Glutamic Acid Decarboxylase 65 and Islet Cell Antigen 512/IA-2 Autoantibodies in Relation to Human Leukocyte Antigen Class II DR and DQ Alleles and Haplotypes in Type 1 Diabetes Mellitus

Abstract

The frequencies of autoantibodies against glutamic acid decarboxylase 65 (GAD65) and islet cell antigen (ICA) 512/IA-2 (512/IA-2) are functions of the specific human leukocyte antigen (HLA) in type 1 diabetes mellitus (T1D). We investigated the association of HLA class II (DR and DQ) alleles and haplotypes with the presence of GAD and IA-2 autoantibodies in T1D. Autoantibodies were tested in 88 Tunisian T1D patients and 112 age- and gender-matched normoglycemic control subjects by enzyme immunoassay. Among T1D patients, mean anti-GAD antibody titers were higher in the DRB1*030101 allele (P < 0.001), together with the DRB1*030101/DQB1*0201 (P < 0.001) and DRB1*040101/DQB1*0302 (P = 0.002) haplotypes, while lower anti-GAD titers were associated with the DRB1*070101 (P = 0.001) and DRB1*110101 (P < 0.001) alleles and DRB1*070101/DQB1*0201 (P = 0.001) and DRB1*110101/DQB1*030101 (P = 0.001) haplotypes. Mean anti-IA-2 antibody titers were higher in the DRB1*040101 allele (P = 0.007) and DRB1*040101/DQB1*0302 (P = 0.001) haplotypes but were lower in the DRB1*110101 allele (P = 0.010) and the DRB1*110101 (P < 0.001) and DRB1*110101/DQB1*030101 (P = 0.025) haplotypes. Multinomial regression analysis confirmed the positive association of DRB1*030101 and the negative association of DRB1*110101 and DQB1*030101, along with the DRB1*070101/DQB1*0201 and DRB1*110101/DQB1*030101 haplotypes, with anti-GAD levels. In contrast, only the DRB1*040101/DQB1*0302 haplotype was positively associated with altered anti-IA-2 titers. Increased GAD65 and IA-2 antibody positivity is differentially associated with select HLA class II alleles and haplotypes, confirming the heterogeneous nature of T1D.

INTRODUCTION

Type 1 diabetes mellitus (T1D) is an endocrine disease characterized by autoimmune destruction of pancreatic β islet cells (2, 27). Several autoantigens have been implicated in triggering this process, including insulin; the 65-kDa isoform of glutamic acid decarboxylase (GAD) (2, 19, 26), an enzyme involved in the synthesis of the inhibitory neurotransmitter γ-aminobutyric acid in pancreatic β islet cells; and islet cell antigen 2 (IA-2), a tyrosine phosphatase expressed in β islet cells (12). Up to 90% of newly diagnosed T1D cases are positive for anti-IA-2 and/or anti-GAD antibodies, compared to the very low prevalence of these autoantibodies in nondiabetic control populations (∼1%) (27, 28). Most peptides derived from GAD and IA-2 autoantigens can bind to T1D-predisposing and -protective HLA molecules, although some exceptions were observed (8). Functionally, T cell-proliferative responses from T1D patients identified GAD65 as a likely candidate in the development of anti-β islet cell immunity (16).

We and others have confirmed the association of select HLA class II alleles and haplotypes with increased risk of T1D (1, 21, 29) and have identified both susceptible (DRB1*030101, DRB1*030101-DQB1*0201, and DRB1*040101-DQB1*0302) and protective (DRB1*070101, DRB1*110101, DRB1*070101-DQB1*0201, and DRB1*110101-DQB1*030101) alleles and haplotypes at the class II DR and DQ loci. Homozygosity for DRB1*030101-DQB1*0201 represented the highest-risk genotype, while DRB1*070101-DQB1*0201 homozygosity was associated with significantly decreased risk. Several studies addressed the likely relationship between HLA risk loci and T1D-associated autoantibodies in both children and adults with newly diagnosed T1D. The presentation of β islet cell autoantigens by high- and low-risk alleles needs to be studied in order to elucidate the mechanism underlying the effect of HLA class II polymorphism on disease risk (2, 26).

It was suggested that high-risk HLA antigens function in binding and later presenting autoantigens (including GAD65) to autoreactive T cells (6, 16, 26). This was highlighted by the HLA-DR2- and HLA-DR4-restricted T cell lines from new-onset T1D patients, which bind to naturally processed GAD65 (12, 23); no similar T cell clones could be generated from healthy controls. While lymphocytes could be generated to synthetic GAD65 peptides from T1D patients (3), their reactivity to naturally processed GAD65 could not be demonstrated. In this study, we investigate the association between anti-GAD and anti-IA-2 antibody titers and HLA class II (DR and DQ) alleles and haplotypes.

