c-Met is a novel tumor associated antigen for T-cell based immunotherapy against NK/T cell lymphoma

Background: The expression of c-Met and its ligand HGF plays a critical role in cell proliferation and is involved in numerous malignancies. Because c-Met expression and its role in NK/T-cell lymphoma remain unclear, we studied the expression and function of c-Met in NK/T-cell lymphoma cells. In addition, we investigated the possibility that c-Met could function as a tumor-associated antigen for helper T lymphocytes (HTLs). Methods: We evaluated whether HGF and c-Met were expressed in NK/T-cell lymphoma and the capacity of predicted c-Met HTL epitopes to induce antitumor responses in vitro. In addition, c-Met inhibitor was evaluated for the ability to inhibit TGF-β production in tumor and subsequently increase HTL recognition. Results: c-Met and HGF were expressed in NK/T-cell lymphoma cell lines, nasal NK/T-cell lymphoma specimens and patient serum samples. Moreover, HGF was shown to promote NK/T cell lymphoma (NKTCL) proliferation in an autocrine manner. Furthermore, we have identified three novel c-Met HTL epitopes that were restricted by several HLA-DR molecules. Notably, peptide-induced HTL lines directly recognized and killed c-Met expressing NK/T-cell lymphomas and various epithelial solid tumors. The c-Met specific HTLs could also recognize dendritic cells (DCs) pulsed with c-Met expressing tumor cell lysates. In addition, we observed that c-Met inhibition augmented HTL recognition by decreasing TGF-β production by tumor cells. Lastly, autophagy partly regulated the HTL responses against tumors. Conclusions: We identified novel c-Met HTL epitopes that can elicit effective antitumor responses against tumors expressing c-Met. Our results provide the rationale of combining c-Met targeting therapy and immunotherapy for NKTCLs and epithelial tumors.


Introduction
The contribution of Epstein-Barr virus (EBV) infection to oncogenesis has been widely accepted in gastric cancer and some other malignancies. 1 Among them, NKTCL is one of the EBV related hematological malignancies, which exhibits an aggressive behavior. 2 The most common treatment for NKTCL is chemoradiation therapy and its efficacy is somewhat unsatisfactory which is accompanied with severe adverse events. 3 Thus, there is an urgent need for novel NKTCL treatments that can improve survival with fewer side effects.
It has been reported that many hematological malignancies and solid tumors including head and neck squamous cell carcinoma (HNSCC) can utilize c-Met as signaling receptor for hepatocyte growth factor (HGF). 4,5 Both tumor cells and the tumor-surrounding stroma can produce HGF to enhance tumor proliferation and induce epithelial-mesenchymal transition, which is a key event in tumor metastasis. 6 c-Met can be considered an oncogenic protein that functions as tumor associated antigen (TAA) for CD8 C cytotoxic T cells. 7 Because CD4 C HTLs are important for the generation of effective antitumor immunity, 8 it may be advantageous to assess whether HTLs are capable of recognizing tumor-derived c-Met epitopes. Thus, in this study, we evaluated whether c-Met can function as a TAA for CD4 C T-cells against NKTCL. In addition, we report that HGF/c-Met signaling contributes to NKTCL cell proliferation in an autocrine manner, and that c-Met reactive CD4 C T cells can exhibit cytotoxic activity against NKTCL and epithelial solid tumors. Moreover, we observed that c-Met plays a key role in the production of TGF-b by NKTCL cells suggesting that c-Met inhibitors can be used to improve the efficacy of CD4 C T cell immunotherapy. Lastly, we report that autophagy was involved in regulating of MHC class II antigen processing and presentation in NKTCL cells, and observed significant enhancement of HTL responses by the induction of autophagy.

c-Met and HGF expression by NKTCL
Although expression of c-Met has been reported in epithelial solid tumors including HNSCC, 4,9 its expression in NKTCL remains unknown. Thus, we first evaluated c-Met protein expression in an EBV-infected NK cell line (KAI3) and in NKTCL cell lines (SNK6, SNT8). As shown in supplementary Fig. 1A, NKTCL cells expressed a substantial amount of c-Met protein.
As with previous reports, c-Met protein was found expressed in HNSCC cells (HPC92Y, HSC2, HSC4, and Sa-3) and lung cancer cell (Calu-1). c-Met expression in six clinical samples of NKTCL were also confirmed by using immunohistochemistry (example of one sample is shown in supplementary Fig. 1B. left  panel). HGF functions as a ligand for c-Met, activating its signaling pathway. As shown in the example in Supplementary Fig. 1B (right panel), the expression of HGF protein was evident in the cytoplasm of a clinical NKTCL sample. In addition, we observed that 2 out of 3 NKTCL cell lines produced HGF (Supplementary Fig. 1C). Next, we examined serum HGF levels in NKTCL patients and found that these levels were significantly higher in NKTCL patients as compared to healthy donors ( Supplementary  Fig. 1D). These data suggest that HGF can be produced by NKTCL (or tumor surrounding cells) and that it may trigger c-Met signaling events affecting the tumor.

