CD4⁺ T Effectors Specific for the Tumor Antigen NY-ESO-1 Are Highly Enriched at Ovarian Cancer Sites and Coexist with, but Are Distinct from, Tumor-Associated Treg

The balance between T-cell subsets that play opposite immune functions is a determinant factor in antitumor immunity. The presence of high proportions of T effectors at tumor sites is generally associated with efficient antitumor responses (1), whereas that of regulatory/suppressor CD4⁺ T cells (Treg) has been often reported to be associated with less favorable clinical outcomes (2). Some studies, however, have instead reported an association between high levels of FOXP3⁺ Treg and improved clinical outcomes (3, 4), a contradictory finding that could possibly result from the fact that certain subsets of Treg comigrate with T effectors and are therefore enriched in tumors that are also highly infiltrated by T effectors (5). In support of their favorable role in antitumor immunity, T effectors in tumors have been found to contain cells that recognize antigens expressed by cancer cells, leading to their elimination (6).

On the basis of this evidence, strategies to improve antitumor immunity currently aim at increasing tumor antigen-specific T effectors in patients with cancer through various approaches, including vaccination, adoptive cell transfer therapy, and/or checkpoint blockade (7–10). Other potentially complementary approaches aim instead at impacting Treg by eliminating/inactivating them or converting them into effectors (11). However, because the antigen specificity of Treg infiltrating human tumors is largely unknown, an outstanding question in the development of all these approaches is whether, and to what extent, tumor antigen-specific CD4⁺ T effectors present at tumor sites overlap with tumor-infiltrating Treg. This, however, has been difficult to assess, partly because Treg are anergic, i.e., unable to proliferate or produce cytokines following T-cell receptor (TCR)-mediated stimulation, which complicates the assessment of their antigen specificity.

NY-ESO-1, a human tumor-specific antigen of the cancer/testis group expressed in germ cells and in cancer, but not in normal somatic tissues (12), is a major candidate for the development of immunotherapy in cancers of different histologic types. NY-ESO-1 is frequently expressed in high-grade serous ovarian cancer, a tumor type in which Treg are present in high proportions and may significantly contribute to dampen antitumor immunity (2, 13, 14). We have previously identified an immunodominant peptide, NY-ESO-1_119-143, recognized by NY-ESO-1–specific CD4⁺ T cells, and have generated MHC class II/NY-ESO-1 peptide tetramers that allow the direct assessment of antigen-specific CD4⁺ T cells by flow cytometry, independent of their function (15). Here, we have used the tetramers to directly assess NY-ESO-1–specific cells ex vivo among conventional CD4⁺ T cells and Treg present at sites of ovarian cancers in patients with spontaneous serologic responses to the antigen.

Surgical tumor specimens and peripheral blood samples were obtained at the Institut de Cancérologie de l'Ouest (Nantes-Saint Herblain, France) and Roswell Park Cancer Institute (Buffalo, NY) from patients with ovarian cancer upon approval by the institutional review boards and signed informed consent. Tumor specimens were processed by mechanical dissection to single-cell suspensions. Cells from ovarian cancer ascites were isolated by centrifugation. Peripheral blood mononuclear cells (PBMC) were isolated by density gradient sedimentation using LSM 1077 lymphocyte separation medium (PAA Laboratories GmbH). CD4⁺ T cells were enriched by positive selection using magnetic cell sorting (Miltenyi Biotec) from PBMC and single-cell suspensions from ovarian solid tumors or ascites. Isolated CD4⁺ T cells were assessed phenotypically by staining with fluorochrome-labeled monoclonal antibodies (mAb; from BD Biosciences unless indicated otherwise) specific for CD4, CD25 (Beckman Coulter), CD127 and FOXP3 (eBioscience), as indicated, and analyzed by flow cytometry (FACSAria, BD Biosciences). For assessment of NY-ESO-1–specific cells, isolated CD4⁺ T cells were stained with phycoerythrin-labeled HLA-DR52b/His-tag-NY-ESO-1_119–143 tetramers (TCMetrix) and anti-CD4, -CD25, and -CD127 mAb as previously described (15). For ex vivo flow cytometry cell sorting, enriched CD4⁺ T cells were stained with mAb specific for CD4, CD25, and CD127 and separated into Tconv (CD4⁺CD25^low) and Treg (CD4⁺CD25^highCD127^low) to high purity (>97%) by flow cytometry cell sorting (FACSAria, BD Biosciences).

