Dm Determines the Cryptic and Immunodominant Fate of T Cell Epitopes

The T cell response in the H-2^d haplotype mice immunized with hen egg white lysozyme (HEL) is almost exclusively directed towards a single immunodominant epitope of HEL, peptide 106–116, bound to E^d MHC molecules (106–116/E^d; references 1–4). Immunization with HEL does not elicit T cells specific for HEL 11–25 or HEL 20–35 bound to A^d MHC 2,3,4. However, HEL 11–25/A^d– and to some degree, HEL 20–35/A^d–specific T cells, can be activated by immunization with the peptides 11–25 and 20–35, respectively 2. Peptides 11–25 and 20–35, which require no further processing in order to be efficiently presented in the groove of the A^d molecules, thus encode cryptic determinants in the BALB/c (H-2^d) mice. Immunodominance is a distinctive and classic characteristic of most T cell immune responses 1,5,6. Nevertheless, the molecular mechanisms dictating the immunodominant and cryptic fate of T cell epitopes remain unknown.

Processing and presentation of exogenous protein antigens taken up via the endocytic pathway involve the trafficking of MHC class II molecules from the constitutive secretory pathway to the endocytic compartment (containing the antigen-derived peptides), en route to the cell surface 7,8. This task is accomplished by a glycoprotein called invariant chain (Ii) that enhances the transport of MHC class II chains to the endocytic compartment 7,8. In addition, a fragment of Ii, called class II–associated invariant chain peptide (CLIP), occupies the peptide-binding groove of MHC class II molecules 9 until DM, a nonclassical MHC class II heterodimer, induces its release and replacement with peptides derived from foreign or endogenous antigens 7,10,11,12. Due to a lysosomal targeting signal encoded within the β chain of DM heterodimers, DM is localized to endocytic MHC class II compartments 13 where it has been shown to transiently bind MHC class II molecules 12,13. DM, in addition to catalyzing the release of CLIP, has been shown to influence the binding of several non-CLIP peptides to the MHC class II molecules 11,12,14,15,16,17. DM mutant mice and APC lines have been used to show that DM is required for presentation of certain foreign antigens 12. We have shown previously that expression of DM can increase, diminish, or have a neutral effect on the presentation of different intracellular self proteins 14,18,19. But the biological consequence of the DM-mediated modification of the repertoire of peptides displayed by the MHC molecules on expression of the epitope specificity of T cell immune responsiveness to a foreign or self antigen remains unexplored. We present evidence here that DM directs the cryptic and immunodominant fate of antigen determinants.

L cells were initially transfected with genes encoding A^d alone, and a positive clone was isolated and then supertransfected with genes encoding genomic Ii (Ii or gIi) or Ii and murine DM. Derivation and characterization of these cells have been described previously 14,18. A^d-positive cells expressing either no cofactor, Ii alone, or Ii and H-2M (or murine DM, hereafter referred to as DM) were then additionally transfected with genes encoding E^d, and positive cells were obtained by preparative sorting using the Eα-specific Ab 14-4-4S and magnetic beads. Sorted and other transfected cells were grown in selection media with appropriate drugs: A^d and A^d-Ii L cells were cultured in the presence of G418 (200 μg/ml; GIBCO BRL), A^d-Ii+DM, in G418 and blasticidin (100 μg/ml; Invitrogen), A^dE^d and A^dE^d-Ii, in G418 and HAT (GIBCO BRL), and finally A^dE^d-Ii+DM, in HAT, G418, and blasticidin. Cells were transferred into normal media 14 containing no drugs 24–48 h before their use in T cell assays.

mAbs reacting with A^d (MKD6) or E^d (14-44S) were obtained from the American Type Culture Collection and used as culture supernatants. Cell surface expression of MHC class II molecules was evaluated by flow cytometry as described 19. In brief, cells were incubated successively with mAbs, followed by secondary step FITC-conjugated goat anti–mouse Igs (IgA, IgG, and IgM, and FITC-conjugated goat anti–mouse; Sigma-Aldrich), and were then analyzed on a FACScan™ cytofluorimeter (Becton Dickinson). Background fluorescence was determined by incubating cells with media alone or an irrelevant mAb, and then with FITC-conjugated goat anti–mouse.

