Mapping of novel autoreactive epitopes of the diabetes-associated autoantigen IA-2

SUMMARY

IA-2, a member of the tyrosine phosphatase family, has been identified as a dominant autoantigen in type 1 diabetes. To define humoral IA-2 epitopes, we generated a panel of IA-2 deletion mutants and chimeric proteins using the highly homologous tyrosine phosphatase-like protein IA-2β. Analysis of autoantibody reactivity in 111 IA-2 antibody positive sera from patients with type 1 diabetes revealed that humoral epitopes cluster to several domains of the intracytoplasmic part of IA-2 [IA-2ic, amino acid (aa) 604–979]. Immunodominant epitopes were found in the first N-terminal 73 amino acids (56% positive), in the middle domain residing between residues 699–874 (45% positive) and the C-terminus depending on the presence of aa 931–979 (at least 37% positive). Competition experiments with overlapping peptides revealed that autoantibody binding towards the N-terminus was dependent on residues 621–628. In the C-terminal domain, two novel conformation-dependent epitopes were identified. The first epitope requires the presence of the C-terminal part of IA-2 (aa 933–979) and an IA-2-specific region between residues 771–932. Reactivity against the second epitope was dependent on intact C-terminal domains as well as residues in the middle (aa 887–932) and N-terminal regions (aa 604–771) which are conserved in IA-2 and IA-2β. We here defined novel autoantigenic determinants in the N-terminus of IA-2 and characterized conformational epitopes residing in the C-terminal region or spanning from C-terminal residues to the N-terminal domain of IA-2ic. The identification of dominant target regions of diabetes-specific autoantibodies may help to elucidate the molecular mechanisms involved in the autoimmunity towards IA-2.

tyrosine phosphatase IA-2, tyrosine phosphatase IA-2β, epitopes, autoantibodies, type 1 diabetes

Introduction

Type 1 diabetes is an autoimmune disease resulting from progressive destruction of the pancreatic beta cells. The disease is characterized by the appearance of autoantibodies directed to several islet cell autoantigens, including insulin, glutamic acid decarboxylase (GAD), ICA69, gangliosides, and two members of the tyrosine phosphatase (PTP) family designated as IA-2/ICA512, and IA-2β/phogrin [1]. The protein tyrosine phosphatase-like molecule IA-2 is a transmembrane protein expressed in neurones and endocrine cells within secretory granules and are highly conserved in evolution, suggesting that they may be important for the function of neuroendocrine cells [2,3]. IA-2 was identified as a major target antigen of the cytoplasmic islet cell antibodies (ICA) which represent the classical serological markers in type 1 diabetes [4]. Since antibodies to IA-2 (IA-2A) represent highly sensitive and specific markers for type 1 diabetes [5–7], the combined screening for IA-2A and antibodies to GAD was established as first line diagnostic markers in type 1 diabetes [8,9]. Recent findings also indicate that the presence of IA-2 antibodies is correlated with an increased risk for rapid development of type 1 diabetes in subjects with and without family history of type 1 diabetes [7,10–12]. In addition to their important role as diagnostic and predictive markers, autoantibodies may directly influence the initiation and natural course of T cell autoimmunity towards the respective target antigen. It was shown that B cell processing and presentation of peptides is dependent on the epitope recognized by their surface immunoglobulin [13]. Indeed, in non obese diabetic (NOD) mice, the development of autoimmune diabetes requires the presence of B cell that present autoantigenic peptides to T cells [14,15]. Therefore, it is of interest to analyse humoral IA-2 immune response in order to characterize the molecular mechanisms involved in IA-2 autoimmunity. Previous studies have shown that IA-2 antibodies are directed to N-terminal, middle, and C-terminal domains of the intracytoplasmic part of IA-2 [16–18]. Blocking studies indicate that the reactivity against IA-2β is mediated by cross-reactive antibodies directed to epitopes shared by both PTP-like proteins [19,20]. While most immunoreactive domains described thus far represent relatively large regions (78–209 amino acids), the precise location of autoantigenic determinants remains obscure.

