Frequent Cytolytic T-Cell Responses to Peptide MAGE-A10_254–262 in Melanoma¹

MAGE genes are expressed in several types of human cancer but not by normal tissues, with the exception of male germ-line cells and, to a lesser extent, placental throphoblast cells(1). Because both latter cells are devoid of surface HLA class I molecules, specific CD8⁺ CTL responses to antigens encoded by MAGE genes can be considered strictly tumor specific. Several CTL-defined antigenic peptides derived from individual MAGE gene products have been identified, and some of them are presently being tested as candidates for vaccine-based immunotherapy of cancer. Despite extensive studies (2, 3, 4, 5),however, CTL responses to MAGE antigenic peptides have been detected only rarely in cancer patients, even after vaccination(6), possibly due to a very low frequency of CTL precursors (7). Taken together, these observations raise concerns on the immunogenicity of MAGE antigens and, hence, on their usefulness as vaccines. Whereas all of the autologous CTL clones used initially to identify antigenic peptides encoded by the MAGE-A1and A3 genes were generated from a single melanoma patient, a CTL clone derived recently from another melanoma patient was found to recognize a MAGE-A10-encoded nonapeptide (254–262) that is presented by HLA-A2.1 (8). In the present study, we used fluorescent tetramers of HLA-A2/peptide MAGE-A10_254–262 complexes to determine whether CTL responses to this peptide could be detected in HLA-A2⁺ melanoma patients. In contrast to the findings mentioned above, we found that blood samples from a large proportion of HLA-A2⁺ melanoma patients contained detectable levels of CTL precursors directed against MAGE-A10_254–262.

Tumor samples and PBMCs³used in this study were obtained from melanoma patients with stage III-IV disease. Melanoma cell lines were established in our laboratory from surgically excised melanoma metastases and cultured in RPMI 1640/10% FCS. The melanoma cell lines NA8-MEL and MZ2-MEL 3.0 were kindly provided by Dr. F. Jotereau (Institut National de la Santéet de la Recherche Médicale, Nantes, France) and Dr. T. Boon(Ludwig Institute for Cancer Research, Brussels, Belgium),respectively.

Expression of MAGE-A10 in frozen tumor samples was analyzed at the mRNA level by reverse transcription-PCR as described previously, using MZ2-MEL 3.0 as a reference cell line for a semiquantitative evaluation of the results (9). Analysis of MAGE-A10 protein expression in melanoma cell lines was performed by Western blotting using MAGE-A10-specific polyclonal antibodies as described previously(9).

Phycoerythrin-labeled HLA-A2/peptide tetramers were synthesized as described previously (10), using peptides MAGE-A10_254–262 (GLYDGMEHL) and NY-ESO-1_86–94 (RLLEFYLAM). Staining of cells with tetramers and fluorescein-conjugated anti-CD8 antibody (Becton Dickinson) and flow cytometric analysis were performed as described previously (10).

For peptide stimulation experiments, CD8⁺lymphocytes were positively selected by magnetic cell sorting from PBMCs of HLA-A2⁺ melanoma patients and used for peptide stimulation experiments as detailed previously(11). Briefly, highly enriched CD8⁺lymphocytes were stimulated with peptide MAGE-A10_254–262 (1 μm) and irradiated autologous CD8⁻ cells as antigen-presenting cells in medium containing human recombinant interleukin 2 and human recombinant interleukin 7. Where indicated, cells were restimulated with peptide-pulsed and irradiated T2 cells and cultured for an additional 6 days before tetramer analysis. CD8⁺ tetramer⁺cells were isolated by fluorescence-activated cell sorting as described previously (11). Antigen recognition was assessed using a standard 4-h chromium release assay. Briefly, chromium-labeled target cells (1000 cells) were incubated in the presence or absence of the indicated peptides for 15 min at room temperature before the addition of effector cells at the indicated ratio. Chromium release was measured in aliquots of supernatants harvested after an incubation of 4 h at 37°C. The percentage of specific lysis was calculated as follows:100 × [(experimental − spontaneous release)/(total − spontaneous release)].

