Vaccination with Recombinant NY-ESO-1 Protein Elicits Immunodominant HLA-DR52b-restricted CD4⁺ T Cell Responses with a Conserved T Cell Receptor Repertoire

Purpose: ESO is a tumor-specific antigen with wide expression in human tumors of different histologic types and remarkable spontaneous immunogenicity. We have previously shown that specific T_H1 and antibody responses can be elicited in patients with no detectable preexisting immune responses by vaccination with rESO administered with Montanide ISA-51 and CpG ODN 7909. The purpose of the present study was to characterize vaccine-induced ESO-specific CD4⁺ T cell responses.

Experimental Design: We generated CD4⁺ T cell clones from patient C2, who had the highest CD4⁺ T cell response to the vaccine, and analyzed their fine specificity and HLA class II restriction to determine the recognized epitope. We then assessed the response to the identified epitope in all vaccinated patients expressing the corresponding HLA class II allele.

Results: We found that ESO-specific CD4⁺ T cell clones from patient C2 recognize peptide ESO_119-143 (core region 123-137) presented by HLA-DR52b (HLA-DRB3*0202), a MHC class II allele expressed by about half of Caucasians. Importantly, following vaccination, all patients expressing DR52b developed significant responses to the identified epitope, accounting for, on average, half of the total CD4⁺ T cell responses to the 119-143 immunodominant region. In addition, analysis of ESO-specific DR52b-restricted CD4⁺ T cells at the clonal level revealed significant conservation of T cell receptor usage among different individuals.

Conclusions: The identification of a DR52b-restricted epitope from ESO that is immunodominant in the context of vaccine-elicited immune responses is instrumental for the immunologic monitoring of vaccination trials targeting this important tumor antigen.

To complement standard therapy, active elicitation of tumor-specific immune responses through vaccination, ideally in combination with immunomodulation, is presently viewed as a strategy that could potentially prevent disease recurrence or/and lead to stabilization of disease, improving the long-term outcome of the current treatment of cancer. Among human tumor antigens (Cancer Immunity Peptide Database⁷

), cancer/testis antigens are characterized by their expression pattern being restricted, in adult tissues, to testis and tumor tissues (1). ESO, a member of the cancer/testis antigen group, is expressed in a variety of tumors of different histologic origin and displays significant spontaneous immunogenicity in patients bearing antigen-expressing tumors (2). Candidate anticancer vaccines using various ESO-based immunogens are currently under trial (Cancer Vaccine Collaborative⁸).

Several studies have analyzed ESO-specific T cell responses in patients with natural anti-ESO immunity and reported the identification of regions of the ESO protein frequently recognized by T cells from different patients. For CD4⁺ T cells, two main immunodominant regions have been identified: ESO_81-100, recognized by CD4⁺ T cells from about half of the patients with spontaneous immunity to ESO, and ESO_119-143, recognized by CD4⁺ T cells from the large majority of the patients (3–7). CD4⁺ T cell responses to a third region, ESO_157-170, frequently recognized by ESO-specific CTL (2), were also initially reported to be immunodominant but have not been frequently reported in subsequent studies (8).

In a recent vaccination trial using as immunogen rESO administered with Montanide ISA-51 and CpG ODN 7909 to patients with no detectable preexisting immunity to ESO, we obtained induction of significant T_H1 ESO-specific CD4⁺ T cell responses in all patients (9). Analysis of the fine specificity of vaccine-induced CD4⁺ T cells revealed that in all patients a sizable fraction of these responses were directed against ESO_119-143. Starting from the analysis of one patient with the highest ex vivo detectable CD8⁺ and CD4⁺ T cell responses, we show in this study that vaccination with the full-length recombinant protein induces ESO_119-143-specific CD4⁺ T cells restricted by HLA-DR52b (DRB3*0202), an allele that is expressed by about half of the Caucasian population. Furthermore, we show that such responses can be detected in all vaccinated patients expressing DR52b, showing the immunodominant nature of DR52b-restricted ESO_119-143-specific CD4⁺ T cell responses after vaccination with rESO. Finally, we found significant conservation of T cell receptor (TCR) usage for ESO-specific DR52b-restricted CD4⁺ T cells from different individuals.

