To study the role of subdominant epitopes in tumor rejection we have used EL4 tumor cells and their ovalbumin (OVA)-transfected counterpart E.G7. Immunization of mice with irradiated EL4 cells conferred protection against challenge with EL4 and E.G7. Surprisingly, immunization with irradiated E.G7 cells did not protect against a subsequent challenge with EL4 or E.G7. Growth of E.G7 tumors in E.G7 immunized mice was not due to loss of expression of OVA or MHC I by the tumor cells in vivo. Adoptive transfer of OVA-specific transgenic T cells, immunization of mice with native or heat-denatured OVA or infection with a recombinant virus expressing OVA also failed to induce rejection of E.G7 tumors. Lack of immunogenicity of the OVA-expressing tumor could not be overcome by combination of a CD40 activating antibody with immunization against E.G7 or OVA. Our results suggest that immunization against subdominant epitopes is more effective than vaccination against dominant epitopes.
This article was published in Cancer Immunity, a Cancer Research Institute journal that ceased publication in 2013 and is now provided online in association with Cancer Immunology Research.
T cells play an important role in tumor rejection. Demonstration in cancer patients of a cytotoxic T lymphocyte response against a number of molecules and the identification of CTL epitopes on tumors and normal cells have encouraged the belief that tumors can be controlled by the immunization with such epitopes. Regardless, tumors have been generally observed to grow progressively even in the presence of a T cell response (1, 2, 3, 4, 5).
One mechanism by which tumor cells escape the anti-tumor immune response is the down-regulation, mutation or loss of tumor antigens (6, 7, 8). Immunization with complex antigens elicits CTLs only against a small number of dominant epitopes out of a broad range of potential epitopes. CTL responses against subdominant epitopes are weak or not measurable, but can be evoked if a vaccine misses the dominant epitopes (9, 10, 11). For an optimal anti-tumor response it might be useful to induce CTLs against subdominant epitopes for two reasons: (a) A combined vaccination against more than one epitope, including subdominant epitopes, could prevent the outgrowth of epitope loss variants of the tumor cells, and (b) many of the human tumor 'antigens' characterized thus far are dominant 'antigens' and have also been detected in normal cells (12, 13). Tolerance to these self-molecules is likely to exist. Subdominant epitopes may not elicit such tolerance (14, 15).
In this backdrop, we have explored the role of subdominant epitopes in tumor rejection. As model tumors, we chose the EL4 thymoma and its OVA-transfected counterpart E.G7 (16). OVA is a well-characterized antigen with two defined dominant CTL epitopes: ova257-264 (17) and ova176-183 (18). The parental tumor line EL4 presents one dominant epitope that has not yet been sequenced (15). However, in E.G7 cells OVA-epitopes are dominant over the endogenous EL4 epitopes (16, 19, 20, 21). Previous studies have shown that rejection of EL4 (15) and E.G7 (19, 22) is mediated by CD8+ T cells.
E.G7 cells induce OVA-specific CTL but can not protect against E.G7 tumor growth
We examined the hierarchy of T cell epitopes in E.G7 cells by a CTL assay. C57BL/6 mice were immunized with irradiated E.G7 cells as described in Materials and Methods, and splenocytes from the immunized mice were stimulated in vitro with the dominant OVA epitope 257-264. In CTL assays, the mixed tumor lymphocyte cultures showed lytic activity against E.G7, but not against EL4 tumor cells. Vaccination with EL4 cells induced CTLs against both EL4 and E.G7 (Fig. 1A). We deduced that the OVA epitope SIINFEKL is dominant over the endogenous EL4 epitopes, as measured in vitro. In studies in vivo, immunization with EL4 cells conferred protection against EL4 and against E.G7. However, immunization with irradiated E.G7 cells did not protect against a subsequent challenge with EL4 nor E.G7 (Fig. 1B). This was surprisingly, as vaccination with E.G7 was observed to elicit OVA-specific CTLs as shown in Fig.1A.
Lack of immunogenicity of E.G7 is not due to antigen loss
We isolated cells from tumors growing in E.G7-immunized and E.G7-challenged mice and used them without further cultivation as targets in a CTL assay. Cells derived from five of six tumors were recognized by OVA-specific CTLs to the same extent as ova257-264 peptide-pulsed cells from the same tumor (Fig. 2 A, C-F). Thus, loss of expression of OVA or MHC class I does not explain the growth of E.G7 tumors in E.G7 immunized mice.
