EWS-FLI-1-Targeted Cytotoxic T-cell Killing of Multiple Tumor Types Belonging to the Ewing Sarcoma Family of Tumors

Immunotherapy is a promising approach for the treatment of established neoplastic disease. Many tumors share the expression of tumor-specific antigens not found in healthy cells, raising the possibility of invoking immune responses that selectively target diseased cells. Ewing sarcoma and the historically named primitive neuroectodermal tumor (pPNET), Biphenotypic sarcoma, and Askin's tumor, all share recurrent translocations that involve fusion of the EWS gene (22q-12) to genes encoding ETS transcription factors, of which, fusion to the FLI-1 gene occurs in about 90% of cases (1). For this reason, these tumors are now more broadly known as the Ewing sarcoma family of tumors (ESFT).

ESFTs are among the most aggressive tumors encountered in children and young adults, both morphologically and clinically. They require equally aggressive therapy, including surgery or radiation of the primary tumor site, and intensive multiagent chemotherapy (2). Presently, approximately 70% of patients with Ewing sarcoma localized disease (3), and just 20% to 30% of patients with metastatic disease (4), survive for 5 years after initial diagnosis; additional therapeutic options are badly needed.

Tumor recurrence is a serious problem following ESFT therapy, and 30% to 40% of patients with Ewing sarcoma localized tumors develop recurrent disease (5). Such patients have a 5-year survival rate of just 13% (6). The addition of adjunctive, immune-activating, therapies delivered during the first remission has the potential to reduce recurrence rates (7), both through the capacity of the immune system to mop up residual disease, and through the stimulation of immune “memory.” The junction region where the EWS and FLI-1 genes fuse, also known as the breakpoint, confers a region of novelty in the EWS-FLI-1 protein, and therefore provides a means to target selectively the immune system to EWS-FLI-1⁺ tumor cells. At least 11 subtypes of the EWS-FLI-1 gene exist in which exons 7, 9, or 10 from EWS are combined with exons 4, 5, 6, 7, or 8 from FLI-1 (8). Although the breakpoint sequences of each subtype are different, EWS-FLI-1 type 1 fusions (EWS exon 7 fused to FLI-1 exon 6) are present in approximately 57% of cases (9). Thus, immunotherapies targeting the EWS-FLI-1 type 1 protein may provide clinical benefit to approximately half of all patients with ESFT.

In this study, we investigated the capacity of peptides derived from EWS-FLI-1 type 1, presented on a common HLA-type, HLA-A2.1, to activate CTL capable of recognizing and killing ESFT carrying the EWS-FLI-1 type 1 translocation. Anticipating that native peptides would be poorly immunogenic (10), we also studied the immune-stimulating properties of native peptides modified at key binding residues to improve the binding affinity for HLA-A2.1. We show that EWS-FLI-1 type 1-derived peptides have a poor capacity to activate CTL, resulting in weak CTL killing of ESFT cells. The modification of key binding residues, in contrast, results in potent CTL activation, strong CTL killing across a range of ESFT types, and increased survival of mice with established Ewing sarcoma xenografts.

HLA-A2/K^b transgenic mice were bred in our facility with the kind consent of Dr. L. Sherman (The Scripps Research Institute; ref. 11). SCID/beige mice were purchased from Taconic. Experiments were conducted under an animal protocol approved by the Animal Care and Use Committee at Beth Israel Deaconess Medical Center.

Anti-HLA-A2.1 mAb (BB7.2) and anti-CD8 (53–6.7) were purchased from BD biosciences. Anti-CD40 (3/23) was from Serotec. PE-HLA-A2.1/YLNPSVDSV and PE-HLA-A2.1/GILGFVFTL MHC Pentamers were from ProImmune. All peptides were synthesized by Thinkpeptides (ProImmune Ltd). Modified peptide sequences were compared against a human protein sequence database (http://web.expasy.org/blast/) conducted at the Swiss Institute of Bioinformatics using the BLAST network service and the NCBI BLAST 2 software; no homology to human proteins was observed.

