Enhancement of Tumor-Reactive Cytotoxic CD4⁺ T-cell Responses after Ipilimumab Treatment in Four Advanced Melanoma Patients

Cytotoxic T-lymphocyte antigen-4 (CTLA-4) is a coinhibitory molecule expressed on activated T cells and regulatory T cells. CTLA-4 is essential in maintaining immune homeostasis and contributes to immune tolerance by downregulating T-cell activation (1–3). Ipilimumab is a fully humanized immunoglobulin G1 (IgG1) monoclonal antibody (mAb) that blocks CTLA-4. Two pivotal phase III trials in patients with advanced melanoma showed a survival benefit for ipilimumab as monotherapy or when added to chemotherapy (4, 5), leading to its approval by the U.S. Food and Drug Administration in March 2011, marking a milestone in the field of immunotherapy. Immunotherapeutic strategies have also focused on targeting antigens expressed exclusively on melanoma cells, for example, melanoma differentiation antigens (such as gp100 or Melan-A) or cancer/testis (CT) antigens. CT antigens are expressed in a variety of human malignancies but not in normal adult tissues except for the testis and placenta (6). To date, the best characterized CT antigen is NY-ESO-1, which is highly immunogenic and elicits spontaneous antibody and T-cell responses in patients with cancer whose tumors express NY-ESO-1 (7). Prior studies by our group and others have shown that ipilimumab treatment of patients with advanced melanoma induces polyfunctional NY-ESO-1- and Melan-A–specific CD8⁺ T-cell responses (8, 9). We have also correlated the induction of polyfunctional NY-ESO-1 CD4⁺ T-cell responses with the development of integrated antibodies and CD8⁺ T-cell responses in patients, who experience clinical benefit (10).

The essential role of CD4⁺ T helper (T_H) cells in enhancing and sustaining CD8⁺ T-cell responses is well established (11). Recent evidence in mouse models, however, suggests more direct roles for tumor-reactive CD4⁺ T cells for cancer immunotherapy (12, 13). Specifically, CTLA-4 blockade has the capacity to expand antigen-specific cytotoxic CD4⁺ T cells in vivo in mice (12). Furthermore, adoptive transfer of CD4⁺ T cells expanded from a single tumor-reactive T-cell clone resulted in a durable complete response in a patient with melanoma (14). However, the cytotoxic function of antigen-specific CD4⁺ T cells during ipilimumab treatment and its intracellular mechanism have not been characterized. We hypothesized that CTLA-4 blockade could result in expansion and/or enhancement of cytotoxic CD4⁺ T-cell responses in human cancer patients through the modulation of T_H1 transcription factors. To address this question, we conducted in-depth immune monitoring of 4 NY-ESO-1 seropositive melanoma patients who were treated with ipilimumab and had availability of properly annotated specimens. Peripheral blood mononuclear cells (PBMC) were analyzed by intracellular cytokine staining (ICS) using multiparametric flow cytometry. Samples were analyzed following in vitro stimulation with NY-ESO-1 overlapping or single peptides. IFN-γ ELISPOT was conducted to define specific CD4⁺ T-cell peptide responses. Transcription factors T-bet and Eomesodermin (Eomes) as well as cytotoxic degranulation markers perforin and granzyme B were analyzed on NY-ESO-1–specific CD4⁺ T cells. NY-ESO-1–specific CD4⁺ T-cell lines were established to confirm their ability to recognize NY-ESO-1–positive tumor cell lines and to induce tumor lysis.

