Identification of HLA-A24-Restricted Novel T Cell Epitope Peptides Derived from P-Cadherin and Kinesin Family Member 20A

We here identified human leukocyte antigen-(HLA-)A∗2402-restricted epitope peptides from Cadherin 3, type 1, P-cadherin (CDH3) and kinesin family member 20A (KIF20A) that were found to be specifically expressed in cancer cells through genome-wide expression profile analysis. CDH3-10-807 peptide and KIF20A-10-66 peptide successfully induced specific CTL clones, and these selectively responded to COS7 cells expressing both HLA-A∗2402 and respective protein while did not respond to parental cells or COS7 cells expressing either HLA-A∗2402 or respective protein. Furthermore, CTL clones responded to cancer cells that endogenously express HLA-A∗2402 and respective protein, suggesting that CDH3-10-807 peptide and KIF20A-10-66 peptide are naturally presented on HLA-A∗2402 molecule of human cancer cells. Our results demonstrated that CDH3-10-807 peptide and KIF20A-10-66 peptide are novel HLA-A24-restricted tumor-associated antigens and would be applicable for CTL-inducing cancer therapies.

We have identified dozens of genes specifically expressed in cancer cells by genome-wide expression profile analysis for cDNA microarray consisting of more than 30,000 cDNAs and expressed sequence tags (ESTs) [14]. Among them, two genes, Cadherin 3, type 1, P-cadherin (CDH3) and kinesin family member 20A (KIF20A), were found to be upregulated in pancreatic cancers [15,16]. CDH3 is one of the classic cadherin family that plays a critical role in cell-cell adhesion and epithelial morphogenesis [17]. We reported that overexpression of CDH3 promoted the motility of cancer cells and blocking of CDH3 by anti-CDH3 antibody inhibited the migration of CDH3-expressing cells [15]. KIF20A is a member of the kinesin family, which is characterized to be a motor protein in cancer cells [18], and northern analysis indicated no expression of KIF20A among examined 23 normal tissues except testis. Furthermore, knock down of KIF20A expression with small interfering RNA suppressed the proliferation of pancreatic ductal adenocarcinoma cells [16].

Journal of Biomedicine and Biotechnology
Thus, both CDH3 and KIF20A would play oncogenic functions in pancreatic cancer cells and are attractive target molecules for cancer therapies including immunotherapy.
We here identified CDH3-and KIF20A-derived novel HLA-A * 2402-restricted epitope peptides that can induce peptide-specific cytotoxic T lymphocyte (CTL), suggesting that these epitope peptide would be applicable to peptidebased cancer vaccine therapies for HLA-A * 2402 positive pancreatic cancer patients.

Materials and Methods
2.1. Peptides. CDH3 and KIF20A-derived 9-mer and 10-mer peptides that have high binding affinity (binding score > 10) to HLA-A * 2402 were predicted by the binding prediction software "BIMAS" (http://www-bimas.cit.nih.gov/molbio/ hla bind/) and were synthesized by Sigma-Aldrich Japan KK (Ishikari, Japan) according to a standard solid-phase synthesis method and purified by reversed-phase high-performance liquid chromatography (HPLC). HIV-A24 epitope peptide (RYLRDQQLL) [19] was also synthesized as a negative control. The purity (>90%) and the identity of the peptides were confirmed by analytical HPLC and mass spectrometry analysis, respectively. Peptides were dissolved in dimethylsulfoxide at 20 mg/mL and stored at −80 • C.

