The immune response to insulin is regulated by MHC class II genes. Immune response (Ir) gene-linked low responsiveness to protein Ags can be mediated by the low affinity of potential antigenic determinants for MHC molecules (determinant selection) or by the influence of MHC on the functional T cell repertoire. Strong evidence exists that determinant selection plays a key role in epitope immunodominance and Ir gene-linked unresponsiveness. However, the actual measurement of relative MHC-binding affinities of all potential peptides derived from well-characterized model Ags under Ir gene regulation has been very limited. We chose to take advantage of the simplicity of the structure of insulin to study the mechanism of Ir gene control in H-2b mice, which respond to beef insulin (BINS) but not pork insulin (PINS). Peptides from these proteins, including the immunodominant A(1–14) determinant, were observed to have similar affinities for purified IAb in binding experiments. Functional and biochemical experiments suggested that PINS and BINS are processed with similar efficiency. The T cell response to synthetic pork A(1–14) was considerably weaker than the response to the BINS peptide. We conclude that the poor immunogenicity of PINS in H-2b mice is a consequence of the T cell repertoire rather than differences in processing and presentation.
The specificity and magnitude of immune responses are influenced heavily by polymorphic proteins encoded in the MHC through their role in binding and presenting peptide Ags to T lymphocytes in the thymus and periphery. The magnitude of the T cell-dependent Ab response to small proteins and synthetic peptides is often strikingly different in individuals as a consequence of genetic polymorphisms that map to the class II region of the MHC (1, 2, 3, 4). Two general mechanisms have been proposed to account for MHC-linked control of T cell responses. The determinant selection model is based on the idea that MHC class II proteins bind and present selected Ag determinants and that polymorphisms determine the specificity of Ag presentation (5, 6). Proteins containing no determinants capable of binding to the MHC class II molecules expressed in a given individual would not elicit an immune response. The concept that MHC molecules influence the development of the T cell repertoire through selection in the thymus led to the alternative hypothesis that MHC-linked unresponsiveness may be a consequence of “holes in the T cell repertoire” (7). In this model, there is no defect in the capacity of class II molecules to present Ag to T cells. Instead, functional T cells capable of recognizing the available MHC-Ag complexes are absent because of clonal deletion, anergy, active suppression, or lack of positive selection in the thymus.
It is evident from a large number of studies beginning with the first direct demonstration of sequence-specific binding of peptides to MHC (8, 9) that polymorphisms in class II molecules have a major influence on the determinant specificity of CD4+ T cell responses. Sette et al. demonstrated that a strong correlation exists between the in vitro binding pattern of denatured proteins and the pattern of restriction of T cell responses elicited by immunization with native Ags (10). Schaeffer et al. measured the MHC binding affinities of overlapping peptide sequences from staphylococcal nuclease and demonstrated that MHC binding is an absolute requirement for immunogenicity (11). Other reports have also provided evidence that the efficiency of MHC-peptide complex formation determines whether or not an immune response is generated to a given protein Ag (12, 13). These studies, coupled with other lines of evidence (14), have led to a general perception that determinant selection accounts for most examples of immune response (Ir)3 gene-linked unresponsiveness. However, MHC class II molecules are very promiscuous in their peptide binding specificities, and even small proteins often contain multiple determinants capable of stable binding to a given class II molecule.
The murine immune response to insulin is controlled by MHC-linked Ir genes (15, 16, 17). H-2b mice respond to beef insulin (BINS) but not pork insulin (PINS), which differs from BINS by only two amino acids in the A-chain disulfide loop (Fig. 1). In contrast, H-2d mice respond to both Ags. Previous studies demonstrated that the minimal immunodominant T cell determinant recognized by BINS-primed H-2b and H-2d mice and PINS-immune H-2d mice is contained within the A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) segment of insulin (18, 19). Disulfide reduction is sufficient to bypass the Ag-processing requirements of insulin, creating A- and B-chain peptides that can directly associate with cell surface IA molecules without further processing to stimulate T cells (18, 20). Early studies provided evidence that PINS can be recognized by CD4+ T cells in nonresponder H-2b mice. Bucy and Kapp demonstrated a heteroclitic response to BINS in PINS-primed mice (21); the induction of a latent population of CD4+ T cells capable of driving secondary in vitro Ab responses to haptenated PINS was also observed (22). However, the extent to which determinant selection contributes to the differential response to BINS vs PINS in H-2b mice has never been examined.