Blood and tissue samples were analyzed from 4 patients (09-079-1, 09-079-7, 09-079-10, and 09-079-17) treated on a clinical trial at Memorial Sloan-Kettering Cancer Center (MSKCC) evaluating the pharmacokinetics of two different biosynthetic formulations of ipilimumab (CA184-087, NCT00920907). All patients received 4 doses of antibody at a dose of 10 mg/kg i.v. administered every 3 weeks for 4 doses during induction therapy. Patients without dose-limiting toxicity and with evidence of clinical benefit (in this case, 09-079-1, 09-079-10, and 09-079-17) then received maintenance ipilimumab at the same dose every 12 weeks starting at week 24. Responses were adjudicated by the recently proposed immune-related response criteria (15). Toxicity was assessed using National Cancer Institute Common Terminology Criteria for Adverse Events, version 3.0. All patients provided informed consent for the clinical studies and additional consent for the collection of blood and tumor tissue for investigational purposes on a separate MSKCC biospecimen utilization protocol. All studies were approved by the MSKCC Institutional Review Board.

NY-ESO-1 overlapping peptides (17 peptides with ∼20-mer length and 10 aa overlap) (16) and NY-ESO-1_92–100 peptide (LAMPFATPM), NY-ESO-1_94–102 peptide (MPFATPMEA), NY-ESO-1_94–104 peptide (MPFATPMEAEL), NY-ESO-1_96–104 peptide (FATPMEAEL), and NY-ESO-1_157–165 peptide (SLLMWITQC) were purchased from JPT Peptide Technologies. Peptides were dissolved in dimethyl sulfoxide at a concentration of 1 mg/mL and stored in aliquots at −80 °C before use. The following autologous or MHC-matched melanoma cell lines were used as target cells: SK-MEL-381 (from patient 09-079-7) and SK-MEL-351 (from patient 09-079-10, NY-ESO-1 negative). Autologous B-lymphoblastoid cell lines (LCL) were generated in our laboratory from the patients' PBMCs, using EBV-containing supernatants and also used as target cells.

Phytohemagglutinin (PHA)-stimulated CD4⁺ T cells (T-antigen presenting cells or T-APCs) were prepared as described previously (17–19). CD4⁺ T cells were separated from PBMCs using Dynabeads (Invitrogen) according to the manufacturer's instruction and seeded into 48-well plates (NUNC) at a density of 1–2 × 10⁶ cells per well in 1 mL RPMI-1640 medium supplemented with 25 μmol/L HEPES, 10% heat-inactivated human AB serum (Gemini Bio-Products), 2 μmol/L l-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin (complete medium). A total of 10 μg/mL of PHA (REMEL) was added to the culture at day 0. Half of the medium was replaced with complete medium containing interleukin (IL)-2 (20 IU/mL, Roche) and IL-7 (40 ng/mL; R&D systems) at day 3, which was then repeated twice weekly. The activated CD4⁺ T-APCs were pulsed with peptides and used as target cells in various assays.

In vitro sensitization was conducted as described previously (17–20). After separating CD8⁺ T cells or CD4⁺ T cells from PBMCs using Dynabeads, CD4⁺ T cells (5 × 10⁵ cells/well) were cultured with antigen peptide-pulsed and irradiated (30 Gy) autologous CD4⁻ CD8⁻ PBMCs (5 × 10⁵ cells/well) in 96 round bottom-well plates in complete medium. From day 4 onward, half of the medium was replaced with complete medium containing IL-2 and IL-7 twice a week. These CD4⁺ T cells in the culture wells were used as effector cells in various assays on day 20.

NY-ESO-1–specific CD4⁺ T cells were isolated by CD154 (CD40L) expression sorting as described with some modification (21, 22). Presensitized CD4⁺ T cells were restimulated for 6 hours in 500 μL X-VIVO15 (BioWhittaker-Lonza) with T-APCs that were pulsed overnight with NY-ESO-1 peptide (1 μmol/L) and labeled with carboxyfluorescein succinimidyl ester (CFSE; Invitrogen-Molecular Probes) in the presence of 20 μL of phycoerythrin (PE)-conjugated anti-CD154 mAb and 0.3 μL GolgiStop (BD Biosciences). CFSE⁻CD154⁺ NY-ESO-1–specific effector T cells were sorted using a FACSAria instrument and FACS Diva software (BD Biosciences). Sorted cells were stimulated with plate-bound anti-CD3 (0.5 μg/mL) and CD28 mAbs (0.5 μg/mL; eBioscience) in the presence of irradiated allogeneic PBMCs and were expanded for about 20 days in the presence of 20 IU/mL of IL-2 and 20 ng/ml of IL-7 in complete medium as described with some modification (23). Reactivity of the CD4⁺ T-cell lines was tested against autologous melanoma cell lines or autologous LCL pulsed with peptides.