TC-71 and TC-32 cells were a kind gift from Dr. Aykut Üren, Georgetown University Medical Center, Washington, DC (12). Rh1 (EW8) cells were a kind gift from Dr. Peter Houghton, Nationwide Children's Hospital, Columbus, Ohio (13). TC-174 and TTC-547 were a kind gift from Prof. Timothy Triche, University of Southern California, Los Angeles, California (14). T2, SK-PN-DW, and SK-ES-1 were from the ATCC, initially authenticated by STR profiling, and passaged for less than 6 months following resuscitation. A673 and EWS502 were provided by S.L., and WE-68, SK-N-MC, STA-ET-1, and TC-83 by F.V.V. RD-ES was a kind gift from Dr. Lee Helman, National Cancer Institute, Bethesda, Maryland.

HLA-A2.1 expression was determined by FACS. Cells with low or undetectable HLA-A2.1 expression were cultured for 48 hours in the presence of 300 U/mL rhIFN-γ (R&D Systems) to increase surface MHC expression, and reanalyzed. The presence of the EWS-FLI-1 type 1 (EWS exon 7 fused to FLI-1 exon 6) or EWS-FLI-1 type 2 (EWS exon 7 fused to FLI-1 exon 5) translocations was determined by RT-PCR as described (15).

The online “SYFPEITHI” database (http://www.syfpeithi.de/) was used to predict HLA-A2.1:peptide binding affinities, and the “BIMAS” database (http://www-bimas.cit.nih.gov/molbio/hla_bind/) to predict HLA-A2.1: peptide dissociation half-times. Binding to HLA-A2.1 was assessed using the REVEAL MHC-Peptide Binding Assay service of Proimmune Ltd.

Peptide stabilization of cell-surface MHC molecules was assessed using human T2 cells (TAP⁻, HLA-A2.1⁺) as described (16).

HLA-A2.1/K^b mice were immunized i.d. (100 μL; both flanks) with 200 μg test peptides in conjunction with CASAC (Combined Adjuvant for Synergistic Activation of Cellular Immunity)—a potent adjuvant for the activation of CD8 T cells (17) comprising 50 μg CpG1826 (Invitrogen), 25 μg anti-CD40, and 100 ng rmIFN-γ (R&D Systems) emulsified into the Sigma Adjuvant System (Sigma). After 8 days, splenocytes and inguinal lymph nodes cells were harvested and restimulated in vitro for 5 days with 1 μg/mL peptide. CD8 cells were negatively selected using MACS beads (Miltenyi biotech), and cultured with CFSE-labeled human ESFT targets. After 5 hours, cell death was assessed by 7-AAD costaining of CFSE-labeled target cells using a FACSCanto II Analyzer as described (18). An irrelevant peptide, GILGFVFTL (Influenza A matrix peptide 58–66), was used to determine nonspecific CTL killing, which was subtracted from all subsequent test data.

Characterized, cryopreserved human peripheral blood mononuclear cells (PBMC) were purchased from Cellular Technology Ltd. HLA-A2⁺ donors were selected on the basis of known recall reactivity to Influenza A peptides. PBMCs were cultured for 4 weeks to generate CTL according to an established protocol (19). To assess CTL activity, PBMCs were cultured with CFSE-labeled TC-71 cells as above at a ratio of 20:1.

A lentiviral vector encoding both green fluorescent protein (GFP) and firefly luciferase was used for the transduction of TC-71 cells. The vector was generated using a third-generation packaging system (20), cotransfecting 293T cells (System Biosciences) with the lentiviral expression plasmid pGreenFire1-CMV (System Biosciences) and packaging plasmids pMDLg/pRRE, pRSV.Rev, and pMD.G (Addgene) using Lipofectamine 2000 (Invitrogen). A 1:1 mix of 293T supernatant and fresh medium containing Polybrene (Sigma; final concentration of 8 μg/mL) was added to TC-71 cells at approximately 25% confluence; plates were spun at 1,000 g for 60 minutes at 32°C to enhance transduction efficiency (21). The virus-containing media were then replaced with fresh media and the cells were cultured for 48 hours before assessment of GFP expression by FACS.

