The tumor antigen NY-ESO-1 is a promising cancer vaccine target. We describe here a novel HLA-B7–restricted NY-ESO-1 epitope, encompassing amino acids 60-72 (APRGPHGGAASGL), which is naturally presented by melanoma cells. The tumor epitope bound to HLA-B7 by bulging outward from the peptide-binding cleft. This bulged epitope was not an impediment to T-cell recognition, however, because four of six HLA-B7+ melanoma patients vaccinated with NY-ESO-1 ISCOMATRIX vaccine generated a potent T-cell response to this determinant. Moreover, the response to this epitope was immunodominant in three of these patients and, unlike the T-cell responses to bulged HLA class I viral epitopes, the responding T cells possessed a remarkably broad TCR repertoire. Interestingly, HLA-B7+ melanoma patients who did not receive the NY-ESO-1 ISCOMATRIX vaccine rarely generated a spontaneous T-cell response to this cryptic epitope, suggesting a lack of priming of such T cells in the natural anti–NY-ESO-1 response, which may be corrected by vaccination. Together, our results reveal several surprising aspects of antitumor immunity and have implications for cancer vaccine design. [Cancer Res 2009;69(3):1046–54]
Tumor cells express a range of antigens that can be recognized by cytotoxic T cells. The family of cancer-testis antigens are promising targets as they are expressed in a wide range of malignancies but not in normal tissues except for the immunologically privileged germ cells of the testis and placental trophoblasts ( 1). NY-ESO-1 is a CT antigen that has been the subject of several recent human cancer vaccine trials ( 2, 3). We completed a phase I clinical trial in which patients with fully resected melanoma at high risk of relapse were vaccinated with full-length NY-ESO-1 protein formulated with ISCOMATRIX adjuvant (CSL Limited; ref. 4), which can deliver antigens to DCs for efficient presentation of both MHC class I– and class II–restricted epitopes ( 5, 6). The outcomes of our vaccination included high-titer antibody responses, strong delayed-type hypersensitivity reactions, and abundant circulating CD4+ and CD8+ T cells specific for a broad range of NY-ESO-1 epitopes, including many previously unidentified determinants ( 4, 7).
The presentation of antigen on MHC class I first requires cleavage of the antigenic protein by proteasomes and other proteases to generate a library of short peptides ( 8). Such peptides may be further trimmed before ( 9, 10) or after being transported into the endoplasmic reticulum (ER; ref. 11) through the action of the transporter associated with antigen processing (TAP). The peptides bind class I in the ER before cell surface presentation.
DCs constitutively express the immunoproteasome and are considered the major antigen presenting cell type to initiate antitumor immune responses, via cross-presentation of tumor antigens ( 12). However, tumor cells express the “house-keeping” proteasome under nonimmune conditions and present tumor antigens directly. Due to the cleavage preference of the two proteasomes and potential differences between the two antigen-presenting pathways, it is expected that some tumor-specific T cells may not recognize tumor cells due to lack of direct presentation of the specific “cryptic” epitopes ( 13); on the other hand, there might be direct presentation of certain tumor antigenic epitopes by the tumor cells, but no antitumor immune response can be initiated due to lack of cross-presentation of such epitopes by dendritic cell (DC). The later category of epitopes would be of great potential for vaccine intervention. For instance, peptide-pulsed DC may rescue these otherwise cryptic immune responses ( 14). Interestingly, there are few well-demonstrated cases for either scenario.
Due to structural constraints, the peptide-binding cleft of MHC class I generally accommodates peptides of 8 to 10 amino acids (AA). However, a growing number of longer epitopes have been reported, ranging from 11 to 14 AA ( 15). Structural analysis has revealed that these long peptides are frequently accommodated within the peptide-binding cleft by adopting a “bulge” in the center ( 16– 21). Originally, this was thought to make such long epitopes poorly immunogenic, as the TCR was expected to have difficulty approaching the peptide-MHC complex ( 15, 22). However, two long (11AA and 13AA) epitopes from the BZLF1 antigen of EBV, despite adopting a bulged conformation, were shown to be immunodominant ( 18, 23, 24), meaning that T-cell responses become focused on these particular epitopes at the expense of the hundreds of other potential determinants from the same antigen ( 25).
