Correspondence: M. E. R. Silva, Laboratório de Investigação Médica 18 (LIM-18), Hospital das Clínicas, Faculdade de Medicina da Universidade de São Paulo, Departamento de Endocrinologia, Av. Dr Arnaldo 455, Sala 3324, São Paulo, SP, Brazil 01246-903.
Interleukin (IL)-21 and protein tyrosine phosphatase non-receptor 22 (PTPN22) regulate lymphocyte function and have been implicated in the pathogenesis of autoimmune diabetes. We sequenced the proximal promoter of the IL-21 gene for the first time and analysed the PTPN22 1858T polymorphism in type 1A diabetes (T1AD) patients and healthy controls (HC). We correlated the frequencies of islet and extra-pancreatic autoantibodies with genotypes from both loci. The case series comprised 612 T1AD patients and 792 HC. Genotyping of PTPN22 C1858T was performed on 434 T1AD patients and 689 HC. The −448 to +83 base pairs (bp) region of the IL-21 gene was sequenced in 309 Brazilian T1AD and 189 HC subjects. We also evaluated human leucocyte antigen (HLA) DR3/DR4 alleles. The frequencies of glutamic acid decarboxylase (GAD65), tyrosine phosphatase-like protein (IA)-2, anti-nuclear antibody (ANA), thyroid peroxidase (TPO), thyroglobulin (TG), thyrotrophin receptor autoantibody (TRAb), anti-smooth muscle (ASM) and 21-hydroxylase (21-OH) autoantibodies were higher in T1AD patients than in HC. The PTPN22 1858T allele was associated with an increased risk for developing T1AD [odds ratio (OR) = 1·94; P < 0·001], particularly in patients of European ancestry, and with a higher frequency of GAD65 and TG autoantibodies. HLA-DR3/DR4 alleles predominated in T1AD patients. A heterozygous allelic IL-21 gene variant (g.-241 T > A) was found in only one patient. In conclusion, only PTPN22 C1858T polymorphism and HLA-DR3 and/or DR4 alleles, but not allelic variants in the 5′-proximal region of the IL-21 gene were associated with T1AD risk. Patients with T1AD had increased frequencies of anti-islet-cell, anti-thyroid, anti-nuclear, anti-smooth muscle and anti-21-OH autoantibodies. The C1858T PTPN22 polymorphism was also associated with a higher frequency of GAD65 and TG autoantibodies.
Type 1A diabetes (T1AD), characterized by T cell-mediated autoimmune destruction of pancreatic beta cells, is believed to result from a complex interplay between genetic predisposition, the immune system and environmental factors [1-3]. The major determinant of T1AD genetic susceptibility is conferred by the human leucocyte antigen (HLA)-DR and HLA-DQ alleles [4, 5]. Another important non-HLA gene, the protein tyrosine phosphatase non-receptor 22 (PTPN22), regulates T cell receptor signalling. The PTPN22 C1858T variant, which corresponds to the lymphoid protein tyrosine phosphatase-LYP-Arg620Trp variant associated with pathogenic T cell responses [6-9], has emerged recently as an important risk factor for type 1 diabetes and other autoimmune diseases [10, 11].
Cytokines also play an important role in T1AD pathogenesis. They are the central mediators of inflammation and control innate and adaptive immune responses as well as tissue damage, defence, repair and remodelling . Interleukin (IL)-21, a new member of the type 1 cytokine superfamily and a critical regulator of T and B cell function, is produced by various subsets of CD4+ T cells. IL-21 has been implicated in the pathogenesis of type 1 diabetes on the basis of the knowledge of the immune pathophysiology of a non-obese diabetic (NOD) mouse strain [13, 14]. IL-21 stimulates the proliferation of both T and B cells and terminal differentiation of natural killer (NK) cells, enhances the cytotoxic activity of CD8+ T cells [15-17], counteracts the suppressive effects of regulatory T cells  and stimulates non-immune cells to generate inflammatory mediators . Recently, the importance of IL-21  and its related T helper type 17 (Th17) cells [21, 22] has emerged in the pathogenesis of type 1 diabetes as well in other autoimmune diseases [23, 24] in humans.
The Th-cell-subset-specific expression of the IL-21 proximal promoter is controlled via the action of several transcription factors, including nuclear factor-activated T cells, cytoplasmic 2 (NFATc2), T-bet and leucine-zipper transcription factor Maf (c-MAF) [25, 26]. Due to the pleiotropic effects of IL-21 on immune regulation, it is important to elucidate the genetically driven changes in its function and regulation that might affect the autoimmune process and cause beta cell destruction.
