Influences of Transporter Associated with Antigen Processing (TAP) on the Repertoire of Peptides Associated with the Endoplasmic Reticulum–resident Stress Protein gp96

BALB/c, B10.D2, BALB.B, 129/Sv, and C57BL/6 mice were obtained from Charles River WIGA (Sulzfeld, Germany) and were maintained in the animal facilities at the Institute for Cell Biology. Antibodies to gp96 (anti-grp94, SPA-850, clone 9G10) were obtained from Stressgen Biotechnologies Corp. (Victoria, BC, Canada).

gp96 was purified as described (4). In brief, P815, P13.1, RMA, RMA-S cells, or their β-gal transfectants were grown in roller bottles to generate a 60-ml cell pellet. This pellet was homogenized in 200 ml of hypotonic buffer (30 mM NaHCO₃, 0.5 mM PMSF, pH 7.1) and separated from the supernatant by a 100,000 g centrifugation. The supernatant was applied to a Con A–Sepharose column and eluted with PBS containing 1 mM MgCl₂, 2 mM KCl, and 6% α-methyl–mannoside. The eluate was dialyzed against 5 mM phosphate buffer, pH 7.0, and separated on FPLC using a MonoQ column (5/5, Pharmacia) with a NaCl gradient from 0 to 1 M. Fractions were tested in SDS-PAGE and Western blot analysis using a mAb specific for gp96. Only fractions containing gp96 and no other detectable contaminating proteins were pooled and concentrated using centricon 10 microconcentrators (Amicon). The approximate concentration was determined by measuring the OD at 280 nm using an extinction coefficient of 1.0.

8- to 10-wk-old female BALB/c and 129/Sv mice were immunized intraperitoneally with 10⁷ irradiated C57BL/6 spleen cells (33 Gy) in 300 μl PBS or with 30 μg gp96 purified from different cell lines in 300 μl PBS. 10 d later, spleens were removed and the splenocytes were stimulated with irradiated (33 Gy) spleen cells from B10.D2 or C57BL/6 mice or synthetic β-gal peptide (TPHPARIGL; 1 μM) for 5 d.

RMA, RMA-S, P815, and P13.1 (12) cells were cultured in RPMI-1640 containing 10% FCS, 2-ME, l-glutamine, and antibiotics. CTL lines were generated by weekly restimulations with irradiated spleen cells from B10.D2 or C57BL/6 mice and synthetic peptides, if applicable, as described (13). CTL assays were performed as described (13) using RMA, RMA-S, P815, P13.1, and Con A blasts of spleen cells as target cells. Con A blasts were generated as described (14).

RMA and RMA-S cells were cotransfected with 2 μg of the plasmid pCH110 (Pharmacia) expressing the Escherichia coli–derived β-gal under the control of the SV40 promotor and 10 μg of the plasmid pBJ1neo to allow G418 selection of the transfectants. Transfectants were selected with 1 mg/ml G418 (GIBCO BRL, Gaithersburg, MD). Surviving cells were cloned and tested for their expression of β-gal by their ability to cleave X-gal (5-bromo-4-chloro-3-indolyl-β-d-galactoside). Cleavage of X-gal was determined by incubating serial dilutions of an 1% NP-40 cell lysate of 10⁶ transfectants with 2.5 μg X-gal in 100 μl PBS at 37°C and measuring the absorbance at 620 nm after 8 h. Clones with comparable β-gal activity were selected and expanded for the purification of gp96.