To define B cell epitopes in detail, we generated novel IA-2 deletion mutants and chimeric IA-2/IA-2β proteins. By stepwise substitution of IA-2β amino acids residues with IA-2 specific domains in the N- and C-terminus, we were able to identify amino acid regions which are crucially involved in the formation of immunodominant humoral epitopes.

Materials and methods

Patients

Sera were obtained from 111 patients (51 females, 60 males, age 2–50 years, mean 16·4 years, median 14·0 years) with newly diagnosed type 1 diabetes. All the patients were detected positive for antibodies to IA-2 using a sensitive radioligand assay as described previously [10]. In the IDW Combined Autoantibody Workshop, the disease sensitivity of the IA-2A assay was 74% and the specificity was 99% [21]. Fifty sera from healthy individuals served as controls.

Construction of IA-2 and IA-2β deletion mutants and chimeras

IA-2 cDNA clone in pSP64poly(A) vector (Promega, Madison, WI, USA) encoding the intracytoplasmic domain of human IA-2 (IA-2ic, amino acids 604–979) (a kind gift of Dr M. Christie) [22] and mouse IA-2β cDNA in pCRII vector (Invitrogen, San Diego, CA, USA) coding for the homologous intracytoplasmic region (amino acids 629–1001) (a kind gift of Dr A.L. Notkins) [23] were used to generate several deletion mutants. Deletion mutants were produced by polymerase chain reaction (PCR) (annealing temperature 58°C, 26 cycles) using anchored 5′ and 3′ primers corresponding to the published IA-2 and IA-2β sequences including a Kozak sequence and start codon (5′-GCCGCCACCATG … -3′) and/or a translation termination codon (5′-ACT … -3′) at predetermined sites [start at amino acid (aa) IA2: aa 604, aa 699 and aa 771; IA-2ß: aa 629 and aa 793 corresponding to IA-2 position aa 604 and aa 771] (stop at aa IA-2: aa 676, aa 772, aa 854, aa 874 aa 930 and aa 979; IA-2β: aa 908, aa 954 and aa 1001 corresponding to IA-2 position aa 886, aa 933 and aa 979). To analyse antibody binding to the N-terminal and C-terminal regions of IA-2, we used IA-2β deletion mutants comprising aa 629–908 (mutant B1) or aa 629–954 (mutant B2) and exchanged IA-2β aa by the homologous IA-2 residues. N-terminal IA-2/IA-2β-B1 chimeras were constructed by sequential PCRs using overlapping primers with additional sequences added at the 5′-ends corresponding to the homologous 5′-sequence of IA-2ic. The same procedure was applied using 3′-primers to generate C-terminal IA-2β-B2/IA-2 chimeras. Each PCR product was blunted with Klenow, ligated into the HincII site of pGEM 4Z vector (Promega) and sequenced using an automated sequencing apparatus (ABI, Applied Biosystem, Forster City, CA, USA). Deletion mutants and chimeras are illustrated in Figs 1 and 2.

Fig. 1.

IA-2 and IA-2β constructs used for the analysis of humoral IA-2 epitopes. Numbers represent amino acid position in IA-2 and IA-2β. A summary of reactivity patterns of 111 sera from patients with type 1 diabetes are indicated on the right margin.

Fig. 2.

IA-2/IA-2β chimeric proteins used for the analysis of N- and C-terminal epitopes. Numbers represent amino acid position in IA-2 and IA-2ß. A summary of reactivity patterns of 105 sera from patients with type 1 diabetes which show no reactivity to IA-2β deletion mutants B1 and B2 is indicated on the right margin.

