HLA-DQA2 (DX alpha) Polymorphism and Insulin Dependent Diabetes J. R. Rowe, M. H. Neme de Gimenez, C. A. Emler, and M. J. Sheehy
A certain HLA-DQA2 1ecus TaqI fragment, DXa”U”, has been reported to be associated with insulin-dependent diabetes mellitus (IDDM). Reports of various studies in this vein have ranged from stating that the association of DQA2”U”witb IDDM exists even among subjects positive for HLA-DR3 and -DR4 to stating that the association of DQA2”U” with diabetes can be attributed to linkage disequilibrium between with DQA2”U” and some component(s) on the affected baplotypes. Using a synthetic 97-base probe corresponding to a portion of an intron of DQA2, in a Southern blot analysis of IDDM and controlsubjects from Wisconsin, we were able to confirm the association of DQA2“U” with diabetes. However, among DR3 subjects there was no significant association between DQA2”U” and diabetes (p = 0.26). Although there was a (nonsignt$cant) association of IDDM with DQA2”U”among DR4-positivesubjects (p = 0.141, this can be completely attributed to linkage disequilibrium between DQA2”V” and DQw8. We also sequenced most of the second exon (corresponding to the (Y 1 domain of the DQA2 glycoprotein) from five individuals that were bomozygous for either DQA2”U” or DQA2’2.” The onlypolymorpbisms observed were a “silent” mutation at position 36 and one example of a difference that would result in a change of amino acid at position 41.
ABSTRACT:
ABBREVIATIONS HTC
homozygous
IDDM
insulin-dependent mellitus
typing cell
diabetes
LCL RFLP
lymphoblastoid cell line restriction fragment length polymorphism
INTRODUCTION Much attention has been focused on the genetics of insulin-dependent diabetes mellitus (IDDM) with the hope of identifying the “disease susceptibility” locus. The most consistent associations are with various HLA class II genes. Alleles of the DRB 1 locus (DR3, Dw4, and Dw 10) and the DQB 1 locus (DQw8) have been shown to be significantly associated with IDDM in Caucasoids (reviewed in ref. 1) while the DQAl locus also seems to play a major role in a negroid population [2 J. A variety of studies have been performed concerning a DQA2 TaqI restriction fragment length polymorphism (RFLP) (first reported by Spielman et al. [3]). The results of these studies have varied from suggesting that the DQA2”U”/IDDM association supersedes previously reported associations to stating that it is the
From the Research Department, American RedCross BloodServi~es (J.R.R.; M.H.N.d.G; C.A.E.; M.J.S.) and the Department of Medicine, University of Wiscomin (M./S.), Madison, Wisconsin. Address reprint request1 toJ. R. Rowe, The American Red Cross, 4860 Sheboygan Avenue, Madison, WI r3705. Received Janua y 25, 1990; accepted May 15, I990
256 0198.8859190/$3.50
0 American
Human Immunology 29, 256-262 Society for Histocompatibility and Immunogenetics,
(1990) 1990
DQA2 Polymorphism
and IDDM
257
result of linkage disequilibrium with them [4-111. We performed the current study to assess whether DR3 and DR4 are associated with IDDM because of linkage disequilibrium with a DQA2 allele or vice versa. We also compared DNA sequences of the second exons of several DQA2 alleles. Although very little polymorphism has been reported for the DQA2 or DQB2 locus to date, we wanted to see if the DQA2 U and L alleles had any corresponding differences in the coding sequence of the second exon. This is the most polymorphic exon in other class II loci, and encodes the probable peptidebinding portion of the class II molecule [12]. MATERIALS
AND
METHODS
Subjects. The 79 Wisconsin diabetic subjects were either newly diagnosed IDDM cases, age 29 years or less at onset (n = 49) from the Wisconsin Diabetes in Youth Study, a population-based epidemiologic study conducted in a 14-county area of southern Wisconsin, or patients attending the Pediatric Diabetes Clinic at University Hospital in Madison, Wisconsin (age 20 years or less at onset, n = 30). All diabetic subjects meet the World Health Organization criteria for IDDM. The 63 Wisconsin controls (nondiabetic) were age-strata and sex-matched controls from the Wisconsin Diabetes in Youth Study (n = 34) and Wisconsin-born, Madison-area residents (n = 29). HLA-A/B typing was performed with locally produced microcytotoxicity sera trays (Tissue Typing Laboratory, American Red Cross Blood Services, Badger Region) and DR typing was performed with trays from One Lambda Co. (Los Angeles, CA). Two cell lines (TAV and 4092) derived from NIH Dw2 homozygous typing cells (HTCs) were also included in the sequencing portion of this study. DNA isolation and so&em blotting. DNA was extracted from lo* Epstein-Barrvirus-transformed lymphoblastoid cell lines (LCLs) of each subject using standard protocols 1131. DNA samples (25 pg) were digested with 4 UIpg of TaqI (Promega) and electrophoresed on 1.5% agarose gels in TAB buffer 1131. Gels were depurinated in 0.15 M HCI for 15 min at room temperature, rinsed with distilled water, and denatured with 0.4N NaOH for 30 min before being transferred to GeneScreen Plus membranes by capillary blotting. The DQA2 locus-specific probe corresponding to the first 97 bases of sense-strand DNA following the DQA2 second exon (based on published sequence data [14]), was synthesized by P-cyanoethyl amidite method (Pharmacia Gene Assembler). The probe was labeled by random primer extension using [~x-~*P] dCTP (3000 Ci/mmol; DuPont) and the Oligolabelling’” kit (Pharmacia). DNA sequencing. DNA from five subjects, each homozygous DQA2 U/U or L/L, was digested with 6 U/pg of Hind111 (Promega) and electrophoresed in 0.8% agarose along with appropriate size markers. DNA between 2 and 3 kb (containing the DQA2 but not DQAl genes [151) was excised and electroeluted from the agarose using an ISCO Sample Concentrator. From 10 ng of these samples, the second exons of the DQA2 genes were amplified by polymerase chain reaction using the “linker primers” LPl (gtgctcaGGT GTG AAC TIC TAC CAG) and LP2 (cacggatccGGT GGC AGC GGT AGA GTT G). The resulting products were cloned into M 13mp 18 and M 13mp 19 phage vectors and screened using the following DQA2 specific oligonucleotides: probe A (antisense, to screen mp 19 clones) G GAC CTC TGC TIT CTC TGA and probe B (sense, to screen mp18 clones) TCA GAG AAA GCA GAG GTC C. Sequencing was performed by the dideoxy chain termination method using the T7Sequencing Kit from Pharmacia.