The ability of tumor-associated CD4⁺ T cells to produce IFN-γ was assessed in a 6-hour intracellular cytokine staining assay using mAb specific for IFN-γ and FOXP3 following stimulation with PMA (100 ng/mL; Sigma-Aldrich) and ionomycin (1 μg/mL; Sigma-Aldrich), as previously described (5). The suppressive capacity of ex vivo-sorted tumor-associated Treg and Tconv was assessed by coculture with CFSE-labeled responder CD4⁺ T cells in the presence of irradiated monocytes and phytohemagglutinin (PHA; Sigma-Aldrich). Growth of responder cells was assessed by flow cytometry analysis of CFSE dilution in day-5 cultures. The growth (percentage of divided cells) of the wells with the test population (experimental group) was compared with that of the wells without the test population (control). The percentage of suppression was determined as follows: 100 − [(growth of experimental group/growth of control) × 100; ref. 5]. NY-ESO-1-tetramer⁺ T-cell lines were obtained by stimulation of tetramer⁺ cells, isolated by flow cytometry cell sorting from tumor-associated CD4⁺ T cells, with PHA and irradiated allogeneic feeder cells as previously described (15). To assess antigen recognition, the lines were stimulated with serial dilutions of NY-ESO-1_119–143 in the presence of HLA-DR52b⁺ antigen-presenting cells (APC), and secreted IFN-γ was measured in 24-hour supernatants by ELISA. The ability of the lines to produce cytokines was assessed in a 4-hour intracellular cytokine staining assay following stimulation with PMA/ionomycin using mAb specific for IFN-γ, TNF-α, interleukin (IL)-2, IL-4, IL-10, and IL-17 (eBioscience).

A phenotypic and functional analysis of ovarian cancer-associated conventional CD4⁺ T effectors (Tconv) and Treg was initially carried out using cell suspensions obtained ex vivo from ovarian cancer ascites (TAL) or from mechanically dissociated ovarian tumor tissue tumor-infiltrating leukocytes (TIL). Treg were identified among CD4⁺ T cells as CD25^highCD127^low, and Tconv were identified as CD25^low (Fig. 1A). Treg identified by this phenotype in tumor samples expressed high levels of FOXP3 and were highly suppressive ex vivo (Fig. 1B and C). In contrast, Tconv had little expression of FOXP3 and no suppressive activity. Tumor-associated Tconv stimulated ex vivo produced, for the most part, high levels of IFN-γ, whereas most tumor-associated FOXP3⁺ Treg failed to secrete IFN-γ (Fig. 1D). Thus, whereas tumor-associated Treg were highly suppressive ex vivo, most tumor-associated CD4⁺ Tconv instead displayed characteristics of effector cells.

Integrated immune responses to NY-ESO-1, including specific antibodies, CD4⁺ and CD8⁺ T cells, spontaneously develop in patients with ovarian cancer bearing antigen-expressing tumors (13). Using MHC class II (DR52b/DRB3*0202)/NY-ESO-1 peptide tetramers, we have previously shown that NY-ESO-1-tetramer⁺ cells are detected among circulating CD4⁺ T cells from patients with spontaneous antibody responses to the antigen, but not in healthy donors or in NY-ESO-1–seronegative patients (15, 16). By assessing a group of 215 patients with ovarian cancer, 20 patients with detectable levels of circulating NY-ESO-1 antibodies were identified. For 2 patients who were DRB3*0202⁺, cryopreserved PBMC and cell suspensions from TAL and TIL were available. From these samples, CD4⁺ T cells were enriched ex vivo by magnetic cell sorting and stained with tetramers in combination with anti-CD25 and -CD127 mAb. NY-ESO-1-tetramer⁺ cells were detected ex vivo in all samples (Fig. 2A). For patient OC1, tetramer⁺ cells were detected among CD4⁺ T cells in TAL and TIL at a frequency of 0.3% and 1%, respectively, i.e., about 20- and 60-fold higher than in circulating CD4⁺ T cells (Fig. 2B). Similarly, for patient OC2 the frequency of tetramer⁺ cells among CD4⁺ T cells in TAL and TIL was 0.1% and 0.2%, respectively, i.e., 5- and 10-fold higher than in the periphery.

We have previously reported that NY-ESO-1-tetramer⁺ cells in circulating CD4⁺ T cells from patients with both spontaneous and vaccine-induced immune responses to NY-ESO-1 are found among Tconv and not Treg (15, 16). Consistent with this, we found that NY-ESO-1-tetramer⁺ T cells in circulating CD4⁺ T cells from patients OC1 and OC2 were CD25^lowCD127^high (Fig. 3A and B). Similarly, NY-ESO-1-tetramer⁺ cells in both TAL and TIL were distinct from tumor-associated CD25^highCD127^low Treg, as they were largely CD25^low and expressed various levels of CD127 (Fig. 3A and B), a phenotype that has been reported to identify differentiated effector cells producing IFN-γ (17).