Immunoprecipitation analysis

For analysis of autoantibody reactivity towards deletion mutants and chimeras, 1 µg purified plasmid cDNA was in vitro transcribed and translated in the presence of [³⁵S]-methionine (Amersham, Braunschweig, Germany) using a rabbit reticulocyte lysate system (Promega) as described previously [11]. Incorporated radioactivity was determined by TCA precipitation and liquid scintillation counting. Aliquots of radiolabelled polypeptides (15 000–20 000 c.p.m. for each construct) were incubated with 20 µl serum diluted in 100 µl 20 m m Tris, pH 7·4, 150 m m NaCl, 2 m m EDTA, 5 m m benzamidine, 5 m m methionine, 0·5% Triton X-100 (buffer A) at 4°C for 12 h. After addition of 100 µl preswollen Protein A Sepharose (50% v/v) for 2 h absorbed immunocomplexes were washed five times in buffer A, eluted and analysed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), fluorography and densitometric scanning. In each experiment, the same positive (serum P1) and negative serum (serum C1) was used as internal control to calculate antibody levels from integrated peak areas as follows: (test serum − C1)/(P1 − C1) × 100. A value corresponding to the IA-2 deletion mutant above mean ± 3 SD of the normal controls was considered antibody positive.

Blocking studies

Peptides of 10 residues with four amino acids overlaps covering potential antibody binding sites in the N-terminal region of IA-2ic (aa 603–640) were synthesized by a multiple peptide synthesiser (PSSM-8, Shimadzu, Duisburg, Germany) using 6-aminohexancarbonyl acid as spacer and fluorenylmethoxycarbonyl-protected amino acids (Biosynthan, Berlin, Germany). High performance liquid chromatography purified peptides 1–8 were diluted in 100 µl buffer A and preincubated with 10 µl serum for 6 h at room temperature (final concentration 20 µg/serum). Then samples were incubated with radiolabelled chimera CH3 (15 000 c.p.m. per serum) and analysed by immunoprecipitation as described above.

Results

Immunoreactivity against IA-2 deletion mutants

To identify sera which display reactivity to the N-terminus or C-terminal domains, we first analysed antibody binding against several IA-2 deletion mutants (fragment A1-A7, Fig. 1). As illustrated in Fig. 1, 82·9% of sera had antibodies directed to the N-terminal half of IA-2ic (A2, aa 604–772) and 72·9% had antibodies reactive to the C-terminal half of the molecule comprising aa 771–979 (A5). Further C-terminal deletion of 96 aa from fragment A2 resulted in a loss of reactivity in 30 of 92 (32·6%) sera. Since the reactivity against the middle part of IA-2ic (aa 699–874) was present in 45·1% of patients, our data indicate the presence of at least one epitope located within the first 76 aa at the N-terminus of IA-2ic and another epitope which requires the presence of aa 699–874. C-terminal deletion of 49 aa from IA-2ic (fragment A7) abolished reactivity by 30 of 81 (37·0%) sera, suggesting the presence of one epitope in the C-terminal part or a major conformational change affecting binding to residues 771–930. These data confirm previous findings that major IA-2 epitopes reside in the N-terminus, the middle region and the C-terminus (Fig. 1). Representative immunoprecipitation patterns obtained with sera from diabetic patients are shown in Fig. 3.

Fig. 3.

Immunoprecipitation analyses of dominant humoral IA-2 and IA-2ß epitopes. Autoradiography of [³⁵S] methionine labelled reticulocyte lysates revealed that IA-2 and IA-2β deletion mutants were expressed as major bands after expression by in vitro transcription and translation. The figure demonstrates representative reactivity patterns against IA-2 deletion mutants (a) and IA-2β deletion mutants (b) obtained by immunoprecipitation using sera from patients with type 1 diabetes. Molecular weight markers (MWM) are given in the left margin.