Cells were harvested from culture by mild trypsin treatment, washed extensively, and then counted. 8 × 10⁶ cells of each type (cells expressing A^dE^d alone, A^dE^d Ii, or A^dE^d Ii+DM) were solubilized in 1.0 ml 0.5% Triton X-100 (Sigma-Aldrich) with protease inhibitors for 45 min on ice as described previously 19. Nuclei and other insoluble debris were pelleted by centrifugation and the supernatant was transferred to a fresh tube. An equivalent aliquot of each cell lysate, equal to 8 × 10⁵ cell equivalents, was then adjusted to 2.0% SDS, 62 mM Tris, pH 6.8, and 10% glycerol, boiled for 3 min, and then fractionated by SDS-10% PAGE. After electrophoresis, proteins were transferred to nitrocellulose, then processed for Western blotting as described previously 19. Blots were first probed and developed using nonspecific rat mAb or normal rabbit serum Abs, and then probed specifically for Ii and DM protein expression, as indicated in the legend to Fig. 1 b. Ii was detected by probing with the rat mAb In-1 20, and DMβ was detected by probing with a rabbit antisera raised against a synthetic peptide corresponding to the COOH-terminal tail of the mouse DMβ chain, conjugated to KLH. Binding of primary Abs to the blot was detected with horseradish peroxidase–coupled secondary goat Abs specific for either rat Ig, purchased from KPL Laboratories, or rabbit Ig, purchased from Life Technologies. Western blots were processed for chemiluminescence and autoradiography as described previously 18.

For analysis of Ii, and DM expression within the APC, 0.5 × 10⁶ cells were plated in two 100-mm tissue culture dishes for 24 h. The adherent cells were then prelabeled for overnight incubation in MEM deficient in leucine with added [³H]leucine obtained from NEN Life Science Products at 25 μCi/ml. Ii, DM, and MHC class I molecules were isolated from 1% Triton detergent lysates by immunoprecipitation with either rabbit antisera specific for the cytosolic tail of DMβ, the rat mAb In-1 20 reactive with the cytosolic tail of Ii, or the class I–specific mAb 16.1.11N 21. Immunoprecipitated proteins were fractionated by SDS-10% PAGE and gels were stained with Coomassie brilliant blue, then destained with 30% methanol/10% acetic acid. Stained and destained gels were then washed with water, treated with Flouro-Hance (Research Products International), dried, and exposed to film at −80°C.

HEL was either purified as described previously 22 or was obtained from Sigma-Aldrich as a purified preparation. Both preparations of HEL gave analogous results. Sperm whale myoglobin (SWM) was obtained form Accurate Chemicals and Scientific Corp. Peptides HEL 11–25, HEL 20–35, HEL 106–116, and SWM 102–118 were synthesized and purified by Macromolecular Resources.

The cloned T cell hybridomas ISG-9-49, ISG-9-69, and 17.B.8.1 were obtained from continuously growing HEL-specific T cell lines derived from LN cells of HEL-primed B10.GD mice as described previously 22,23. The cloned T hybrids D2-4-18 and D2-4-8 were obtained from HEL-specific T cell lines derived from LN cells of B10.D2 mice immunized with HEL 22,23. In brief, T cell blasts from T cell lines were fused with the TCRα⁻β⁻ variant of BW5147 as a fusion partner and cloned by limiting dilution in the presence of HEL. HEL-reactive hybrids were further analyzed for recognition of peptides. The hybrid C7-R14, specific for SWM 102–118 24,25, was a gift from Dr. Jay Berzofsky (National Institutes of Health, Bethesda, MD).

T hybridoma cells (10⁵) were cultured with 1–5 × 10⁴ of either A^d or A^dE^d APCs, A^d-Ii or A^dE^d-Ii APCs, or A^d-Ii+DM or A^dE^d-Ii+DM APCs with different concentrations of HEL, SWM, or peptide 22,23, as indicated in the figure legends (Fig. 2,Fig. 3,Fig. 4,Fig. 5). Fixed APCs, where indicated, were treated with 1% formaldehyde in PBS for 12 min, then washed extensively before use. All cultures were done in triplicate. The culture supernatants collected 24 h later were assayed for IL-2 by ELISA as described previously 26. The substrate used was O-phelylenediamine dihydrochloride (Sigma-Aldrich). Absorption at 490 nm was measured using a microplate reader (Molecular Devices).