The IFN-γ ELISPOT assay was conducted as described previously (18–20). Briefly, 96-well nitrocellulose ELISPOT plates (MAHA S4510; Millipore) were coated overnight at 4°C with 2 μg/mL anti-human IFN-γ mAb (1-D1K) and blocked with 10% human AB serum-containing RPMI-1640 for 2 hours at 37°C. A total of 2 × 10⁴ sensitized CD4⁺ T cells and 2 × 10⁴ peptide-pulsed T-APCs were placed in each well of the ELISPOT plate at a final volume of 200 μL RPMI-1640 medium without serum. After incubation for 22 hours at 37°C in a CO₂ incubator, the plate was developed using 0.2 μg/mL biotinylated anti-human IFN-γ mAb (7-B6-1, Mabtech), 1 μg/mL streptavidin-alkaline phosphatase conjugate (Roche Diagnostics), and 5-bromo-4-chloro-3-indolyl phosphate/NBT (Sigma-Aldrich). The number of spots was evaluated using a C.T.L. Immunospot analyzer and software (Cellular Technology). Results were shown as the number of spot-forming cells (SFC) without subtracting the number of background spots, as the number of SFC in negative control was less than 3 spots per well in all assays. A positive response with more than 10 spot counts per well as well as spot counts ≥3-fold more than background spots obtained with nonpulsed target cells was considered to be significant.

For surface staining, APC-Cy7-CD8, PE-Cy7-CD3, Pacific Blue-CD3 (BD Bioscience), ECD-CD4, ECD-CD45RA, ECD-CD45RO (Beckman Coulter Inc.), and FITC-CCR7 (R&D Systems) were used. For ICS, in vitro stimulated and cultured CD4⁺ T cells were harvested as effector cells. A total of 2 × 10⁵ effector CD4⁺ T cells were cocultured for 6 hours in 500 μL 10% human AB serum RPMI in the presence of 0.35 μL GolgiStop and with PE-Cy5-CD107a (20 μL/mL, BD Bioscience) with 2 × 10⁵ autologous T-APCs that had been pulsed overnight with each antigen peptide pool and labeled with CFSE. Cytoplasmic cytokines were stained using a BD Cytofix/Cytoperm kit (BD Biosciences) according to manufacturer's instructions with FITC, Alexa Fluor 647, PE-Cy7, or Horizon-V450-IFN-γ, PE-Macrophage inflammatory protein (MIP)-1β, APC-IL-2, PE or FITC-IL-4, APC-IL-5, Horizon-V450 or APC-IL-13, Alexa Fluor 700-granzyme B, PerCP-Cy5.5-T-bet (BD Bioscience), PE-Cy7-TNF-α, PE-IL-10, Alexa Fluor 647-IL-17A, Alexa Fluor 647-Eomesodermin (Eomes; eBioscience), PE-perforin (Cell sciences).

CFSE-negative effector CD4⁺ T cells were analyzed using a CyAn flow cytometer with Summit software (Dako Cytomation Inc.) or a LSR Fortessa with FACS Diva software (BD Biosciences). All analyses were carried out using FlowJo software (TreeStar, Inc.). Results were shown as a frequency (%) of antigen-specific cytokine responses and CD107a mobilization (functions⁺) in CD4⁺ T cells after subtracting the frequency of background functionpositive cells. A positive response ≥3-fold more than the negative control obtained with nonpulsed target cells (% function-positive cells/sample in all assays) was considered to be significant. All background intracellular responses (function-positive) were <0.1%, except for the transcription factors T-bet and Eomes, which were <0.5%.