One tumor whose cells could not be killed by OVA-specific CTLs had lost the expression of the ova257-264 epitope, but not MHC class I, as it could be lysed after being pulsed with the SIINFEKL epitope (Fig. 2B). From this tumor we established a cell line "E.G7 OVA LOSS". Vaccination of C57BL/6 mice with irradiated E.G7 OVA LOSS cells was observed to confer protection against a subsequent challenge with live E.G7 cells (Fig. 3A). This observation was similar to the results obtained by immunization with EL4. Thus, the failure of E.G7 cells to elicit a protective immune response against itself could be attributed to an immune response against the dominant epitope of OVA.
Adoptive transfer of OVA-specific T cells and immunization with OVA fail to induce protective immunity against E.G7
We adoptively transferred naive and in vitro stimulated spleen cells from the transgenic mouse line OT-1 that expresses a T cell receptor specific for ova257-264 in the context of H-2Kb molecules (23, 24) into normal C57BL/6 mice. Recipient mice could not reject a subsequent challenge with live E.G7 cells (Fig. 4A and C) although the in vitro stimulated OT-1 spleen cells demonstrated high cytolytic activity in a CTL assay (data not shown).
To investigate the fate of OVA-specific T cells in vivo we adoptively transferred CD45.1-positive OT-1 spleen cells into normal C57BL/6 mice (CD45.1-negative). The mice were then challenged with buffer or E.G7 or EL4 cells or were immunized with 5 mg OVA protein i.p. The spleens and draining and non-draining lymph nodes of recipient mice were examined by cytofluorometry. Six days after tumor challenge, spleens and lymph nodes of mice challenged with EL4 or E.G7 cells had the same numbers of CD45.1+ CD8+ cells as the control mice that had received buffer after adoptive transfer (data not shown). In contrast, we observed expansion of OT-1 cells in mice immunized with OVA (1.76% CD45.1+CD8+ lymph node cells 6 days after OVA injection compared to 0.26% in control mice), as described earlier (25).
We tested the ability of immunization with OVA protein to elicit resistance to E.G7 and CTLs against SIINFEKL/Kb. Mice were immunized with native or heat-denatured OVA or infected with recombinant vesicular stomatitis virus-expressing OVA (VSV-OVA) (25) and tested for CTL activity. All forms of antigen elicited potent CTL response to SIINFEKL/Kb (Fig. 5A). Nevertheless, all groups of OVA-immunized mice failed to reject the E.G7 tumor (Fig. 3B and 3C, 6A). Surprisingly, mice infected with wild type VSV rejected the E.G7 tumor (Fig. 6B). Thus, the introduction of the OVA antigen into the vaccine in fact abrogated the protective unspecific immune response conferred by VSV infection in this set of experiments.
Vaccination of mice with OVA in combination with anti-CD40 antibodies has been shown to induce memory CTLs against OVA (26). However, combination of immunization with irradiated E.G7 or heat-denatured OVA with injection of a CD40-activating antibody also failed to elicit rejection of E.G7 tumors (Table 1).
The experiments presented here indicate that the presence of the dominant antigen OVA in a complex tumor vaccine abrogates an otherwise protective immune response against the OVA-expressing tumor E.G7. Irradiated EL4 cells can protect against the growth of EL4 tumors and the OVA-expressing cell line E.G7. However, if mice are immunized with E.G7 cells, they can not reject EL4 or E.G7 tumors. Our results are consistent with other studies that show failure of irradiated E.G7 cells to protect against a subsequent E.G7 challenge (4, 19). As also shown by other groups, immunization with free ova257-264 peptide in incomplete Freund's adjuvant or PBS (19, 27) and with native OVA (28) fails to protect from E.G7 tumor growth. Previous studies in other systems have also shown protection from tumor growth by CTLs against subdominant epitopes (15, 29, 30) or viral infection (31, 32).
In contrast, protection against E.G7 tumors was observed after i.p. injection of OVA by Fenton et al. (33) and el-Shami et al. described rejection of E.G7, but not of EL4, tumors following immunization with ova257-264 peptide-pulsed RMA-S.B7 cells (21). Interestingly, the initial CTL response in the latter study was confined to the dominant OVA epitope 257-264; however, after rejection of E.G7, epitope spreading of the CTL response to two other OVA epitopes and to endogenous EL4 epitopes occurred, and was presumably responsible for rejection. Our results are thus consistent with the latter study.