The presence of autoantibodies against islet-cell antigens is the first indication of diabetes development and is a well-established fact. Currently, four autoantibodies are used to predict the development of T1AD: antibodies against glutamic acid decarboxylase (GAD65), tyrosine phosphatase-like protein (ICA512, also termed IA-2), insulin and the recently discovered zinc T8 transporter (ZnT8) [1, 2, 27]. T1AD is also associated frequently with other immune-mediated disorders [27, 28] such as autoimmune thyroiditis [29, 30], Addison's disease , pernicious anaemia [32, 33] and coeliac disease [30, 34].
During the past few years, extensive research has been conducted to predict the occurrences of autoimmune diseases through the detection of organ-specific antibodies in T1D patients [27, 35]. Early detection of antibodies and latent organ-specific dysfunction is important to alert physicians to take appropriate measures to prevent the progression to full-blown disease.
Several autoimmune diseases are related to T1AD and elevated IL-21 expression in both human and animal models, as well as to a high frequency of the PTPN22 C1858T polymorphism. The Brazilian population is one of the most heterogeneous in the world, composed mainly of European (Caucasian descent, 0·771), African (0·143) and Amerindian (Native South American, 0·085) ancestry . We hypothesized that the variants of these genes that regulate immune function would influence not only diabetes risk, but also the expression of other tissue-specific autoantibodies among patients with T1D in a Brazilian population. Therefore, we studied a variant of the PTPN22 gene with a well-documented influence on T cell receptor signalling and diabetes risk, and searched for variants in the proximal promoter region of the IL-21 gene related to autoimmune risk in T1AD patients and healthy controls in São Paulo, Brazil, which has a population with high genetic diversity. HLA-DR3/DR4 alleles were also analysed.
Materials and methods
All T1AD patients satisfied the American Diabetes Association (ADA) classification criteria for type 1A diabetes . This project was approved by the Ethics Committee for Research Project Analysis of Hospital das Clínicas, University of São Paulo School of Medicine. All the samples were collected after the patients were provided with guidance and had signed a consent form.
Autoantibodies against insulin (IAA), glutamic acid decarboxylase (GAD65), tyrosine phosphatase (IA2) and 21-hydroxylase (21-OH) were assessed by radioimmunoassay (RSR Limited, Cardiff, UK). Autoantibodies against thyroid peroxidase (TPO) and thyroglobulin (TG) were evaluated by fluorometry (AutoDELPHIA, Turku, Finland). Anti-nuclear antibody (ANA), anti-liver/kidney microsomal type 1 antibody (LKM1) and anti-smooth muscle (ASM) antibody were quantified using indirect immunofluorescence. Rheumatoid factor (RF) was evaluated using nephelometry, and TSH receptor autoantibody (TRAb) was assessed using iodine radioreceptor assay (RSR Limited).
Sequencing and genotyping
Genomic DNA was extracted by salting-out in blood leucocytes. The region encompassing −448 to +83 base pairs (bp) of the IL-21 gene was amplified and sequenced from samples of 309 Brazilian T1AD patients and 189 control individuals. The following primers were used for the IL-21 gene: (−448) forward: 5′-CCTTATGACTGTCAGAGAGAACA-3′ and (+83) reverse: 5′-CTTGATTTGTGGACCAGTGTC-3′. Direct sequencing of polymerase chain reaction (PCR)-amplified products was performed using an ABI 3100 capillary sequencer (Applied Biosystems, Tokyo, Japan) with the ABI PRISM BigDye Terminator version 3·1 cycle sequencing kit (Applied Biosystems) and analysed using an ABI PRISM 3730 genetic analyser (Applied Biosystems).
Restriction fragment length polymorphism (RLFP) genotyping of PTPN22 C1858T
The following PCR amplification reaction primers were used: PTPN22 forward: 5′-TCACCAGCTTCCTCAACCACA-3′ and PTPN22 reverse: 5′-GATAATGTTGCTTCAACGGAATTT-3′. PCR amplification products were digested enzymatically using the Xcml restriction enzyme (Uniscience-New England BioLabs, Inc., Ipswich, MA, USA), which resulted in a 215-bp product for the CC variant (wild-type); 215-bp, 169-bp and 46-bp products for the CT variant; and 169-bp and 46-bp products for the TT variant. PTPN22 genotyping was performed in 689 controls and 434 T1AD patients. All results were confirmed using an RsaI restriction enzyme assay (Uniscience).