A cytotoxicity assay was conducted as described previously with some modifications (12). To determine the in vitro killing of tumor targets, target cells were labeled with 0.5 μmol/L CFSE, whereas control target cells were labeled with 5 μmol/L CFSE. NY-ESO-1–specific CD4⁺ T cell lines were cocultured at different ratios (24) with a 50:50 mixture of target cells and control target cells for 6 hours. Then, cells were acquired on a CyAn flow cytometer with Summit software and analyzed by FlowJo software. A total of 20 μg/mL of anti-human HLA-DR, DP, or DQ antibody (BD Bioscience) and mouse IgG2a, κ antibody (BD Bioscience) as a control were used for blocking of MHC class II on target cells. Cytotoxicity was calculated using the following formula: cytotoxicity (%) = [1 − (%live cell_{target cells}/%live cell_{control target cells})] × 100.

Data were analyzed using Prism 5.0 (GraphPad Software, Inc.). All the experiments were repeated two or three times. Statistical significance was determined by a Student t test. P < 0.05 was considered a statistically significant difference.

Patient profiles and clinical demographics are summarized in Table 1.

Patient 09-079-1 is a 63-year-old man with metastatic melanoma to soft tissue, liver, lymph nodes, bone, and left adrenal gland. He had experienced progressive disease (PD) on multiple chemotherapy regimens. In August 2009, he initiated ipilimumab therapy. After completing induction therapy, imaging at week 12 revealed PD but repeat imaging at week 24 showed a partial response (PR), although the adrenal lesion was noted to have increased in size. He initiated maintenance ipilimumab and subsequently underwent resection of the adrenal metastasis during week 27, as a presumed immunologic “escape” lesion. Since that time, he has maintained a durable PR.

Patient 09-079-7 is a 49-year-old man who was diagnosed with metastatic melanoma to the lung, lymph nodes, pancreas, retroperitoneum, bone, soft tissue, thoracic nodes, and spleen in April 2009. He received first-line therapy with ipilimumab. Imaging after induction therapy revealed PD. Further imaging at week 16 indicated further significant PD. He was taken off the study protocol, transitioned to supportive care, and died in March 2010.

Patient 09-079-10 is a 39-year-old woman with stage IV melanoma to the lung, liver, lymph nodes, spleen, and bone, with prior PD on a cisplatin/vinblastine/temozolomide. She initiated ipilimumab therapy in September 2009. Imaging after completing induction therapy revealed PD at week 12 but stable disease (SD) at week 24. She then received maintenance ipilimumab, with her last treatment in July 2011. She continues to have SD.

Patient 09-079-17 is a 54-year-old woman with recurrent melanoma to the back, abdomen, and pelvis. This patient has a history of melanoma dating back to 1990, when she had a 2.4-mm melanoma with deep dermal invasion removed from her left lower back. She was disease free for approximately 15 years. She was noted to have a recurrence in 2005 and underwent multiple resections, including dermal, axillary node, and mesenteric masses. She previously experienced progression of disease on temozolomide, a gp100 DNA vaccine, talimogene laherparepvec (recombinant attenuated herpes simplex virus expressing GM-CSF) and high-dose IL-2. She then initiated protocol therapy with ipilimumab in October 2009. Computed tomographic imaging revealed PD at week 12 and SD at week 24. She received maintenance ipilimumab, and approximately 2.5 years into this treatment, she continues to have SD.

Sera collected from these 4 patients were analyzed for NY-ESO-1 antibody responses. All four patients were seropositive at baseline (Patients 09-079-7, -10 and -17) or following ipilimumab treatment (Patient 09-079-1). NY-ESO-1 antibody titers increased in all 4 patients, especially in patient 09-079-1, with subsequent ipilimumab treatments (Supplementary Table S1) as we have reported previously (10). We confirmed baseline NY-ESO-1 antigen expression in tumor tissue from all 4 patients by immunohistochemistry and reverse-transcriptase PCR.