Rejection of E.G7 tumors has also been achieved by immunization with dendritic cells pulsed with ova257-264 peptide or OVA (19) or with dendritic cells infected with an OVA-expressing adenovirus (34). Surprisingly, direct infection of mice with this virus could not induce rejection of a subsequent E.G7 challenge, although it elicited a strong CTL response (34). Our results that vaccination against OVA can not protect from E.G7 tumors, despite the induction of an effective OVA-specific CTL response, are consistent with that experience.
Several mechanisms have been described to explain the failure of the immune system to induce an effective anti-tumor response. Tumor cells can escape the immune response by down-regulation of their MHC expression or by in vivo selection of tumor cells that have lost the expression of the target antigen (35, 36, 37). In our system, the majority of tumors in E.G7 challenged mice were recognized by OVA-specific CTLs in a CTL assay, suggesting that growth of E.G7 tumors after E.G7 immunization can not be explained by antigen loss of the tumor cells.
Tolerance can be due to ignorance of the antigen (38, 39) or to anergy if the antigen is not presented with a sufficient co-stimulatory signal (40, 41). Deletion or suppression of activated CTLs can occur (42, 43, 44, 45, 46, 47) as well as migration of CTLs away from the tumor site (4). The precursor frequency of CTLs can be too low, so that the T cells loose the competition with the growing tumor, or the tumor cells could produce humoral factors that locally block cytotoxicity (reviewed in 1). We exclude the possibility that the OVA antigen is ignored by the immune system since we observe effective priming of OVA-specific CTLs after immunization with irradiated E.G7 cells or with OVA protein and after infection with VSV-OVA. Moreover, the CTL response against OVA after immunization with irradiated E.G7 cells dominates the response to endogenous EL4 epitopes. The failure to induce a CTL response against the subdominant EL4 epitopes better explains the growth of EL4 tumor after immunization with E.G7 in our study.
The adoptive transfer of high numbers of naive or in vitro stimulated OVA-specific T cells into normal C57BL/6 mice was not able to protect against E.G7 tumor growth, suggesting that the failure of the immune system to control E.G7 tumor growth is not a quantitative problem of OVA-specific precursor T cells. Although injection of native OVA was able to trigger vigorous expansion of OVA-specific CD8+ T cells in vivo, neither native nor heat-denatured OVA could protect mice from E.G7 tumor growth. To sum up these data, effective priming of CTLs as measured by CTL assays in vitro does not correspond with protection against tumor growth in vivo.
Why do CTLs against epitopes that are subdominant in OVA or E.G7 protect against OVA-expressing tumors whereas dominant CTLs do not? A possible explanation is a transient activation of dominant CTLs after antigen contact followed by a down-regulation of this response due to suppressor cells or deletion of these CTLs (42, 43, 44, 45, 46, 47). Failure to maintain an anti-tumor CTL response induced by viral infection despite the presence of antigenic tumor cells has been previously described (48). These mechanisms of down-regulation may be operative for the stronger, and not for the weaker, response against subdominant epitopes.
The results presented here indicate that vaccination against a dominant antigen, by itself or in context with tumor cells or viral infection, may not necessarily confer tumor protection, and may indeed obstruct generation of effective and protective immune response to other antigens. Immunization against subdominant epitopes may circumvent this hurdle. They also indicate that CTL response is not a surrogate for protective tumor immunity. Some of these points have been argued previously (49).
Materials and methods
Mice and cell lines
C57BL/6 and bm1 mice (50) were purchased from The Jackson Laboratory (Bar Harbor, ME). C57BL/6-Ly-5.2 mice were obtained from Charles River (Wilmington, MA) through the National Cancer Institute animal program. The OT-1 mouse line (23) was generously provided by W. R. Heath (WEHI, Parkville, Australia) and F. Carbone (Monash Medical School, Prahran, Victoria, Australia) and was maintained as a C57BL/6-Ly-5.2 line.
EL4 cells are a thymoma of C57BL/6 origin. The E.G7 cell line was derived from EL4 by transfection with chicken OVA cDNA (16). Cells were maintained in RPMI 1640 without glutamine (Gibco, USA) supplemented with 5% fetal calf serum, 1% L-glutamine, 1% penicillin/streptomycin, 1% sodium pyruvate and 1% non-essential amino acids (all from Gibco, USA). Media for E.G7 cells also contained 400 µg/ml G418 (Gibco, USA).
The cell line "E.G7 OVA LOSS" was established from tumors growing in mice immunized and subsequently challenged with E.G7 cells. C57BL/6 mice were vaccinated and challenged with E.G7 cells as described below. Tumors from these mice were isolated 18 days later and a single cell suspension was prepared. Cells were grown in complete medium without G418 as described above.
