This commentary reviews the data on HLA-A2-restricted CD8 T cells specific for peptide540-548 derived from hTERT (human telomerase reverse transcriptase). Several studies have reported the successful generation of such T cells (1, 2, 3). However, tumor recognition was observed in some, but not all, studies. More data are required to elucidate whether hTERT peptide540-548-specific T cells can indeed recognize and destroy tumor cells. It would be highly useful if telomerase would emerge as a universal tumor antigen that can be targeted in the cancer immunotherapy of HLA-A2 positive patients.
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.
As potential targets of immunotherapy, tumor antigens need to confer efficient interactions between tumor cells and the corresponding antigen-specific T cells. Antigens that are expressed by many different tumor types but not in normal tissue are of great interest. Moreover, these antigens should be functionally involved in oncogenesis, resulting in the reduction of malignant potential if tumor cells escape T cell recognition by antigen downregulation or mutation.
Telomerase is a ribonucleoprotein complex responsible for the maintenance of the length of telomeres, which are shortened during cell division. Telomerase expression has been directly linked to tumor development, and inhibition of telomerase in human tumor cells may arrest growth (4, 5, 6). The human telomerase catalytic subunit hTERT is expressed in the vast majority of human cancer cells derived from multiple tissues, but infrequently in normal cells (7). Thus, hTERT is an attractive candidate target for tumor therapy.
To evaluate whether cytotoxic T cells can recognize and kill telomerase-expressing cells, one needs to identify peptide antigens that are presented by HLA class I antigens and recognized by CD8 T cells. A recent review has summarized the currently identified MHC class I binding peptide antigens from hTERT (8), and clinical trials are underway to test whether these antigens can elicit tumor-specific T cell responses (9).
Identification of the hTERT peptide540-548
The hTERT peptide540-548 was identified based on the computerized prediction of binding to HLA-A*0201 (1, 2). The peptide was synthesized and used to stimulate human peripheral blood mononuclear cells (PBMCs). Such lymphocyte cultures were tested in cytotoxicity assays. The studies demonstrated lysis of target cells loaded with specific peptide but not antigen negative targets, indicating the presence of peptide540-548-specific cytotoxic T lymphocytes (CTLs). Subsequently, the same cultures were tested against various selected tumor cells. Positive correlations were found between tumor cell killing and the expression of telomerase / HLA-A*0201. The authors concluded that HLA-A*0201-restricted hTERT peptide540-548-specific CD8 T cells could efficiently recognize and kill telomerase-expressing A*0201+ tumor cells (Table 1).
Inefficient antigen processing of the hTERT peptide540-548
Recently, we have reassessed the usefulness of hTERT peptide540-548 as a CTL-defined tumor antigen (3). In agreement with previous studies (1, 2), peptide540-548-specific CTLs could be generated using PBMCs from HLA-A*0201 positive melanoma patients. CTL clones established from these cultures showed strong cytotoxicity against HLA-A*0201 target cells in the presence of exogenously added peptide. However, no cytotoxicity was observed against melanoma and colon cancer cells, despite their expression of hTERT and HLA-A*0201. Transfection of melanoma cells with a plasmid containing an hTERT minigene resulted in recognition by the specific CTLs, but only for the minimal minigene (coding for hTERT amino acids 540-548). Longer minigenes with C- and/or N-terminal flanking regions did not confer recognition. Subsequently, a long hTERT precursor peptide (amino acids 534-554) was incubated with proteasomes and the product was analyzed by mass spectrometry and cytotoxic assays. The results suggested inefficient antigen processing of hTERT resulting in lack of peptide presentation on the cell surface, explaining why tumor cells were not recognized by the CTLs. This finding is in disagreement with previous data (1, 2).
T cell clones and transfected target cells as tools to validate tumor antigens
When stimulating PBMCs with MHC class I binding peptides, one can frequently observe an enrichment of antigen-specific CD8 T cells, while other CD8 T cells, CD4 T cells, B cells, and NK cells are diluted and progressively lost. However, even in cultures maintained for several weeks, significant percentages of cells other than specific CD8 T cells remain. These cells may account for "nonspecific" cytotoxic activity. In particular, NK cells can be cytotoxic against a variety of cells, whereby some but not all target cells are susceptible. After exclusion of NK cells, one still needs to consider the "nonspecific" cytotoxicity that can be exerted by T cells. This may be due to different receptors, such as the alphabetaTCR-mediated recognition of HLA-E positive peptides derived from viruses or allogeneic HLA class I signal sequences (10). Activatory and inhibitory NK receptors may also play a role (11, 12).