HGF/c-Met signaling mediates NKTCL proliferation
To examine the function of HGF/ c-Met signaling in NKTCL, we evaluated whether c-Met inhibition could affect the tumor cell proliferation. As shown in Fig. 1A, while the addition of anti-c-Met antibody had no significant effect on the cell proliferation for the three lines tested (KAI3, SNK6, and SNT8) the tyrosine kinase inhibitor (TKI) ARQ197 reduced the proliferative response by approximately 50% in 2 of the 3 cell lines. On the other hand, when exogenous HGF was added to the cultures, the level of proliferation was increased in all three cell lines and both the anti-c-Met antibody and the TKI were able to suppress this enhancement (Fig. 1B). These results suggest that HGF/c-Met signaling is partly responsible for NKTCL proliferation.

c-Met functions as a TAA for HTLs
Previously, we have reported that predicted MHC-II binding peptides from various TAAs can elicit CD4 C HTL responses displaying effective antitumor immunity. 10 Because HTL epitopes from c-Met have not yet been described, we utilized computer peptide-MHC-II predictive algorithms to assist in the identification of such epitopes (see Materials & Methods). According to the prediction scores, which represent the theoretical binding affinity of peptides to diverse, common MHC-II alleles, three peptide sequences, c-Met 278-292 , c-Met 817-831 and c-Met 1244-1258 were selected as candidate HTL epitopes and used for further experiments. We then evaluated the capacity of synthetic peptides containing these sequences to stimulate HTL responses using purified CD4 C T cells obtained from two HLA-typed healthy blood donors (Donor 1: HLA-DR 9/12, DR53 and Donor 2: HLA-DR 9/13, DR53). After four cycles of peptide stimulation, these peptides generated several c-Met antigen specific HTL lines that responded with the stimulating peptide in a dose responding manner ( Fig. 2A). From Donor 1, six HTL lines were isolated, two of them responding to peptide c-Met 278-292 and four with c-Met 817-831 . On the other hand, three HTL lines were isolated from Donor 2 and all of them responded to peptide c-Met 1244-1258 . Representative HTL lines reactive with the three epitopes were selected for further studies. Next, we proceeded to study the MHC class II restriction elements (HLA-DR, -DP or -DQ) of the HTL lines toward their corresponding peptide epitopes. We first determined that the selected HTL lines recognized their corresponding peptides in the context of HLA-DR molecules since treatment with a specific anti-HLA-DR antibody (that does not cross react with the HLA-DP and -DQ molecules) inhibited all the HTL responses while a control antibody reactive with HLA class I molecules did not inhibit the HTL responses (Fig. 2B). Next, using a panel of gene-transduced mouse fibroblasts (Lcells) expressing HLA-DR9 or -DR53 and an LCL expressing HLA-DR12 as APCs, we determined that c-Met 278-292 reactive HTL line 1-B was restricted by HLA-DR9 (Fig. 2C).  Table S2).
We noted that one of the three c-Met peptides that was selected for the present studies, c-Met 1244-1258 bears considerable homology to peptides sequences found in EGFR, HER-2, HER-3 and FAK (Table 1). Thus, two c-Met 1244-1258 reactive HTL lines (2-A and 2-C) were evaluated for their responses against the analog peptides. As shown in Fig. 2D, both HTL lines effectively recognized the analog peptides even though they differed on 6 out of the 15 residues.