Immunization, virus infection and tumor challenge
For the immunization of mice, tumor cells (EL4, E.G7 or E.G7 OVA LOSS) were irradiated with 7500 rad and washed three times. 2 x 107 irradiated tumor cells were injected twice s.c., with a one-week interval. For the two simultaneous subcutaneous immunizations with E.G7 on one flank and EL4 on the other, we used 1 x 107 cells of each line.
Immunization with OVA was performed with 5 mg of native OVA (Grade VI, Sigma) injected i.p. or 1 mg of heat-denatured (boiled for 5 min) OVA administered by s.c. injection. For CTL induction we injected 1 mg native or heat-denatured OVA s.c.
For virus infection, 1 x 106 PFU of VSV or VSV-OVA were injected i.v. The production of VSV-OVA has been described previously (25). One week after the last immunization or two weeks after virus infection mice were challenged s.c. with 2.5 x 106 EL4 or E.G7 cells, respectively.
Measurement of cytolytic activity
Mice were splenectomized one week after the last immunization or after viral infection. 8 x 106 spleen cells from immunized mice were cultured in 24-well plates as a mixed lymphocyte tumor culture with 8 x 104 irradiated (7500 rad) tumor cells. Alternatively, spleen cells of immunized mice were stimulated in vitro with 10-6 M ova257-264 peptide. After 5 days, cytolytic activity of spleen cells was measured in a 51Cr release assay.
To test the expression of OVA by tumors growing in E.G7 immunized and E.G7 challenged mice, we used spleen cells from VSV-OVA infected mice stimulated in vitro with 10-6 M ova257-264 peptide for 5 days. A single cell suspension was prepared from tumor tissue of mice immunized and challenged with E.G7, and cells were immediately used as targets in a 51Cr release assay. As a positive control, we used these tumor cells pulsed with 10-6 M ova257-264.
Adoptive transfer and analysis by flow cytometry
This method was adapted from Kearney et al. (51). Spleen cells from OVA-T cell receptor-transgenic mice OT-1 (Ly5.2-positive) were injected through the tail vein into normal C57BL/6 mice (Ly-5.2-negative) either directly or following 2 days of in vitro stimulation with OVA. For in vitro stimulation, OT-1 spleen cells were incubated with 2.5 mg/ml OVA in a 24-well plate (1 x 106 cells/well). After 2 days in culture, cells were washed three times and injected i.v. (4 x 106 stimulated or 5 x 106 naive OT-1 spleen cells per recipient mouse). Two days after adoptive transfer, recipient mice were challenged with 2.5 x 106 live E.G7 or EL4 tumor cells or injected with 5 mg OVA i.p. As a control, we adoptively transferred naive or in vitro stimulated spleen cells from normal C57BL/6 mice. On days 2 and 7 of in vitro culture, the cytolytic activity of OT-1 spleen cells was measured in a 51Cr release assay.
Six days after tumor cell challenge, spleen, draining and non-draining lymph nodes were collected, homogenized, filtered, washed in PBS and incubated with Fc Block (CD16/CD32- antibody, PharMingen, USA) at 4°C for 10 min followed by antibody incubation at 4°C for 45 min. Cells were washed three times in PBS and fluorescence intensity was measured by FACScan (Becton Dickinson, USA). Transferred OT-1 cells were identified as CD45.1 (Ly-5.2)+ cells. Directly conjugated antibodies CD45.1-PE and CD8-FITC were purchased from PharMingen, USA.
C57BL/6 mice were immunized with 2 x 107 irradiated E.G7 cells s.c. or 1 mg heat-denatured OVA (boiled for 5 min) s.c. on day 0. 100 µg of the purified CD40-activating antibody FGK45 in 200 µl PBS or a rat IgG2a control antibody (PharMingen, USA) were given i.v. on days 0, 1 and 2. The antibody FGK45 was generously provided by S. P. Schoenberger (Division of Immune Regulation, La Jolla Institute for Allergy and Immunology, San Diego, CA, USA). Two weeks later mice were challenged with 2.5 x 106 live E.G7 cells and tumor growth was measured.
This work was supported by NIH grant CA84479 to P.K.S. and by a sponsored research agreement with Antigenics Inc. in which one of us (P.K.S.) has a significant financial interest.
- Received March 26, 2002.
- Accepted March 26, 2002.
- Copyright © 2002 by Pramod K. Srivastava