Therefore, for a direct demonstration of specific CTL-mediated cytotoxicity it is necessary to generate T cell clones (in which only one type of T cell receptor is present). Polyclonal T cell lines may also be used, provided that one can demonstrate that they are highly antigen-specific (e.g. using fluorescent HLA/peptide tetramers). Unfortunately, many studies on cancer epitopes have not been carried out with T cell lines and clones. Often, poorly defined bulk lymphocyte cultures are generated through peptide stimulation and directly tested for cytotoxicity against tumor cells.
It is possible that the bulk lymphocyte cultures used in the two studies (1, 2) may have contained (besides peptide540-548-specific T cells) some cells with "nonspecific" cytotoxicity, killing some but not all tumor cells, which by coincidence correlated with antigen expression. Thus lysis may have been triggered by antigens and receptors other than telomerase and its cognate T cell receptors. Alternatively, antigen-specific CD8 T cells obtained in the two studies (1, 2) may have been of higher avidity than those used in our study (3). To assess T cell avidity, we have performed peptide titration experiments (3). It would be useful to do this also with the lymphocyte populations used in the first two studies (1, 2), to evaluate whether T cell avidity can possibly account for the discrepancy in tumor recognition.
An open question is whether dendritic cells (DCs) are required for the generation of efficient effector T cell populations. As compared to PBMCs, DCs are known to activate T cells more efficiently. However, T cells with low and high efficiency of antigen recognition ("avidity") can be obtained through both PBMC and DC stimulation, respectively.
Antigen positive and negative target cell pairs should be genetically identical, with the exception of the antigen or the HLA molecule in question, to ensure that all other parameters that could potentially influence the killing (receptor density, adhesion and costimulatory molecules, HLA molecules and other ligands of natural killer receptors, etc.) are identical. This is not the case when differing tumor cells are compared as target cells. In the studies on peptides540-548, tumor cells of different origins were typed for antigen and HLA expression, and a positive correlation was found with their susceptibility to lysis. Transfection of target cells was performed in two studies. The transfected target cells were tested with polyclonal lymphocytes (1) or with a T cell line and T cell clones (3). The latter study is the only one that combined the use of T cell clones and transfected target cells.
One strategy is to transfect cells with the gene coding for the antigen and to check for susceptibility to CTL-mediated lysis in comparison to cells transfected with irrelevant genes. Another approach is to transfect the restricting HLA molecule in cells endogenously expressing the antigen and to determine whether this results in susceptibility to lysis. The latter allows antigenicity to be assessed without manipulating the antigen providing the peptide epitope, while the former can test productive processing and cell surface presentation of the peptide antigen.
T cell activation in tumor lesions
High percentages of tumor antigen-specific T cells are frequently found in tumor infiltrating lymphocytes (TILs) and tumor-infiltrated lymph nodes (TILNs) (13). Indeed, the isolation of tumor reactive T cells from TIL(N)s was the basis for the identification of many tumor antigens (14, 15). Moreover, the fact that antigen-specific T cells are enriched (i.e. have proliferated) due to in vivo stimulation by endogenous tumor antigens is one of the best signs that the corresponding antigen is recognized in vivo. Enriched telomerase-specific CD8 T cells have so far not been observed ex vivo, even though telomerase is expressed by many tumors (16). Further studies should determine whether there are tumor lesions that may contain telomerase-specific T cells in detectable numbers, for instance using tetramer-based identification of antigen-specific T cells.
Although the computerized prediction of HLA class I binding and proteasome digestion is a powerful tool to identify candidate peptides (17, 18), the majority of these are not CTL-defined epitopes. At present it remains controversial whether hTERT peptide540-548-specific T cells recognize tumor cells in humans. Lack of tumor cell recognition may be due to inefficient antigen processing and presentation, or because of low TCR avidity. In vitro specificity and efficacy of T cell recognition needs to be analyzed carefully to evaluate the potential for in vivo protective immunity and immunotherapy.
- Copyright © 2002 by Daniel Speiser