J. R. Rowe
258
et al.
Data analysis. p values were calculated by the Fisher exact test [ 161 for relationships of HLA alleles and diabetes. The Southern blots permitted unambiguous genotyping for the DQA2 U/L RFLP. No subject groups differed significantly from Hardy-Weinberg equilibrium for the genotypes U/U, U/L, and L/L (all p > 0.3). Therefore the U and L allele frequencies, not genotype frequencies, were used in the tests of association with IDDM. Since HLA-DR genotypes could not be assigned unambiguously in all cases (apparent homozygosity was not verified by family testing), the significance tests for DR3 and DR4 were based on antigen frequencies (70 of individuals positive) instead of allele frequencies.
RESULTS We examined the degree of association of the reported DQA2 RFLP (“DXaU”) 131 and diabetes in our subject panels. We first tested 20 matched pairs of diabetics and controls from the Wisconsin Diabetes in Youth Study with a synthetic DQA2 locus-specific probe in Southern blots (Table 1) and found that the DQA2“U” allele was significantly associated with diabetes in this subject group (p = 0.01) as reported by others [4-6,8-101. To assess the relationship of this association with previously reported associations between IDDM and other HLA alleles, we tested groups of subjects that were selected to be either DR3 or DR4 positive. We also broke the DR4 group down into those subsets that carry previously defined markers identifying diabetes-associated DR4 haplotypes (either Dw4/ DwlO positive or DQw8 positive). These results are summarized in Fig. 1. There was no significant association between DQAZ‘V” and diabetes among the DR3or DR4-positive subjects (p = 0.26 and 0.14, respectively). Among the DR4 subjects that were positive for either Dw4 or DwlO there was a marginally significant association (p = 0.04). However, among the DR4 subjects that were positive for DQw8 there was no association whatsoever between DQA2“U” and diabetes (p = 0.54). Thus the association of DQA“U” with IDDM can be attributed to linkage disequilibrium with DR3 and DQw8, HLA alleles previously implicated in IDDM susceptibility. We also did an analysis that was the mirror image of that described above to see if the DR3 and DR4 associations with IDDM could possibly be attributed to linkage disequilibrium with DQA2“U.” Specifically, we analyzed the original group of 20 pairs (unselected for HLA types) in order to see if DR3 and DR4 were still IDDM associated among the DQA2“U” subset (Fig. 2). For both DR3 and DR4 there was still a significant association with IDDM (p = 0.02 in each
TABLE
1
DQA2 is associated with IDDM among random IDDM subjects and matched controls in Wisconsin
Diabetics (n = 20)
Controls” (n = 20)
30% U/L 45% U/L 25% L/L
5% u/u 40% U/L 55% L/L
* “Best friend” controls,
p = 0.01.
sex and age strata matched
DQA2
Polymorphism
% ‘W
and IDDM
259
Allele
00
q
n
Diabetics
80
Controls I
I
7t-i S”
[
p30.26
p=O.Ol
pro.14
pro.04
pt0.54
1
60 50 40 30 20 10 0 n=20*n=20
n=22 n=20
n=22 n=22
DR3*
DR4+
Unselected
FIGURE 1 DQA2“U,” associated among DR3+, in each group.
n=21 n=lO
Dw4fw
n=lO n=15
10.