Patients samples, cells, and tissue culture. Peripheral blood samples were collected from cancer patients enrolled in a clinical trial of vaccination with recombinant ESO protein, Montanide ISA51, and CpG 7909 (9) upon informed consent and approval by the Institutional Review Boards of Columbia University and New York University medical centers. Study patients received four s.c. injections of rESO/Montanide/CpG vaccine at 3-wk intervals. The patients enrolled had histologic diagnosis of cancer types known to express ESO. Of the 18 patients enrolled in the clinical trial, 11 were diagnosed with melanoma, 3 with breast cancer, 3 with sarcoma, and 1 with ovarian cancer. At study entry one sarcoma patient had a lung metastasis and all other patients had no evidence of disease. With the exception of one melanoma patient, none of the patients had detectable ESO-specific immune responses prior to vaccination, but they all developed specific antibody and CD4⁺ T cell responses following vaccination, as reported previously (9). Peripheral blood samples from healthy donors were obtained from the Etablissement Français du Sang Pays de la Loire (Nantes, France). MHC class II alleles were determined by high-resolution molecular typing. Melanoma cell lines were kindly provided by Dr. D. Rimoldi (Ludwig Institute for Cancer Research, Lausanne, Switzerland) and Prof. F. Jotereau (INSERM U892, Nantes, France). Monocyte-derived dendritic cells (moDC) were generated from enriched CD14⁺ cells, isolated from peripheral blood mononuclear cells (PBMC) using magnetic sorting (Miltenyi Biotech Inc.), by culture in the presence of 1,000 U/mL rhGM-CSF and 1,000 U/mL rhIL-4 (R&D Systems) for 5 d.

Assessment of ESO-specific CD4⁺ T cell responses and generation of specific clones. For ex vivo assessment, cryopreserved total PBMC were thawed, rested overnight, and stimulated for 7 h in the absence or presence of a pool of 20 to 24 amino acid long overlapping peptides (NeoMPS Inc.) spanning the full-length ESO sequence. Brefeldin A was added 2 h after the beginning of the incubation. Cells were then stained with antibodies directed against surface markers (CD3, CD4, and CD8; BD Biosciences), fixed, permeabilized, and stained with anti-IFN-γ, -IL-4, IL-10 (BD Biosciences), or -IL–17 monoclonal antibodies (mAb; eBiosciences), as previously described (9). For assessment of CD4⁺ T cell responses following in vitro stimulation, CD4⁺ cells were enriched from PBMC by magnetic cell sorting (Miltenyi Biotec Inc.) and stimulated with irradiated autologous antigen-presenting cells (APC) in the presence of the NY-ESO-1 peptide pool or the indicated NY-ESO-1 peptides (2 μmol/L each; NeoMPS Inc.), rhIL-2 (10 IU/mL), and rhIL-7 (10 ng/mL). At day 8 cultures were tested for intracellular IFN-γ secretion following stimulation, during 4 h, in the absence or presence of the ESO peptide pool or of individual peptides. Where indicated, CD4⁺ T cell cultures were preincubated for 1 h with anti-HLA-DRw52 mAb (clone 7.3.19.1; Monosan) prior to peptide stimulation. ESO_119-143-specific CD4⁺ T cells were isolated, following 4 h stimulation, using the IFN-γ secretion assay – cell detection kit (Miltenyi Biotech Inc.) and flow cytometry cell sorting and cloned by limiting dilution cultures in the presence of phytohemagglutinin, allogeneic irradiated PBMC, and rhIL-2 (100 IU/mL). Clones were subsequently expanded by periodic stimulation (every 3-4 wk) under the same conditions.

Antigen recognition assays and TCR BV analysis. CD4⁺ T cell clones were stimulated in the absence or presence of peptides, at the indicated concentration, and IFN-γ production was assessed either in a 4-h intracellular cytokine staining assay as described above or by measurement of IFN-γ in 24-h culture supernatant by ELISA as previously described (7). Where indicated, EBV-B cell lines or PBMC from healthy donors were preincubated in the absence or presence of peptide ESO_119-143, washed, and used to stimulate CD4⁺ T cell clones. Blocking experiments were done by preincubating CD4⁺ T cells with anti-HLA-DR (clone G46-6; BD Biosciences), -DP (clone B7/21; Abcam), -DQ (clone SPVL3; Immunotech), or -DRw52 mAb, prior to peptide stimulation. For assessment of reactivity to naturally processed full-length ESO, tumor cell lines or moDC were either incubated for 16 h with recombinant ESO or Melan-A proteins or transfected by electroporation with ESO-encoding pcDNA3.1 vector (Amaxa Inc.) and used to stimulate CD4⁺ T cell clones. TCR variable β chain (BV) usage was determined by flow cytometry using anti-BV mAb (Immunotech) and by molecular analysis as described previously (10) using a panel of previously validated primers (11) and nomenclature according to Arden B. et al. (12).