SCID/beige were inoculated i.v. with 2 × 10⁶ TC-71-Luc cells. Five and 12 days later, PBS (no T-cell control) or 3 × 10⁶ CD8 T cells derived, cultured, and negatively selected as described in CTL killing assays above were adoptively transferred i.v. Tumor growth was monitored by bioluminescence imaging. Blind assessments of animal health were conducted.

The Xenogen IVIS-50™ Bioluminescence Imaging System was used to assess tumor growth. SCID/beige mice were anesthetized with isoflurane and injected i.p. with 150 mg/kg of d-luciferin (Caliper LifeSciences) in PBS. Ten minutes after injection, and under continued isoflurane inhalation, the mice were imaged for a period of 5 minutes. Using Living Image® software (Xenogen), the luminescence signal was represented as a heat map superimposed over a grayscale photograph of the animal.

Mouse tumor images were acquired on an ASPECT Model M2 1T tabletop MRI scanner (ASPECT Magnet Technologies Ltd.). VivoQuant (inviCRO) software was used to visualize DICOM datasets.

Statistical comparisons of mean values were conducted using unpaired Student's t-test. Statistical comparisons of survival curves were conducted using the logrank test with the null hypothesis that treatments did not change survival. P less than 0.05 (*) was considered significant. P less than 0.005 (**) and P less than 0.001 (***) are indicated.

The affinity with which peptides bind to MHC Class-I is critical for the induction of CTL activation. Therefore, we used the SYFPEITHI and BIMAS databases to predict the binding affinity of native peptide sequences that span the breakpoint region of EWS-FLI-1 type 1 to HLA-A2.1. For each peptide that spans the fusion region between EWS and FLI-1, a relatively low-affinity score was predicted (Table 1). Of these, highest SYFPEITHI score, 15, was predicted for QQNPSYDSV.

The low predicted binding scores for these peptides suggested that these sequences would not bind with the affinity required to induce CTL activation, and would therefore not be suitable for use in targeting CTL responses against cells carrying the EWS-FLI-1 type 1 translocation. However, appropriately modified tumor peptides can often prime the immune response to act against native tumor peptides incapable of inducing an immune response on their own (22). We therefore substituted key anchor residues in 3 native peptides with the most promising SYFPEITHI affinity scores: SYGQQNPSY (score = 5), SSYGQQNPS (score = 6), and QQNPSYDSV (score = 15), according to previously described HLA-A2.1-binding principles (23). As expected, the predicted binding affinities of peptides modified in this manner increased substantially both SYFPEITHI and BIMAS scores following the substitution of residues at positions 1, 2, 6, and/or 9 with Tyrosine (Y), Leucine (L), Isoleucine (I) and/or Valine (V: see Table 2).

Because computer models of peptide:MHC binding affinity have only 60% to 80% predictive accuracy (16), we measured empirically the ability of native and modified peptides to bind to HLA-A2.1. Binding to HLA-A2.1 was detected using the REVEAL MHC-Peptide Binding Assay; an antibody staining assay on the basis of the conformational changes that only occur when peptide, MHC class I, and β2m form a complex. As shown in Fig. 1A–C, most of the modified peptides, but not all, displayed a stronger binding capacity than the corresponding native peptides from which they were derived. Together, these data show that the HLA-A2.1-binding of native EWS-FLI-1 type 1 junction peptides can be improved through the modification of anchor residues.