MATERIALS AND METHODS

Subjects.

The study subjects comprised 88 unrelated T1D patients (44 males and 44 females; mean age, 16.4 ± 7.7 years). The diagnosis of T1D was based on clinical features and laboratory data. All T1D patients were ketosis prone, lacked endogenous insulin secretion, and were dependent on insulin for controlling hyperglycemia. T1D patients were not obese, were free of any concomitant complications, and were not receiving additional treatment at the time of blood collection. Patients with other forms of diabetes (latent autoimmune diabetes of adults, maturity onset diabetes of the young, or type 2 diabetes) were excluded. Control subjects consisted of 112 university students and healthy children (65 males and 47 females; age, 28.2 ± 5.8 years) who had normal glucose tolerance and no family history of T1D or other autoimmune diseases. All patients and control subjects were Tunisian Arabs, were from central Tunisia, and were asked to sign a consent form according to the study protocol, and all institutional ethics requirements were met.

HLA-DRB1 and -DQB1 genotyping.

HLA-DRB1 and -DQB1 gene alleles were analyzed using the PCR sequence-specific-priming (SSP) technique, using the Micro SSP Generic HLA Class II (DRB/DQB) DNA Typing kit (lot 05A), according to the manufacturer's specifications (One Lambda, Thousand Oaks, CA). PCR products were analyzed on ethidium bromide-stained agarose gels. HLA allele nomenclature was as previously reported (15). In total, 16 DRB1 and 7 DQB1 alleles were tested.

Autoantibody screening.

IA-2 and GAD-65 autoantibodies were measured at the time of initial diagnosis for T1D patients. Serum samples were obtained from all participants and were stored as small aliquots at or below −40°C. Anti-IA-2 and anti-GAD-65 antibodies were measured on two separate occasions by enzyme-linked immunosorbent assay (ELISA) (Kronus Inc., Boise, ID). Cutoff values for antibody positivity were based on the 99th percentile of antibody levels obtained in nondiabetic controls. The results were expressed as antibody titers (arbitrary units [AU]/ml) or as the percent antibody positive out of the total.

Statistical analysis.

Statistical analysis was performed on SPSS v. 17.0 (SPSS Inc., Chicago, IL). Allele frequencies were determined by the gene-counting method, using HLAStat 2000 software, and haplotype frequencies were determined by the maximum-likelihood method. P values were corrected for the number of different alleles or haplotypes tested (Pc) using the Bonferroni inequality method [Pc = 1 − (1 − P)ⁿ]. Data were expressed as P values, odds ratios (OR), and 95% confidence intervals (CI) between patients and controls. The Spearman correlation coefficient was used to determine correlation between the level of autoantibodies and HLA alleles and haplotypes. Antibody titers were expressed as mean ± standard deviation (SD); differences between cases and controls were made using a two-tailed Student t test. Logistic regression analysis was performed in order to determine the OR and 95% CI associated with T1D risk, taking the controls as the reference group. Statistical significance was set at a P value of <0.05.

RESULTS

HLA allele and haplotype frequencies.

Significant DRB1 allelic differences were seen between T1D patients and controls. These comprised DRB1*030101 (Pc = 0.006), which was higher, and DRB1*070101 (Pc = 0.003) and DRB1*110101 (Pc = 0.027), which were lower among patients (Table 1). Similarly, significant DQB1 allelic differences were seen at the DQB1 locus, which comprised DQB1*0302 (Pc = 0.012), which was higher, and DQB1*030101 (Pc = 0.007) and DQB1*060101 (Pc = 0.041), which were lower among patients than among control subjects (Table 1). In addition, the frequencies of DRB1*030101-DQB1*0201 (Pc < 0.001) and DRB1*040101-DQB1*0302 (Pc = 0.010) were higher, while those of DRB1*070101-DQB1*0201 (Pc = 0.015) and DRB1*110101-DQB1*030101 (Pc = 0.036) were lower in T1D patients than in control subjects, thereby conferring T1D susceptibility and protection, respectively, on these haplotypes (Table 1).

Table 1.