Purified CD4⁺ T cells isolated from PBMCs were stimulated in vitro with NY-ESO-1 overlapping peptides and cultured for 20 days. Reactivity against NY-ESO-1 overlapping peptides and the individual NY-ESO-1 peptides was then determined for these cultured CD4⁺ T cells by ICS and IFN-γ ELISPOT assay. In the 3 patients, who were seropositive at baseline (Patients 09-079-7, -10 and -17), pretreatment NY-ESO-1–specific CD4⁺ T-cell responses were detected (TNF-α in patient 09-079-7; IFN-γ and TNF-α in patient 09-079-10; IFN-γ, TNF-α, IL-2, MIP-1β, and CD107a in patient 09-079-17). In the patient seronegative at baseline (09-079-1), baseline NY-ESO-1–specific CD4⁺ T-cell responses were not detected. After ipilimumab treatment, NY-ESO-1–specific CD4⁺ T-cell responses (IFN-γ, TNF-α, and CD107a with/without IL-2) were detected in all 4 patients. These cells did not secrete T_H2 cytokines (IL-4, IL-5, or IL-13) or T_H17 cytokine (IL-17A) by ICS, suggesting that that they are T_H1 cells. (Fig. 1A)

Because of the low frequency of NY-ESO-1 antigen-specific CD4⁺ T cells in peripheral blood in these 4 patients, we did not detect NY-ESO-1 antigen-specific CD4⁺ T-cell response using ex vivo ICS staining. After 20 days of in vitro cell culture, we were able to obtain approximately 5 × 10⁶ CD4⁺ T cells with an approximately 1 log expansion of CD4⁺ T cells from the initial 5 × 10⁵ CD4⁺ T cells per well. The yield of detectable NY-ESO-1 peptide-specific CD4⁺ T cell ranged from 2,500 to 92,250 per well. Using these parameters, a peptide-specific CD4⁺ T-cell response with 10 IFN-γ spots per 20,000 effector cells in an ELISPOT assay after in vitro stimulation is estimated to yield 2,500 NY-ESO-1–specific CD4⁺ T cells after one 20-day in vitro culture. We studied the breadth and frequency of each peptide-specific T-cell response by IFN-γ ELISPOT. Baseline spontaneous CD4⁺ T-cell responses to two individual peptides (10–19 spots, 2,500–4,750 cells/well after stimulation) were detected in patient 09-079-10 and to 6 peptides: 2 peptides, >60 spots (>15,000 cells/well after stimulation); 4 peptides, 20–59 spots (5,000–14,750 cells/well after stimulation) in patient 09-079-17. Following ipilimumab therapy, a broader spectrum of CD4⁺ T-cell peptide responses was observed with new peptides recognized by 3 patients (all except 09-079-17). The number of peptide-specific responses did not change following ipilimumab therapy (6 peptides) in patient 09-079-17; however, 5 (p81–100, p101–120, p119–143, p131–150, p151–170) of 6 individual T-cell peptides (which were positive at baseline) were increased significantly in IFN-γ spot number (P < 0.05; Fig. 1B).

After ipilimumab treatment, the majority of peptides recognized by CD4⁺ T cells included sequences located in four immunodominant distinct regions of the protein, corresponding to peptides NY-ESO-1_41–60 (2/4 patients), NY-ESO-1_81–100 (4/4 patients), NY-ESO-1_119–143 (3/4 patients), and NY-ESO-1_151–170 (2/4 patients) as previously reported in patients with spontaneous responses as well as those immunized with NY-ESO-1 vaccines (refs. 25–29; Fig. 1B). In summary, polyfunctional NY-ESO-1 CD4⁺ T-cell responses were detected at baseline or posttherapy samples in all 4 patients.