To assess tumor-associated NY-ESO-1–specific CD4⁺ T effectors functionally, we isolated NY-ESO-1-tetramer⁺ T cells from TIL and TAL and expanded them in vitro by mitogen stimulation to generate NY-ESO-1-tetramer⁺ T-cell lines (Fig. 4A and B). The lines obtained following this strategy exhibited a clear T_H1 effector profile as they produced high levels of IFN-γ, TNF-α, and IL-2 but did not produce cytokines associated with other CD4⁺ T-cell subsets, such as IL-17, IL-4, or IL-10 (Fig. 4C).

In sum, our data provide direct evidence that high frequencies of CD4⁺ T cells specific for NY-ESO-1 are found at sites of ovarian cancers of patients with spontaneous immune responses to the antigen, where they coexist with high proportions of Treg. However, NY-ESO-1–specific CD4⁺ T cells in tumors are clearly distinct from Treg and exhibit instead characteristics of T_H1 effectors. Because tumor-associated Treg are highly suppressive ex vivo, and despite the capacity of NY-ESO-1–specific T_H1 cells to expand at tumor sites, it is likely that Treg still affect them and limit their full growth potential and/or effector functions, particularly in solid tumor tissues, where Treg are present in high numbers.

It is noteworthy that, because only a few MHC class II tetramers incorporating peptides from human tumor-specific antigens have been developed thus far, the data reported here represent the first direct ex vivo assessment of tumor antigen-specific CD4⁺ T cells at tumor sites of patients with spontaneous immune responses to a tumor-specific antigen. In our previous assessments of NY-ESO-1-tetramer⁺ cells in circulating CD4⁺ T lymphocytes from patients with cancer with spontaneous or vaccine-induced immune responses (15, 16), and, in this study, in tumor-associated CD4⁺ T cells, we have consistently failed to detect NY-ESO-1–specific Treg. Our results are at variance with those of other studies reporting that a fraction of NY-ESO-1–specific CD4⁺ T cells that spontaneously develop in patients with cancer or are induced through vaccination are Treg (18, 19). These studies, however, did not use ex vivo assessments by tetramers, but indirect strategies that may have confounded results and conclusions.

The origin and specificity of tumor-associated Treg is currently under debate. One study has proposed that accumulation of Treg in tumors is the consequence of an early recruitment/expansion at tumor sites of Treg specific for self-antigens (20) presumably of thymic origin (called natural or nTreg). Other studies, however, have suggested that the accumulation of Treg in tumors results from the local conversion of Tconv, including those directed against tumor-specific antigens, into induced Treg (iTreg; ref. 21). In line with the concept that Treg infiltrating ovarian cancers are mostly nTreg, we have recently reported that a large fraction of them express Helios, a transcription factor of the Ikaros family that has been proposed to distinguish nTreg from iTreg (5, 22). Interestingly, based on their comparison of the TCR sequences of tumor-infiltrating Treg in two different mouse models, Sainz-Perez and colleagues recently reported that Treg–Teff conversion is not an active process at tumor sites (23), a conclusion that is in line with the results of the present study.

Although our results do not rule out that iTreg specific for certain types of tumor-associated antigens, for example tissue differentiation antigens, could be represented among tumor-associated Treg, they nevertheless show that tumor-associated CD4⁺ T cells directed against a highly tumor-specific and immunogenic antigen such as NY-ESO-1 are largely distinct from tumor-associated Treg. The results of this study are highly relevant to the development of cancer immunotherapy in several respects. First, they show that CD4⁺ T cells that recognize highly specific tumor antigens can accumulate in tumors and, even in the presence of very high proportions of Treg, are not converted into Treg, but maintain instead an effector phenotype. These results therefore imply that vaccination strategies based on NY-ESO-1 or other tumor-specific and immunogenic antigens have a high potential for expanding antitumor effectors and not Treg in vivo, even when these cells migrate at tumor sites. On the other hand, because tumor antigen-specific CD4⁺ T effectors at tumor sites are distinct from Treg, our results suggest that a selective elimination of Treg, if achievable, would not erode the tumor antigen-specific component of antitumor immune responses, but could rather potentiate it, and contribute to remove the brakes that prevent antitumor effectors from completely eradicating tumor cells, in vivo.

No potential conflicts of interest were disclosed.

Conception and design: M. Ayyoub, D. Valmori

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Pignon, J.-M. Classe, K. Odunsi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Ayyoub, D. Valmori

Writing, review, and/or revision of the manuscript: M. Ayyoub, D. Valmori

Study supervision: M. Ayyoub, D. Valmori

The study was supported by the Cancer Research Institute, the Ludwig Institute for Cancer Research, and La Ligue contre le Cancer.