmmunogenetics 1990: 31: 383-5. 23. Trueba JL, Garcia Merino JA. Personal communication. 24. Olerup 0, Hillert J. HLA class II-associateid genetic susceptibility in multiple sclerosis: A critical evaluation. Tissue Anrigens 1991: 38: 1-15. 25. Kurdi A, Ayesh I, Abdallat A, et al. Different B lymphocyte alloantigens associated with multiple sclerosis in Arabs and North Europeans. Lancer 1977: i: 1123-5. 26. Marrosu MG, Muntoni F, Murru MR, et al. Sardian multiple sclerosis is associated with HLA-DR4. Neurology 1988: 38: 1749-53. 27. Cerna M, Fernandez-Viiia M, Ivaskova E., Stastny P. Comparison of HLA class I1 alleles in Gypsy and Czech populations by DNA typing with oligonucleotide probes. Tissue Antigens 1992: 39: 1 11-6.
195
de Pablo et a]. 28. Todd JA. Genetic control of autoimmunity in type 1 diabetes. Immunology Today 1990: 11: 122-9. 29. Kohonen-Codsh MRJ, Dunckley H, Serjeantson Sw. HLA-DR and -DQ DNA genotyping in seven populations of Asia-Oceania and Australia. Tissue Antigens 1988: 32: 3240.
196
Address: Rosario de Pablo Servicio de Inmunologa Hospital Puerta de Hierro ClSan Martin de Porres 4 28035 Madrid Spain