We sought to determine whether the NY-ESO-1–specific CD4⁺ T cells detected after ipilimumab treatment had high-affinity T-cell receptors capable of recognizing naturally processed tumor antigen. NY-ESO-1 single peptide-specific CD4⁺ T-cell lines were generated from PBMCs from patients 09-079-1 and -17 before and after the treatment at weeks 7 and 24 or from patients 09-079-7 and -10 at weeks 12 and 24 by a CD154 (CD40L) sorting method (Supplementary Fig. S1A) after stimulation with individual, dominant NY-ESO-1 peptides. We confirmed that there were no contaminating CD8⁺ T cells in CD40L-sorted NY-ESO-1–specific CD4⁺ T cells (Supplementary Fig. S1B). These cells were then cocultured with autologous melanoma cell lines and autologous APC (EBV- infected B cell lines; LCLs) with each corresponding peptide or MAGE-A4 overlapping peptides as the negative control. The NY-ESO-1–specific CD4⁺ T-cell lines established from 4 patients' expressed intracellular IFN-γ, granzyme B, and surface CD107a, as compared with the CD4⁺ T-cell response to the negative control (Fig. 2A and B).

All NY-ESO-1–specific CD4⁺ T-cell lines before and after the treatment were of high avidity to recognize 10 to 100 nmol/L peptide (data not shown). Furthermore, they recognized recombinant NY-ESO-1 protein naturally processed in autologous APCs (LCLs) by ELISPOT assay (Fig. 2C) and lysed autologous LCL pulsed with corresponding NY-ESO-1 protein (100 nmol/L; Fig. 2D) using the cytotoxicity assay described above. NY-ESO-1–specific CD4⁺ T-cell lines lysed the autologous LCL pulsed with each cognate peptides (100 nmol/L) at 5:1 and 25:1 effector:target ratio (Fig. 2E). Among the autologous melanoma cell lines derived from these patients, SK-MEL-381 (from patient 09-079-7) expresses both NY-ESO-1 and MHC class II. The NY-ESO-1_81–100–specific CD4⁺ T-cell line from patient 09-079-7 lysed the autologous tumor cell line (SK-MEL-381), and cytotoxicity was inhibited by MHC class II antibody blocking of the target cells (Fig. 2F). The NY-ESO-1–specific CD4⁺ T-cell lines do not coexpress CD8 markers before and after restimulation in the ICS assay (Supplementary Fig. S1C).

From patient 09-079-17, we were able to establish NY-ESO-1_131–150 peptide-specific CD4⁺ T-cell lines at the pretreatment time point as well as at weeks 7 and 24 by the CD154 (CD40L) sorting method and subsequently analyzed serial changes in cytotoxic function. These cell lines recognized autologous APCs (LCLs) pulsed with NY-ESO-1_131–150 peptide (100 nmol/L) in the ICS assay. In contrast with IFN-γ, which was consistently produced before and after the therapy, accumulation of granzyme B and perforin was observed only after ipilimumab treatment (Fig. 3A–C). In addition, in vitro peptide restimulation induced granzyme B synthesis only in CD4⁺ T cells after the therapy (Fig. 3C). In accordance with the expression of cytotoxic molecules, higher expression of the transcription factor Eomes previously reported as the master regulator of cytotoxic CD8⁺ T cell (30) was observed after ipilimumab treatment at weeks 7 and 24 when compared with expression levels before treatment (Fig. 3D; *, P < 0.05). In contrast, expression of T-bet was not increased (Fig. 3D). Expression of PD-1 became lower at week 24 after treatment when compared with pretreatment and at week 7 (Fig. 3D; *, P < 0.05).

We next assessed whether the ipilimumab-related changes in expression of lytic markers described above in the NY-ESO-1_131–150–specific CD4⁺ T-cell lines from patient 09-079-17 affected their ability to lyse autologous LCL pulsed with NY-ESO-1_131–150 peptides (100 nmol/L) at 5:1 and 25:1 E/T ratio. The degree of cytotoxicity (%) was significantly increased after ipilimumab treatment (peak at week 7), when compared with that of cells obtained pretreatment (*, P < 0.05). This cytotoxicity was inhibited with the use of an MHC class II blocking antibody on target cells (Fig. 4).