Dctw8.
although IDDM associated among random subjects, is not IDDM DR4+, or DQwS+ subjects. * = number of subjects, not alleles,
q
Diabetics
n
Controls
100 00
pg0.05
ps0.02
pro.00
1
p10.02
80 70 80 %
60 40 30 20 10 0 1~20 n=20 D&Among Unselected
FIGURE
2
n=15 n=O DR$Among DQA2
-U
DR3 and DR4 are IDDM
n=20 m-20
n=15 n=O
DFt4* Among
DFt4* Among
unsdected
associated
DQAP ‘W
among DQA2”U”-positive
subjects.
case) even among individuals, all of whom were DQA2“U” positive. Thus the associations of DR3 and DR4 cannot be attributed to linkage disequilibrium with DQA2“U.” Additionally we sequenced most of the second exon (corresponding to the (Y1 domain of the putative DQA2 glycoprotein) from five individuals that were
260
J. R. Rowe et al.
homozygous for either DQA2“U” or DQA2“L.” Three of the subjects were DR 314 diabetic patients (two DQA2 U/U and one DQA2 L/L) and the other two subjects were NIH Dw2 HTCs (one DQA2 U/U and one DQA2 L/L). There was a marked lack of polymorphism in these sequences, with only two differences from the previously published sequences noted. The first was a “silent” mutation at position 36, found in 3/10 haplotypes (resulting in an alternative tyrosine codon, TAC instead of TAT). The second difference occurred in one haplotype of the NIH Dw2 HTC that was DQA2 L/L homozygous, resulting in a change from a threonine codon (ACG) to a methionine (ATG) at position 41. These polymorphisms do not create or eliminate TaqI sites, and neither is strictly associated with the DQA2 U or L allele although all of the differences noted occurred on DQA2 L haplotypes. DISCUSSION In this paper we have focused on a DQA2 TaqI RFLP (first reported by Spielman et al. 131 using a DQAl coding region probe) and its relationship to IDDM and other HLA region markers for IDDM. A variety of studies have been undertaken examining the relationship of DQAZ‘V” with a number of autoimmune diseases. IDDM has comprised the bulk of the interest and most of these studies have shown associations between IDDM and DQA2“U” with specific exceptions in some individuals and populations 14-11). Associations with DQA2“U” have also been reported for rheumatoid arthritis [4], celiac disease Cb], dermatitis herpetiformis [6,17], multiple sclerosis [18], and linear IgA disease [17], while both Graves’ disease Cl93 and narcolepsy [20] have been reported to have no association with DQA2“U.” We feel that it is important to carefully test new associations with those previously reported in order to more fully elucidate the true picture of disease susceptibility and to rule out spurious or secondary associations. In a previous paper {2 11, our laboratory, along with the laboratories of Drs. Henry Erlich and Barbara Nepom, showed that both the DRBl and DQBl loci seem to contribute equally and interactively to IDDM susceptibility. In this paper we have carried out a similar analysis with a quite different result. The observation that DQA2“U” is not IDDM associated among the DR3 subjects or among the DQw8 subjects shows that DQA2“U” is not associated with IDDM in a new or unique way. Rather it shows that DQA2“U” is in linkage disequilibrium with DR3 and DQw8, not surprising since the recombination frequency within the DQ region and between DR and DQ is very low [22]. Also, in light of the lack of DQA2“U”/IDDM association in DQw8 subjects, it is not surprising that the DQA2“U”iIDDM association is stronger among the Dwcl/DwlO subjects than among all the DR4 subjects. This reflects the positive interactions between DRBl and DQBl alleles (Dw4 and DwlO, and DQw8) in their contributions to IDDM susceptibility, as previously reported [ 11,2 l]. Certain studies {S, 10,23] have reported linkage disequilibrium between DQA2“U” and DR3 on certain haplotypes, specifically those carrying HLA-B8 and HLA-DRB3.0 101 (the latter previously called DRw52a) concluding that the association of DXA“U” and IDDM is purely the result of linkage disequilibrium. With this in mind, we found that even the slight and nonsignificant excess of DQA2“U” among DR3+ patients versus DR3+. controls (Fig. 1) disappeared when the analysis was limited to individuals that were B8+DR3+: 74% of patients versus 7 1% of controls being DQA2“U”+ (p > 0.7). Although no product for the DQ2 genes has been described, the DNA sequences, including flanking sequences, suggest that these are not pseudogenes [14,15]. Berdot et al. {24] have already commented on the remarkable level of
DQA2
Polymorphism
261
and IDDM
conservation of the DQB2 locus, suggesting that a protein product exists and that there is a strong selective pressure for the conservation of the putative DQ2 amino acid sequence. Our results for DQA2 are quite parallel. An exceptionally high degree of sequence conservation in the second exon of DQAZ, the only differences noted being a silent mutation seen in 3/10 haplotypes sequenced and only one instance of a change in the putative amino acid sequence (a methionine instead of a threonine at position 4 l), also points to the expression of the DQ2 genes with selective pressures maintaining the amino acid sequence. It is interesting to ponder the circumstances under which expression of the DQ2 gene occurs and what function(s) they perform. sequence
ACKNOWLEDGMENTS
This work was supported in part by Juvenile Diabetes Foundation Grant no. 187699, a matching grant from the American Red Cross Blood Services, and an American Red Cross Established Investigator Award (to M.J.S.).
REFERENCES 1. Wassmuth R, Lernmark Immunol Immunopathol
A: The genetics 53:358, 1989.
of susceptibility
to diabetes
(review).