Detailed mapping of a major N-terminal epitope

Further epitope mapping by progressive N-terminal or C-terminal deletion of fragment A1 or A5 failed to produce reliable results. We decided to generate IA-2/IA-2β chimeric proteins using the mouse IA-2β instead of the human molecule because (i) the mouse molecule display a higher degree of diversity compared to human IA-2 (phogrin); (ii) data from a previous study indicated that residues that are not conserved between murine and human IA-2β may be involved in antibody binding [23]. Indeed, only six (5·4%) sera were found to react with mutant B1 and B2 comprising the N-terminal and middle domains of mouse IA-2β and only 18 (16·2%) sera displayed reactivity against fragment B3 (aa 793–1001). Therefore, the B1 and B2 polypeptides represent useful tools for the design of chimeric proteins to map conformational epitopes potentially involved in autoantibody binding.

Substitution of 52 N-terminal amino acids of IA-2βic with residues 604–655 of IA-2 (chimera 1, CH1) resulted in the recovery of immunoreactivity (Fig. 2). Among 105 sera which did not display reactivity with B1, 58 (55·2%) had autoantibodies against chimera 1 (CH1). To analyse reactivity in more detail, we constructed three novel chimeras comprising IA-2 aa 636–655 (CH2), 618–636 (CH3) and 604–617 (CH4), respectively. No reactivity was observed against CH2 and only 2 (1·9%) sera bound to CH4. In contrast, 39 (37·1%) sera recognized fragment CH3 (corresponding to 67·2% of sera positive for CH1) (Fig. 4a). Since 39 of 62 (62·9%) patients who had antibodies directed to deletion mutant A1 (aa 604–676) also reacted with CH3, our data suggests that the major autoantigenic epitope within the N-terminus of IA-2ic resides between residues 618–636.

Fig. 4.

Analysis of immunoreactivity against IA-2/IA-2β chimeric proteins.Chimeras were expressed and labelled with [³⁵S] methionine by in vitro transcription and translation (upper panel). Antibody binding was analysed by immunoprecipitation with sera from patients with type 1 diabetes. Representative antibody patterns are illustrated against IA-2/IA-β chimeras (a) and IA-2β/IA-2 chimeric proteins (b). Molecular weight markers (MWM) are given on the left margin.

Due to the strong reactivity of some sera with chimera CH3, we speculate that this region might harbour linear epitopes. To address this question, we performed blocking studies with individual peptides covering aa 603–640 to inhibit immunoprecipitation of fragment CH3. Peptide 5 completely blocked antibody binding in all CH3-reactive sera (n = 20) and peptide 6 displayed complete blocking in 40% and partial inhibition in 60% of tested sera (Fig. 5). In contrast, peptides 1–4 and peptides 7–8 did not influence antibody reactivity. Thus, residues 621–628 may harbour the IA-2-specific sequence involved in antibody binding to the N-terminus of IA-2ic.

Fig. 5.

Reactivity against the N-terminal epitope can be blocked by synthetic peptides. Sera were preincubated with 20 µg peptide 1–8 and then used for immunoprecipitation of [³⁵S] methionine labelled chimera CH3. Preincubation of sera with peptide 5 and 6 completely blocked the reactivity to chimera CH3. Peptides 1–4 and 7–8 do not inhibit binding of any sera. The lower panel demonstrates sequence homology between human IA-2 and mouse IA-2β harbouring the epitope for N-terminal IA-2 autoantibodies. Solid lines indicate identical amino acids (aa), points indicate aa with similar charge, polarity, or hydrophobicity. Numbers at peptides refer to the aa residues of human IA-2.