Isolation of ESO_119-143-specific vaccine-induced T_H1 clones. Following vaccination with rESO, all patients developed a specific CD4⁺ T cell response (9). For patient C2, vaccine-induced IFN-γ–producing CD4⁺ T cells were detected in postimmune but not in preimmune PBMC in response to stimulation with a pool of overlapping peptides covering ESO (Fig. 1A). ESO-specific IFN-γ–producing CD4⁺ T cells represented ex vivo close to 1% of total CD4⁺ T cells and displayed a typical T_H1 profile, as they failed to produce IL-4, IL-17, or IL-10 in response to antigen stimulation (data not shown). We isolated specific CD4⁺ T cells based on IFN-γ secretion, followed by cloning under limiting dilution conditions as described (2). We obtained four clones reactive to the ESO peptide pool and tested them for reactivity to immunodominant peptides ESO_81-100 and ESO_119-143. The clones specifically recognized peptide ESO_119-143, but not ESO_81-100 (Fig. 1B and C). As determined by using a panel of BV-specific mAb, all clones used BV2 (Fig. 1D).

Peptide ESO_119-143 is recognized by vaccine-induced CD4⁺ T cells in the context of HLA-DR52b. To determine the HLA-restriction of vaccine-induced ESO_119-143-specific clones from patient C2, we first assessed inhibition of antigen recognition using blocking mAb against HLA-DR, -DP, and -DQ. For all clones, antigen recognition was inhibited in the presence of anti-DR but not of anti-DP and anti-DQ mAb (Fig. 2A). As assessed by high-resolution molecular typing, patient C2 expressed DRB1*0701, DRB1*1201, DRB3*0202, and DRB4*0103 alleles. We first tested the clones for their capacity to recognize peptide ESO_119-143 presented by transfected mouse fibroblasts expressing DRB1*0701 (L-DR7 cells), and detected no reactivity (data not shown). We could not directly assess restriction by DRB1*1201 as no DRB1*1201⁺ APC were available. To establish the frequency of the restricting allele in the population, we assessed the ability of APC from HLA-unselected healthy donors to present peptide ESO_119-143 to clone C2/C4E7. This analysis revealed that APC from 8 of 15 donors were able to present the antigen (Fig. 2B). Therefore, the frequency of the restricting allele (50%) did not correspond to the frequency of DRB1*1201 in the population (about 3%), leaving DRB3 and DRB4 molecules, that are less polymorphic than DRB1, as possible candidates. In line with this, a monoclonal antibody specific for HLA-DR52 abrogated antigen recognition by clone C2/C4E7 but not by a control CD4⁺ T-cell clone (672/33) recognizing an unrelated peptide (SSX-2_37-58) in the context of DR11 (ref. 13; Fig. 2C). To define the restricting allele, we used as APC molecularly typed EBV-immortalized B cell lines EBV14 [DRB3*0202 (DR52b)], COX [DRB3*0101 (DR52a)], and EBV156 [DRB3*0301 (DR52c)]. CD4⁺ T cells recognized peptide ESO_119-143 presented by EBV14, but not by COX or EBV156, thus establishing DRB3*0202 (DR52b) as the restricting allele (Fig. 2D).