Clin
2. Todd JA, Mijovic C, Fletcher J, Jenkins D, Bradwell AR, Barnwett AH: Identification of the susceptibility loci for insulin-dependent diabetes mellitus by trans-racial gene mapping. Nature 3381587, 1989. 3. Spielman RS, Lee J, Bodmer WF, Bodmer JG, Trowsdale J: Six HLA-D region LYchain genes on human chromosome 6: Polymorphism and associations of D&-related sequences with DR types. Proc Nat1 Acad Sci USA 81:3461, 1984. 4. Festenstein H, Awad J, Hitman GA, Cutbush S, Groves AV, Cassell P, Oilier W, Sachs J: New HLA DNA polymorphisms associated with autoimmune diseases. Nature 323:64, 1986. 5. Hitman GA, Sachs J, Cassell P, Awad J, Bottazzo GF, Tarn AC, Schwartz G, Monson JP, Festenstein H: A DR3-related DXa gene polymorphism strongly associates with insulin-dependent diabetes mellitus. Immunogenetics 23:47, 1986. 6. Hitman JN, Fry patients disease.
GA, Niven MJ, Festenstein H, Cassell PG, Awad J, Walker-Smith J, Leonard L, Ciclitira P, Kumat P, Sachs J: HLA class II a-chain gene polymorphisms in with insulin-dependent diabetes meIIitus, dermatitis herpetiformis, and celiac J Clin Invest 79:608, 1987.
7. Serjeantson SW, Ranford PR, Kirk RL, Kohonen-Corish MRJ, Mohan V, Ramachandran A, Snehalatha C, Viswanathan M: HLA-DR and -DQ DNA genotyping in insulindependent diabetes patients in South India. Dis Markers 5:101, 1987. 8. Kohonen-Corish MRJ, Serjeantson SW, Lee HK, Zimmet P: Insulin-dependent tes mellitus: HLA-DR and -DQ genotyping in three ethnic groups. Dis Markers 1987. 9. Fletcher J, Odugbesan 0, Mijovic C, Mackay E, Bradwell HLA DNA polymorphisms in Type 1 (insulin-dependent) Indian origin. Diabetologia 3 1:343, 1988.
diabe5: 15 3,
AR, Barnnett AH: Class II diabetic patients of North
10. Carrier CM, Mollen N, Rothman WC, Rodriguez de Cordoba S, Rey-Campos J, Ginsberg-Fellner F, Carpenter C, Rubinstein P: Definition of IDDM-associated HLA DQ and DX RFLPs by segregation analysis of multiplex sibships. Hum Immunol 24:51, 1989.
262
J. R. Rowe et al.
11. Zerbib A, Molvig J, Thomsen M, Coppin H, Cambon-Thomsen A, Sommer E, Nerup J, De Preval C: Restriction fragment length polymorphism of HLA and non-HLA genes in DR3/4 heterozygous Danish insulin-dependent diabetic patients and healthy individuals: Reassessment of the influence of (Y DX and insulin-linked polymorphic loci and new splits of DQw3 haplotypes. Dis Markers 7:27, 1989. 12. Brown JH, Jardetzky T, Saper MA, Samraoui B, Bjorkman PJ, Wiley DC: A hypothetical model of the foreign antigen binding site of class II histocompatibility molecules. Nature 322:845, 1988. 13. Maniatis T, Fritsch EF, Sambrook J: Molecular Cloning, A Laboratory Spring Harbor, NY, Cold Spring Harbor Laboratory, 1982.
Manual. Cold
14. Auffray C, Lillie JW, Arnot D, Grossberger D, Kappes D, StromingerJL: Isotypic and allotypic variation of human class II histocompatibility antigen o-chain genes. Nature 308~327, 1984. 15. Okada K, Boss JM, Prentice H, Spies T, Mengler R, Auffray C, Lillie JW, Grossberger D, Strominger JL: Gene organization of the human major histocompatibility complex. Proc Nat1 Acad Sci USA 82:3410, 1985. 16. Fisher RA: The Design
of Experiments.
Edinburgh,
Oliver and Boyd,
1960.
17. Sachs J, Leonard JN, Awad J, McLoskey D, Festenstein H, Hitman GA, Fry L: A comparative serological and molecular study of linear IgA disease and dermatitis herpetiformis. Br J Dermatol 118:759, 1988. 18. Haegert DG, Michaud M, Schwab C, Tansey C, Decary F, Francis G: HLA-DRP, DQo, and -DQp restriction fragment length polymorphisms in multiple sclerosis. Neurosci Res 23:46, 1989. 19. Weeyman AP, Roe C, So A: DXa Antigens 31:1, 1988.
gene polymorphism
in Graves’
disease.
J
Tissue
20. Matsuki K, Maeda H, Juji T, Inoki H, Ando A, Tsuji K, Honda Y: TaqI-generated HLA-DQa! polymorphism in Japanese patients with narcolepsy. Immunogenetics 27~87, 1988. 21. Sheehy MJ, Scharf SJ, Rowe JR, Neme de Gimenez MH, Meske LM, Erlich HE, Nepom BS: A diabetes-susceptible haplotype is best defined by a combination of HLA-DR and -DQ alleles. J Clin Invest 83:830, 1989. 22. Hardy DA, Bell JI, Long EO, Lindsten T, McDevitt HO: Mapping of the class II region of the major histocompatibility complex by pulsed-field gel electrophoresis. Nature 323:453, 1986. 23. Gorski J, Niven MJ, Sachs J, Mach B, Cassell PG, Festenstein H, Awad J, Hitman GA: HLA-DRa, -DXo, andDR@III gene association studies in DR3 individuals. Hum Immunol 20:273, 1987. 24. Berdoz J, Tiercy JM, Rollini P, Mach B, Gorski J: Remarkable sequence conservation of the HLA-DQB2 locus (DXP) within the highly polymorphic DQ subregion of the human MHC. Immunogenetics 29:24 1, 1989.