A MAGE-A10 full-length cDNA (8) was used for PCR amplification of the entire coding sequence or a truncated cDNA coding for a protein with a 10-amino acid deletion at the NH₂ terminus. The amplified fragments were then subcloned into a pCR3-derived vector (Invitrogen). Whereas full-length MAGE-A10 is a nuclear protein, the truncated protein displays a cytoplasmic localization.⁴Transient transfection of COS-7 cells was performed using Fugene 6 transfection reagent according to the manufacturer’s instructions(Roche). The following day, transfected cells were tested for their ability to stimulate the release of TNF by MAGE-A10_254–262-specific CTLs as described previously (12). In brief, CTLs were added at the appropriate E:T ratio in the presence or absence of 1μ m MAGE-A10_254–262peptide. After a 24-h incubation at 37°C, supernatants were collected, and TNF content was determined in a functional assay using WEHI-164 clone 13 cells.

To determine whether CD8⁺ CTL precursors directed against peptide MAGE-A10_254–262 could be detected in blood samples from HLA-A2⁺ melanoma patients, we initially selected patient LAU 50. In the course of previous studies, this patient had been found to exhibit relatively strong CTL responses to several other HLA-A2-restricted melanoma-associated antigenic peptides (Refs. 11 and13; data not shown). Highly enriched CD8⁺ T cells isolated from PBMCs of this patient were cultured for 2 weeks in the presence of peptide MAGEA10_254–262, autologous antigen-presenting cells, and cytokines, as described previously(11). To directly enumerate peptide-specific CD8⁺ T cells, we used fluorescent HLA-A2/peptide tetramers (10). A high proportion of HLA-A2/peptide MAGE-A10 _254–262 tetramer⁺CD8⁺ T cells was clearly detected in the cultured population (Fig. 1,A), whereas no positive cells were detected after staining with tetramers containing the unrelated peptide NY-ESO-1_86–94. HLA-A2/peptide MAGE-A10_254–262 tetramer⁺T cells were isolated from the cultured population by tetramer-guided cell sorting. After phytohemagglutinin-driven expansion, this population, which contained >99% tetramer⁺CD8⁺ T cells (Fig. 1,B), efficiently lysed HLA-A2⁺ target cells in the presence of peptide MAGE-A10_254–262, but not in the presence of peptide NY-ESO-1_86–94, which was used as a negative control (Fig. 1 C).

MAGE-A10 is a nuclear protein (8) expressed by different types of tumors, with the highest frequencies (33–50%) observed in melanoma, bladder carcinoma, lung carcinoma, and esophageal and head and neck squamous carcinoma (8). Initial studies on MAGE-A10 transcript levels in tumors suggested a low level of gene expression, which was estimated to be insufficient for the production of enough antigenic peptides to allow recognition by specific CTLs(1, 14). However, using specific antibodies, we observed that the MAGE-A10 protein was expressed in melanoma cells at a level similar to that of the MAGE-A1 protein (9, 15). These results suggested that endogeneous production of MAGE-A10 peptides could lead to CTL recognition. Direct evidence for that was obtained by Huang et al. (8) using a CTL clone derived from an autologous mixed lymphocyte-tumor cell culture. To further substantiate these findings, we assessed the ability of the HLA-A2/peptide MAGE-A10_254–262tetramer⁺ CD8⁺ T cell population to lyse MAGE-A10⁺ tumor cells. As illustrated in Fig. 2, A and B, the T-cell population efficiently lysed MAGE-A10⁺/HLA-A2 ⁺ melanoma cells, regardless of the presence of peptide MAGE-A10_254–262, whereas NA8-MEL and Me 290 melanoma cells (which are HLA-A2⁺ but MAGE-10⁻) were recognized only in the presence of exogenously added peptide. In addition, no significant lysis was observed with melanoma cell line MZ2-MEL 3.0 (which is MAGE-A10⁺ but HLA-A2⁻) in either the presence or the absence of antigenic peptide. Recognition of endogenously produced HLA-A2/MAGE-A10 peptide complexes by the tetramer⁺ cells was directly documented by using COS-7 cells (MAGE-A10⁻,HLA-A2⁻) transiently cotransfected with a plasmid encoding the MAGE-A10 protein and a plasmid encoding HLA-A2.1. As illustrated in Fig. 2,C and in good agreement with the tumor recognition data shown in Fig. 2 A, the tetramer⁺ cells efficiently recognized COS-7 cells transfected with a plasmid encoding the MAGE-A10 protein, but not those transfected with an empty vector. Interestingly,transfection with a construct coding for a truncated form of MAGE-A10 that localized to the cytoplasm (data not shown) also resulted in efficient antigen recognition by the tetramer⁺cells.