HLA class II typing
HLA class II typing for DRB1 was performed using PCR with One Lambda's SSP™ Generic HLA class II (DRB) DNA typing trays (One Lambda, Canoga Park, CA, USA).
The χ2 or Fisher's exact test was used to investigate the associations between PTPN22 pooled genotypes and categorized variables; Student's t-test, for age comparisons; and the Mann–Whitney U-test, for comparisons between patients and controls related to other parameters. All variants followed the Hardy–Weinberg equilibrium (P > 0·05).
The case series comprised 612 T1AD patients (of whom 81·9% were of European ancestry) who were treated with two or more injections of insulin per day, and 792 healthy individuals (of whom 65·4% were of European ancestry) without any family history of types 1 or 2 diabetes or autoimmune diseases and normal glucose and HbA1c levels.
IL-21 proximal promoter region
A heterozygous allelic variant (g.-241 T > A) was found in the 5′-proximal region of the IL-21 gene in only one patient. This patient was female, aged 30 years, at the onset of disease. She was found to be positive for GAD65 autoantibody (22·8 U/ml) and IA-2 autoantibody (36·9 U/ml). This allelic variant was not found in the other 497 individuals (308 T1AD patients and 189 healthy controls).
C1858T PTPN22 polymorphism
Although the CT and TT genotypes at this locus could be distinguished, only two individuals with the TT genotype were found in this sample (one in the T1AD group and one in the control group). The CT and TT genotypes were pooled into a single class for statistical analyses to avoid classes with very small numbers, referred to as CT/TT. The CT/TT genotype frequency was 18·7% in the T1AD patients and 10·6% in the healthy controls [odds ratio (OR) = 1·94; confidence interval (CI): 1·37–2·73; P < 0·001; Table 1]. The distribution was similar in males (12·7%) and females (14·9%), but was more frequent in individuals of European ancestry (15·4 versus 9·6%; P = 0·0116). When the sample was analysed separately for ancestry, the CT/TT genotype was found to be associated with T1AD risk only in the cohort of European ancestry (OR = 1·811; P = 0·0046). The C1858T PTPN22 polymorphism was not associated with the age of diabetes onset (11·6 ± 6·9 CT/TT versus 11·1 ± 7·3 CC; P = 0·5).
Table 1. The frequency of pooled genotypes at the 1858C>T protein tyrosine phosphatase non-receptor 22 (PTPN22) locus in T1AD patients and healthy controls and the relationship with autoantibodies and clinical data.
The following islet and extra-pancreatic autoantibodies were analysed: anti-insulin (IAA), anti-glutamic acid decarboxylase (GAD65), anti-tyrosine phosphatase (IA2), anti-21-hydroxylase (21-OH), anti-thyroid peroxidase (TPO), anti-thyroglobulin (TG) antibodies, anti-nuclear antibody (ANA), anti-liver/kidney microsomal type (LKM1), anti-smooth muscle (ASM), rheumatoid factor (RF) and TSH receptor antibody (TRAb). With the exception of anti-LKM1 (which was very rare in both the groups) and RF, all other autoantibodies were significantly more frequent in T1AD patients than in the healthy controls (P < 0·001). Islet autoantibodies were the most frequent in T1AD, followed by thyroid autoantibodies and ANA (Table 2; Fig. 1).
Table 2. Frequency of pancreatic and extrapancreatic autoantibodies in T1AD patients and healthy controls.
The C1858T polymorphism was associated with a higher frequency of GAD65 (26·5% versus 15·9%; OR = 1·891; CI: 1·254–2·853; P = 0·003) and TG autoantibodies (22·2% versus 12·4%; OR = 2·023; CI: 1·164–3·513; P = 0·02) in the whole group (T1AD patients plus healthy controls). A subset of T1D patients who had had the disease for more than 10 years showed that this variant was not associated with persistent islet autoantibodies. The C1858T polymorphism was not related to the frequencies or titres of the other analysed autoantibodies.
HLA-DR3 and/or HLA-DR4 alleles
In all, 460 T1AD patients and 700 healthy controls were analysed. The HLA-DR3 and/or HLA-DR4 alleles were more common in T1AD patients (84·1% versus 43%; P < 0·001; OR = −7·027, CI: 5·25–9·406).
In this study, three genetic regions were assessed for associations with T1D in a Brazilian population. The populations of this country are highly heterogeneous and composed of an admixture of European, African and native Amerindian descendants. The analyses included studies of the 5′-proximal regions of the IL-21 gene and the PTPN22 C1858T variant, and their association with autoantibodies as well as the HLA-DR and DQ alleles.