For tumor antigens such as NY-ESO-1, epitope discovery has focused largely on peptides of normal length and no long class I epitopes have been identified. Whereas longer epitopes have been described for other tumor antigens (tyrosinase, CAMEL, and MAGE-A1; refs. 15, 26– 28), their position in the immunodominance hierarchy is unknown. We describe here an immunodominant 13AA epitope encompassing AAs 60 to 72 (APRGPHGGAASGL) of NY-ESO-1, and reveal several unique features of the immune response to this epitope that significantly aid our understanding of HLA class I–restricted presentation of NY-ESO-1 determinants after vaccination and cancer vaccine design.
Materials and Methods
Cells and medium. All cell lines are maintained in RF-10 [RPMI 1640 supplemented with 2 mmol/L Glutamax, antibiotics, 10 mmol/L HEPES (Invitrogen), and 10% FCS (Thermo Trace)]. The LM-Mel-26 melanoma cell line was generated from melanoma biopsy, whereas the SK-Mel-14 melanoma line was obtained from the Memorial Sloan-Kettering Cancer Center. Lymphoblastoid cell lines (LCL) 9038 and 9065 were made available from the 10th International HLA Workshop. T242 and T291 were made available by the Victorian Transplantation and Immunogenetics Service (a gift from Dr. B. Tait, Victorian Training and IT Solutions, Melbourne, Australia). Peripheral blood mononuclear cells (PBMC) were prepared from whole blood by Ficoll-Paque centrifugation.
Patients and vaccination. Melanoma patients (listed in Table 1 ) vaccinated by i.m. injection with 100 μg NY-ESO-1 ISCOMATRIX vaccine were described ( 4). 5 Some patients were not vaccinated. All studies were approved by the Human Research Ethics Committees of Austin Health and the Peter MacCallum Cancer Centre. All patients provided written informed consent.
Peptides, antibodies, tetramer, and flow cytometry. Peptides were synthesized by Multiple Peptide Systems, EZBiolab, and Chiron Mimotopes. For initial screening of T-cell responses, libraries of 18mer peptides with 12 AA overlap and 13mers with 11 AA overlap were used. Single Alanine substituted peptides and those used for mapping minimal epitope were >95% in purity. Anti-CD4, CD8, IFN-γ, CD107b (LAMP-2), and HLA class I (W6/32) were obtained from BD Biosciences, antibodies to TCRVβ families were obtained from Serotec, and HB152 [American Type Culture Collection, specific to HLA-Bw6 epitope on HLA-B7 (indicating HLA-B*0702 throughout this study)] culture supernatant was used for HLA-B7 binding assay. The NY-ESO-160-72/HLA-B7 tetramer was synthesized at the Tetramer Production Facility of the Ludwig Institute for Cancer Research. Flow cytometry was performed using the BD FACSCalibur or FACSCanto II instruments, and data were analyzed using FlowJo software (TreeStar, Inc.).
T cells, peptide stimulation, and intracellular cytokine staining/tetramer staining. T-cell expansion with synthetic peptides, tumor-infiltrating lymphocytes (TIL) expansion using phytohemagglutinin (PHA; Sigma), IFN-γ intracellular cytokine staining (ICS), and tetramer staining were performed essentially as described ( 4, 29). Briefly, cryopreserved PBMC were thawed and pulsed with 18mer peptide at 10 μmol/L, then cultured in RF-10 with 25U/mL interleukin (IL)-2 for ∼11 to 15 d unless specifically indicated. Cultured T cells were restimulated with peptide for 5 h in the presence of 10 μg/mL Brefeldin A (BFA). The cells were then stained with tetramer, anti-CD4 and anti-CD8, fixed with 1% formaldehyde, and further stained with anti–IFN-γ in the presence of 0.2% saponin.