A certain HLA-DQA2 1ecus TaqI fragment, DXa”U”, has been reported to be associated with insulin-dependent diabetes mellitus (IDDM). Reports of various studies in this vein have ranged from stating that the association of DQA2”U”witb IDDM exists even among subjects positive for HLA-DR3 and -DR4 to stating that the association of DQA2”U” with diabetes can be attributed to linkage disequilibrium between with DQA2”U” and some component(s) on the affected baplotypes. Using a synthetic 97-base probe corresponding to a portion of an intron of DQA2, in a Southern blot analysis of IDDM and controlsubjects from Wisconsin, we were able to confirm the association of DQA2“U” with diabetes. However, among DR3 subjects there was no significant association between DQA2”U” and diabetes (p = 0.26). Although there was a (nonsignt$cant) association of IDDM with DQA2”U”among DR4-positivesubjects (p = 0.141, this can be completely attributed to linkage disequilibrium between DQA2”V” and DQw8. We also sequenced most of the second exon (corresponding to the (Y 1 domain of the DQA2 glycoprotein) from five individuals that were bomozygous for either DQA2”U” or DQA2’2.” The onlypolymorpbisms observed were a “silent” mutation at position 36 and one example of a difference that would result in a change of amino acid at position 41.
ABSTRACT:
ABBREVIATIONS HTC
homozygous
IDDM
insulin-dependent mellitus
typing cell
diabetes
LCL RFLP
lymphoblastoid cell line restriction fragment length polymorphism
INTRODUCTION Much attention has been focused on the genetics of insulin-dependent diabetes mellitus (IDDM) with the hope of identifying the “disease susceptibility” locus. The most consistent associations are with various HLA class II genes. Alleles of the DRB 1 locus (DR3, Dw4, and Dw 10) and the DQB 1 locus (DQw8) have been shown to be significantly associated with IDDM in Caucasoids (reviewed in ref. 1) while the DQAl locus also seems to play a major role in a negroid population [2 J. A variety of studies have been performed concerning a DQA2 TaqI restriction fragment length polymorphism (RFLP) (first reported by Spielman et al. [3]). The results of these studies have varied from suggesting that the DQA2”U”/IDDM association supersedes previously reported associations to stating that it is the
From the Research Department, American RedCross BloodServi~es (J.R.R.; M.H.N.d.G; C.A.E.; M.J.S.) and the Department of Medicine, University of Wiscomin (M./S.), Madison, Wisconsin. Address reprint request1 toJ. R. Rowe, The American Red Cross, 4860 Sheboygan Avenue, Madison, WI r3705. Received Janua y 25, 1990; accepted May 15, I990
256 0198.8859190/$3.50
0 American
Human Immunology 29, 256-262 Society for Histocompatibility and Immunogenetics,
(1990) 1990
DQA2 Polymorphism
and IDDM
257
result of linkage disequilibrium with them [4-111. We performed the current study to assess whether DR3 and DR4 are associated with IDDM because of linkage disequilibrium with a DQA2 allele or vice versa. We also compared DNA sequences of the second exons of several DQA2 alleles. Although very little polymorphism has been reported for the DQA2 or DQB2 locus to date, we wanted to see if the DQA2 U and L alleles had any corresponding differences in the coding sequence of the second exon. This is the most polymorphic exon in other class II loci, and encodes the probable peptidebinding portion of the class II molecule [12]. MATERIALS
AND
METHODS
Subjects. The 79 Wisconsin diabetic subjects were either newly diagnosed IDDM cases, age 29 years or less at onset (n = 49) from the Wisconsin Diabetes in Youth Study, a population-based epidemiologic study conducted in a 14-county area of southern Wisconsin, or patients attending the Pediatric Diabetes Clinic at University Hospital in Madison, Wisconsin (age 20 years or less at onset, n = 30). All diabetic subjects meet the World Health Organization criteria for IDDM. The 63 Wisconsin controls (nondiabetic) were age-strata and sex-matched controls from the Wisconsin Diabetes in Youth Study (n = 34) and Wisconsin-born, Madison-area residents (n = 29). HLA-A/B typing was performed with locally produced microcytotoxicity sera trays (Tissue Typing Laboratory, American Red Cross Blood Services, Badger Region) and DR typing was performed with trays from One Lambda Co. (Los Angeles, CA). Two cell lines (TAV and 4092) derived from NIH Dw2 homozygous typing cells (HTCs) were also included in the sequencing portion of this study. DNA isolation and so&em blotting. DNA was extracted from lo* Epstein-Barrvirus-transformed lymphoblastoid cell lines (LCLs) of each subject using standard protocols 1131. DNA samples (25 pg) were digested with 4 UIpg of TaqI (Promega) and electrophoresed on 1.5% agarose gels in TAB buffer 1131. Gels were depurinated in 0.15 M HCI for 15 min at room temperature, rinsed with distilled water, and denatured with 0.4N NaOH for 30 min before being transferred to GeneScreen Plus membranes by capillary blotting. The DQA2 locus-specific probe corresponding to the first 97 bases of sense-strand DNA following the DQA2 second exon (based on published sequence data [14]), was synthesized by P-cyanoethyl amidite method (Pharmacia Gene Assembler). The probe was labeled by random primer extension using [~x-~*P] dCTP (3000 Ci/mmol; DuPont) and the Oligolabelling’” kit (Pharmacia). DNA sequencing. DNA from five subjects, each homozygous DQA2 U/U or L/L, was digested with 6 U/pg of Hind111 (Promega) and electrophoresed in 0.8% agarose along with appropriate size markers. DNA between 2 and 3 kb (containing the DQA2 but not DQAl genes [151) was excised and electroeluted from the agarose using an ISCO Sample Concentrator. From 10 ng of these samples, the second exons of the DQA2 genes were amplified by polymerase chain reaction using the “linker primers” LPl (gtgctcaGGT GTG AAC TIC TAC CAG) and LP2 (cacggatccGGT GGC AGC GGT AGA GTT G). The resulting products were cloned into M 13mp 18 and M 13mp 19 phage vectors and screened using the following DQA2 specific oligonucleotides: probe A (antisense, to screen mp 19 clones) G GAC CTC TGC TIT CTC TGA and probe B (sense, to screen mp18 clones) TCA GAG AAA GCA GAG GTC C. Sequencing was performed by the dideoxy chain termination method using the T7Sequencing Kit from Pharmacia.