To test the hypothesis that peptide binding to gp96 occurs inside the ER we transfected RMA and RMA-S cells with E. coli–derived β-gal resulting in cytosolic expression of the protein (12). β-gal–transfected RMA and RMA-S cell clones, called R/gal and RS/gal and expressing similar levels of the protein were selected (Fig. 1) and expanded for the purification of gp96 according to standard protocols (4). BALB/c mice were immunized with these gp96 preparations and as a control with gp96 from P815 or P13.1, the β-gal transfectant of P815. As shown in Fig. 2,A, only the immunization with gp96 from P13.1 but not P815 induces a CTL response against P13.1 or peptide-incubated P815 cells, underlining the specificity of the gp96 immunization. As shown in Fig. 2 B, the induction of β-gal–specific CTL requires the presence of functional TAP molecules, because only the gp96 preparation from R/gal but not RS/gal cells is able to induce CTL recognizing P13.1 or β-gal peptide-incubated P815 cells. This experiment clearly shows that β-gal–derived peptides are loaded onto gp96 molecules inside the ER, as this process is not observed in cells with a TAP defect. These results were reproduced in three independent experiments with a gp96 concentration (30 μg), which is in threefold excess of the amount required for optimal tumor protection (2) and CTL induction, as determined by titrations of gp96 over a range from 2 to 30 μg (data not shown).

In the following experiments, we analyzed whether the TAP-dependent peptide loading of gp96 molecules is a general phenomenon that can be observed for other antigens as well. Therefore, BALB/c mice (H-2^d) were immunized with gp96 molecules from R/gal and RS/gal cells to induce cross-priming against minor H antigens expressed in C57BL/6 mice (H-2^b), from which R/gal and RS/gal cells were generated. The spleen cells from the immunized mice were stimulated with B10.D2 splenocytes (H-2^d), sharing most of the minor H antigens with C57BL/6. As shown in Fig. 3, spleen cells from the untreated mice showed no CTL activity, because the induction of minor H-specific CTL requires in vivo priming (15), but spleen cells from mice immunized with gp96 preparations from both R/gal and RS/gal displayed CTL activity against B10.D2 Con A blasts. A contamination of gp96 molecules from RS/gal cells with those from R/gal cells can be excluded because in some experiments spleen cells from mice immunized with RS/gal gp96 were also stimulated with β-gal peptides and never showed β-gal–specific CTL activity. The results could also be reproduced with gp96 molecules from untransfected RMA and RMA-S cells (data not shown). Therefore, these experiments show that the loading of peptides, reflecting some of the minor H antigens that differ between BALB and C57BL mice, onto gp96 molecules is independent of TAP. The origin of these minor H antigens is not known, but obviously their peptides enter the ER and are loaded onto gp96 molecules by a TAP-independent mechanism. This is not a rare phenomenon because some minor H antigens can be also recognized by H-2^b–restricted CTL in a TAP-independent fashion, as shown earlier (14) and again in Fig. 4. RMA and RMA-S cells are lysed comparably by CTL specific for minor H antigens differing between BALB.B and C57BL/6 mice, now presented by MHC molecules of the H-2^b haplotype. TAP-independent loading of peptides onto gp96 and MHC class I molecules is not observed in all minor H incompatible strain combinations. When we immunized 129/Sv mice with gp96 molecules from RMA or RMA-S cells to induce a CTL response against minor H-antigens that differ between 129/Sv and C57BL/6 mice, we observed that only gp96 molecules from RMA but not RMA-S cells were able to generate a minor H-specific CTL response (Fig. 5,A). This is in contrast with the experiments shown in Fig. 3 where the identical RMA-S gp96 preparation induced a CTL response in BALB/c mice against B10.D2 Con A blasts. In line with the above results, only RMA but not RMA-S cells are recognized by the minor H-specific CTL induced by immunization of 129/Sv mice with C57BL/6 cells or RMA-derived gp96 molecules (Fig. 5 B). This supports the notion that peptide loading onto MHC and gp96 molecules happens in the same intracellular compartment.

