Islet Amyloid Polypeptide Is a Target Antigen for Diabetogenic CD4⁺ T Cells

NOD, NOD.scid, and NOD.RIP-TAg mice were bred and maintained in the Biological Resource Center at National Jewish Health. NOD.IAPP^−/− mice were bred in our colony by backcrossing C57BL6.IAPP^−/− mice (6) onto the NOD background; these mice have been fully backcrossed to the NOD mouse (10 generations). All experimental procedures were in accordance with the guidelines of the institutional animal care and use committee, National Jewish Health.

NOD.RIP-Tag mice (7) were used as a source of β-cell tumors for the preparation of antigen; the enrichment of membrane proteins from β-cells isolated from NOD.RIPTAg adenomas has been previously described (8). In brief, single-cell suspensions of β-cell adenomas were prepared by pressing whole-tumor tissue through a mesh screen, followed by several centrifugation steps to obtain a cellular membrane pellet; this pellet (β-Mem) is highly antigenic for the T-cell clones and serves as a positive control antigen in T-cell assays as well as the source of antigen for further purification. Membrane protein preparations were solubilized in detergent-containing buffer (20 mmol/L Tris, pH 8.0, and 1% octyl-β-glucoside), followed by centrifugation at 18,400g (10 min at 4°C) to remove insoluble debris. Size-exclusion chromatography (SEC) was carried out on a Superdex 200 16/60 column (Amersham Biosciences) at 21°C (flow rate: 1 mL/min; fraction size: 1.25 mL; and injection volume: 2.0 mL), eluting with SEC buffer (20 mmol/L Tris, pH 8.0, and 150 mmol/L NaCl). Fractions from SEC were assayed for antigenic activity with the T-cell clones, and a peak antigenic fraction (950 μL) was acidified with 20 μL trifluoroacetic acid before adding 30 μL acetonitrile. A total of 800 μL of this mixture was then applied to a reverse-phase high-performance liquid chromatography (RP-HPLC) mRP-C18 high-recovery protein column (Agilent). A buffer gradient was used to elute proteins from the column, and a total of 36 fractions were collected between 0 and 72 min at a flow rate of 0.75 mL/min and a constant column temperature of 80°C. Antigenic peaks were again determined through assays with T-cell clones.

The antigenicity of islet cells, cellular and biochemical fractions, or peptides was assessed through the IFN-γ responses of T-cell clones. In 96-well microtiter plates, assay cultures contained 2 × 10⁴ responder T cells, 2.5 × 10⁴ NOD peritoneal exudate cells as antigen-presenting cells, and β-cell antigen. IFN-γ production by the responder T cells was determined by ELISA (BD Biosciences). IFN-γ standards were added to the assay plates at concentrations between 0 and 100 ng/mL, but concentrations of the standard >50 ng/mL were above the assay detection range. An aliquot of β-Mem or whole NOD islet cells were used as positive controls; test wells contained SEC or RP-HPLC fractions, peptides, or islet cells. Organic solvents were removed from HPLC fractions by vacuum centrifugation prior to assay. Synthetic peptides were obtained from CHI Scientific.

The protein-identification strategy we have previously described (5) was used for mass spectrometry. Proteins were digested with trypsin, and extracted peptides were resolved by chromatography online on a C18 column and a 1200 Series HPLC system (Agilent Technologies). Analysis was carried out with a 6340 LCMS ion-trap mass spectrometer in the National Jewish Health facility. Raw data were extracted and used to search the Swiss-Prot or National Center for Biotechnology Information databases with the Spectrum Mill search engine (Rev A.03.03.038 SR1; Agilent Technologies). Data were evaluated and protein identifications were considered significant if the following confidence thresholds were met: individual peptide scores of at least 10 and scored percent intensity of at least 70%. A reverse (random) database search was simultaneously conducted, and manual inspection of spectra was used to validate the match of the spectrum to the predicted peptide fragmentation pattern.

Cell cultures of BDC-2.5 and BDC-5.2.9 were expanded in interleukin-2, and after harvesting, T cells (1 × 10⁷) were injected intraperitoneally into age-matched 6- to 14-day-old NOD or NOD.IAPP^−/− recipient mice. Mice were monitored on a daily basis for urine glucose using Diastix (Bayer). The presence of urine glucose was followed by blood glucose monitoring using a OneTouch Ultra glucometer (LifeScan). Mice were considered diabetic when blood glucose levels were >18 mmol/L (324 mg/dL). For disease-transfer experiments, a Wilcoxon rank sum test was used to determine statistical significance.