Materials and Methods
Cell lines and lymphokine assays
The murine B cell line LB27.4 was generously provided by Dr. R. Azofski (National Institutes of Health, Bethesda, MD). T2BB (23) is a human HLA-DM-deficient B cell line transfected with IAb. B1A4 is an IAb-restricted hybridoma that recognizes a B-chain epitope and was generated from C57BL/10 SnJ mice immunized with PINS (24). Cells were maintained in RPMI 1640 supplemented with 10% FCS. B1A4 T cells (1 × 105/well) were incubated with LB27.4 APCs (1 × 105/well) and various concentrations of Ag in flat-bottom, 96-well tissue culture plates for 24 h. Lymphokine production was quantified by using the IL-2-dependent T cell line CTLL-2. Culture supernatants (100 μl) were transferred to flat-bottom, 96-well tissue culture plates, freeze-thawed, and cultured with 104 CTLL cells/well for 24 h. Each well was pulsed with 1 μCi of [3H]thymidine during the final 10 h of culture, and units of IL-2 were determined by comparison with a titration of rIL-2. The results represent the mean ± SD units IL-2 from triplicate cultures.
Protein and peptide Ags
BINS and PINS were purchased from Elanco Products (Indianapolis, IN). Sulfonated insulin A-chain (A(SSO3)4) and B-chain (B(SSO3)2) were prepared as described previously (18, 25), and the individual chains were purified by reverse-phase HPLC using a C4 column and an acetonitrile gradient in trifluoroacetic acid. NαA1, NαB1, NεB29 tri-biotinyl-insulin was prepared by reaction with a 10-fold molar excess of biotin amidocaproate N-hydroxysuccinimide in dimethylformamide containing 10 mM triethylamine (26). After 24 h, the reaction was terminated by acidification with excess 1 M acetic acid followed by extensive dialysis using Spectra/Por 3 membranes (Fisher Scientific, Phoenix, AZ). Peptides were synthesized in the Emory Microchemical Facility as described previously (27). Peptide sequences were: pork A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14), GIVEQCCTSICSLY; beef A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14), GIVEQCCASVCSLY; mouse A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14), GIVDQCCTSICSLY; Eα(52–68), ASFEAQGALANIAVDKA; OVA(323–339), ISQAVHAAHAEINEAGR; and hen egg lysozyme (46–60), NTDGSTDYGILQINS. Some peptides were labeled through the α amino group with fluorescein or biotin by reaction with biotin amidocaproate N-hydroxysuccinimide (27).
Peptide binding assays
IAb was purified from C12E9 detergent-solubilized T2BB membrane preparations using a Y-3P (IAb) mAb immunoaffinity column as described previously (27). HLA-DM was purified from human B cell lines as described previously (28). IAb (50 nM) was incubated with biotin-peptide in 0.2% Nonidet P-40 and 100 mM citrate/phosphate (pH 4.5) for 18 h at 37°C in the presence or absence of 4 mM DTT. Purified HLA-DM (200 nM) was included unless otherwise indicated to enhance binding. In some experiments, peptide affinity was measured by competition inhibition assay, in which 50 nM of IAb was incubated with 0.5 μM of biotin-Eα(52–68) with varying concentrations of competitor peptide. Observed peptide affinities were not changed by the presence or absence of DM. After incubation, IAb was captured on microtiter assay plates coated with Y-3P; bound biotin-peptide was quantified in duplicate or triplicate samples with europium-streptavidin fluorescence as described previously (29). The dissociation rates of Fl-A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) peptides were measured by incubating preformed peptide complexes (1 μM of IAb) in 100 mM of citrate/phosphate buffer (pH 4.5) containing 0.2% Nonidet P-40, 400 nM of DM, and 200 μM of unlabeled Eα(52–68) at 37°C. Aliquots were analyzed at various timepoints by high performance size exclusion chromatography and fluorometry as described previously (28).