Detecting natural presentation of the NY-ESO-160-72 epitope by melanoma cells. NY-ESO-155-72 (GPGGGAPRGPHGGAASGL) expanded T cells were incubated with a 5-fold excess of melanoma cells in the presence of BFA for 5 h. The cells were stained with tetramer followed by ICS for IFN-γ. CD107b mobilization was used to show degranulation by antigen activated T cells; 2 μmol/L monensin and anti–CD107b-FITC were added to the 5-h T cell–melanoma cocultures and the cells were further stained with tetramer followed by anti-CD8.
Detecting cross-presentation of the NY-ESO-160-72 epitope by monocyte-derived DCs. CD14+ monocytes were isolated from PBMCs collected from HLA-B*0702 and HLA-A*0201 double positive volunteers using magnetic beads (Miltenyi Biotec) and cultured with granulocyte macrophage colony-stimulating factor and IL-4 for 6 to 7 d ( 5). Monocyte-derived DCs (MoDC) at 5 × 105/mL were incubated with NY-ESO-1 ISCOMATRIX for 16 h at 37°C. In some assays, the serine peptidase inhibitor Ala-Ala-Phe-chloromethylketone (AAF-cmk) was added to MoDCs for 45 min before and throughout the antigen pulse period. MoDCs pulsed with NY-ESO-1157-165A (SLLMWITQA) or NY-ESO-160-72 (APRGPHGGAASGL) served as positive controls. After washing, MoDCs and NY-ESO-1 peptide–specific T-cell lines were cocultured in the presence of 10μg/mL BFA for 4 h. Cells were then washed and stained for specific tetramers and ICS (see above).
Production of recombinant NY-ESO-160-72/HLA-B7 complexes and structure determination. Soluble HLA-B*0702 complexed to NY-ESO-160-72 peptide was produced by expressing a truncated form of the B7 heavy chain and human β2m in Escherichia coli as inclusion bodies. Soluble complexes were purified after oxidative refolding of solubilised HLA-B7 and human β2m as described ( 30). The refolded soluble MHC/peptide complexes were concentrated and purified using standard methods ( 30). Crystals of the purified NY-ESO-160-72/HLA-B7 complexes were grown at 20°C using the hanging drop vapor diffusion technique, by mixing equal volumes of protein at 2 mg/mL with 28% PEG4000, 0.1 mol/L Tris (pH 8.3) and 0.2M MgCl2. Crystals were flash frozen using 10% glycerol as a cryoprotectant. A 2.1 Å resolution data set was collected using a RU-H3R rotating anode and an R-AXIS IV++ detector. Data were processed and scaled with programs from the CCP4 suite ( 31). The structure was solved by molecular replacement implemented by Phaser ( 32) using HLA-B8 (PDB ID 1M05) as the search model with all differences mutated to Alanine. The progress of refinement was monitored by the Rfree value (5% of the data) with neither a σ nor low-resolution cutoff being applied to the data. The structure was refined with Refmac ( 31) interspersed with rounds of model building in “Coot” ( 33) to final Rfactor and Rfree of 19.0% and 25.9%, respectively. Other relevant details regard to data collection and refinement statistics are summarized in Supplementary Table S1.
Novel responses to NY-ESO-1 identified in patients vaccinated with NY-ESO-1 ISCOMATRIX vaccine. As part of the immunologic monitoring of the NY-ESO-1 ISCOMATRIX vaccine trial ( 4), PBMC were obtained from patients before and at various times after the initial vaccination. These cells were cultured for 11 to 15 d with 18mer peptides, which collectively cover the entire NY-ESO-1 sequence. The expanded T cells were then assessed with individual peptide by ICS for IFN-γ. The results for PBMC prepared from Patient 9 at day 70 after vaccination are shown in Fig. 1A . This analysis revealed that the patient mounted major responses to 2 NY-ESO-1 18mer peptides [55-72 and 97-114 (ATPMEAELARRSLAQDAP)], whereas no responses above background to the other 26 peptides were detected. Neither of these 18mer peptides contains previously identified NY-ESO-1 epitopes, suggesting that each peptide contains at least one novel epitope.