Cloned populations of L cells deficient in MHC class II, Ii, and DM heterodimers (DM) were transfected with genes encoding A^d MHC molecules followed by either Ii or Ii and DM (Ii+DM), as described previously 14,19. The A^d, A^d-Ii, and A^d-Ii+DM L cell lines were further transfected with genes for E^d MHC molecules. Fig. 1 shows the characterization of L cells transfected with A^d and/or A^dE^d MHC molecules, together with genes for either Ii only or with Ii and the DM heterodimers (Ii+DM). It is clear that the expression of MHC class II is optimal in cells expressing both the Ii and DM (Fig. 1 a), as has been observed previously 14,27. Protein expression of Ii and DM are shown in Fig. 1b and Fig. c, confirming that the original cell lines used for transfection lack the constitutive expression of Ii and DM glycoproteins, and that protein expression of Ii and DM was readily detected in the appropriate cells.

Fig. 2 shows the display of the cryptic HEL determinant 11–25 by A^dE^d L cells used as APCs at different concentrations of HEL. The determinant display was examined by analyzing activation of two different cloned T cell hybridomas, ISG9-49 and ISG-9-69, that recognize HEL 11–25/A^d and were derived as described previously (22; see Materials and Methods). It is clear that the A^dE^d-Ii APCs, lacking expression of DM heterodimers, can efficiently process HEL and present the cryptic epitope 11–25. The display of 11–25 epitope by the A^dE^d-Ii APCs can be observed at even low (0.2–2 μM) concentrations of HEL. Most strikingly, the presentation of 11–25 was nearly extinguished, even at >30-μM concentrations of HEL, when presented by APCs that express DM in addition to Ii (Fig. 2). The peptide 11–25, which requires no further processing, is equivalently presented by APCs with Ii alone and with Ii+DM (insets, Fig. 2, a and b), despite the fact that APCs that express Ii alone display lower levels of MHC class II molecules than Ii+DM APCs (Fig. 1 a). To further confirm DM-mediated antagonism of presentation of the epitope 11–25, we used an additional set of APCs, the A^d APCs, expressing either Ii alone or Ii+DM. Once again, the presentation of the cryptic epitope 11–25 was dramatically diminished when DM molecules were expressed by the APCs. A^d-Ii and A^dE^d-Ii APCs lacking DM molecules were clearly very active at presenting this cryptic epitope in the absence of DM (Fig. 2 and Fig. 3). In addition, we used fixed APCs for presentation of the cryptic epitope 11–25 to address the question whether the A^d-Ii APCs were able to effectively present the cryptic HEL 11–25 simply by virtue of being more efficient in taking up either denatured HEL or any other contaminants of HEL. As is evident in Fig. 3b and Fig. d, the fixed A^d-Ii or A^d-Ii+DM APCs are unable to present HEL even at the highest concentration, 100 μM, of HEL. The competence of the A^d-Ii APCs to present the cryptic epitope is thus solely due to proficient intracellular processing of HEL. Results obtained using a third 11–25/A^d–specific cloned T hybrid, GD-5-42, additionally validated the DM-mediated antagonism of presentation of 11–25/A^d, even at high concentrations (25–100 μM) of HEL (data not shown).

Fig. 4 shows the display of an additional cryptic epitope of HEL, 20–35/A^d, by A^d APCs. Interestingly, presentation of this second cryptic epitope of HEL was also antagonized by the expression of DM within APCs. Once again, A^d-Ii APCs were able to present this epitope very efficiently to 20–35–specific T cells, 17B.8.1. However, A^d-Ii+DM APCs were severely compromised in their ability to display this epitope. The fixed A^d-Ii or A^d-Ii+DM APCs, as shown for the 11–25/A^d cryptic epitope, were unable to present this epitope from HEL at even the highest concentration of HEL tested (Fig. 4 b), thus ruling out the possibility that the expression of the epitope 20–35/A^d was a result of efficient uptake of denatured HEL.