HLA-DRB1* and -DQB1* allele distribution

Allele/haplotype	Allele/haplotype frequency		Pa	Pcb
	Patients (n = 88)	Controls (n = 112)
DRB1*030101	0.3641	0.147	3.6 × 10−4	0.006
DRB1*040101	0.239	0.116	7.5 × 10−3	0.120
DRB1*070101	0.068	0.214	1.8 × 10−4	0.003
DRB1*090102	0.028	0.000	0.023	0.362
DRB1*100101	0.023	0.063	0.051	0.816
DRB1*110101	0.028	0.107	1.7 × 10−3	0.027
DQB1*0201	0.426	0.290	0.054	0.377
DQB1*030101	0.097	0.201	0.001	0.007
DQB1*0302	0.216	0.094	0.002	0.012
DQB1*060101	0.080	0.179	0.006	0.041
DRB1*030101-DQB1*0201	0.287	0.065	<0.001	<0.001
DRB1*040101-DQB1*0302	0.184	0.073	0.001	0.010
DRB1*070101-DQB1*0201	0.068	0.177	0.002	0.015
DRB1*110101-DQB1*030101	0.023	0.097	0.005	0.036

^aDetermined by Fisher's exact test; boldface indicates significant differences.

^bPc is the corrected P value for the number of alleles tested, calculated using the Bonferonni method.

Correlation studies.

We examined the functional attributes of HLA alleles and haplotypes on autoantibody levels among T1D patients. Table 2 summarizes the correlation between the level of autoantibodies and HLA alleles and haplotypes among T1D patients positive for a specific allele and haplotype; patients negative for that allele or haplotype served as controls. Anti-GAD levels were positively correlated with the DRB1*030101 allele (r² = 0.378; P < 0.001) and the DRB1*030101/DQB1*0201 (r² = 0.572; P < 0.001) and DRB1*040101/DQB1*0302 (r² = 0.284; P = 0.001) haplotypes. Anti-GAD titers were negatively associated with the protective DRB1*070101 (r² = −0.322; P = 0.001), DRB1*110101 (r² = −0.345; P < 0.001), and DQB1*030101 (r² = −0.294; P = 0.002) alleles, along with the DRB1*070101/DQB1*0201 (r² = −0.328; P = 0.001) and DRB1*110101/DQB1*030101 (r² = −0.346; P = 0.001) haplotypes. In comparison, anti-IA-2 levels were positively correlated only with the DRB1*040101/DQB1*0302 (r² = 0.312; P = 0.013) haplotype but were negatively correlated with the DRB1*110101 allele (r² = −0.329; P = 0.009) and the DRB1*110101/DQB1*030101 (r² = −0.343; P = 0.006) haplotype.

Table 2.

Correlation studies^c

Allele/haplotype	Anti-GAD		Anti-IA2
	r2a	P	r2	P
Allelesb
DRB1*030101	0.378	6.5 × 10−5	0.074	0.563
DRB1*040101	0.122	0.211	0.219	0.084
DRB1*070101	−0.322	0.001	−0.185	0.148
DRB1*110101	−0.345	3.2 × 10−4	−0.329	0.009
DQB1*030101	−0.294	0.002	−0.220	0.083
DRB1*060101	−0.129	0.189	−0.009	0.945
Haplotypes
DRB1*030101/ DQB1*0201	0.572	1.5 × 10−10	0.104	0.419
DRB1*040101/ DQB1*0302	0.284	0.003	0.312	0.013
DRB1*070101/ DQB1*0201	−0.328	0.001	−0.176	0.168
DRB1*110101/ DQB1*030101	−0.346	2.8 × 10−4	−0.343	0.006

^aSpearman correlation coefficient.

^bOnly alleles and haplotypes significantly associated with altered risk of T1DM in Tunisians (7).

^cBoldface indicates statistical significance.

Anti-GAD and anti-IA-2 antibody titers.

The results in Table 3 show that the mean anti-GAD antibody titers were higher in T1D patients positive for the DRB1*030101 allele (P < 0.001) and the DRB1*030101/DQB1*0201 (P < 0.001) and DRB1*040101/DQB1*0302 (P = 0.002) haplotypes. Mean anti-GAD antibody titers were lower in carriers of the DRB1*070101 (P = 0.001) and DRB1*110101 (P < 0.001) alleles and in the DRB1*070101/DQB1*0201 (P = 0.001) and DRB1*110101/DQB1*030101 (P = 0.001) haplotypes. In comparison, mean anti-IA-2 antibody titers were higher in T1D patients positive for the DRB1*040101 (P = 0.007) allele and the DRB1*040101/DQB1*0302 (P = 0.001) haplotypes but were lower in carriers of the DRB1*110101 (P = 0.010) and DRB1*110101 (P < 0.001) alleles and the DRB1*110101/DQB1*030101 (P = 0.025) haplotype.