We also sought to determine whether modified peptides would show an improved capacity to stabilize MHC molecules on the live cell surface, as the ability of peptide/MHC complexes to remain on the cell surface is critical to the initiation of T-cell activation. We selected modified peptides with a predicted SYFPEITHI score greater than 21 and improved binding to HLA-A2.1 for further study. Human T2 cells express HLA-A2.1 but lack Transporters associated with Antigen Processing (TAP); consequently their cell surface MHC molecules are empty. The association of exogenously added peptides with the empty HLA-A2.1 molecules stabilizes them and results in increased levels of cell surface HLA-A2.1 molecules, recognizable by monoclonal antibodies and flow cytometry. Unlike the native peptide sequences SYGQQNPSY (#1) and SSYGQQNPS (#9) from which they were derived, each of the modified peptides (#4, #6, #8, and #11, #12, and #13, respectively) showed improved stabilization of cell surface MHC (Fig. 1D and E). The native sequence QQNPSYDSV (#14) was found to stabilize cell surface MHC molecules at a high concentration (50 μg/mL), but not at low concentrations less than 10 μg/mL (Fig. 1F). Two of the modified QQNPSYDSV peptides (#18 and #22) did not show improved MHC stabilization when compared with the native sequence QQNPSYDSV (#14). The remaining 4 peptides (#20, #21, #23, and #24) revealed an excellent capacity for MHC stabilization, even at a 50-fold reduction in peptide concentration (1 μg/mL). One peptide in particular, YLNPSVDSV (#24), was found to stabilize MHC to a similar extent as 50 μg/mL of the native peptide QQNPSYDSV (#14) at a 500-fold lower peptide concentration (0.1 μg/mL). These data show that modifying native EWS-FLI-1 type 1 junction peptides can result in better retention of peptide/MHC complexes on the cell surface.

A critical question when modifying tumor peptides to induce CTL activation is whether or not CTL raised against the modified peptide will recognize and kill tumor cells expressing the native peptide. We conducted CTL assays using purified CD8 T cells derived from HLA-A2.1/K^b mice immunized with native or modified peptides. We used an unmodified human Ewing sarcoma cell line: TC-71, as a tumor target, which we confirmed as HLA-A2.1⁺ by flow cytometry (data not shown). The EWS-FLI-1 type 1 translocation was also confirmed by RT-PCR using TaqMan probes (data not shown). As shown in Fig. 1G, the native peptide SYGQQNPSY (#1) was not able to induce a detectable CTL response, and the CTL responses induced by the 3 modified SYGQQNPSY peptides (#4, #6, and #8) were weak (≤15% killing). The native peptide SSYGQQNPS (#9) also induced weak CTL activity (<10%: Fig. 1H), although each of the 3 modified SSYGQQNPS peptides (#11, #12, and #13) induced an appreciable increase in CTL activity, with up to approximately 25% killing. Interestingly, these 3 peptides all induced strong CD8 T-cell proliferation during in vitro restimulation (data not shown). However, this did not appear to correlate with a high level of target cell killing. Of the native peptides, QQNPSYDSV (#14) induced the strongest CTL response with approximately 15% to18% killing (Fig. 1I). Surprisingly, its modified derivative, YLNPSYDSV (#20), did not induce CTL killing despite displaying an increased capacity to stabilize cell-surface MHC (see Fig 1F). The strongest CTL activity was displayed by modified peptides QINPSVDSV (#21) and YLNPSVDSV (#24) which both displayed over 40% killing. Peptide YLNPSVDSV (#24) was found to be considerably more immunogenic than native peptide QQNPSYDSV (#14), as evidenced by the consistently increased size of draining lymph nodes following vaccination (Supplementary Fig. S1).

Collectively, these data show that the CTL-inducing capacity of native EWS-FLI-1 type 1 junction peptides presented on HLA-A2.1 is low, but modification of key MHC-binding anchor residues can result in the activation of CTL clones capable of recognizing and killing unmodified tumor cells expressing native EWS-FLI-1 type 1 junction peptides. Because of its consistent ability to induce strong immune activation, peptide YLNPSVDSV (#24) was selected for further studies.