Identification of the minimum epitope within peptide 55-72. The response to peptide 55-72 was chosen for further detailed study. NY-ESO-155-72–expanded T cells from Patient 9 were assessed with 13mer peptides within 55-72 sequence ( Fig. 1B). Of these, only peptide 60-72 stimulated a comparable response, suggesting that this 13mer contained the epitope. In addition, comparison of T cells expanded from PBMC collected before and after vaccination revealed that the response was vaccine-induced.
To identify the minimum epitope, additional HPLC-purified peptides with single AA extension or truncation at either end of 60-72 were tested at various concentrations in the absence of serum to avoid peptide trimming by serum proteases ( Fig. 1C). Three peptides gave equivalent responses when tested at 1 × 10−5M: the 13mer 60-72, as well as 2 14mers (59-72 and 60-73). However, at lower concentrations, the longer peptides were presented less efficiently. Similarly, truncated peptides (61-72 and 60-71) were far less efficient at activating these T cells. Thus, the 13mer NY-ESO-160-72 is the optimal epitope.
The NY-ESO-160-72 epitope is presented by HLA-B*0702. To determine the HLA restriction of the NY-ESO-160-72 epitope, a panel of EBV-transformed LCL expressing defined HLA molecules was used as antigen-presenting cells for T cells from Patient 9 ( Fig. 2A ). LCL line 9038 expresses HLA-Cw*0701 and HLA-A*0201 in common with the patient yet failed to present NY-ESO-160-72. In contrast, LCL lines 9065 and T242 efficiently presented NY-ESO-160-72 due to the shared expression of HLA-B*0702 with patient 9. Together, these results show that NY-ESO-160-72 is presented on HLA-B*0702.
The ability of HLA-B7 to present NY-ESO-160-72 was further confirmed by MHC binding assays using T2-B7 cells [a kind gift from Dr. P. Cresswell (Yale University, New Haven, CT)]. As these cells are TAP-deficient, they fail to express significant levels of surface HLA-B7 unless they are induced at lower temperature and then stabilized with exogenous peptide. Using this approach, the 13mer 60-72 was shown to bind to T2-B7 ( Fig. 2B) but not T2 cells (data not shown). Interestingly, the 14mer (59-72) also bound well, although less antigenic ( Figs. 1C and 2B). Other related peptides (61-72, 60-73 and 60-71) did not display detectable binding. Finally, an NY-ESO-160-72/HLA-B*0702 tetramer was synthesized, which stained a discrete population of CD8+ T cells within a peptide-expanded culture from Patient 9 ( Fig. 2C–D), thereby further confirming NY-ESO-160-72 to be the minimum epitope.
The NY-ESO-160-72 epitope is naturally processed and presented by melanoma cells. As the novel NY-ESO-160-72 epitope was identified using synthetic peptides, it was important to show its natural presentation on tumor cells. To this end, peptide-expanded T cells from Patient 9 were cultured for 5 hours with HLA-B*0702+ melanoma cell lines expressing or lacking NY-ESO-1. The ability of epitope-specific T cells to recognize the tumor cells was then determined by staining with the NY-ESO-160-72/HLA-B7 tetramer in combination with ICS for IFN-γ. As shown in Fig. 2C, the tetramer-positive cells produced IFN-γ in response to tumor cells that expressed NY-ESO-1 but not those lacking NY-ESO-1 expression, confirming natural presentation of this epitope by tumor cells. As expected, this response was significantly enhanced by pretreating the tumor cells with IFN-γ, which is known to up-regulate many components of the MHC class I presentation pathway, including surface class I expression ( 34). After IFN-γ treatment, the class I expression level on these lines increased 2- to 3-fold (detected by W6/32; data not shown).