ESO_123-137 is the minimal optimal peptide recognized by DR52b-restricted CD4⁺ T cell clones. To define the DR52b epitope within the ESO_119-143 region, we used the MHC class II peptide prediction algorithm RankPep (http://imed.med.ucm.es/Tools/rankpep.html) to identify ESO sequences with significant predicted binding capacity to DR52b. Only two 9-mer core sequences were identified (Table 1). In particular, peptide ESO_127-135 was predicted to bind DR52b with an affinity only 3-fold inferior to that of the consensus sequence. Based on the identification of ESO_127-135 as the putative core region, we designed truncated peptides by sequential removal of amino acids at either the NH₂- or COOH-terminus of the original 24-mer and assessed their relative recognition efficiency by peptide titration (Fig. 3). Removal of amino acids up to position 123 at the NH₂-terminus did not significantly affect recognition, whereas further truncation significantly decreased recognition. At the COOH-terminus, truncation up to position 137 did not affect recognition, whereas further truncation decreased it. These results identified the 15-mer ESO_123-137 as the minimal optimal peptide recognized by DR52b-restricted CD4⁺ T cells.

DR52b-restricted CD4⁺ T cell clones recognize natural ESO antigen exogenously processed by APC. To assess the recognition of natural ESO antigen by DR52b-restricted CD4⁺ T cells, we tested their ability to recognize rESO processed by professional APC (DR52b⁺ moDC; Fig. 4A, left panel). MoDC efficiently processed rESO and presented the DR52b-restricted epitope to specific CD4⁺ T cells (Fig. 4A, right panel). To assess if DR52b-restricted CD4⁺ T cells could also directly recognize the ESO antigen endogenously expressed by tumor cells, we selected two ESO⁺ DR52b⁺ melanoma cells lines (Me252 and Me312; ref. 14). Both lines expressed significant levels of DR52 (Fig. 4B) and presented peptide ESO_119-143 to specific CD4⁺ T cells (Fig. 4C). However, we failed to detect significant direct recognition of tumor cells by ESO-specific DR52b-restricted CD4⁺ T cells even after treatment with IFNγ (Fig. 4C). Similarly, A2⁺DR52b⁺ moDC, transfected with a plasmid encoding ESO, failed to be recognized by specific DR52b-restricted CD4⁺ T cells, although they were recognized by A2-restricted CD8⁺ T cells (Fig. 4D). Thus, ESO_119-143-specific DR52b-restricted CD4⁺ T cells were able to recognize exogenously but not endogenously processed ESO antigen.

ESO_119-143-specific DR52b-restricted CD4⁺ T cell responses are immunodominant following vaccination with ESO protein. To evaluate the prevalence of ESO-specific DR52b-restricted CD4⁺ T cell responses in patients vaccinated with rESO, we isolated CD4⁺ T cells from postimmune samples of 15 patients and stimulated them during 10 days with the pool of ESO peptides. We then assessed the presence of specific CD4⁺ T cells by intracellular IFN-γ staining after stimulation with peptide ESO_119-143. To determine the proportion of DR52-restricted CD4⁺ T cells in the cultures, we did the assay in the absence or presence of anti-DR52 specific antibody. Six of the analyzed patients expressed DR52b and nine were negative. Significant proportions of CD4⁺ T cells specifically producing IFN-γ in response to ESO_119-143 were detected in postvaccine samples from all patients (Fig. 5A). Their frequency in cultures from different patients ranged from 1.8% to 18.3% of CD4⁺ T cells and was similar, in average, for DR52b⁺ and DR52b⁻ patients. However, whereas no significant inhibition of antigen recognition was observed for cultures from DR52b⁻ patients in the presence of the anti-DR52 mAb, the latter blocked antigen recognition by ESO_119-143-specific CD4⁺ T cells in cultures from all DR52b⁺ patients, to different extents (range, 21-88%; mean 44.7% ± 23.8%; Fig. 5B). Together, our results show that DR52b-restricted ESO_119-143-specific CD4⁺ T cell responses are immunodominant in DR52b-expressing patients vaccinated with the rESO.