We then examined the display of the immunodominant determinant of HEL, 106–116/E^d, by A^dE^d APCs using the cloned T cell hybrids D2-4-18 and D2-4-8. Fig. 5 shows that in sharp contrast to our observations for the HEL 11–25 and 20–35 determinants (Fig. 2), A^dE^d-Ii APCs, lacking DM molecules, were very poor at displaying this immunodominant epitope (Fig. 5), whereas cells additionally expressing DM showed efficient presentation. The half-maximal response of the T cell hybrid D2-4-18 was obtained at 2 μM HEL using A^dE^d-Ii+DM APCs. However, the activation of this T hybrid essentially remained at the baseline levels at concentrations of up to 33.3 μM HEL, using A^dE^d-Ii APCs lacking DM. Optimal processing and display of this immunodominant determinant is thus dependent on the expression of DM molecules. Furthermore, these results attest that the A^dE^d-Ii+DM APCs are functionally competent and the abrogation of presentation of the cryptic 11–25 determinant in the presence of DM was not due to some non-DM–related factor expressed by the Ii+DM APCs. The data presented in the insets (Fig. 5, a and b) show that the A^dE^d-Ii and A^dE^d-Ii+DM APCs were able to stimulate the 106–116/E^d T cells equivalently when the peptide HEL 106–116, which requires no further processing (data not shown), was used. These results indicate the lack of any intrinsic deficiency in A^dE^d-Ii APCs to interact with 106–116/E^d–specific T cells.

To pursue the argument that one function of the DM molecules is to play a role in the manifestation of immune dominance upon immunization with protein antigens, we examined the effect of DM on the presentation of a dominant determinant of SWM 23. Fig. 6 shows that just like the immunodominant epitope 106–116/E^d of HEL, the optimal display of the immunodominant 102–118/A^d determinant of SWM 23,24 by A^d APCs was dependent on expression of DM in these APCs. The half-maximal response of SWM 102–118–specific C7-R14 T cell hybrids stimulated by the A^d-Ii+DM APCs was induced at only ∼1 μM SWM, while a similar response by the A^d-Ii APCs was induced at 60 μM SWM. Again, the results in Fig. 6 not only support the conclusion that the immunodominant fate of T cell determinants is dependent on the expression of DM, but they also affirm that the failure of our set of A^d-Ii+DM APCs (Fig. 3 and Fig. 4) to present the cryptic epitope 11–25 is unrelated to any non-DM–related attribute of these cells.

The focusing of T cell responses directed to protein antigens to only a limited number of potential T cell epitopes is a classical feature of antigen-specific immune responses 1,5,6. The evolutionary development of this basic principle appears to be to accomplish (a) avoidance of activation of an army of T cells unnecessary for a successful attack on a pathogen; and (b) construction of an effective regulatory T cell system directed towards a small subset of T cells 28 specific for the immunodominant epitopes. T cell responses directed towards complex pathogens such as Leishmania have been shown to be, astonishingly, focused on a single immunodominant epitope 29,30. A variety of concepts including determinant capture, nature of APCs, proteolytic enzymes in distinct intracellular microenvironments, lymphokines, and general antigen processing have been proposed to explain this classic feature of the immune system 1,3,31,32,33. The molecular mechanisms to explain these hypotheses still remain undetermined. We now show that DM, an MHC-encoded heterodimer, previously shown by us and others 14,16,17 to influence the repertoire of self-peptides loaded onto MHC class II molecules, may well be the guide that directs the antigen-derived peptides to have either an immunodominant fate or a cryptic fate. The two immunodominant determinants (HEL 106–116/E^d and SWM 102–118/A^d) examined by us are dependent on DM for presentation (Fig. 5 and Fig. 6). A third, the Leishmania-derived, immunodominant determinant Leishmania homologue of receptors for activated C-kinase (LACK [29, 30]) 161–173/A^d, also requires the expression of DM molecules in order to be displayed on the surface of APCs (data not shown). In the case of HEL (and perhaps also of other protein antigens), DM determines the immunodominant fate of epitopes by strongly augmenting the presentation of the dominant determinant, 106–116/E^d, on APCs, while simultaneously extinguishing the expression of the two cryptic epitopes studied in this report, 11–25/A^d and 20–35/A^d. DM-mediated antagonism was also observed for the display of 74–96, a minor cryptic epitope 4,23 of HEL (our preliminary observations, not shown). The results reported here also indicate that the cryptic fate of the 11–25/A^d epitope is clearly not due to “determinant capture” 1. The concept of determinant capture predicts that during the creation of the dominant 106–116/E^d determinant 1, the denatured HEL molecule would bind to E^d MHC, and the portion of HEL outside the binding groove (including the 11–25 region) would get digested by proteolytic enzymes 1,31. This outcome of the determinant capture is expected to be independent of expression of Ii and DM within APCs. Since the A^dE^d-Ii APCs are able to process and present this epitope remarkably efficiently (Fig. 2), it is clear that the E^d MHC molecules, while “capturing” the dominant epitope HEL 106–116/E^d, are not responsible for destroying the 11–25 epitope. Instead, it is the expression, or lack of DM molecules in APCs, that determines the presentation of the cryptic epitope.