These results are supported by analysis of CD107b mobilization, a marker of antigen-specific degranulation of cytotoxic T cells ( 35). As shown in Fig. 2D, tetramer-positive cells mobilized CD107b to the cell surface in response to NY-ESO-1–expressing tumor cells (especially when the tumor cells had been pretreated with IFN-γ) but displayed no response to tumor cells lacking NY-ESO-1 expression. Collectively, these results confirm the natural presentation of NY-ESO-160-72 on the surface of tumor cells. Furthermore, this recognition leads to degranulation (cytotoxicity) and production of IFN-γ, reflecting functional recognition of physiologically relevant levels of the determinant.
T-cell responses to NY-ESO-160-72 occur frequently in vaccinated patients but rarely occur naturally. To assess the prevalence of T-cell responses against the NY-ESO-160-72 epitope, five additional HLA-B*0702+ melanoma patients who had been vaccinated with NY-ESO-1 ISCOMATRIX vaccine were screened for the presence of NY-ESO-160-72–specific T cells. The data in Table 1 show that, 70 days after vaccination, 3 of these patients had readily detectable responses to this epitope. In contrast, when we analyzed PBMC samples collected from the same patients before vaccination, only one of the five had a weak response, suggesting that the NY-ESO-160-72 epitope is poorly immunogenic in the absence of vaccination. This concept is supported by analysis of seven additional melanoma patients that did not receive the vaccine. Despite all of these patients having anti–NY-ESO-1 antibodies in the serum that is reported to closely associate with cellular immunity ( 36), only one had a detectable, weak T-cell response to NY-ESO-160-72. Thus, the overall response rate in the absence of vaccination was 2 of 13 (15%), whereas after vaccination a total of 4 of 6 patients (67%) responded.
In Patient 9, responses were detected to only two epitopes across the entire length of the NY-ESO-1 protein (i.e., NY-ESO-160-72 and the epitope within NY-ESO-197-114; see Fig. 1A), suggesting that the NY-ESO-160-72 epitope may be immunodominant. To further assess this possibility, the three additional vaccinated patients who had detectable responses to NY-ESO-160-72 were screened for responses to other NY-ESO-1 epitopes ( Fig. 3A ). One of these individuals (Patient 120) responded to a range of NY-ESO-1 peptides in addition to NY-ESO-160-72. In contrast, for the two other patients, the response to the 18mer encompassing NY-ESO-160-72 was completely immunodominant, as no other responses to NY-ESO-1 could be detected. Tetramer staining ( Fig. 3B) confirmed that each of these three patients had T cells specific for NY-ESO-160-72 presented by HLA-B7. Thus, the NY-ESO-160-72 epitope was immunodominant in 3/4 HLA-B*0702+ patients receiving the NY-ESO-1 ISCOMATRIX vaccine, suggesting that this determinant is highly relevant to the anti–NY-ESO-1 immune response.
The NY-ESO-1 ISCOMATRIX vaccine formulation allows DCs to cross-present the NY-ESO-160-72 epitope efficiently. The results presented in Table 1 show that NY-ESO-160-72–specific T cells are rarely generated naturally. However, they are frequently elicited after vaccination with NY-ESO-1 ISCOMATRIX vaccine, suggesting that the vaccine formulation may be a particularly efficient way for directing NY-ESO-1 protein into an appropriate antigen processing pathway for cross-presentation. To investigate the ability of DCs to cross-present the NY-ESO-160-72 epitope, MoDCs were loaded with various concentrations of the full-length NY-ESO-1 ISCOMATRIX vaccine and tested for their ability to stimulate IFN-γ production by T cells specific for NY-ESO-160-72 or, as a control, the well-characterized HLA-A*0201-restricted NY-ESO-1157-165 determinant ( 5, 6). As shown in Fig. 3C, both epitopes could be efficiently cross-presented to T cells by MoDCs. Furthermore, exposure of MoDCs to the serine peptidase inhibitor AAF-cmk, an inhibitor of tripeptidyl peptidase II (TPP II), which was reported to be involved in cytosolic peptide trimming ( 9, 10), resulted in inhibition of cross-presentation of the NY-ESO-1157-165 epitope yet simultaneously and significantly enhanced presentation of NY-ESO-160-72 ( Fig. 3D). Thus, DCs can efficiently cross-present NY-ESO-160-72 to T cells when the antigen is provided together with the ISCOMATRIX adjuvant.