Conserved TCR usage of ESO_119-143-specific DR52b-restricted CD4⁺ T cell clones. T cell clones recognizing defined MHC/peptide complexes can display conserved structural features. To assess TCR usage of clones recognizing peptide ESO_119-143 in the context of DR52b, we derived a panel of ESO_119-143-specific clones from vaccinated patients expressing DR52b. We obtained 62 clones from 4 different patients (50 clones from patient C2, 8 from patients N13, 3 from C5, and 1 from patient N10). Of these, 33 (53%) were DR52b-restricted, as determined by using molecularly typed APC (data not shown). Because the CD4⁺ T cell clones initially obtained from patient C2 used BV2, we assessed BV2 expression by all other clones using specific mAb. This analysis revealed that >70% of the DR52b-restricted clones (including clones from three patients) expressed BV2. To further assess the structural features of DR52b-restricted TCR, we sequenced the TCR β chains of the BV2-expressing clones. We identified 13 distinct clonotypes (Table 2), 5 of which used the same TCR β chain joint segment (2.1) whereas the remaining 8 used 6 other BJ. In addition, the 13 distinct CDR3 regions were variable both in terms of length (10-13 amino acids) and amino acid composition. Some conservation was nevertheless noticeable, such as the presence of A at position 1 of the CDR3 of 11 of the 13 clonotypes and R at position 2 for 7 of them.

Here, we have reported the identification of an ESO-derived DR52b-restricted epitope recognized by CD4⁺ T cells induced by vaccination with a rESO vaccine administered with the immunologic adjuvants Montanide ISA-51 and CpG ODN 7909, a formulation that predominantly elicits T_H1 responses. Previous studies from us and others have identified ESO_119-143 as an immunodominant region, recognized by CD4⁺ T cells from virtually all patients with spontaneous or vaccine-induced immunity to ESO (3, 5–7). Several overlapping epitopes contained within the ESO_119-143 region and restricted by multiple HLA-DR alleles have been identified (refs. 3, 4; summarized in Cancer Immunity Peptide Database⁷). Surprisingly, the DR52b-restricted epitope identified in this study has not been reported thus far. Our group, however, has previously reported the identification of two other DR52b-restricted epitopes from the tumor antigens SSX-4 and Melan-A (16, 17).

At variance with the β chain of the mouse I-E molecule (homolog to human HLA-DR), encoded by a single gene, the β chain of human HLA-DR is encoded by multiple genes. In addition to the DRB1 gene encoding the prevalent β chain of the DR isotype, additional genes code for other β chains, namely DRB3 (DR52), DRB4 (DR53), and DRB5 (DR51). They are less polymorphic than DRB1 and generally expressed at lower levels, but code for DR molecules that are fully functional with respect to antigen presentation (18, 19). These genes have strong linkage disequilibrium with defined DRB1 alleles. In particular, DR52 is very frequently expressed in the population, as DRB3 alleles are associated through linkage disequilibrium to some of the most common DRB1 alleles (20). Lower expression and linkage disequilibrium with DRB1 alleles may account for the fact that T cell epitopes restricted by these alternate DR molecules have been described less frequently than those restricted by DRB1-encoded molecules, or have been reported as restricted by the associated DRB1 allele.

In general, alternate DR molecules have been less well investigated as compared with those encoded by DRB1. However, expression of several DRB3-encoded molecules has been recently reported to associate with different autoimmune diseases, which has resulted in an increased interest in investigating their structure and peptide-binding specificity. Expression of DRB3*0202 (DR52b), one of the main DRB3 alleles, has been associated with Grave's disease, multiple sclerosis, and essential hypertension caused by infection with Chlamydia pneumoniae (21–23).

Using truncated overlapping peptides, we defined the minimal optimal sequence recognized by ESO DR52b-restricted CD4⁺ T cells as the 15-mer 123-137. Within this sequence, a screening of the entire ESO sequence, using the MHC class II peptide-binding prediction algorithm RankPep (http://imed.med.ucm.es/Tools/rankpep.html), identified a sequence with high predicted binding capacity to DR52b, corresponding to peptide 127-135 (TVSGNILTI). Although the crystal structure of DR52b has not been yet resolved, some structural consideration on the potential contribution of single amino acids in the identified peptide to binding can be drawn based on previous analyses of natural peptides isolated from DR52 molecules and on the recently reported structure of the highly homologous DR52c molecule bound to a self-peptide derived from the Tu elongation factor (24, 25). The most salient feature of the identified peptide is the amino acid N located in the central part of the sequence. Together with DR52c, and at variance with most other DR molecules, DR52b has a Q at position β74, that together with other residues in the P4 pocket, limits the amino acids binding at this position to N or D, whereas the P1 and P6 pockets are expected to be rather permissive and can accommodate many different residues.