The TC-71 Ewing sarcoma cell line is known to be tumorigenic in SCID/beige mice (12), and has been used for preclinical in vivo studies for Ewing sarcoma (12, 24). Following the stable transduction of the firefly luciferase gene into TC-71 cells (TC-71-Luc), TC-71-Luc cells were transferred i.v. into SCID/beige mice. On days 5 and 12 after tumor challenge, we adoptively transferred CD8 T cells derived from HLA-A2.1/K^b mice immunized with YLNPSVDSV (#24) or irrelevant control peptide (Flu: GILGFVFTL). Bioluminescence imaging of the mice 21 days after tumor challenge indicated fewer mice with detectable tumors in the YLNPSVDSV-treated group (Fig. 2A). Treatment of the xenografts with T cells specific for YLNPSVDSV significantly increased mouse survival time when compared with untreated mice or mice receiving T cells recognizing an irrelevant peptide (Flu: GILGFVFTL; Fig. 2B, P < 0.05). Ovaries and kidneys/adrenals were a strong metastatic site for these tumors (Fig. 2C). Figure 2D depicts a paraspinal suprarenal tumor imaged in a mouse recipient of EWS-FLI-1 type 1-targeting T cells. These data are consistent with a recent study (25) in which Rag2^−/−;γc^−/− mice were used as hosts for TC-71 Ewing sarcoma xenografts. We did not see a correlation between treatment group and tumor location. The results indicate that the adoptive transfer of YLNPSVDSV-specific CTL, but not irrelevant control CTL, into mice with established Ewing sarcoma xenografts results in enhanced survival.

The ESFTs are grouped together because they share expression of fusion genes like the EWS-FLI-1 translocation. We therefore asked whether multiple ESFT members would be susceptible to EWS-FLI-1-targeted CTL killing. An array of human ESFT tumor lines were screened to identify lines that are both HLA-A2.1⁺ and carry the EWS-FLI-1 type 1 translocation. As shown in Table 3, 11 suitable lines were identified spread across multiple different EWS-FLI-1⁺ ESFT tumor types. We identified a further 3 EWS-FLI-1 type 1⁺ cell lines of unknown tumor specificity (unknown). Three negative control lines were also identified: RD-ES (EWS-FLI-1 type 1⁻, HLA-A2.1⁻), SK-N-MC (EWS-FLI-1 type 1⁺, HLA-A2.1⁻), and SK-ES-1 (EWS-FLI-1 type 1⁻, HLA-A2.1⁺). We conducted CTL assays using purified CD8 T cells derived from HLA-A2.1/K^b mice immunized with native (QQNPSYDSV #14) or modified (YLNPSVDSV #24) peptides. As shown in Fig. 3A, killing by CTL specific for EWS-FLI-1 type 1 (above 10%, and after subtraction of nonspecific CTL killing) was displayed against 8 of the 11 target lines, including at least one example of killing in each of the ESFT tumor types tested. Furthermore, YLNPSVDSV-specific CTL (#24) showed significantly improved killing against 7 of these 8 lines when compared with QQNPSYDSV-specific CTL (#14). CTL killing directed against the negative control lines (RD-ES, SK-N-MC, and SK-ES-1), in which EWS-FLI-1 type 1-peptides are not presented on HLA-A2.1 molecules, was not significantly above background, as expected. These data show that CTL which target the EWS-FLI-1 type 1 translocation display potent killing of multiple EWS-FLI-1 type 1⁺ expressing cells in an HLA-A2.1 dependent manner. These data also underline the improved cytotoxic capacity of CTL raised using a modified (YLNPSVDSV #24) EWS-FLI-1 type 1 peptide in comparison to CTL raised using the native (QQNPSYDSV #14) peptide.

Because our results suggested that YLNPSVDSV (#24) peptide could induce EWS-FLI-1-directed CTLs in the HLA-A2.1/K^b mouse, we expected this peptide to be an epitope for human CTL. In the absence of available blood from patients with HLA-A2.1⁺ known to carry the EWS-FLI-1 type 1 translocation, we attempted to generate CTL from the PBMCs of healthy donors.

Precharacterized PBMCs from 3 different HLA-A2.1⁺ donors with known recall responses Influenza A peptides were stimulated with native (QQNPSYDSV #14), modified (YLNPSVDSV #24), or positive control (GILGFVFTL; Flu) peptide. The degree of peptide-specific CD8 T-cell clonal expansion in the modified- and control-peptide populations was assessed through CD8 and MHC pentamer costaining. Because of the low binding affinity of the native peptide to HLA-A2.1, we were unable to make HLA-A2.1/QQNPSYDSV pentamer.