J. R. Rowe
258
et al.
Data analysis. p values were calculated by the Fisher exact test [ 161 for relationships of HLA alleles and diabetes. The Southern blots permitted unambiguous genotyping for the DQA2 U/L RFLP. No subject groups differed significantly from Hardy-Weinberg equilibrium for the genotypes U/U, U/L, and L/L (all p > 0.3). Therefore the U and L allele frequencies, not genotype frequencies, were used in the tests of association with IDDM. Since HLA-DR genotypes could not be assigned unambiguously in all cases (apparent homozygosity was not verified by family testing), the significance tests for DR3 and DR4 were based on antigen frequencies (70 of individuals positive) instead of allele frequencies.
RESULTS We examined the degree of association of the reported DQA2 RFLP (“DXaU”) 131 and diabetes in our subject panels. We first tested 20 matched pairs of diabetics and controls from the Wisconsin Diabetes in Youth Study with a synthetic DQA2 locus-specific probe in Southern blots (Table 1) and found that the DQA2“U” allele was significantly associated with diabetes in this subject group (p = 0.01) as reported by others [4-6,8-101. To assess the relationship of this association with previously reported associations between IDDM and other HLA alleles, we tested groups of subjects that were selected to be either DR3 or DR4 positive. We also broke the DR4 group down into those subsets that carry previously defined markers identifying diabetes-associated DR4 haplotypes (either Dw4/ DwlO positive or DQw8 positive). These results are summarized in Fig. 1. There was no significant association between DQAZ‘V” and diabetes among the DR3or DR4-positive subjects (p = 0.26 and 0.14, respectively). Among the DR4 subjects that were positive for either Dw4 or DwlO there was a marginally significant association (p = 0.04). However, among the DR4 subjects that were positive for DQw8 there was no association whatsoever between DQA2“U” and diabetes (p = 0.54). Thus the association of DQA“U” with IDDM can be attributed to linkage disequilibrium with DR3 and DQw8, HLA alleles previously implicated in IDDM susceptibility. We also did an analysis that was the mirror image of that described above to see if the DR3 and DR4 associations with IDDM could possibly be attributed to linkage disequilibrium with DQA2“U.” Specifically, we analyzed the original group of 20 pairs (unselected for HLA types) in order to see if DR3 and DR4 were still IDDM associated among the DQA2“U” subset (Fig. 2). For both DR3 and DR4 there was still a significant association with IDDM (p = 0.02 in each
TABLE
1
DQA2 is associated with IDDM among random IDDM subjects and matched controls in Wisconsin
Diabetics (n = 20)
Controls” (n = 20)
30% U/L 45% U/L 25% L/L
5% u/u 40% U/L 55% L/L
* “Best friend” controls,
p = 0.01.
sex and age strata matched
DQA2
Polymorphism
% ‘W
and IDDM
259
Allele
00
q
n
Diabetics
80
Controls I
I
7t-i S”
[
p30.26
p=O.Ol
pro.14
pro.04
pt0.54
1
60 50 40 30 20 10 0 n=20*n=20
n=22 n=20
n=22 n=22
DR3*
DR4+
Unselected
FIGURE 1 DQA2“U,” associated among DR3+, in each group.
n=21 n=lO
Dw4fw
n=lO n=15
10.
Dctw8.
although IDDM associated among random subjects, is not IDDM DR4+, or DQwS+ subjects. * = number of subjects, not alleles,
q
Diabetics
n
Controls
100 00
pg0.05
ps0.02
pro.00
1
p10.02
80 70 80 %
60 40 30 20 10 0 1~20 n=20 D&Among Unselected
FIGURE
2
n=15 n=O DR$Among DQA2
-U
DR3 and DR4 are IDDM
n=20 m-20
n=15 n=O
DFt4* Among
DFt4* Among
unsdected
associated
DQAP ‘W
among DQA2”U”-positive
subjects.