Our results are intriguing to consider in light of an earlier report that the dendritic cells, but not B cells, focus T cell responses against the immunodominant peptide of HEL and are unable to present the peptide HEL 7–31 34. One possible explanation for this differential determinant display could be the cell type–specific expression within B cells, but not dendritic cells, of DO 35, an additional MHC class II heterodimer that has been shown to inhibit the function of DM molecules 35,36. It has been shown that DM specifically induces the release of the MHC-bound peptides that display a lower binding stability, while favoring the binding of peptides that show a higher stability of binding MHC class II molecules 17. In view of the optimal DM function and expression in the acidic endosomal or lysosomal compartments 11,16,17, we examined MHC binding and display of these peptides using fixed APCs 37 at a variety of pH. We found that at a pH between 4 and 5, a 10-fold lower concentration of the immunodominant peptide (106–116/E^d) was able to result in half-maximal stimulation compared with the stimulatory requirements for the cryptic peptide (11–25/A^d). However, at pH 7.0, equivalent concentrations of both peptides were needed for half maximal stimulation (data not shown). These observations could be the result of distinct pH optima of peptide binding to E^d (∼pH 4.5) and A^d molecules (∼pH 5.5), differential off-rates for the two peptides at the lower pH 38, or effects of DM expressed on the surface of cells, as has been shown in a recent report 39.

It has been proposed that the DM-induced editing of the repertoire of MHC class II–bound peptides could be a function of differential off-rates of the peptide–MHC binding, leading to differences in the stability of the peptide–MHC complexes 17,40. In addition, it can be envisaged that DM would selectively enhance the peptide binding within the MHC–peptide complexes that show optimal binding at lower pH, leading to the immunodominance of such epitopes. In this context, it is interesting that the only immunodominant epitope in both the HEL/H-2^d and cytochrome c/H-2^k systems 41, and one of two dominant epitopes in the SWM/H-2^d system 24, are restricted by I-E molecules (the latter with lower pH optima than the I-A molecules). The precise features that contribute to the DM-mediated effects on the differential display of peptides as seen in the current report, including the relative stability of the 11–25/A^d and 106–116/E^d complexes using purified MHC molecules, which is yet undetermined, are avenues of future investigations.

In summary, we propose that the cryptic and immunodominant fate of antigenic epitopes is a consequence of the ability of DM to skew the repertoire of peptides displayed by the MHC class II molecules. Our results that expression of DM within APCs is able to dictate the fate of two cryptic and three dominant epitopes testify that the DM-mediated decree of the nature of an antigenic epitope could be a general phenomenon. This function of DM may have evolved in parallel with the “immunodominance” attribute of the immune system as a mechanism to trigger only selective T cell responses essential to fight a pathogen. It is now becoming clear that one is tolerant to only the dominant determinants of self-proteins and that the T cell repertoire to the cryptic determinants remains intact in the host 5,42,43. It is the T cell repertoire to the cryptic self-epitopes that becomes activated in autoimmune disease 5,42,43. Any variation in the levels or activity of DM heterodimers thus may have consequences for autoimmunity as well as response to pathogens.

We thank Georgia Angelakos, Frederick Krafcik, and Nasa Valentine for excellent technical assistance, and John Katz for assistance in the analyses of the L cell transfectants. We are grateful to Dr. J. Berzofsky for providing us with the SWM-specific T cell hybrids, and Dr. E. Sercarz for purified HEL. We also thank Drs. E. Sercarz and S. Scheider for providing us with recloned, recharacterized 17B.8.1 T cell hybrid.

This work was supported in part by National Science Foundation grant MCB 9808841 (to N.K. Nanda), and by National Institutes of Health grant PO DK49799 and Juvenile Diabetes Foundation International (to A.J. Sant).