Characterization of the C-terminal epitopes

The dominant epitopes located at the C-terminus of IA-2 were mapped by fusion of fragment B2 with IA-2 aa 933–979 (CH5). This allows us to exclude the reactivity of IA-2 specific antibodies directed to the middle and N-terminal domains of IA-2 (aa 604–932). Chimera CH5 was recognized by 61 of 105 (58·9%) IA-2 antibody positive, fragment B2 negative sera (Fig. 4b, patterns 1 and 2) indicating that at least one epitope is preserved in this fusion protein. As expected, the majority of CH5 reactive sera also recognized mutant A5 and full length IA-2βic (Table 1). The importance of the conformation of CH5 was emphasized by the observation that deletion of IA-2β residues 909–954 (corresponding to IA-2 aa 887–932, chimera 6) and deletion of the last 22 C-terminal aa (chimera 8) abolished reactivity by 97% of CH5-reactive sera. In addition, none of the sera reacted with the last 22 aa (chimera 7) which share 100% identity between IA-2 and IA-2ß (Figs 2 and 4). Interestingly, deletion of the 164 N-terminal aa from CH5, yielding a fusion protein in which residues 771–932 of IA-2 are replaced by the corresponding domain of IA-2β (aa 793–954, chimera 9), resulted in a loss of antibody binding in 75·4% and 82·7% of sera which displayed a positive reactivity against CH5 and mutant A5 (aa 771–979), respectively. These findings suggest that at least two distinct autoantigenic epitopes are dependent on the intact C-terminus. One epitope is located within aa 771–979 of IA-2 (fragment A5) depending on the IA-2-specific residues 887–932 (only two of 81 (2·5%) A5 positive sera recognized CH6) and 931–979 (30 of 81 (37·0%) A5 positive sera failed to react with fragment A7) (Table 1). The comparison of antibody reactivities against CH5 with CH6 and CH8 and CH5 with CH9 revealed that the second epitope requires the intact C-terminus of IA-2 (aa 887–932 and aa 958–979) and a region in the N-terminal domain of the molecule (between aa 604–770) which is not affected by the introduction of the homologous region of IA-2ß (aa 629–792), despite a high degree of diversity in the N-terminus. The requirement of a small C-terminal segment (22 aa) and regions in the middle and N-terminal part of IA-2ic suggests that the latter autoantibodies may target a conformational epitope spanning from the C-terminus to the N-terminal region (upstream aa 771) of IA-2.

Table 1.

Correlation of autoantibodies against different C-terminal deletion mutants and IA-2β/IA-2 chimeric proteins in 105 IA-2A positive patients

Sera	Number	Positive reactivity against
		CH5	CH9	Frag. A5	Frag. A7	IA-2βic
CH5 positive	61	–	15 (22·1%)	53 (86·9%)	40 (65·6%)	58 (95·1%)
CH9 positive	15	15 (100%)	–	14 (93·3%)	13 (86·7%)	14 (93·3%)
Frag. A5 positive	75	53 (70·7%)	14 (18·7%)	–	45 (60·0%)	58 (77·3%)
Frag. A7 positive	45	40 (88·9%)	13 (28·9%)	45 (100%)	–	37 (82·2%)
IA-2βic positive	64	58 (90·6%)	14 (21·9%)	37 (57·8%)	58 (90·6%)	–

Six sera of the total population were excluded because of the presence of antibodies directed to IA-2β fusion partners B1 and B2. This made it impossible to analyse specific reactivities against chimeric proteins.

Correlation between antibody specificities

Among 111 IA-2 antibody positive patients with type 1 diabetes, 29 (26·1%) had antibodies directed against the N-terminus (CH1), the middle region (aa 699–874) and the C-terminus (CH5). Nineteen (17·1%) were found positive for CH1 antibodies only, eight (7·2%) had antibodies to the middle region only and 12 (10·8%) had antibodies exclusively directed against the C-terminal region. The other 43 patients displayed different combinations of antibody reactivities against several domains of the C-terminus (CH5 and/or A5) and middle region or the N-terminus (CH1 and/or A2) and middle region, respectively. There was no significant association with the age at onset of diabetes or the gender with any specific epitope pattern.

We also compared IA-2 levels with the antibody reactivities against dominant epitopes in the N-terminus (CH1 and CH3), the middle region (fragment A6, aa 699–874) and the C-terminus (CH5). IA-2 antibody levels were graded in three groups: (i) low level, 3·5–50 AU; (ii) intermediate level, 51–100 AU; and (iii) high level, > 100 AU. As shown in Table 2, epitope recognition was not restricted to patients with intermediate or high level antibodies to IA-2ic. The frequency of antibodies against mutant A6 (aa 699–874) and chimera CH5 tended to be higher in subjects with high IA-2 antibody levels. There was no significant relationship between IA-2 antibody levels and antibodies directed against the N-terminus.