case) even among individuals, all of whom were DQA2“U” positive. Thus the associations of DR3 and DR4 cannot be attributed to linkage disequilibrium with DQA2“U.” Additionally we sequenced most of the second exon (corresponding to the (Y1 domain of the putative DQA2 glycoprotein) from five individuals that were
260
J. R. Rowe et al.
homozygous for either DQA2“U” or DQA2“L.” Three of the subjects were DR 314 diabetic patients (two DQA2 U/U and one DQA2 L/L) and the other two subjects were NIH Dw2 HTCs (one DQA2 U/U and one DQA2 L/L). There was a marked lack of polymorphism in these sequences, with only two differences from the previously published sequences noted. The first was a “silent” mutation at position 36, found in 3/10 haplotypes (resulting in an alternative tyrosine codon, TAC instead of TAT). The second difference occurred in one haplotype of the NIH Dw2 HTC that was DQA2 L/L homozygous, resulting in a change from a threonine codon (ACG) to a methionine (ATG) at position 41. These polymorphisms do not create or eliminate TaqI sites, and neither is strictly associated with the DQA2 U or L allele although all of the differences noted occurred on DQA2 L haplotypes. DISCUSSION In this paper we have focused on a DQA2 TaqI RFLP (first reported by Spielman et al. 131 using a DQAl coding region probe) and its relationship to IDDM and other HLA region markers for IDDM. A variety of studies have been undertaken examining the relationship of DQAZ‘V” with a number of autoimmune diseases. IDDM has comprised the bulk of the interest and most of these studies have shown associations between IDDM and DQA2“U” with specific exceptions in some individuals and populations 14-11). Associations with DQA2“U” have also been reported for rheumatoid arthritis [4], celiac disease Cb], dermatitis herpetiformis [6,17], multiple sclerosis [18], and linear IgA disease [17], while both Graves’ disease Cl93 and narcolepsy [20] have been reported to have no association with DQA2“U.” We feel that it is important to carefully test new associations with those previously reported in order to more fully elucidate the true picture of disease susceptibility and to rule out spurious or secondary associations. In a previous paper {2 11, our laboratory, along with the laboratories of Drs. Henry Erlich and Barbara Nepom, showed that both the DRBl and DQBl loci seem to contribute equally and interactively to IDDM susceptibility. In this paper we have carried out a similar analysis with a quite different result. The observation that DQA2“U” is not IDDM associated among the DR3 subjects or among the DQw8 subjects shows that DQA2“U” is not associated with IDDM in a new or unique way. Rather it shows that DQA2“U” is in linkage disequilibrium with DR3 and DQw8, not surprising since the recombination frequency within the DQ region and between DR and DQ is very low [22]. Also, in light of the lack of DQA2“U”/IDDM association in DQw8 subjects, it is not surprising that the DQA2“U”iIDDM association is stronger among the Dwcl/DwlO subjects than among all the DR4 subjects. This reflects the positive interactions between DRBl and DQBl alleles (Dw4 and DwlO, and DQw8) in their contributions to IDDM susceptibility, as previously reported [ 11,2 l]. Certain studies {S, 10,23] have reported linkage disequilibrium between DQA2“U” and DR3 on certain haplotypes, specifically those carrying HLA-B8 and HLA-DRB3.0 101 (the latter previously called DRw52a) concluding that the association of DXA“U” and IDDM is purely the result of linkage disequilibrium. With this in mind, we found that even the slight and nonsignificant excess of DQA2“U” among DR3+ patients versus DR3+. controls (Fig. 1) disappeared when the analysis was limited to individuals that were B8+DR3+: 74% of patients versus 7 1% of controls being DQA2“U”+ (p > 0.7). Although no product for the DQ2 genes has been described, the DNA sequences, including flanking sequences, suggest that these are not pseudogenes [14,15]. Berdot et al. {24] have already commented on the remarkable level of
DQA2
Polymorphism
261
and IDDM
conservation of the DQB2 locus, suggesting that a protein product exists and that there is a strong selective pressure for the conservation of the putative DQ2 amino acid sequence. Our results for DQA2 are quite parallel. An exceptionally high degree of sequence conservation in the second exon of DQAZ, the only differences noted being a silent mutation seen in 3/10 haplotypes sequenced and only one instance of a change in the putative amino acid sequence (a methionine instead of a threonine at position 4 l), also points to the expression of the DQ2 genes with selective pressures maintaining the amino acid sequence. It is interesting to ponder the circumstances under which expression of the DQ2 gene occurs and what function(s) they perform. sequence
ACKNOWLEDGMENTS
This work was supported in part by Juvenile Diabetes Foundation Grant no. 187699, a matching grant from the American Red Cross Blood Services, and an American Red Cross Established Investigator Award (to M.J.S.).
REFERENCES 1. Wassmuth R, Lernmark Immunol Immunopathol
A: The genetics 53:358, 1989.
of susceptibility
to diabetes
(review).
Clin
2. Todd JA, Mijovic C, Fletcher J, Jenkins D, Bradwell AR, Barnwett AH: Identification of the susceptibility loci for insulin-dependent diabetes mellitus by trans-racial gene mapping. Nature 3381587, 1989. 3. Spielman RS, Lee J, Bodmer WF, Bodmer JG, Trowsdale J: Six HLA-D region LYchain genes on human chromosome 6: Polymorphism and associations of D&-related sequences with DR types. Proc Nat1 Acad Sci USA 81:3461, 1984. 4. Festenstein H, Awad J, Hitman GA, Cutbush S, Groves AV, Cassell P, Oilier W, Sachs J: New HLA DNA polymorphisms associated with autoimmune diseases. Nature 323:64, 1986. 5. Hitman GA, Sachs J, Cassell P, Awad J, Bottazzo GF, Tarn AC, Schwartz G, Monson JP, Festenstein H: A DR3-related DXa gene polymorphism strongly associates with insulin-dependent diabetes mellitus. Immunogenetics 23:47, 1986. 6. Hitman JN, Fry patients disease.