Table 3.

Antibody titers in HLA class II alleles and haplotypes

Allele/haplotype	Antibody titera
	Anti-GAD				Anti-IA2
	Positive	Negative	Z	P	Positive	Negative	Z	P
DRB1*030101	73.10 ± 59.51	28.16 ± 17.08	3.87	<0.001	30.90 ± 24.52	21.89 ± 20.49	0.58	0.521
DRB1*040101	64.83 ± 52.68	52.93 ± 40.52	1.26	0.209	63.44 ± 56.44	25.20 ± 25.34	1.73	0.007
DRB1*070101	37.34 ± 21.42	60.08 ± 31.45	−3.30	0.001	20.94 ± 11.27	42.08 ± 31.45	−1.45	0.146
DRB1*110101	11.24 ± 4.35	58.62 ± 43.92	−3.51	<0.001	27.42 ± 12.80	58.48 ± 28.71	−2.57	0.010
DQB1*030101	17.22 ± 14.03	56.79 ± 39.74	−3.01	0.001	13.12 ± 12.87	29.77 ± 27.29	−1.73	0.083
DRB1*060101	26.68 ± 24.27	53.22 ± 49.19	−1.32	0.031	16.57 ± 15.21	29.41 ± 27.52	−0.07	0.365
DRB1*030101/DQB1*0201	114.72 ± 42.99	24.09 ± 25.39	2.77	<0.001	43.61 ± 44.71	29.14 ± 31.79	1.01	0.257
DRB1*040101/DQB1*0302	111.29 ± 49.97	52.12 ± 38.92	0.786	0.007	69.22 ± 53.07	25.17 ± 25.01	2.52	0.001
DRB1*070101/DQB1*0201	19.34 ± 15.36	60.24 ± 56.90	−3.24	0.001	25.17 ± 13.81	40.98 ± 30.42	−0.92	0.359
DRB1*110101/DQB1*030101	17.32 ± 15.56	98.30 ± 50.10	−3.25	0.001	15.89 ± 10.52	31.83 ± 23.81	−2.21	0.025

^aPositive, patients carrying the specific allele/haplotype; negative, other patients; Z, Z score. The values are mean ± SD antibody titers (AU/ml); differences in antibody titers between the patient groups were determined using a two-tailed Student t test.

Regression analysis.

The selective association of DRB1 and DQB1 alleles and haplotypes with altered anti-GAD and anti-IA-2 elevated antibody titers were confirmed by regression analysis after controlling for potential covariates. Multinomial regression analysis confirmed the positive association of the DRB1*030101 allele and the DRB1*030101/DQB1*0201 haplotype and the negative association of the DRB1*110101 and DQB1*030101 alleles and the DRB1*070101/DQB1*0201 and DRB1*110101/DQB1*030101 haplotypes with altered anti-GAD levels. In contrast, only the DRB1*040101/DQB1*0302 haplotype was positively associated with altered anti-IA-2 titers (Table 4).

Table 4.

Matched odds ratios associated with antibody titers in class II alleles and haplotypes^c

Allele/ haplotypea	Anti-GAD		Anti-IA2
	P	aOR (95% CI)b	P	aOR (95% CI)
DRB1*030101	<0.001	5.23 (2.21–12.40)	0.582	1.37 (0.60–3.09)
DRB1*040101	0.478	1.52 (0.61–3.78)	0.100	2.31 (0.95–5.52)
DRB1*070101	0.016	0.30 (0.11–0.82)	0.424	0.56 (0.21–1.70)
DRB1*110101	<0.001	0.75 (0.66–0.86)	0.269	0.34 (0.09–1.67)
DQB1*030101	0.006	0.21 (0.07–0.67)	0.424	0.56 (0.21–1.70)
DRB1*060101	0.070	0.39 (0.14–1.06)	0.144	0.33 (0.11–1.27)
DRB1*030101/DQB1*0201	<0.001	15.76 (5.83–42.57)	0.151	1.99 (0.87–4.44)
DRB1*040101/DQB1*0302	0.102	2.45 (0.86–7.02)	<0.001	5.60 (2.01–14.33)
DRB1*070101/DQB1*0201	0.006	0.21 (0.07–0.67)	0.168	0.28 (0.08–1.40)
DRB1*110101/DQB1*030101	<0.001	0.75 (0.66–0.86)	0.380	0.54 (0.20–1.63)

^aOnly alleles and haplotypes significantly associated with altered risk of T1DM in Tunisians (7).