Measurment of IAb-peptide complexes in APCs cultured with biotin-insulin
LB27.4 cells were washed twice with medium and cultured at 1 × 106 cells/ml for 4 h at 37°C with 300 μg/ml biotin-labeled insulin followed by three washes in HBSS. Cells were pelleted, resuspended in 50 μl of 0.5% Nonidet P-40 lysis buffer (0.15 M NaCl, 50 mM Tris (pH 7.5), 0.01% azide, and protease inhibitor mixture), and incubated for 40 min on ice with intermittent vortexing. Lysates were cleared by centrifugation for 10 min at 10,000 × g. Samples were transferred to prepared microtiter plates for capture and quantification of bound biotin-peptide as described above.
Lymph node proliferation assays and Ab responses
C57BL/10 SnJ mice (6 to 10 wk of age, The Jackson Laboratory, Bar Harbor, ME) were immunized s.c. with 25 μg of peptide in CFA. After 10 days, draining lymph nodes were removed and single-cell suspensions were prepared. Lymph node cells (5 × 105/well) were cultured in 96-well tissue culture plates in RPMI 1640 containing 0.5% normal mouse serum and 5 μM 2-ME for 3 days at 37°C and pulsed during the final 18 h with 1 μCi of [3H]thymidine. Results represent the mean ± SD cell-associated cpm from triplicate cultures. Ab responses were measured by immunizing mice s.c. with 50 μg of insulin in CFA. Serum samples (14 day) were diluted serially into 96-well plates that had been coated with BINS or PINS (10 μg/ml in borate-buffered saline, pH 8.6) and blocked with excess protein. Wells were washed after incubation for 1.5 h at 4°C and further incubated with 10 ng/ml europium-labeled anti-mouse IgG (Wallac Oy, Turku, Finland) followed by measurement of europium fluorescence as described previously (29).
Peptides from PINS and BINS bind IAb with similar affinities
PINS is markedly less immunogenic than BINS in H-2b mice as measured by Ab response or T cell proliferation (Fig. 1, Refs. 15, 16, 17, 30). Competitive binding experiments with purified IAb were performed to determine the affinities of peptides from BINS and PINS. Intact BINS and PINS do not inhibit peptide binding to IAb (Fig. 2,A). However, both insulins inhibit the binding of a high-affinity peptide, biotin-Eα(52–68), to IAb in the presence of the reducing agent, DTT, which liberates A- and B-chains and reduces the A-chain loop structure (Fig. 2 B). Thus, one or both chains of BINS and PINS can bind the IAb peptide binding groove with similar affinity.
Isolated B-chain, which is identical in BINS and PINS, binds to IAb with high affinity (Table I). Full-length beef and pork A-chains derived from sulfonated insulin bind IAb with affinities similar to the previously characterized determinant, OVA(323–339). Experiments were also done with synthetic A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) to rule out potential hindering effects mediated by residues outside of the core immunodominant T cell determinant that could obscure differences in affinity for IAb (31, 32) and to exclude the possibility that a determinant other than A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) is primarily responsible for the observed binding. However, beef, pork, and mouse A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) peptides all bind to IAb with similar affinities that are slightly lower than those of full-length A-chains (Table I). The affinities of biotinylated beef and pork A-chains were directly measured in saturation binding experiments (Fig. 3). The Kd of sulfonated beef and pork A-chains were determined to be 1.0 ± 0.2 and 1.1 ± 0.4 μM, respectively, from Scatchard analysis of three independent experiments. Taken together, these results demonstrate that the differential immunogenicity of BINS and PINS in H-2b mice is not a consequence of differences in IAb binding affinity.
|Peptide .||IC50 (μM)a .|
|Peptide .||IC50 (μM)a .|
Data represent the peptide concentration required to inhibit the binding of indicator peptide by 50%.
