Differential Splicing of the IA-2 mRNA in Pancreas and Lymphoid Organs as a Permissive Genetic Mechanism for Autoimmunity Against the IA-2 Type 1 Diabetes Autoantigen

IA-2 is a tyrosine phosphatase-like protein with a single transmembrane (TM) region (residues 577–600) and extracellular and intracellular domains (Fig. 1) (1). IA-2 is enriched in the secretory granules of pancreatic islet cells and neuroendocrine cells, including peptidergic neurons, pituitary cells, and adrenal chromaffin cells (2). It is unclear why the immune system reacts against IA-2 as part of the autoimmune responses associated with type 1 diabetes (3–6). The IA-2 gene is also expressed in the human thymus (7) in a manner similar to other genes coding for self-molecules with tissue-restricted expression, including insulin and glutamic acid decarboxylase (GAD) (7–11), two other autoantigens in type 1 diabetes. Self-antigen expression in thymus may play a key role in promoting immunological self-tolerance, and variation in such expression may have dramatic effects on tolerance. We have also obtained evidence that the self-antigens are detected not only in thymus but in peripheral lymphoid organs as well (spleen and lymph nodes) (12).

An alternatively spliced variant of the IA-2 transcript has been recently discovered through the sequencing of the ICA512.bdc clone (13) derived from a human pancreas library routinely translated in vitro as source of antigen in a specific assay for the detection of type 1 diabetes–associated IA-2 autoantibodies. This variant transcript lacks exon 13 (Δ exon 13), which codes for 73 amino acids (aa 557–629) encompassing the TM and JM domains of the IA-2 protein (Fig. 1). Several epitopes recognized by autoimmune responses in type 1 diabetes overlap with these domains; thus, we hypothesized that splicing of the IA-2 gene could be differentially regulated in lymphoid organs and pancreas, because such differences may influence immune responsiveness to IA-2. We investigated IA-2 expression in islets, thymus, and spleen from nondiabetic human tissue donors (no tissues were available from patients) of different ages (from 21 weeks of gestation to 52 years of age) and from diverse racial groups (Table 1). IA-2 mRNA expression and splicing were evaluated by reverse transcriptase–polymerase chain reaction (RT-PCR) using primers that amplify both full-length and Δ exon 13 transcripts, and the identity of the products was verified by sequencing. All nine pancreatic samples (four islet and five whole pancreas samples) expressed full-length IA-2 message, and five of the nine also expressed the Δ exon 13 transcript (Table 2). IA-2 mRNA was also detected in lymphoid organs, including 9 of 15 thymus and 11 of 15 spleen specimens (Table 2, Fig. 2A). Consistent with previous literature, not all of the lymphoid tissues tested expressed IA-2 message (7,10). This finding may have several explanations, including RNA degradation and genetic variability of trans- or cis-acting factors that may affect the expression of IA-2. Of the lymphoid tissues expressing IA-2, all thymus (9 of 9) and spleen (11 of 11) specimens exclusively expressed the Δ exon 13 transcript. The above also included five cases for which we could simultaneously evaluate IA-2 expression in pancreas, thymus, and spleen (cases 3222, 3236, 3444, 3446, and 3450). The analysis of tissues from the same subject also indicates that the observed differences between pancreas and lymphoid organs reflect tissue-specific splicing and not individual variation in the regulation of IA-2 splicing in lymphoid organs. We confirmed the above findings using a second primer pair and also detected another transcript intermediate in size between the full-length and the Δ exon 13 transcript (Fig. 2C). Sequence analysis confirmed that this is another alternatively spliced IA-2 transcript in which 129 bp of exon 14 are spliced out, resulting in the deletion of 43 amino acids (aa 653–695) in the intracellular domain (Figs. 1 and 2C and Table 3). This transcript was detected in 4 of 9 pancreatic samples compared with 0 of 11 thymic samples and 1 of 15 samples from spleens (very weak signal, case 3222) (Table 2). Thus, our findings show that IA-2 splicing results in the exclusive expression of the Δ exon 13 transcript in lymphoid organs, but they also show that individual variation in IA-2 splicing and expression exists in pancreas. Of note is the observation that individual variation was also found in five adrenal samples: three of five adrenal samples expressed full-length IA-2 mRNA, and four of five expressed the exon 13 variant; thus, two of five adrenal samples tested exclusively expressed the exon 13 variant (however, none of the adrenal samples expressed the exon 14 variant [data not shown]). Moreover, two brain samples expressed all three variants (data not shown). The presence of all IA-2 variants in pancreas, adrenal, and brain samples suggests that exclusive expression of the Δ exon 13 transcript may be typical of lymphoid tissues.

Because both alternatively spliced transcripts are read in frame downstream of the alternative splicing sites, they should be translated into protein. To verify the in vivo translation of the Δ exon 13 transcript, we stained frozen thymus and spleen sections with two rabbit antibodies recognizing either intracellular (aa 601–979) or extracellular portions (aa 389–576) of IA-2 (2). Distinct cells in thymus (Fig. 2D) and spleen (Fig. 2E) were specifically stained with both sera, demonstrating that IA-2 mRNA is translated into protein in lymphoid organs. Because the Δ exon 13 transcript is essentially the only IA-2 message detected in lymphoid organs, the staining observed can only derive from the translation of the Δ exon 13 transcript into protein. The IA-2–positive cells observed correspond to a subset of apparently tolerogenic self-antigen–presenting cells that we have recently characterized in lymphoid organs (12), and similar cells with tolerogenic function exist in the mouse thymus (9). This finding further supports the concept that differential expression may affect immune responsiveness to IA-2. We also performed Western blot analysis to verify that the alternative IA-2 transcripts are translated into proteins in the pancreas. Consistent with the mRNA expression studies, three bands of the predicted size for full-length IA-2 and the two splicing variants were detected by Western blotting pancreas protein extracts (Fig. 2F) with a rabbit antibody recognizing the extracellular domain (aa 389–575) (14).