T cells recognizing NY-ESO-160-72 use a broad TCR repertoire. Several reports have shown that T cells responding to longer class I epitopes tend to have a highly skewed TCR repertoire ( 18, 24, 37– 39). To determine if this was also true of T cells recognizing NY-ESO-160-72, we stained peptide-expanded T cells with a panel of antibodies to various TCR Vβ families in combination with either IFN-γ or tetramer staining ( Fig. 4 ). Analysis of bulk T-cell cultures from Patient 9 revealed that T cells specific for NY-ESO-160-72 expressed a surprisingly broad TCR repertoire, as at least 10 distinct Vβ chain families were used. A similar level of TCR diversity was noted for PHA-expanded TIL from Patient 9. Furthermore, analysis of a second HLA-B*0702+ patient (Patient 120) revealed a similarly diverse TCR Vβ usage among NY-ESO-160-72–specific cells.
NY-ESO-160-72 adopts a bulged conformation when bound to HLA-B7. To address how the 13-mer NY-ESO-1 epitope bound HLA-B7, we determined the structure of the complex to 2.1 Å resolution to an Rfactor and Rfree of 19.0% and 25.9%, respectively (see Supplementary Table 1 for details). Electron density for the peptide ligand was mostly unambiguous, although density around the P6-His residue was relatively poor, suggesting some flexibility in this side chain.
The structure of NY-ESO-160-72 bound to HLA-B7 (green) is shown in Fig. 5A superimposed with a published gp100280-288 (orange) bound to HLA-A2 (blue; PDB ID 1TVB) to reveal a super-bulged conformation, in which the middle segment of the NY-ESO-160-72 protruded away from the peptide binding cleft, consistent with other complexes with peptides of similar size ( 18, 19, 22, 24). The contacts the NY-ESO-160-72 makes with HLA-B7 are detailed in Supplementary Table S2. Additionally, due to two hydrogen bond interactions made by P6 and P7 of the NY-ESO-160-72 peptide with Gln155 and Glu152 of HLA-B7, this bulged loop was pulled toward the α1 helix of the MHC molecule ( Fig. 5A).
Most of the contacts made between the 13mer peptide and the peptide-binding cleft of HLA-B7 were close to the NH2 and COOH terminal ends of the peptide. From the NH2 terminus of the peptide, P1 (Ala) was stabilized by a series of H-bonds and van der Waals forces, mainly by HLA-B7 residues Tyr7, Tyr171, and Tyr159. Peptide residue P2 (Pro) and P3 (Arg) participated in water-mediated hydrogen bonds and salt bridges to Tyr99 and Asp114. Similarly, the COOH-terminal end of the peptide exhibited an extensive network of hydrogen bonds between P10 (Ala) to P13 (Leu) of the peptide and Glu152, Arg156, Trp147, Ser77, Asn80, and Thr143 on HLA-B7. Thus, the NY-ESO-160-72 epitope bound to HLA-B7 via tethering at the NH2 and COOH termini and bulging centrally from the Ag-binding cleft.
Alanine-substitution peptides identify the residues critical for class I MHC binding and T-cell recognition of NY-ESO-160-72. A panel of Alanine-substitution peptides was synthesized to identify residues within the NY-ESO-160-72 determinant critical for MHC binding and/or TCR recognition. Initially, each peptide was tested for its ability to activate peptide-expanded T cells ( Fig. 5C). Substitution at P4, P6, P8, or P11 all completely abolished T-cell recognition. Substitution at P2, P3, or P13 reduced (but did not prevent) T-cell recognition. This was manifest as both a reduction in the percentage of cells responding and an increase in the dose at which the half-maximal response was observed (the EC50), indicating that these peptides are less potent than the native one. On the other hand, substitution of the glycines at P7 or P12 actually enhanced recognition compared with the native peptide, not only by slightly increasing the percentage of T cells able to recognize the peptide but also by decreasing the EC50. Only P5 seemed to have no effect, as substitution of this residue had no detectable effect on IFN-γ production compared with the native peptide.