A heterozygous single nucleotide polymorphism (g.-241 T > A) was detected in the 5′-proximal region (−448 to +83) of the IL-21 gene in a T1AD patient; this polymorphism was not present in the healthy controls or reported in databases. This variant is located outside the known NFATc2 and T-bet controller regions , and does not affect any known transcription factor-binding sites. The patient's sister, who does not have diabetes, showed the same allelic variant. A functional study might be necessary to define the effect of this variant on diabetes susceptibility.
No other polymorphism in the proximal IL-21 gene promoter region was observed in either group, including the single nucleotide polymorphism (SNP) rs77935281 GT, which was reported previously within this region in databases (http://www.ensembl.org). Thus, sequence variants are rare in the 5′-proximal region of the IL21 gene, suggesting that it has a biologically important function or that it is a relatively new molecule from an evolutionary viewpoint .
Conversely, a higher frequency of the C1858T PTPN22 gene polymorphism was observed in T1AD patients: CT/TT genotypes in 18·7% versus 10·6% of controls (OR = 1·94; CI: 1·37–2·73; P < 0·001). This association has been shown across different Caucasian populations, in which the frequency of the CC/CT-pooled genotypes was found to range from 26·8% (United States)  to 42·1% (Finland) . However, the *T1858 allele is almost absent in the African American and Asian populations [40, 41]. In our study, the frequency of *T1858 allele was lower than that in the European ancestry samples, due probably to our ethnic heterogeneity, which includes African, Amerindian, Asian and European descendants. In accordance with this, the C1858T allele frequency was higher in the European descendants (15·4%) than in those of other ancestries (9·6%; P = 0·0116). The risk of T1AD was conferred by the CT/TT genotypes in the European ancestry cohort (OR = 1·811; P = 0·0046). This effect was not significant in our subsample of patients of non-European ancestry (OR = 1·482; P = 0·383), suggesting that ethnicity affected the T1AD susceptibility conferred by this variant. However, the small number of healthy controls (n = 242) and T1AD patients (n = 81) of non-European ancestry is lower than the necessary sample size defined previously to have 80% statistical power (n = 276), limiting further conclusions about the ethnic effect. The CT and TT genotypes were pooled to avoid classes with very small numbers, because only two individuals had the TT genotype (one in the case group and one in the control group). This type of pooling was also used in other studies. Therefore, distinguishing between the dominant or co-dominant model of inheritance for the C and T alleles at this locus and their effect on the studied variables is difficult.
However, as expected, the effect of ethnicity was not observed in the HLA-DR3 /DR4 allele frequency, because these alleles usually confer high susceptibility to T1AD in all populations [4, 5].
The association of C1858T polymorphism with T1AD and other autoimmune diseases was proposed to depend upon the pathogenic LYP-W620 variant that shows increased phosphatase activity and is a gain-of-function inhibitor of T cell signalling . In our study, this polymorphism was associated with an increased frequency of GAD65 autoantibody and TG autoantibody when the entire cohort (T1AD patients + healthy controls) was considered. Although the T1AD patients had higher frequencies of pancreatic and non-pancreatic autoantibodies than the healthy controls, there was no association between the *T1858 allele and other islet and organ-specific autoantibodies. Thus, although the frequency of organ-specific autoantibodies in our population was similar to what has been reported previously for Caucasians, this frequency did not depend on the presence of the T1858 allele, except for the autoantibodies against the pancreas and thyroid.
The C1858T PTPN22 polymorphism was associated with T1AD susceptibility and autoimmune thyroid disease. Autoantibodies specific to other organs and tissues were frequent in T1AD carriers, predominantly the thyroid glands. The 1858T PTPN22 polymorphism was associated with a higher frequency of GAD65 and TG autoantobody. Allelic variants in the 5′-proximal region of the IL-21 gene were not related to T1AD susceptibility and other autoimmune diseases. The HLA-DR3 and/or DR4 alleles predominated in T1D patients.
We thank Dr George S. Eisenbarth of the Barbara Davis Center for review of the manuscript. We thank Greci S. Paula, Adriana Rosa, Maria de Fátima Sanches and Maria José Pegoraro of the Laboratório de Investigação Médica LIM 18 and to LIM-25, LIM-42, LIM-56 and Hospital das Clínicas da Faculdade de Medicina da USP for technical assistance. This work was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo-FAPESP, process 2006/06390-1.
All authors declare they have no conflicting interests.