We demonstrated that alternative splicing is responsible for differential IA-2 expression in pancreas compared with lymphoid organs. The exclusive expression of the Δ exon 13 transcript in lymphoid organs and the finding of individual variation in the pancreatic expression of the IA-2 variants suggest that differential IA-2 splicing in these tissues may be a permissive mechanism for the potential development of autoimmunity against IA-2. Such differences may affect immune responsiveness to specific epitopes and help explain why IA-2 and not many other islet proteins become targets of autoimmunity in type 1 diabetes. We speculate that immunological tolerance to linear or conformational epitopes typical of the full-length protein or of the Δ exon 14 variant may not be achieved if these are expressed in islets but not in thymus and spleen, where they appear to be expressed by tolerogenic self-antigen–presenting cells. The specific lack of expression of the TM/JM domains (exon 13) in lymphoid organs helps explain why epitopes from these domains are often targeted by autoimmune responses in type 1 diabetes. Indeed, autoantibodies against IA-2 epitopes encoded by exons 13 and 14 have been reported, and in some patients they can precede the appearance of autoantibodies against other intracellular epitopes (epitope spreading). In particular, the 605–682 and 605–620 epitopes (Fig. 1) are early autoantibody targets and are recognized by 56% of patients’ sera (15–18). Two additional epitopes exist in the JM domain between residues 601–642 (Fig. 1), and a third epitope extends to the region preceding the 601 residue (poster presentation [17]). Moreover, type 1 diabetes in children is reportedly more strongly associated with autoantibodies against JM rather than intracellular epitopes (15). Only a few T-cell epitopes have been identified because of the limitations of current T-cell assays, which include poor reproducibility, limited ability to distinguish patients from control subjects, and the sometimes arbitrary choice of overlapping peptides to measure T-cell reactivity (19–23). Moreover, T-cell reactivity against IA-2 epitopes encoded by exons 13 and 14 or against epitopes generated by alternative splicing of these exons has not yet been analyzed. However, there is evidence that the HLA-DR4–restricted naturally processed 654–674 epitope (exon 14) is recognized by autoreactive T-cells (24). Although the Δ exon 13 IA-2 transcript exclusively expressed in lymphoid organs contains this sequence, the lack of the TM domain may affect its processing and presentation in lymphoid organs. It is also possible that immunity against exon 14 epitopes may be driven by the islet-specific expression of the Δ exon 14 protein lacking amino acid residues 653–695. In theory, partial sequence sharing and cross-reactivity between Δ exon 14 and full-length proteins may result in a spreading of the immune response to exon 14 epitopes expressed by the full-length molecule.

Notably, all of the 25 subjects in this study were nondiabetic tissue donors who were randomly selected from the general population and who belonged to diverse races and age-groups. Because of its common occurrence in the population, differential IA-2 splicing and expression may help explain the reported difficulties in distinguishing patients from unaffected subjects when studying stimulated T-cell reactivity in vitro (19,20). In fact, measurable responses could be elicited even from normal subjects if the population is not fully tolerant to IA-2, as suggested by our findings. The observation that only minorities of individuals develop spontaneous autoimmune responses against IA-2 and diabetes may have several explanations. First, IA-2 autoantibodies are usually detected during the later stages of the prodromic period, suggesting that IA-2 autoimmunity may be secondary to inflammatory damage taking place during the earlier stages of the disease process. This may also be one of the reasons why type 1 diabetic patients do not apparently react against other tissues expressing IA-2 (i.e., adrenal), because TM and intracellular IA-2 epitopes may be unavailable to the immune system without inflammation and tissue damage. Second, other genetic factors (e.g., genes coding for HLA antigens, cytokines, etc.) and environmental exposures may be necessary to trigger autoimmunity. Consistent with our findings and this interpretation is the observation that differential splicing affects the expression of the proteolipid protein autoantigen in brain and thymus, which in turn affects susceptibility to experimental autoimmune encephalomyelitis, a mouse model of multiple sclerosis (25,26). Although differential splicing is similarly observed in both susceptible and control strains, only the susceptible mice bearing a predisposing major histocompatibility complex develop disease. Third, the observed individual variation in pancreatic IA-2 splicing and expression (only four of nine pancreatic samples expressed the Δ exon 14 variant that was not expressed in lymphoid organs) may also be a factor because individuals expressing the Δ exon 14 variant in the pancreas may be more prone to develop autoimmune responses.

In conclusion, we suggest that differential splicing of the IA-2 gene in pancreas compared with lymphoid organs could affect the immune system reactivity to IA-2 and help explain why IA-2 is an autoantigen in type 1 diabetes. Similar to the parent-of-origin effects affecting insulin gene expression in thymus (7,8) and peripheral lymphoid organs (27), differential IA-2 splicing appears to function as a mechanism regulating gene expression independent of inherited alleles at the insulin and IA-2 loci. Although recent investigations had excluded linkage with IA-2 polymorphisms (28), our findings show that expression studies for selected candidate genes in tissues relevant to the disease process can help in dissecting the complex genetics of a multifactorial disease such as type 1 diabetes.