Decreased T-cell activation could be due to either a reduced ability of the TCR to recognize the peptide or reduced MHC binding. To distinguish between these two possibilities, class I binding assays were conducted with the substituted peptides ( Fig. 5D). The four peptides that were completely unable to activate NY-ESO-160-72 specific T cells still bound (albeit with reduced efficiency) to HLA-B7. This is not surprising, as P4, P6, P8, and P11 are located within the bulge of the peptide (see Fig. 5B) and expected to interact with the TCR but minimally involved in binding MHC. In contrast, no binding to HLA-B7 was detected for the peptides containing P2, P3, and P13 substitutions at the HLA-B7 anchoring positions (Supplementary Table S2; Fig. 5B). It is interesting, however, that these peptides can still activate some T cells. Possibly, the binding of these peptides to HLA-B7 is below the limit of detection of the binding assay but still sufficient to allow recognition by some T cells. Taken together, the functional data are in good agreement with the structure.
The NY-ESO-1 ISCOMATRIX vaccine has proven highly immunogenic in patients with early stage melanoma ( 4). The success of this approach is probably related to the exceptional immunogenicity of the NY-ESO-1 antigen coupled with the unique properties of the ISCOMATRIX adjuvant, which is particularly efficient at targeting antigen for cross-presentation of class I–restricted epitopes ( 5, 6). In the present study, we show that this vaccination strategy promotes an immunodominant T-cell response to a novel epitope, which is rarely part of the natural immune response to NY-ESO-1 in the majority of melanoma patients tested. However, this response occurred with high frequency in patients vaccinated with the NY-ESO-1 ISCOMATRIX vaccine, albeit relatively small numbers of patients were assessed in this study, and the epitope was naturally generated by melanoma cells. This determinant is therefore likely to be relevant for the antitumor immune response in melanoma.
The novel NY-ESO-1 13AA epitope described in this report becomes part of a rapidly growing list of long class I epitopes derived from various tumor antigens including M-CSF gene ( 16), tyrosinase, MART-2, CAMEL, and p53 ( 15), suggesting that class I epitopes of noncanonical length form an important component of antitumor immune responses. Furthermore, the computer-based epitope prediction algorithms currently available do not consider epitopes longer than 10 AA, implying that the significance of such long epitopes in antitumor immunity, as well as immune responses to pathogens, may be significantly underestimated ( 15) and this has important implications for vaccine design ( 14). Interestingly, the data in Fig. 1 show that T cells from Patient 9 recognized a second NY-ESO-1 epitope within the region 97-114, and although full characterization of this epitope is beyond the scope of the present study, preliminary results indicate that this second determinant is also of noncanonical length (most likely 12AA).
The binding of NY-ESO-160-72 to MHC is similar to other long class I epitopes, in that the extra length is accommodated by the peptide bulging out in the center. For example, two long (11AA and 13AA) epitopes from the EBV antigen BZLF1 were shown to adopt a similar bulged conformation upon binding to class I MHC. In both cases, a TCR interacting with the peptide-MHC complex accommodated the bulge, with the TCR either “perching” on top of the bulged region ( 40) or flattening the bulge to enable more extensive interaction with the MHC ( 41). Both determinants were recognized by T cells displaying a highly restricted TCR repertoire ( 18, 24), suggesting a limitation in TCR availability when responding to such unusual peptide-MHC structures ( 42). Strikingly, however, we have shown that the NY-ESO-160-72 epitope was recognized by a very broad TCR repertoire. The fact that such an unusual peptide-MHC structure was recognized by so many different TCRs raises the intriguing possibility that the complex may possess some degree of flexibility, adopting different conformations upon being recognized by different TCRs.