As previously described (5), to identify candidate antigens for the T-cell clones, we carried out biochemical fractionation of β-cell tumors and mass spectrometric analysis of peak antigenic fractions. Similar to the other clones from the panel, BDC-5.2.9 responds in vitro to pancreatic islet cells or cell extracts of β-cell adenomas presented in the context of I-A^g7 (2). Our starting source of antigen was β-cell tumor tissue excised from NOD.RIPTAg mice (7) and separated by differential centrifugation to yield a highly antigenic β-cell membrane pellet (β-Mem); this preparation also served as the positive control in T-cell activity assays. After lysis, the membrane pellet was separated by sequential chromatography, first through an SEC column and then by RP-HPLC. Illustrated in Fig. 1, the presence of antigen was tracked through the IFN-γ response of the T-cell clone BDC-5.2.9 (top axis, gray lines) to protein (bottom axis, black lines) in individual fractions. Figure 1A is the chromatogram from SEC and shows a single antigenic peak that corresponds to a low–molecular weight protein region. The antigenic fractions from SEC were further purified by RP-HPLC, yielding another single peak of antigen activity (Fig. 1B).

The highly enriched and antigenic RP-HPLC fractions were subjected to in-solution tryptic digest and were then analyzed on an ion-trap mass spectrometer. Resulting spectra were sequenced and matched to proteins by searching the Swiss-Prot protein database using the search program Spectrum Mill. We identified a total of 21 proteins distributed over HPLC fractions 11–14 (Fig. 2A). Comparison of the antigen distribution profile with individual protein spectral-intensity values, an approximate indicator for relative protein abundances, singled out IAPP as the only candidate. Other proteins did not match the antigen distribution profile; for example, secretogranin 1 was detected only in nonantigenic fractions 11 and 14. We identified two IAPP peptides, representing 28% amino acid coverage of prepro-IAPP and 70% coverage of the 37-residue peptide hormone (Fig. 2B and C). The peptide LANFLVR was matched with high confidence and the peptide SSNNLGPVLPPTNVGSNTY was matched with medium confidence.

To investigate the peptide(s) targeted by BDC-5.2.9, we analyzed a series of overlapping IAPP peptides, spanning the entire propeptide region (Fig. 3A). One peptide, KS20, elicited a vigorous response from BDC-5.2.9 and could stimulate the clone at concentrations as low as 10 ng/mL (4.6 nmol/L) (Fig. 3B). BDC-5.2.9 did not respond to any other IAPP peptide tested or to the insulin peptide B9–23. Two neighboring peptides, RT20 and GR20, share either the C- or N-terminus sequence with KS20, but neither of these peptides can stimulate BDC-5.2.9, indicating that the T-cell receptor of this clone recognizes the adjoining sequences. An insulin-reactive clone, PD-12.4.4 (9), was used as a negative control and showed no response to IAPP peptides. The fact that BDC-5.2.9 is stimulated by very low concentrations of the IAPP peptide KS20 suggests that this peptide may contain the natural ligand for BDC-5.2.9.

To further validate IAPP as the antigen for the BDC-5.2.9 T-cell clone, we analyzed IFN-γ responses to islet cells from either NOD (IAPP^+/+) or NOD.IAPP^−/− mice. The T-cell clone BDC-2.5 was used as a positive control because it responds to a different β-cell antigen, ChgA (5). As illustrated in Fig. 4, both clones responded to antigen present in wild-type islets from NOD mice (IAPP^+/+) (Fig. 4A), and the ChgA-reactive clone BDC-2.5 also responded to islets from NOD.IAPP^−/− mice (Fig. 4B). In contrast, there was no stimulation of BDC-5.2.9 by IAPP-deficient islets (Fig. 4B), indicating that islets from IAPP^−/− mice lack the antigen for this clone.

The highly aggressive nature of BDC-5.2.9 in disease induction can be demonstrated by adoptive transfer into young NOD and NOD.scid mice. To investigate the in vivo islet reactivity of BDC-5.2.9 in IAPP-deficient mice, adoptive transfers with this clone were made into young (aged <2 weeks) IAPP^−/− recipients. BDC-2.5 was again used as a positive control and rapidly induced disease in both IAPP^+/+ and IAPP^−/− recipients (Fig. 5A). When BDC-5.2.9 was used in transfers, disease developed in most of the NOD.IAPP^+/+ mice within 3 weeks, but there was no diabetes in IAPP-deficient mice for the duration of the experiment, up to 45 days. We also conducted histological analysis of pancreatic sections from IAPP-deficient mice to assess leukocyte infiltration and granulation of islet β-cells (Fig. 5B). The spontaneous development of insulitis and hyperglycemia in NOD.IAPP^−/− mice is typical and very similar to that in NOD.IAPP^+/+ mice. Transfer experiments were terminated when recipients were ~2 months of age and the pancreatic histology of NOD.IAPP^−/− mice receiving BDC-5.2.9 closely resembled that of aged-matched, uninjected IAPP^−/− controls, with at least 60% of islets being well granulated and no infiltrate apparent. In contrast, histology from IAPP^−/− mice that received the BDC-2.5 T-cell clone showed complete degranulation of β-cells. These results further demonstrate that islets lacking IAPP do not contain the antigen for BDC-5.2.9.