Evidence that BINS and PINS are processed similarly
Recent studies indicate that HLA-DM (H2-M) may edit the repertoire of peptide complexes displayed by APCs, selectively favoring the most stable complexes (28, 33, 34, 35). Differences in sensitivity to DM-catalyzed peptide dissociation do not always correlate with affinities measured in competitive peptide binding assays (28). Therefore, experiments were done to determine whether beef and pork A-chain peptides are differentially sensitive to DM-catalyzed release from IAb. The dissociation rates of fluorescein-labeled beef and pork A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) peptides, measured in the presence of a high concentration of purified DM, were nearly identical (Fig. 4,A). The possibility that DM may selectively influence the binding of A-chain peptides to IAb at different pH was also evaluated (Fig. 5 B). The extent of binding of beef and pork A-chains at different hydrogen ion concentrations was very similar.
The relative efficiency of processing BINS vs PINS was investigated in experiments with an IAb-restricted T cell hybridoma that recognizes a B-chain determinant, B(17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30) (24). T cell hybridomas with this specificity are isolated with low frequency from PINS-primed H-2b mice. The dose responses to PINS and BINS were identical, indicating that the proteins are processed with equal efficiency to generate IAb-peptide complexes containing the B(17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30) determinant (Fig. 5,A). We are not able to directly evaluate the efficiency of presentation of A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) peptides using a similar approach because we have not been able to isolate IAb-restricted T cell hybridomas that recognize pork A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14). However, we observed similar processing of BINS and PINS in biochemical assays measuring the formation of IAb-peptide complexes in APCs cultured with insulin labeled with biotin through amino groups in both the A- and B-chain (Fig. 5 B).
The above experiments indicated that peptides from PINS and BINS bind IAb with equal affinity; no evidence was obtained for differences in Ag processing or editing by DM. However, we could not formally rule out the possibility that IAb complexes containing the A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) determinant are inefficiently generated during intracellular processing of PINS in APCs. The capacity of T cells to respond to this determinant was evaluated by immunizing mice with synthetic pork or beef A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) peptides and measuring responses in a secondary in vitro lymph node proliferation assay. Under these conditions, Ag presentation does not require intracellular processing. Pork A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) was considerably less immunogenic than the BINS peptide (Fig. 6), indicating that there is a relative deficiency in the capacity to respond to the pork A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) determinant expressed at the level of the T cell repertoire.
It is now clear that the relative affinities of Ag sequences for expressed MHC class II proteins have a dominant impact on the determinants that are made available for recognition by T cells. Thus, the original determinant selection hypothesis (5, 6) withstood the test of time with accompanying advances in our understanding of the molecular basis of Ag presentation and T cell recognition. However, the phenomenon of Ir gene control was observed with many complete protein Ags (15, 36, 37, 38, 39), and it remains unclear what relative contribution is made by determinant selection vs the composition of the T cell repertoire in controlling the immunogenicity of these proteins. In the present study, we have reinvestigated this issue using insulin, concluding that determinant selection is not responsible for MHC-linked regulation of CD4+ T cell responses to insulin.
H-2b mice respond vigorously to BINS but weakly to PINS, which differs by only two amino acids in the A-chain loop. Peptide binding experiments demonstrate that beef and pork A-chain peptides bind with equal affinity to IAb. In H-2k mice, sheep insulin is immunogenic, whereas BINS and PINS are not. Preliminary experiments have demonstrated that peptides from BINS and PINS bind to IAk with the same affinities as those from sheep insulin. Thus, it appears that, in general, differences in MHC binding affinity are not responsible for MHC-linked differences in the immunogenicity of species variants of insulin. However, we have observed that insulin A- and B-chain peptides bind with very low affinity to purified IEk and IEd molecules. Thus, MHC binding affinity does appear to be responsible for the exclusive IA-restriction of T cell responses to insulin in mice expressing both IE and IA.