A considerable body of evidence now supports the concept that CD8+ T-cell responses are initiated by DCs through a process known as cross-priming ( 12). This process involves the uptake of exogenous antigen by DCs, which then escapes the endosomal compartment to enter the cytoplasm and access the class I presentation pathway, ultimately resulting in priming of naïve CD8+ T cells ( 43). In the present study, T-cell responses to NY-ESO-160-72 were frequently observed in patients vaccinated with NY-ESO-1 ISCOMATRIX vaccine but rarely observed in unvaccinated patients, implying that T cells specific for this epitope were not efficiently cross-primed by DCs in the setting of a natural immune response. One explanation for this observation is that the default class I antigen processing pathways in DCs are inappropriate for processing this epitope. For instance, it may be overprocessed (cleaved), as its cross-presentation by DCs was significantly enhanced in the presence of AAF-cmk, an inhibitor of the TPP II enzyme, which is involved in cytosolic peptide trimming ( Fig. 3D; refs. 9, 10). In contrast, previous studies have shown that formulating NY-ESO-1 with ISCOMATRIX adjuvant allows it to access an unusual proteolysis pathway in DCs that is independent of proteasome degradation ( 5), and which may be more appropriate for generation of the NY-ESO-160-72 epitope. Interestingly, this alternative proteolysis pathway involved TPP II for generation of the HLA-A*0201–restricted epitope NY-ESO-1157-165, 6 whereas in our study, TPP II inhibition led to enhanced cross-presentation of NY-ESO-160-72, indicating that processing pathways may vary for different NY-ESO-1 CD8+ T-cell epitopes. Further studies are required to assess the processing of NY-ESO-160-72 by DCs under various conditions. However, our results suggest that formulation of protein antigens with the ISCOMATRIX adjuvant may allow the development of useful T-cell responses that would otherwise not be generated during the course of natural immunity.
Characterization of the novel NY-ESO-160-72 epitope has revealed a number of interesting features of antitumor immunity and recognition of class I determinants. First, unusually long class I peptides that bulge out from the MHC molecule do not necessarily represent a challenge to T-cell recognition, as NY-ESO-160-72 was recognized by a broad TCR repertoire and as the most immunodominant epitope, indicating that the T-cell response to long epitopes such as this can be remarkably robust. Second, our results suggest that certain antitumor immune responses are not efficiently generated, despite the existence of an appropriate naïve T-cell repertoire and despite the fact that these are biologically useful responses, which was reflected by natural presentation of NY-ESO-160-72 by tumor cell lines in vitro as well as tumor infiltration by vaccinated, NY-ESO-160-72–specific T cells in vivo. Finally, at least in the case of the NY-ESO-160-72 epitope, vaccination in the context of ISCOMATRIX adjuvant seems to correct such a defect, possibly by increased antigen delivery efficiency and by targeting alternative antigen processing pathways in DCs. We expect vaccination with peptide-based approaches ( 14) would also lead to effective NY-ESO-160-72–specific T-cell priming. Together, our studies highlight the importance of studying longer class I epitopes in the context of antitumor immunity and suggest that vaccines formulated with full-length protein antigen and ISCOMATRIX adjuvant represent a promising approach to cancer treatment.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: National Health and Medical Research Council (NHMRC) project grants 433608 (W. Chen), 491117 (A.W. Purcell and J. Rossjohn), and grants from Cancer Council Victoria 381409 (W. Chen) and 433626 (L. E). J. Rossjohn is an ARC Federation Fellow; C.S. Clements is an ARC QEII fellow and A.W. Purcell is a NHMRC Senior Research Fellow.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵5 Klein et al., Clinical Cancer Research, in press 2009.
↵6 Schnurr et al., Journal of Immunology, in press.
- Received August 3, 2008.
- Revision received September 28, 2008.
- Accepted October 17, 2008.
- ©2009 American Association for Cancer Research.