In this study, we conducted in-depth immune monitoring on four NY-ESO-1–seropositive melanoma patients treated with ipilimumab, based upon the availability of suitable PBMCs. We found that NY-ESO-1 antigen-specific T_H1 responses against multiple epitopes were induced or expanded after ipilimumab treatment in all 4 patients. NY-ESO-1 antigen-specific CD4⁺ T-cell lines established from all 4 patients were of high avidity and recognized naturally processed NY-ESO-1 protein in APCs, as indicated by the production of perforin, granzyme B, and the upregulation of CD107a. Finally, we showed that one of these NY-ESO-1 antigen-specific CD4⁺ T-cell lines directly lysed autologous melanoma cells in an MHC class II-restricted manner. We believe this is the first demonstration of the induction or enhancement of tumor-reactive cytotoxic CD4⁺ T cells after ipilimumab treatment and is therefore a novel property of CTLA-4 blockade.

Induction of antigen-specific T cells against NY-ESO-1 has previously been accomplished in many ways. NY-ESO-1–specific CD8⁺ T cells could be induced in HLA-A2⁺ patients with cancer vaccinated with NY-ESO-1 peptides p157–165/p157–167 (31). These T cells were highly reactive with the peptides used for vaccination, but only rarely recognized HLA-matched, NY-ESO-1–expressing tumor cell lines. Vaccine-induced antigen-specific T cells were heterogeneous in functional activity, especially in terms of natural tumor recognition. The frequency of antigen-specific T cells does not always equate with functional tumor reactivity. Therefore, precise and multiparametric immune monitoring assays are critical to identify the proportion of tumor-reactive T cells within the population of vaccine-induced antigen-specific effector cells. An NY-ESO-1 helper peptide vaccine has been reported to induce NY-ESO-1 antigen-specific CD4⁺ T cells with lower avidity (>1 μmol/L) compared with NY-ESO-1–specific preexisting naïve CD4⁺CD25⁻ T-cell precursors or spontaneously induced CD4⁺ T-cell effectors in NY-ESO-1–seropositive cancer patients. These cells were only able to recognize NY-ESO-1 helper peptide, but not naturally processed NY-ESO-1 protein in APCs from patients with NY-ESO-1–expressing epithelial ovarian cancer (32). Our analysis showed that most of the NY-ESO-1–specific CD4⁺ T-cell lines generated after ipilimumab treatment were of high avidity and recognized naturally processed NY-ESO-1 protein in APCs in specimens of the 4 patients analyzed, distinguishing these results from those of exogenous vaccination strategies. Our observation in this study was based upon the results collected from either in vitro culture of CD4⁺ T cells or CD40L sorted NY-ESO-1–specific CD4⁺ T-cell lines. Because of the limited PBMC availability and low frequency of NY-ESO-1 antigen-specific CD4⁺ T cells in peripheral blood from these patients, we were not able to characterize directly these NY-ESO-1 antigen-specific CD4⁺ T cells in ex vivo assays.

The optimal T-cell receptor (TCR) affinity threshold for natural tumor recognition and maximal antitumor T-cell response has been measured using human CD8⁺ T cells transduced with TCR variants (33). Thus, the relative lack of high affinity/avidity TCRs for CD8⁺ T cells is considered to be one reason why immune responses toward self-tumor antigens have not been protective. It is possible that the cytotoxic capacity of CD4⁺ T cells might be subject to an optimal TCR affinity threshold as well. In this study, NY-ESO-1–specific CD4⁺ T-cell lines were directly sorted from bulk CD4⁺ T-cell cultures after one round of in vitro presensitization. Therefore, these CD4⁺ T-cell lines, not “clones,” reflect a spectrum of the NY-ESO-1–specific CD4⁺ T-cell population in the immune repertoire induced after ipilimumab treatment. The clinical activity of CTLA-4 blockade may mechanistically involve expansion of naturally occurring antigen-specific T cells (NY-ESO-1 cells, as an example) with high avidity. The ability to recognize antigenic epitopes naturally processed endogenously in tumor cells and to lyse tumors are logical characteristics to expect from an effective T-cell response.