Direct recognition of NKTCL and epithelial solid tumors by c-Met reactive HTLs
Next we evaluated whether the c-Met reactive HTLs would be effective in directly recognizing c-Met positive, HLA-DR expressing NKTCLs and epithelial solid tumor cells. As shown in Fig. 3, c-Met reactive HTLs produced IFNg when stimulated with c-Met positive NKTCL and epithelial solid tumor cell lines that expressed the corresponding restricting HLA-DR alleles and these responses were inhibited by the addition of anti-HLA-DR antibody. Moreover, c-Met negative tumors or tumors expressing irrelevant HLA-DR alleles did not stimulate the HTL responses. Direct recognition of c-Met expressing tumors was also evaluated in cytotoxicity assays. The data presented in Fig. 4 demonstrates that c-Met reactive HTLs were very effective in killing HLA-DR matched tumor cells expressing c-Met in an antigen-specific and a dose-dependent manner. HTLs can also recognize tumorderived antigens presented by APCs such as DCs that take up dead or apoptotic tumor cells. Therefore, c-Met reactive HTLs were also assessed for their capacity to recognize autologous DCs that were pulsed with c-Met expressing tumor lysates. The results shown in Fig. 5 demonstrate that the four HTL lines (1-B, 1-D, Table 1. c-Met helper epitopes and analogous c-Met epitopes. c-Met helper epitopes were predicted by computer based epitope prediction analysis. Analogous epitopes from EGFR, HER-2, HER-3 or FAK were identified by comparing the protein sequences. Underlined bold letters indicate amino acids that are different from c-Met helper epitope peptides. No analogous epitopes to c-Met 278-292 or c-Met 817-831 were detected in the EGFR, HER-2, HER-3, and FAK protein sequences.  1-F, and 2-A), were stimulated by DCs that were pulsed with the corresponding synthetic peptide or c-Met positive tumor lysates and that recognition of tumor lysate pulsed DCs was blocked by the addition of anti-HLA-DR antibody. The overall results demonstrate that the newly described c-Met HTL epitopes can be presented directly on tumor cells or through exogenous antigen-processing machinery of DCs in an MHC class II restricted manner.

c-Met inhibition suppresses tumor TGF-b production, enhancing HTL responses
Tumor derived soluble factors such as TGF-b inhibit the antitumor activity of innate and adaptive immune cells and serve as a mechanism of tumor immune evasion. 11 As shown in Fig. 6A, the HTL response to peptidepulsed APCs was reduced by approximately 50% by the addition of TGF-b and similar inhibitory activity was evident by the addition of tissue culture from KAI3 NKTCL cells. Next, we examined the production of TGF-b by several NKTCL cells and observed a substantial production of this immune inhibitory cytokine by 2 of the 4 tumor cell lines (Fig. 6B). Interestingly, the production of TGF-b was significantly reduced by the addition of the c-Met inhibitor ARQ197. In view of these findings, we next evaluated whether the addition of ARQ197 would increase HTL responses to the TGF-b producing NKTCL cells (KAI3 and SNK6). As shown in Fig. 6C, the c-Met inhibitor ARQ197 increased the capacity of HTLs (lines 1-B and 1-F) to recognize the NKTCL cells that produce TGF-b. Similar enhancement of the HTL response was observed by the addition of anti-TGF-b mAb. As predicted, the c-Met inhibitor or the anti-TGF-b mAb did not improve HTL responses against the c-Met positive tumors (SNT8 and HPC92Y) that do not produce TGF-b.

Autophagy mediators influence helper T cell recognition of tumors
Since autophagy (chaperon-dependent or -independent) is known to play a role in tumor survival, autophagy inhibitors have been tested as therapeutics for various malignancies. 12,13 Autophagy is also involved in MHC class I tumor antigen crosspresentation, 14 but the effect of autophagy inhibitors in HTL tumor recognition has not been examined. Thus, we determined whether pretreatment of tumor cells with the chaperon-independent autophagy inhibitor 3-methyladenine (3MA), the chaperon-independent autophagy enhancer valproic acid (VPA) or chaperon-dependent autophagy inhibitor 17-DMAG had any effect in the recognition of the tumor cells by c-Met specific HTLs. Because autophagy mediators inhibit proliferation in NKTCL cells and lead to cell death, 15,16 we used the HNSCC cells (HPC92Y and Sa-3) for these assays. Expression of the LC-3 II (as an autophagy marker) and HSP70 (as an adverse surrogate marker of 17-DMAG), were used to demonstrate that autophagy changes were induced by VPA, 3-MA and 17-DMAG in the HNSCC tumor cells (Fig. 7A). Next, we evaluated effects of VPA, 3-MA and 17-DMAG in the recognition of HNSCC by HTLs. The results show that enhancement of autophagy by VPA increased HTL responses to HNSCC cells and that both autophagy inhibitors, 3MA and 17-DMAG decreased the HTL response (Fig. 7B). These results indicate that autophagy plays a role in MHC class II antigen processing and presentation affecting HTL recognition of tumor cells.