DR52b is expressed by about half of Caucasians, which is similar to the frequency of expression of the most investigated human MHC class I molecule: HLA-A*0201. The frequent expression of HLA-A*0201 has allowed extensive analysis, including assessment of ex vivo frequency and phenotype of tumor antigen–specific HLA-A*0201-restricted CTL (including Melan-A and ESO-specific) using HLA-A*0201/peptide fluorescent tetramers (2, 26, 27). The identification of the ESO DR52b-restricted epitope may allow the development of a similar approach using MHC class II/peptide tetramers to assess CD4⁺ T cell responses to ESO. This is particularly relevant taking into account the immunodominant character of ESO-specific DR52b-restricted CD4⁺ T cell responses, following vaccination. Indeed, by assessing the quantitative contribution of DR52b-restricted responses to the overall response to the immunodominant 119-143 region, following vaccination, we show that, although variable among different individuals, these represented on average about 50% of total specific CD4⁺ T cells. The prevalence of DR52b-restricted CD4⁺ T cells in patients with spontaneous responses to ESO, however, might be different and remains to be determined.

ESO-specific DR52b-restricted CD4⁺ T cell clones isolated in this study efficiently recognized the natural exogenous ESO antigen after processing and presentation by APC but failed to recognize endogenously expressed ESO. We have previously obtained similar results with CD4⁺ T cell clones specific for another cancer/testis antigen, SSX-4 (16), whereas we have observed recognition of both exogenous and endogenously expressed antigen using Melan-A–specific CD4⁺ T cells (17). The ability of CD4⁺ T cells to recognize endogenously expressed tumor antigens may be epitope-dependent (28, 29) and can significantly vary for different tumor antigens, depending on their intracellular localization. At variance with cancer/testis antigens, melanocyte differentiation antigens such as Melan-A have a natural access to the endogenous MHC class II processing and presentation pathway, as they are localized in melanosomes, or, in their absence, in lysosomes (30). Generation of ESO-specific CD4⁺ T cells prevalently recognizing exogenously processed antigen is expected following vaccination with recombinant ESO protein. However, as direct recognition of tumor cells is most likely not the dominant mechanism through which tumor antigen–specific CD4⁺ T cells contribute to tumor rejection (31–33), lack of recognition of endogenous ESO antigen by CD4⁺ T cells does not necessarily imply a decreased importance of this epitope in the effector phase of antitumor immune response to ESO-expressing tumors. To assess this point, it would be of interest to assess the presence of specific DR52b-restricted CD4⁺ T cells among tumor-infiltrating lymphocytes from patients with spontaneous immunity to ESO.

By assessing the TCR of ESO_119-143-specific DR52b-restricted CD4⁺ T cell clones from different individuals, we could show conserved TCR usage, with frequent usage of BV2 often in association with BJ 2.1. The CDR3 region of the different clonotypes identified, however, was variable, both in terms of length and amino acid composition, which could indicate a certain degree of heterogeneity in the fine specificity and/or avidity of antigen recognition among different clones. We have previously reported conserved but distinct TCR usage for ESO-specific HLA-A*0201-restricted CTL that occur naturally or are induced through peptide vaccination (34), which was associated with their ability to recognize or not the naturally processed antigen. To our knowledge, however, this is the first study assessing TCR usage by ESO-specific CD4⁺ T cells. It will therefore be of interest to compare the TCR usage of ESO_119-143-specific DR52b-restricted CD4⁺ T cells elicited by vaccination with that of CD4⁺ T cells naturally occurring in patients with spontaneous immunity to ESO. Interestingly, Kudela P. et al. have recently reported the existence of promiscuous clonal CD4⁺ T cells able to recognize peptide ESO_119-143 in the context of several distinct MHC class II molecules (35). Although the CD4⁺ T cell clones isolated in the present study did not display promiscuous antigen recognition (data not shown), it will be clearly of great interest, in future studies, to compare their TCR usage with those of monogamous or promiscuous CD4⁺ T clones recognizing other epitopes in the 119-143 region.

No potential conflicts of interest were disclosed.

We are grateful to the clinical research teams at New York University and Columbia University as well as the members of the Ludwig Institute Clinical Trial Office who have been involved in conducting the previously reported clinical study (9).