GA, Niven MJ, Festenstein H, Cassell PG, Awad J, Walker-Smith J, Leonard L, Ciclitira P, Kumat P, Sachs J: HLA class II a-chain gene polymorphisms in with insulin-dependent diabetes meIIitus, dermatitis herpetiformis, and celiac J Clin Invest 79:608, 1987.
7. Serjeantson SW, Ranford PR, Kirk RL, Kohonen-Corish MRJ, Mohan V, Ramachandran A, Snehalatha C, Viswanathan M: HLA-DR and -DQ DNA genotyping in insulindependent diabetes patients in South India. Dis Markers 5:101, 1987. 8. Kohonen-Corish MRJ, Serjeantson SW, Lee HK, Zimmet P: Insulin-dependent tes mellitus: HLA-DR and -DQ genotyping in three ethnic groups. Dis Markers 1987. 9. Fletcher J, Odugbesan 0, Mijovic C, Mackay E, Bradwell HLA DNA polymorphisms in Type 1 (insulin-dependent) Indian origin. Diabetologia 3 1:343, 1988.
diabe5: 15 3,
AR, Barnnett AH: Class II diabetic patients of North
10. Carrier CM, Mollen N, Rothman WC, Rodriguez de Cordoba S, Rey-Campos J, Ginsberg-Fellner F, Carpenter C, Rubinstein P: Definition of IDDM-associated HLA DQ and DX RFLPs by segregation analysis of multiplex sibships. Hum Immunol 24:51, 1989.
262
J. R. Rowe et al.
11. Zerbib A, Molvig J, Thomsen M, Coppin H, Cambon-Thomsen A, Sommer E, Nerup J, De Preval C: Restriction fragment length polymorphism of HLA and non-HLA genes in DR3/4 heterozygous Danish insulin-dependent diabetic patients and healthy individuals: Reassessment of the influence of (Y DX and insulin-linked polymorphic loci and new splits of DQw3 haplotypes. Dis Markers 7:27, 1989. 12. Brown JH, Jardetzky T, Saper MA, Samraoui B, Bjorkman PJ, Wiley DC: A hypothetical model of the foreign antigen binding site of class II histocompatibility molecules. Nature 322:845, 1988. 13. Maniatis T, Fritsch EF, Sambrook J: Molecular Cloning, A Laboratory Spring Harbor, NY, Cold Spring Harbor Laboratory, 1982.
Manual. Cold
14. Auffray C, Lillie JW, Arnot D, Grossberger D, Kappes D, StromingerJL: Isotypic and allotypic variation of human class II histocompatibility antigen o-chain genes. Nature 308~327, 1984. 15. Okada K, Boss JM, Prentice H, Spies T, Mengler R, Auffray C, Lillie JW, Grossberger D, Strominger JL: Gene organization of the human major histocompatibility complex. Proc Nat1 Acad Sci USA 82:3410, 1985. 16. Fisher RA: The Design
of Experiments.
Edinburgh,
Oliver and Boyd,
1960.
17. Sachs J, Leonard JN, Awad J, McLoskey D, Festenstein H, Hitman GA, Fry L: A comparative serological and molecular study of linear IgA disease and dermatitis herpetiformis. Br J Dermatol 118:759, 1988. 18. Haegert DG, Michaud M, Schwab C, Tansey C, Decary F, Francis G: HLA-DRP, DQo, and -DQp restriction fragment length polymorphisms in multiple sclerosis. Neurosci Res 23:46, 1989. 19. Weeyman AP, Roe C, So A: DXa Antigens 31:1, 1988.
gene polymorphism
in Graves’
disease.
J
Tissue
20. Matsuki K, Maeda H, Juji T, Inoki H, Ando A, Tsuji K, Honda Y: TaqI-generated HLA-DQa! polymorphism in Japanese patients with narcolepsy. Immunogenetics 27~87, 1988. 21. Sheehy MJ, Scharf SJ, Rowe JR, Neme de Gimenez MH, Meske LM, Erlich HE, Nepom BS: A diabetes-susceptible haplotype is best defined by a combination of HLA-DR and -DQ alleles. J Clin Invest 83:830, 1989. 22. Hardy DA, Bell JI, Long EO, Lindsten T, McDevitt HO: Mapping of the class II region of the major histocompatibility complex by pulsed-field gel electrophoresis. Nature 323:453, 1986. 23. Gorski J, Niven MJ, Sachs J, Mach B, Cassell PG, Festenstein H, Awad J, Hitman GA: HLA-DRa, -DXo, andDR@III gene association studies in DR3 individuals. Hum Immunol 20:273, 1987. 24. Berdoz J, Tiercy JM, Rollini P, Mach B, Gorski J: Remarkable sequence conservation of the HLA-DQB2 locus (DXP) within the highly polymorphic DQ subregion of the human MHC. Immunogenetics 29:24 1, 1989.