Discussion
A growing body of evidence indicates that the HGF/c-Met signaling pathway plays a role in tumor proliferation, invasion and metastases, making this pathway an attractive target for developing tumor-targeted therapeutics. 17 c-Met TKIs such as ARQ197 are being evaluated against solid tumors showing promising results. 18 In this study, we observed that not only solid tumors such as HNSCC but also some hematological malignancies such as NKTCL express c-Met, suggesting that this signaling pathway may also be a target for treating these tumors. Moreover, NKTCL were found to produce HGF and the HGF/c-Met signaling pathway was involved in an autocrine manner in the tumor cell proliferation. Most importantly, we found that c-Met expression and HGF production were evident in clinical specimens and in sera from NKTCL patients. Moreover, because NKTCL proliferation was inhibited by ARQ197 or c-Met blocking antibody, it is possible that these drugs may be effective in the treatment of NKTCL. In conditions of low HGF concentrations, the dependence on HGF/c-Met signaling in tumor proliferation might not be prominent and thus c-Met inhibition has little influence over tumor proliferation. However, since the tumor surrounding stroma can produce HGF in addition to the tumor cells, 19 the dependence on HGF/c-Met signaling in cancer proliferation might become more relevant and c-Met inhibition could play a major role in the proliferation of tumor.
The main goal of the present study was to evaluate whether c-Met could function as a TAA for CD4 C HTLs. Three peptide sequences from c-Met predicted to bind to HLA-DR molecules were shown to function as HTL epitopes, eliciting effective responses against HNSCC and NKTCL. Recently, CD4 C T cells have been shown to exhibit cytotoxic activity and have been considered as responsible effector cells in antitumor immunity. 8,20 Results presented here indicate that c-Met reactive CD4 C T cells were also capable of killing tumor cells in an antigen-specific and MHC-restricted manner. We believe that the naturally processed c-Met peptide epitopes described here could be used in the design of peptide or DNA therapeutic vaccines against HNSCC and NKTCL. Furthermore, the c-Met 278-292 , c-Met 817-831 and c-Met 1244-1258 epitopes were able to induce HTL responses that were restricted by HLA-DR9, DR12, and DR53, respectively, indicating that the combined use of these epitopes in cancer immunotherapy may broaden population coverage as compared to the use of a single epitope.
We found that the amino acid sequence of the DR53 restricted c-Met 1244-1258 epitope was highly conserved with EGFR, HER-2, HER-3, and FAK homologous sequences ( Table 1). Interestingly, these homologous peptides were all efficient in stimulating the c-Met 1244-1258 HTL response, suggesting that c-Met 1244-1258 peptide may be able to generate effective responses against tumors expressing EGFR, HER-2, HER-3, and/or FAK.
Over the last several decades, great efforts have been devoted to the development of therapeutic vaccines for numerous cancer types. However, most clinical studies evaluating these vaccines have yielded unsatisfactory results with a few exceptions. 21,22 Therefore, ascertaining new approaches to improve the efficacy of cancer vaccines has become a priority in tumor immunotherapy. One of the main reasons why cancer vaccines and other forms of cancer immunotherapy fail is due to the immunosuppressive nature of the tumor microenvironment, which inhibits T cell responses. It is known that inflammatory cell infiltrates including T and B lymphocytes are commonly observed in NKTCL tissue samples. 23 Moreover, there is a high likelihood that NKTCL cells secret immune suppressive factors such as TGF-b that prevent these lymphocytes from inhibiting tumor growth. Thus, blocking TGF-b production at the tumor site would be a reasonable strategy to improve immunotherapy. We have demonstrated here that the c-Met inhibitor ARQ197 improved HTL responses by reducing TGFb production by the tumor cells. TGF-b has many biological activities such as converting na€ ıve helper T cells to regulatory T cells 24 and its inhibitory effects in Th1 helper T cells, 25 thus, TGF-b inhibition mediated by ARQ197 may increase antitumor immune responses through more than one mechanism. Since ARQ197 has already been in clinical trials, 18 we believe that the combined use of this agent with CD4 C T cell immunotherapy would be synergistic against c-Met positive tumors that produce TGF-b.
Antigen-specific T cell function is dependent on the antigenprocessing machinery of APCs including target tumor cells. Because chaperon-dependent and -independent autophagy plays Volume 4 Issue 2 OncoImmunology a significant role in antigen processing, 14,26 we evaluated change in the susceptibility of tumor cells that were pretreated with autophagy mediator to stimulate c-Met reactive HTLs. As expected, both chaperon-dependent and -independent autophagy mediators influenced the recognition of HNSCC tumor cells by HTLs. Both the chaperon-independent autophagy activator, VPA and chaperon dependent autophagy inhibitor, 17-DMAG have been reported to suppress the cell growth of NKTCL directly. 15,16 However, we did not confirm the effect of the inhibitors for modification of autophagy in NKTCL cells. Nevertheless, we did observe that 17-DMAG significantly reduced NKTCL T cell recognition (data not presented).
Recently, it has been reported that macroautophagy and chaperon-mediated autophagy plays a role in endogenous MHC class II processing and presentation of tumor or viral antigen. 14 Our results suggest that chaperoning by HSPs and/or macroautophagy participate in the c-Met derived peptide presentation by endogenous MHC class II molecules for direct tumor recognition by HTLs. Collectively, these observations imply that one must consider the possibility of inhibiting antitumor immunity when treating cancer patients with autophagy inhibitors. On the other hand, autophagy activators could be utilized to enhance antitumor immunity.