As shown in Fig. 3B, the baseline level of YLNPSVDSV-specific CD8 T cells in unstimulated PBMC was low (∼0.4%) compared with GILGFVFTL peptide (∼0.5%–1.4%); this is expected because of the prior exposure of these patients to Influenza A. After 4 weeks of peptide stimulation, however, the level of CD8 T cells specific for YLNPSVDSV peptide expanded to between 2.57% and 2.86%. We further examined the cytotoxic activity of these CTLs against TC-71 cells. As shown in Fig. 3C, for each of the 3 donors tested, YLNPSVDSV-specific CTL displayed substantially improved killing of TC-71 Ewing sarcoma cells compared with QQNPSYDSV-specific CTL. We also saw killing of TC-71 cells using YLNPSVDSV-specific CTL derived from a further 2 healthy donors, although comparative native-peptide-specific CTL assays were not conducted because of lack of expansion of these cells in culture (data not shown). These results identify CD8 T cells capable of recognizing a modified (YLNPSVDSV #24) peptide from the junction region of the EWS-FLI-1 type 1 translocation in the peripheral blood of healthy donors, and show that they can be readily activated by peptide, to kill efficiently cells bearing the EWS-FLI-1 type 1 translocation.

The EWS-FLI-1 oncogene is well recognized as a highly promising therapeutic target for ESFT (26). The complete lack of EWS-FLI-1 expression in healthy cells, coupled with the absolute requirement of diseased cells to maintain expression of this gene (27, 28), have led to the development of several EWS-FLI-1-targeting therapies in recent years. While many of these studies have concentrated on disrupting the function of this oncogene (29, 30), studies are also being conducted which attempt to activate the immune system specifically to target cells carrying EWS-FLI-1.

In a pilot clinical trial conducted by Dagher and colleagues (31), EWS-FLI-1-targeted immunotherapy was investigated using native junction peptides. In this study, patient peripheral blood leukocytes (PBL) were pulsed with long (15–18aa) peptides spanning the breakpoint region for both EWS-FLI-1 type 1 and type 2, irradiated, and injected intravenously into patients, who were then put on IL-2 therapy. No patients showed any evidence of peptide-specific cytokine release, and all patients suffered toxicity as a result of IL-2 therapy. However, this study was primarily a safety study incorporating immune compromised patients with bulky, rapidly progressing tumors. Importantly, no evidence of peptide-induced toxicity was found.

A subsequent study by the same group (7) showed improved 5-year survival rates in patients following the transfer of autologous T cells, harvested prechemotherapy and reintroduced postchemotherapy, in combination with long breakpoint peptide-pulsed dendritic cell vaccines. However, survival did not correlate with immune responses to peptide vaccines, which were poor in comparison to responses induced by a commercially available Flu vaccine delivered around the same time as the dendritic cell/peptide vaccines. Thus, improved survival may have been attributed to immune reconstitution rather than EWS-FLI-1-targeted immunotherapy. Subsequently, the authors highlighted the need for the discovery of more immunogenic EWS-FLI-1 antigens and more potent vaccine-delivery systems.

Because Caucasians have the highest incidence of Ewing sarcoma (32), and HLA-A2.1 is present in approximately 40% of Caucasians (33), we focused our study on HLA-A2.1-restricted presentation of EWS-FLI-1 type 1 peptides. Our results suggest that native peptides from the breakpoint region of EWS-FLI-1 type 1 bind with weak affinity to HLA-A2.1, resulting in poor stability of peptide/MHC complexes on the cell surface. Despite coimmunizing HLA-A2/K^b mice with native peptides in combination with a potent CTL-inducing adjuvant (CASAC; refs 17, 34), strong CTL activity could not be induced. In contrast, modified peptides, in which we substituted key binding residues of the native peptides, displayed improved HLA-A2.1-binding affinity, an enhanced capacity to stabilize cell-surface HLA-A2.1 molecules, and induction of potent CTLs targeting EWS-FLI-1 type 1. Importantly, CTL raised using modified peptides were able to recognize and kill tumor cells endogenously expressing native EWS-FLI-1 type 1 peptides.