^bThe reference group was patients negative for the specific allele or haplotype. aOR, adjusted odds ratio, controlled for gender, age of onset, and fasting glucose.

^cBoldface indicates statistical significance.

DISCUSSION

T1D is an organ-specific autoimmune disease resulting in T cell-mediated destruction of pancreatic β islet cells. T1D is also distinguished by the presence of a number of autoantigens (3, 6, 12, 27). GAD and IA-2 are two of the major and best-characterized autoantigens. Several studies have demonstrated that HLA-DQ and -DR alleles influence T1D susceptibility (1, 20, 21, 29), and T1D-predisposing or -protective HLA-DQ and -DR alleles were identified (1, 29). Apart from the regulation of the autoreactive T cell repertoire by susceptible and protective HLA molecules, presentation of autoantigenic peptides by specific HLA molecules or induction of regulatory T cells (Treg) explains, at least in part, the modulatory capacity of HLA variants on the overall T1D risk (8).

This study is the first report on the association of HLA class II alleles and haplotypes and antibody titers with T1D in Tunisia. We previously demonstrated that the contribution of HLA haplotypes to T1DM genetic susceptibility among Tunisians depends on specific HLA class II haplotypes (29). The results showed that anti-GAD antibody levels were higher in DRB1*030101 allele carriers, in agreement with recent findings in the Saudi population (14) but in partial agreement with an earlier study on Taiwanese subjects, in which anti-GAD positivity was associated with DR3 and DR4 (4). These differences are likely attributable to racial differences in the contribution of HLA class II alleles and haplotypes to T1D pathogenesis (2, 8, 11). Our study demonstrated that the prevalence and increased titer of anti-GAD antibodies are associated with DRB1*030101 and DRB1*030101/DQB1*0201, while anti-IA-2 positivity was associated with the DRB1*040101/DQB1*0302 haplotype. Studies on Swedes reported the same association (7). This association between IA-2 and the DRB*04 allele was reported in previous studies (11, 13, 22, 24).

Multinomial regression analysis revealed a positive association of the haplotype DRB1*040101/DQB1*0302 with altered anti-IA-2 titers in our population. These results are concordant with previous studies on Canadian (24), Japanese (20), German (10), Finnish (27, 28), Italian (13), and Belgian (5) populations that showed that the prevalence and level of IA-2 autoantibody are increased in patients with HLA DR4/DQ8. However, IA-2 was negatively associated with DRB1*110101/DQB1*030101 in titers and in prevalence. Graham et al. showed the same association among Swedes (7).

One of the characteristics of major histocompatibility complex (MHC) genetic associations with T1D is that primary susceptibility is determined by HLA-DQB1*0302, with penetrance of clinical disease significantly influenced by the specific subtype of DR4 present on the same haplotype (30). It is tempting to speculate that this is based on the selection of specific dominating epitopes favored during disease progression (12, 21, 30). Other genetic modifiers of susceptibility, which are also attributed to HLA molecules, the dominant protection conferred by HLA-DQ alleles, may also have a primary mechanistic influence by biasing or adding to the specific epitope recognition cascade (12, 17, 30).

In this way, T1D can be viewed as a disease of epitope selection and immunological focusing, whereby autoimmunity may progress toward tissue destruction or protection (17, 30). While specific HLA alleles and haplotypes clearly impact susceptibility to T1D, it is likely that anti-GAD positivity and T1D-susceptible locus HLA types translate into increased T cell reactivity. HLA class II molecules expressed on antigen-presenting cells have critical roles for the activation of helper T cells (mainly CD4⁺ T cells) and may influence antibody status in the periphery. This was highlighted by the findings of Itoh et al., in which HLA DR9/X-positive T1D patients had significantly higher numbers of GAD-reactive IFN-γ-producing T cells than those with other loci (9).

HLA polymorphism affects autoimmune responses according to the binding capacities of antigenic peptides, and the repertoire of T cell receptors of reactive T cells (25) and different HLA-DR/DQ molecules might have different binding affinities to disease-associated peptides (18). These facts may explain our results showing unique and different HLA associations with autoantibodies from the Caucasoid populations. In conclusion, our study of T1DM among Tunisians reveals that both GAD and IA-2 antibodies and the HLA-DR and -DQ alleles are critical in determining the risk for the disease.