Table 2.

Correlation of IA-2 antibody reactivities against chimera 1 (CH1, IA-2[604–655]/IA-2β[678–908]), chimera 3 (CH3, IA-2[618–636]/IA-2β[661–908]), chimera 5 (CH5, IA-2β[629–954]/IA-2[933–979]) and fragment A6 (IA-2 aa 699–874) with IA-2 antibody levels

IA-2 antibody level (AU)	Number	Antibody reactivity against
		CH1	CH3	Frag. A6	CH5
3·5–50	12	5 (41·7%)	3 (25·0%)	2 (16·7%)	1 (8·3%)
51–100	66	37 (56·0%)	24 (36·3%)	26 (39·4%)	42 (63·6%)
> 100	27	16 (59·3%)	12 (44·4%)	16 (59·3%)	18 (66·7%)
Total	105	58 (55·2%)	39 (37·1%)	44 (41·9%)	61 (58·1%)

Six sera of the total population were excluded because of the presence of antibodies directed to IA-2β fusion partners B1 and B2. This made it impossible to analyse specific reactivities against chimeric proteins.

Discussion

The tyrosine phosphatase-like protein IA-2 represents a major humoral and cellular target autoantigen in type 1 diabetes. Although type 1 diabetes may be mediated by a cellular autoimmune response, a detailed analysis of immunodominant humoral epitopes is essential to study the interaction between B cell and T cells which may be crucial to trigger autoimmunity [13–15]. To define B cell epitopes, we designed novel chimeric polypeptides by the introduction of IA-2-specific amino acid segments in the homologous regions of IA-2βic. This approach allowed us to overcome some limitations using deletion mutants (low specific radioactivity, poor separation of small polypeptides on SDS-PAGE) and has the advantage of preserving the two and/or three dimensional conformation which has been described to affect antibody binding [24]. Indeed, this study defines one novel immunodominant humoral epitope located in the N-terminal domain of IA-2ic and characterizes two distinct epitopes which requires the presence of C-terminal residues and regions in the middle and N-terminus to maintain the correct epitope conformation.

Our findings on a heterogeneous IA-2 autoantibody response against multiple epitopes within IA-2ic are in line with previous studies which reported on IA-2 antibodies directed to the juxtamembrane region (aa 605–682 or aa 605–693) and the PTP domain (aa 687–979, aa 693–979, aa 771–979) in 33–56% and 82–100%, respectively [16,17,19,24, 25]. In agreement with our results, Zhang and coworkers described antibody binding against a middle domain (aa 692–875) in 40% of sera from patients with type 1 diabetes [17]. It should be noted that most autoantigenic regions described in previous studies represent relatively large polypeptides whose precise location was still unknown. We demonstrated that the transfer of the first 52 aa from IA-2 to mutant B1 can indeed recover the immunoreactivity by 55% of IA-2 antibody positive sera. This major humoral N-terminal epitope of IA-2ic was characterized in detail by combination of three chimeras covering nonoverlapping IA-2 residues between aa 604–655 and blocking studies with 10-mer synthetic peptides. Previously, one study reported on the presence of one IA-2 epitope within aa 601–620 [24]. In addition, it has been shown that deletion of aa 601–635 abolished the binding of IA-2 antibodies directed to the juxtamembrane region [16]. Using the chimeric IA-2/IA-2β proteins, we were able to map one N-terminal epitope. We observed only few sera with antibodies reactive to aa 604–617, suggesting that aa 601–604 and/or aa 618–620 may be important to maintain reactivity to the first 20 juxtamembrane residues of IA-2ic. However, we clearly demonstrate that 37·1% of all IA-2 antibody positive sera and 62·9% of sera reactive to the N-terminus (aa 604–676) displayed a positive reactivity towards a chimeric protein comprising IA-2 residues 618–636. Further mapping in competition experiments using synthetic peptides strongly suggest that residues 621–628 are crucially involved in the formation of this N-terminal epitope.