CTL induced with modified peptide (YLNPSVDSV) and recognizing EWS-FLI-1 type 1 were adoptively transferred into immune compromised mice bearing established TC-71 xenografts. This produced enhanced survival, but not a cure. In the absence of host regulatory T-cell (T-reg) activity in our mice, it is likely that the numbers of transferred T cells were simply insufficient to control the aggressive growth of TC-71 cells. Although not investigated in the present study, this model might be an attractive one with which to examine the combined influence of conventional cytoreductive therapies, such as chemotherapy, radiation, and surgery to reduce significantly the tumor load, with subsequent immunotherapy to prevent tumor recurrence.

YLNPSVDSV-specific CTLs were found to be capable of killing multiple ESFT tumor types. In 7 of the 8 target lines sensitive to CTL-killing, these CTL significantly improved killing when compared with CTL raised against the native peptide QQNPSYDSV (#14). Interestingly, we did not see convincing CTL killing against 3 cell lines: WE-68, STA-ET.1, and TTC-547, despite the preincubation of these cells with 300 U/mL rhIFN-γ to increase surface MHC expression. This raises the possibility that these tumor lines have a defect in the processing or presentation of endogenous EWS-FLI-1 peptides on their cell surface. Alternatively, it has been shown that some ESFT lines downregulate caspase-8 in culture, rendering them resistant to Fas-mediated killing (35).

The successful activation of CD8 T cells from healthy human PBMC using the modified YLNPSVDSV peptide indicates that CD8 T-cell clones capable of recognizing this sequence exist and are not depleted during selection in the thymus. Furthermore, YLNPSVDSV-specific CTL can be readily activated by peptide to result in the potent killing of cells bearing the EWS-FLI-1 type 1 translocation. Subsequent experiments need to be carried out to ensure that YLNPSVDSV-specific T cells present in the blood of patients with EWS-FLI-1 type 1⁺ HLA-A2.1⁺ are similarly responsive to stimulation with YLNPSVDSV. For peptide-based ESFT vaccines to reach their full potential, a library of peptides known to activate EWS-FLI-1-targeting CTL across multiple HLA-restrictions may need to be created. In addition, immunogenic MHC Class II-restricted peptides spanning the EWS-FLI-1 breakpoint region (10) merit further investigation for their capacity to induce tumor-specific helper T-cell responses, as ultimately a multipronged approach is likely to yield the most effective antitumor immunity.

Because of the aggressive growth rates of pediatric sarcomas, the propensity of advanced-stage tumors to downregulate MHC expression (36), and the importance of MHC class I expression for patient survival (37), both the timing and partnering of immunotherapies with cytoreductive therapies for ESFT warrant further investigation.

No potential conflicts of interest were disclosed by the other authors.

Conception and design: C. H. Evans, J. W Wells

Development of methodology: R. M. Porter, K. Robichaud, M. C Gebhardt, J. W Wells

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. H. Evans, J. W Wells

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Liu, R. M. Porter, R. P. O' Sullivan, T. Merghoub, E. P. Lunsford, K. Robichaud, F. van Valen, J. W Wells

Writing, review, and/or revision of the manuscript: C. H. Evans, F. Liu, R. M. Porter, R. P. O' Sullivan, T. Merghoub, S. L. Lessnick, J. W Wells

Study supervision: C. H. Evans, J. W Wells

Contributed expertise regarding EWS-FLI-1 and Ewing sarcoma: S. L. Lessnick

We thank Linda Sherman for permission to use HLA-A2/K^b transgenic mice. We also thank Aykut Üren, Georgetown University Medical Center, Washington, DC; Peter Houghton, Nationwide Children's Hospital, Columbus, Ohio; Timothy Triche, University of Southern California, Los Angeles, California; and Lee Helman, National Cancer Institute, Bethesda, Maryland, for their generous contribution of rare ESFT cell lines.

This work was supported by a grant from the Children's Cancer Research Fund to J. W. Wells.

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.