Cytotoxic effector cells kill tumor cells through FAS, TRAIL, or granzyme-perforin-dependent mechanisms (34). The T-box transcription factor Eomes is critical for inducing the expression of granzyme B and other lytic granules (30). Dual costimulation of OX40 plus 4-1BB was shown to induce Eomes in HA-specific CD4⁺ T cells toward cytotoxic T_H1 differentiation (35). Recently our group showed that OX40 agonist antibody in combination with cyclophosphamide treatment and adoptive transfer of tumor‐specific CD4⁺ T cells could produce cytotoxic CD4⁺ T cells through Eomes and T-bet (36). We analyzed Eomes and T-bet by flow cytometric staining and found higher expression of Eomes on NY-ESO-1 antigen-specific CD4⁺ T cells after ipilimumab treatment; however, no change was noted for T-bet. Higher secretion of IFN-γ, granzyme B, and perforin were detected consistently after ipilimumab treatment at weeks 7 and 24 when compared with those present pretreatment. These findings suggest that CTLA-4 blockade may induce the expression of lytic granules on NY-ESO-1–specific cytotoxic CD4⁺ T cells via Eomes, but further detailed mechanistic studies are needed to conclusively show this.

In summary, previous studies from several groups have shown the following important points: (i) CTLA-4 blockade enhances regression of spontaneous mouse tumors by expansion of transferred effector CD4⁺ T cells (12); (ii) adoptive transfer of autologous CD4⁺ T cells directed against NY-ESO-1 has resulted in tumor regression in humans (14); and (iii) ipilimumab promotes the expansion of infused CD8⁺ CTLs as memory CTLs (37). Our data now bring these observations together by revealing that CTLA-4 blockade with ipilimumab induces the expression of Eomes as well as markers of lytic degranulation on high-affinity cytotoxic-CD4⁺ T cells. These cytotoxic NY-ESO-1–specific CD4⁺ T cells are able to lyse autologous tumor cells, which naturally express the cognate NY-ESO-1 antigen. It brings forth a potentially novel therapeutic mechanism in which CTLA-4 blockade or inhibition with other immunomodulatory antibodies might be combined with CD4⁺ T cells to enhance cytolytic function in cancer immunotherapy clinical trials.

J.P. Allison has ownership interest (including patents) and J.D. Wolchok has a commercial research grant and is a consultant/advisory board member for Bristol-Myers Squibb. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Kitano, S. Gnjatic, J. Yuan, J.D. Wolchok

Development of methodology: S. Kitano, T. Tsuji, C. Liu, Z. Mu, S. Gnjatic, J. Yuan, J.D. Wolchok

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Kitano, T. Tsuji, C. Liu, Z. Mu, J.D. Wolchok

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Kitano, C. Liu, D. Hirschhorn-Cymerman, C. Kyi, J.P. Allison, S. Gnjatic, J. Yuan, J.D. Wolchok

Writing, review, and/or revision of the manuscript: S. Kitano, C. Liu, D. Hirschhorn-Cymerman, C. Kyi, S. Gnjatic, J. Yuan, J.D. Wolchok

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Gnjatic

Study supervision: J. Yuan, J.D. Wolchok

The authors thank Teresa Rasalan and Sapna Tandon for their excellent technical assistance. This article is dedicated to the memory of Dr. Lloyd Old, our mentor and friend and one of the true leaders in investigational cancer medicine. His contributions to tumor immunology inspired much of this work.

This work was supported by grants from the NIH (RC2CA148468; to J.D. Wolchok), the Melanoma Research Alliance (to S. Gnjatic and J.D. Wolchok), Swim Across America (to J.D. Wolchok), the Cancer Research Institute (to S. Gnjatic and J.D. Wolchok), the Virginia and D.K. Ludwig Fund for Cancer Research (to J.P. Allison), the Lita Annenberg Hazen Foundation (to J.D. Wolchok), and the Commonwealth Foundation for Cancer Research (to J.D. Wolchok).

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.