The activation of autoreactive CD4 T cells through T-cell receptor responses to peptide/major histocompatibility complexes is a major event in autoimmune diseases such as type 1 diabetes. The identification of β-cell antigens, and the peptide epitopes from these antigens, could have important implications for both diagnosis and therapy of type 1 diabetes. Our results reported here demonstrate that IAPP is now the third secretory granule protein in islet β-cells, after insulin and ChgA, to be targeted by pathogenic CD4 T cells, and it is possible that other secretory granule proteins (e.g., other secretogranins) also could be antigenic targets in type 1 diabetes. Similar to insulin, IAPP is uniquely expressed within the pancreatic β-cells. This peptide hormone is cosecreted with insulin from the β-cell granules at a molar ratio of ~1:100 IAPP:insulin (10). Similar to insulin and ChgA, IAPP is expressed as a larger precursor molecule (pro-IAPP), which is then posttranslationally processed within the secretory granules by the enzymes prohormone convertase 2 and carboxypeptidase E (11).

IAPP is well known as a component of the amyloid plaques found in the pancreatic islets of patients with type 2 diabetes (12–14). In type 1 diabetes, there has been documentation of autoantibodies to IAPP in patients, but there was no correlation to disease (15). To date, there has been little association of IAPP with type 1 diabetes. One major exception was the report from Panagiotopoulos et al. (16) in which it was found that a significant proportion of CD8 T cells from HLA-A*0201 patients with recent-onset type 1 diabetes respond to a peptide derived from the leader sequence of prepro-IAPP. This group also contributed to a later study in which it was found that a second sequence of prepro-IAPP (ppIAPP 9–17) elicited responses from cytotoxic T-lymphocytes of type 1 diabetic patients (17). The other indicator that IAPP might have a role in type 1 diabetes, and the first evidence from the NOD mouse, was a genetic mapping study in which we reported linkage of the iapp gene with the antigen locus for a CD4 T-cell clone BDC-6.9 (18). Although we have not identified an IAPP peptide that stimulates this T-cell clone, strong corroborating evidence for IAPP as its antigen target was provided by islets from the NOD.IAPP^−/− mouse. As observed with the BDC-5.2.9 clone, IAPP-deficient islets were unable to activate BDC-6.9 (K.H., unpublished data). In our new report, we have provided definitive evidence that diabetogenic CD4 T cells in the NOD mouse react to IAPP as an antigen, including the identification of a peptide containing a natural ligand for one of these T cells. The sequence of secreted IAPP is quite similar between mouse and human forms of the molecule, and there is only one amino acid difference between the KS20 mouse and human sequences. It is not unreasonable, therefore, to assume that human CD4 T cells also may react to IAPP peptides presented, for example, by HLA-DQ8.

A major function of secretory granule proteins is to participate in the biogenesis of the secretory granules and in the process, these proteins tend to form aggregates (19,20). IAPP is of particular interest in this regard because of its propensity to form amyloid deposits in type 2 diabetic patients. Such deposits are not found in type 1 diabetic patients, but it has been demonstrated by others that the initial step in amyloid deposition is the formation of toxic IAPP oligomers (21). Soluble oligomers of IAPP also have been shown to be responsible for activation of the NLRP3 inflammasome that leads to increased interleukin-1β production in type 2 diabetes (22). In type 1 diabetes, similar smaller aggregates of IAPP could form under conditions of inflammatory stress. It is quite conceivable that such oligomers would be taken up by antigen-presenting cells and processed in a way that leads to the formation of neoantigenic epitopes.

This work was supported by Juvenile Diabetes Research Foundation grant 1-2008-132 and National Institutes of Health (NIH) grant RO1-DK-50561 to K.H., NIH Training grant T32-AI-007405, a shared instrument grant from the National Center for Research Resources (S10RR023703) to N.R., and NIH grant UL-1-RR-O25780 to N.R. Islets were obtained from the DERC Islet Core at the Barbara Davis Center, supported by NIH grant P30-DK-057516.

No potential conflicts of interest relevant to this article were reported.

T.D. performed the biochemical purification and antigen analysis and wrote the manuscript. R.L.B. performed disease transfer experiments and analyzed histology. N.R. and R.R. supervised the mass spectrometry and contributed to discussion. R.L.P. and M.A. assisted with mass spectrometry and HPLC. G.B. maintained and assayed T-cell clones. B.B. runs the mouse colony and contributed to in vivo experiments. K.H., as the principal investigator, held overall responsibility and oversight of the project and wrote and edited the manuscript.