Immunohistochemistry
An immunohistological staining was performed using the Envision HRP System (Dako, #K5361) as described. 29 Formalin-fixed, paraffin-embedded sections were obtained from NKTCL and HNSCC patients. Samples were boiled in citrate buffer (pH 6.0) for antigen retrieval and incubated in 1% H 2 O 2 in methanol to inhibit endogenous peroxidase activity. After this, the sections were incubated with mouse anti-human c-Met monoclonal antibody (clone 3D4, Invitrogen, #37-0100) diluted 1:50 or goat anti-human HGF antibody (R&D Systems, #AF-294-NA) diluted 1:20, for 1 h in a room temperature, followed by incubation with HRP conjugated secondary antibodies and substrate. The institutional ethics committee approved this study (approval number 977), and the appropriate written informed consent was obtained from all patients providing tissue samples.
In vitro induction of peptide-specific CD4 C T-cell clones The method utilized for the generation of c-Met reactive HTL clones using peptide-stimulated lymphocytes from PBMCs of human healthy donors has been described in detail. 20 Briefly, DCs were produced from purified CD14 positive cells that were cultured for 7 d with GM-CSF (50ng/mL, PeproTech, #300-03) and IL-4 (1,000IU/mL, PeproTech, #200-04). Peptide-pulsed (3mg/mL) DCs were irradiated (4,200 rad) and cocultured with autologous purified CD4 C T cells in 96-flat-bottomed-well culture plates and one week later, the cells were restimulated with autologous irradiated PBMCs pulsed with peptide. Ten IU/mL Recombinant human IL-2 (PeproTech, #200-02) was added 2 d after restimulation. One week later, CD4 C T cells were tested for antigen reactivity using IFNg release assay, described below. Those microcultures exhibiting a significant response to peptide (at least 3-fold over background) were cloned by limiting dilution and cultured in 24well plates by weekly peptide restimulation with irradiated autologous PBMCs. 3% human male AB serum supplemented AIM-V (Invitrogen/GIBCO, #12055091) was used as complete culture medium. Written informed consent was obtained before blood sampling.

Cytotoxicity assays
Cytotoxic activity of HTL clones was measured using a colorimetric CytoTox assay (Promega, #G1780). This system measures the release of lactate dehydrogenase (LDH) from target cells. HTL cells were mixed with 2 £ 10 4 target cells at different effector to target ratios (0.1-30) in 96-round-bottomed-well plates. After 6 h incubation at 37 C, supernatant samples were collected to measure LDH content.

Measurement of HGF production
Sera from five healthy donors, six untreated NKTCL patients, six untreated stage IV HNSCC patients and tumor supernatant were tested for HGF levels using HGF ELISA kits (R&D systems, #DHG00). The institutional ethics committee approved this study (approval number 1066), and the appropriate written informed consent was obtained from all patients.

Statistical Analysis
All data are showed as mean with standard error of the mean. In all experiments, statistical differences were analyzed by using a two-tailed Student's t test and p < 0.05 was considered as statistically significant.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.