Altogether, these results show that MAGE-A10-specific CTLs were readily detectable by staining with HLA-A2/peptide MAGEA10_254–262 fluorescent tetramers of a peptide-stimulated CD8⁺ T-cell population derived from patient LAU 50. Importantly, the isolated tetramer⁺ cells specifically recognized peptide MAGE-A10_254–262 with high avidity and specifically lysed HLA-A2⁺ melanoma cell lines expressing MAGE-A10.

The occurrence of a CTL response to peptide MAGE-A10_254–262 was further assessed in 21 additional HLA-A2⁺ melanoma patients. These patients were subdivided into two groups according to MAGE-A10 expression of their tumor lesions and/or in vitro cultured melanoma cell lines. Highly enriched CD8⁺ T cells from each patient were initially stimulated with peptide MAGE-A10_254–262 as described for patient LAU 50. One week after the first stimulation, cultures were restimulated with peptide-pulsed T2 cells, cultured for an additional week, and then stained with HLA-A2/peptide MAGE-A10_254–262tetramers. As shown in Table 1, tetramer⁺ CD8⁺ T cells were clearly detected in cultures derived from 8 of 12 patients with a MAGE-A10⁺ tumor. Interestingly, cultures derived from 3 of the 10 patients whose tumors showed no detectable MAGE-A10 expression also contained tetramer⁺CD8⁺ T cells. The latter findings prompted us to investigate whether MAGE-A10_254–262-specific T-cell responses could also be detected in some HLA-A2⁺ normal donors. Indeed,tetramer⁺ CD8⁺ T cells could be detected in cultures from 2 of 10 normal donors tested (Table 2).

These results indicate that the CTL response to peptide MAGE-A10_254–262 is readily detected in two-thirds of melanoma patients bearing a MAGE-A10-expressing tumor lesion. Such a response is also detectable in a smaller proportion(20–30%) of melanoma patients bearing MAGE-A10⁻ lesions or in normal HLA-A2⁺ individuals. These figures are in striking contrast with those obtained for other peptide antigens derived from MAGE and MAGE-related genes(16, 17) and are closer to those found for epitopes derived from melanocyte differentiation antigens (18). Thus, peptide MAGE-A10_254–262 appears to be among the most immunogenic melanoma-associated antigens described thus far. The high immunogenicity of peptide MAGE-A10_254–262 could be due to the existence of a relatively high frequency of specific CTL precursors in blood, in contradistinction to previously investigated MAGE peptides(7). To assess the frequency and the phenotype of MAGE-A10_254–262 tetramer⁺T-cell precursors in unstimulated PBMCs, samples from 14 melanoma patients and 2 normal donors were stained with HLA-A2/MAGE-A10_254–262 tetramers in combination with anti-CD8 and anti-CD45RA mAbs. Overall, the frequency was close to or below the limit of detection of tetramer staining (≅ 1 in 10,000 CD8⁺ T cells; data not shown). As a consequence of this, the phenotype of MAGE-A10_254–262-specific precursors could not be clearly determined in these samples. A higher frequency of tetramer⁺ cells (≥1 in 2,500 CD8⁺ T cells) has been detected ex vivo in the case of the immunodominant tumor-associated antigen Melan-A (peptide MelanA_26–35; Ref.19). However, similar to what we found here for MAGE-A10_254–262, we have previously observed that the frequency of HLA-A2/tyrosinase_368–376tetramer⁺ T cells was also close to or below the tetramer detection limits, although HLA-A2⁺melanoma patients frequently respond to in vitro stimulation with peptide tyrosinase_368–376.(20). Thus, a frequency of specific precursors close to or below 1 in 10,000 CD8⁺ T cells could still result in frequent responses. The data reported in the present study provide new insight into the immunogenicity of MAGE antigens that should stimulate the implementation of peptide MAGE-A10_254–262-based cancer vaccination trials. The availability of HLA-A2/peptide MAGE-A10_254–262 tetramers will be essential for the monitoring of such trials.

We thank Dr. K. Servis for peptide synthesis; N. Montandon, S. Salvi, and K. Muehlethaler for excellent technical assistance; and M. van Overloop for assistance in manuscript preparation. We are grateful to the melanoma patients for their generous participation in this research project.