Previous studies suggest the presence of epitopes in C-terminus of IA-2. Bonifacio and coworkers have identified a region between aa 795–889 and residues aa 877 and aa 911 to be important for antibody binding [24]. Comparison of reactivity patterns against C-terminal deletion mutants (aa 605–937, aa 389–948) with the binding patterns to full length IA-2ic gave indirect evidence that the last 32–42 aa of IA-2ic may be involved in antibody binding [19,25]. To further define these epitopes, we generated several deletion mutants and chimeric proteins comprising the C-terminal and middle regions of IA-2ic. Comparison of reactivity patterns between CH5–CH9 and fragments A5 and A7 revealed that IA-2 antibodies did not directly bind to the C-terminus. Instead, two conformational epitopes were identified which both depend on the presence of the intact C-terminus. One epitope, recognized by at least 27% of IA-2 positive sera, was found to require the presence of residues 887–932 and 931–979. A second epitope was identified after substitution of the last 47 aa of IA-2 to the IA-2β fragment B2. This approach was sufficient to recover immunoreactivity by 58·9% of IA-2 positive sera from patients with type 1 diabetes. Stepwise C-terminal, internal and N-terminal deletion of this chimera revealed that the last 22 residues, a IA-2 specific sequence in the middle region (aa 887–932) and a domain within the first 164 N-terminal residues which is preserved in IA-2β may be involved in the formation of this discontinous epitope. The present data support the concept that part of the IA-2 epitopes are strongly dependent on the correct folding of the molecule [26]. Our findings on the C-terminal epitopes seem to be in sharp contrast to data presented by Kawasaki and coworkers, who reported that only a few sera (1 out of 37) recognize IA-2 after swapping the middle (aa 762–887) and/or the N-terminal region (aa 601–762) between IA-2ic and human IA-2βic (phogrin) [25]. However, it is important to note that these authors analysed antibody reactivity after blocking of cross-reactive antibodies by preincubation with phogrin. We also observed that more than 90% of IA-2β antibodies cross-reacted with common epitopes of IA-2β and IA-2 (data not shown). Since approximately two-thirds of patients had IA-2βA, these antibodies represent a relevant subgroup of IA-2 antibodies. Therefore, we do not exclude these antibody specificities. Using overlapping IA-2β mutants, we provide evidence that the majority of antibodies against IA-2β are not directed to linear epitopes. Comparison of antibody patterns against IA-2β and chimeric proteins (CH5 and CH9), as well as homology analysis between IA-2 and IA-2β, suggest that the C-terminus and the middle domain are essential for antibody binding to IA-2β. Therefore, it can be speculated that chimera CH5 may harbour at least one epitope spanning between the C-terminus and the N-terminal third of the molecule which is also expressed in IA-2β.

In the present study, we performed a detailed characterization of epitope boundaries in the IA-2 molecule. Although the majority of sera displayed immunoreactivity against multiple conformational determinants, stepwise substitution of aa residues from IA-2 into the IA-2β molecule allows us to define three novel immunodominant B cell epitopes. These findings may provide the basis to improve the analysis of epitope spreading in the prediabetic phase and determine common and distinct autoantigenic determinants of the humoral and cellular autoimmune response against IA-2.

Acknowledgments

The authors thank Dr M. Christie for the gift of human IA-2ic cDNA and Professor A.L. Notkins for providing of the mouse IA-2ßic cDNA clone. Cordula Wünsche, Ulrike Wohlrab and Melanie Lettmann are acknowledged for their excellent technical assistance. The study was supported by grants from the Lilly Foundation International (JS) and the Deutsche Forschungsgemeinschaft (Se 725/2–1).