There are a number of potential mechanisms through which MHC-linked differences in Ag processing can influence the presentation of T cell determinants (40, 41). Amino acid sequence differences outside of the core determinant can influence processing by altering the stability of the protein, protease cleavage sites, or the affinity of competing determinants for available MHC class II molecules. These mechanisms are unlikely to account for the differences in the immunogenicity of insulins in H-2b mice, because PINS and BINS differ only in the A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) determinant and H-2b mice only express IAb, excluding the potential effects of other class II molecules on the processing of insulin. It is possible that the PINS A-chain loop is more sensitive to endopeptidase cleavage than the corresponding BINS sequence. This is unlikely, however, because PINS and BINS are equally immunogenic in H-2d mice; these proteins have similar potency in stimulating T cells that cross-react with pork and beef A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) determinants.
Our experiments with a B-chain-specific T cell hybridoma and direct quantification of IAb-peptide complexes in APCs suggest that PINS and BINS are processed with similar efficiency. We could not directly measure the formation of IAb molecules bearing the PINS A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) determinant. However, disulfide reduction is both necessary and sufficient to generate A- and B-chain peptides from insulin that can bind class II molecules and stimulate T cells without further proteolytic cleavage. Thus, the liberation of A- and B-chains and their availability for binding to IAb molecules in APCs are expected to occur with the same kinetics. Recent studies have demonstrated that DM can differentially influence the cell surface expression of MHC-peptide complexes by selectively editing less stable complexes (28, 33, 34, 35). It is unlikely that PINS A-chain peptides are selectively edited by DM (or H2-M), because PINS and BINS A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) peptides are observed to have nearly identical rates of dissociation from IAb in the presence of a high concentration of DM. Despite these results, we cannot formally exclude the possibility that PINS and BINS are differentially processed in H-2b APCs. However, the reduced immunogenicity of pork A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) as compared with the corresponding BINS peptide as measured by T cell proliferation indicates that differences in Ag processing are not required to explain the difference in immunogenicity of PINS and BINS.
The conclusion that Ir gene regulation of immune responses to insulin is a consequence of the influence of MHC on the functional T cell repertoire rather than determinant selection should, perhaps, come as no surprise given the similarity between mouse insulin (MINS) and other mammalian insulins. Pork and mouse A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) have identical loop sequences and differ only at the A4 position. We have demonstrated that mouse A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) binds to IAb. Thus, the MINS A-chain loop is likely to be available for T cell recognition, necessitating the generation of active tolerance. However, in H-2d mice, PINS is immunogenic despite its sequence similarity to MINS. The A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) determinant is immunodominant, and we have observed that pork and mouse A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14) peptides bind to IAd with similar affinity (data not shown). However, it is evident that a large number of T cells are present in H-2d mice that selectively recognize pork and not mouse A(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14), which differ at only one position (Fig. 1).
It is likely that self tolerance rather than determinant selection is responsible for most examples of Ir gene-linked unresponsiveness to complete protein Ags. In general, Ir gene control has been observed only for relatively small proteins with limited polymorphism among mammals. Large proteins and proteins with extensive sequence divergence from self homologues tend to be highly immunogenic. Despite a clear level of specificity, MHC class II molecules are highly promiscuous in their peptide binding specificity. This promiscuity is amplified compared with MHC class I molecules because class II molecules can bind determinants in polypeptides of unrestricted length, making it possible to “scan” a polypeptide to find a region with an appropriate distribution of anchor residues. Thus, even relatively small proteins are likely to contain determinants that can bind to a given class II molecule with sufficient affinity to generate TCR ligands.
We thank Dr. Dominique A. Weber and Larry E. Westerman for advice and technical assistance.
This work was supported by National Institutes of Health Grants AI30554 and AI33614.
Abbreviations used in this paper: Ir gene, immune response gene; BINS, beef insulin; PINS, pork insulin; MINS, mouse insulin.