Our results clearly show that the loading of gp96 molecules with T cell–recognized antigens is TAP dependent. Thus, gp96 must associate with peptide ligands inside the ER and its ability to induce specific CTL responses against antigens expressed in the cells from which it has been purified is not an artifact due to cell lysis during gp96 purification. This is evident from the immunization experiments with gp96 molecules isolated from cells with or without a defect in the TAP2 molecules and expressing a defined antigen, β-gal, exclusively in the cytosol. Only gp96 molecules isolated from TAP-competent cells induced a specific CTL response, indicating the association of β-gal–derived peptides with gp96 (Fig. 2). The same was observed for the induction of a CTL response against C57BL/6 cells in 129/Sv mice. Gp96 from RMA but not from RMA-S cells was able to immunize against C57BL/6 minor H antigens (Fig. 5,A). This effect is not due to variations in the quality of the gp96 preparations from the different cell lines because identical gp96 preparations from RS/gal or RMA-S cells that failed to induce TAP-dependent β-gal or 129/Sv anti-C57BL/6 minor H-specific CTL responses were able to induce TAP-independent BALB/c anti C57BL/6 minor H-specific CTL (Fig. 3). In some cases, this was observed in CTL cultures derived from the same mouse (Fig. 2,B; Fig. 3). We also observed that CTL induced by immunization with C57BL/6 spleen cells or RMA-derived gp96 molecules recognized RMA but not RMA-S target cells (Fig. 5 B). This shows that not only the gp96 but also the MHC molecules of RMA-S cells completely lack those minor H antigen-derived peptides that are recognized by 129/Sv anti C57BL/6 CTL. These experiments, revealing a TAP-dependent loading of peptides onto gp96 molecules, are complemented by results obtained in vitro where radiolabeled peptides were added to streptolysin-O-permeabilized cells. In this study, it was shown that peptides associated with gp96 molecules only when the cells expressed functional TAP molecules and ATP was present (7, 16).

On the other hand, not all peptides associated with gp96 molecules depend on the presence of functional TAP molecules. This is evident from the ability of both RMA and RMA-S gp96 preparations to induce cross-priming against B10.D2 cells in BALB/c mice (Fig. 3). For at least a subset of these H-2^d–restricted peptides, the association with gp96 molecules is TAP independent. Interestingly, the presentation of H-2^b–restricted peptides in a similar combination of minor H antigens was also found to be TAP-independent: BALB.B-derived CTL specific for C57BL/6 cells recognize RMA and RMA-S cells to a similar extent (Fig. 4). RMA-S cells have been previously reported to present other peptide antigens including minor H antigens despite their mutation in the TAP2 gene (14, 17). This presentation seems to be due to luminal production or to unknown intrinsic properties of these peptides, allowing either TAP-independent or just TAP1-mediated transport (16).

The above experiments indicate that gp96 does not exclusively receive its peptides from TAP molecules but also binds peptides that are present inside the ER due to TAP-independent translocation or that are generated inside the ER directly, possibly from ER-resident proteins. Our experiments might also provide an answer to the question how MHC molecules are loaded with peptides in TAP-deficient cell lines. In this situation, a direct loading of MHC molecules via their association with TAP (8, 18), or mediated by tapasin in human cells (19), is not possible. It seems feasible that gp96 or other ER-resident peptide-binding proteins like a 60-kD protein identified by Lammert et al. (7) or Marusina et al. (16), could be involved in these events. This might also include MHC class I loading of those peptides that require further processing before their binding to MHC molecules (20) and the removal of peptides from the ER that do not posses the constraints imposed by the ligand motif of the MHC molecules present in the ER. These functions were proposed for BiP (21) based on its ability to bind peptides in vitro (22). But it was clearly shown recently that BiP is not able to bind peptides translocated by TAP in streptolysin-O-permeabelized cells (7, 16). Therefore, new candidates, like gp96 or the 60-kD protein, are required for these services.

Even though the physiological role of gp96 is not defined yet, our results show that its peptide repertoire seems to reflect all the peptides that have entered the ER, either by TAP-dependent or -independent mechanisms. Combined with its ability to induce protective immune responses in vivo against a wide variety of antigens, including tumor antigens that are not further identified, our results provide the theoretical foundation for applying gp96 for the induction of immune responses against all peptides present in the ER, including those that are able to associate with MHC molecules.

This work was supported by a grant from the Bundesminister für Forschung und Technologie, the European Union (Biomed 2), and by the Deutsche Forschungsgemeinschaft (Leibnizprogramm Ra 369/4-1).