In Vitro Induction of Peptide-Specific CTL.
To examine the ability to induce peptide-specific CTL, purified CD8 + T cells were cocultured with autologous monocyte-derived mature dendritic cells (DCs) pulsed with peptide. Both CD8 + T cells and DCs were prepared from peripheral blood mononuclear cells (PBMCs) of same HLA-A * 2402-positive healthy volunteers. Briefly, PBMCs were isolated by Ficoll-Paque solution (GE Healthcare, Uppsala, Sweden), then cells were cultured in AIM-V medium (Invitrogen) containing 2% heat-inactivated autologous serum (AS). After the over night incubation, nonadherent cells were washed out, then 1000 U/mL of granulocyte-macrophage colony-stimulating factor (GM-CSF; R&D Systems, Minneapolis, MN) and 1000 U/mL of interleukin (IL)-4 (R&D Systems) were added in the culture to induce monocyte-derived DCs. To mature DCs, 0.1 KE/mL of OK-432 (Chugai Pharmaceutical Co., Tokyo, Japan) was added in the culture on day 5. Seven days later, DCs were pulsed with 20 μg/mL of synthesized peptide in AIM-V medium containing 3 μg/mL of β2-microglobulin (Sigma-Aldrich) at 37 • C for 3 h [20] and incubated in the media containing 30 μg/mL of Mitomycin C (MMC) (Kyowa Hakko Kirin Co. Ltd., Tokyo, Japan) for 30 min. Following washing out residual peptide and MMC, cells were used as antigen-presenting cells to induce peptide-specific CTL. Generated monocyte-derived mature DCs expressed CD80, CD83, CD86, and HLA class II on their cell surface (data not shown). Autologous CD8 + T cells were prepared from PBMCs derived from the same HLA-A * 2402-positive donor by positive selection with Dynal CD8 positive isolation kit (Invitrogen) according to the manufacturer's instructions. 1.5 × 10 4 of peptide-pulsed DCs and 3 × 10 5 of CD8 + T cells were cocultured in 0.5 mL of AIM-V medium supplemented with 10 ng/mL of IL-7 (R&D Systems) and 2% AS on 48well plates (Corning Inc., Corning, CA). IL-2 (CHIRON, Emeryville, CA) was added to the culture at 20 IU/mL 3 days after coculture, and peptide-pulsed DCs were additionally supplied into the culture on days 7 and 14. Eight wells were prepared for CTL induction by every peptide in a single experiment. On day 21, interferon-(IFN-) γ production was examined by IFN-γ enzyme-linked immunospot (ELISPOT) assay under the stimulation with peptide-pulsed TISI cells.

IFN-γ Enzyme-Linked Immunospot (ELISPOT) Assay.
T cell response to epitope peptide was measured by ELISPOT assay using IFN-γ ELISPOT kit and AEC substrate set (BD Pharmingen, San Diego, CA) according to the manufacturer's instruction. Briefly, TISI cells were pulsed with 20 μg/mL of respective peptide at 37 • C for 20 h, and the residual peptide that did not bind to TISI cells was washed out to prepare peptide-pulsed TISI cells as the stimulator cells. 200 μL of cell culture suspension were distributed to two wells (100 μL each) on Multiscreen-IP 96-well plate (Millipore, Bedford, MA) following removing 500 μL of supernatant from each well from culture of "in vitro induction of peptide-specific CTL." Cells were coincubated with peptide-pulsed TISI cells (1 × 10 4 cells/well) at 37 • C for 20 h. The plates were analyzed by the automated ELISPOT reader, ImmunoSPOT S4 (Cellular Technology Ltd, Cleveland, OH) and ImmunoSpot Professional Software  Figure 1: IFN-γ production from CTLs responding to CDH3-or KIF20A-derived peptides. IFN-γ production by CTLs induced with CDH3derived peptides (a) or KIF20A-derived peptides (b) responding to respective peptide-pulsed HLA-A * 2402 positive TISI cells. CTLs were expanded and harvested following "in vitro induction of peptide-specific CTL," and IFN-γ production was examined by IFN-γ ELISA. "Closed bar" indicates the mean IFN-γ production responding to TISI cells pulsed with indicated peptide, and "open bar" indicates the mean IFN-γ production responding to TISI cells pulsed with HIV-A24 peptide (negative control). All experiments were performed triplicate. Similar results were obtained in three to five independent experiments. * P < 0.05, * * P < 0.01.  Binding score was obtained using BIMAS program. CTL induction was indicated as positive (+) or negative (-). Similar results were obtained 3-7 independent experiments using PBMC of 3-7 healthy volunteers.
well was more than 50 spots/well compared with that in the control well, we estimated that peptide-specific CTL were induced (positive) and subsequently expanded CTL from the positive well. Sensitivity of our ELISPOT assay was estimated as approximately average level by ELISPOT panel of Cancer Immunotherapy Consortium [CIC (http://www.cancerresearch.org/consortium/assay-panels/)].

CTL Expansion.
Peptide-specific CTL obtained from CTL positive well of "in vitro induction of peptide-specific CTL" were expanded by the modified protocol based on the previously described methods [21][22][23][24]. Briefly, 5 × 10 5 of CTLs were cocultured with 5 × 10 6 of MMC-treated (30 μg/mL at 37 • C for 30 min) EB-3 and Jiyoye cells in 25 mL of AIM-V containing 5% AS and 40 ng/mL of anti-CD3 mAb. The cultures were supplemented with IL-2 (final concentration: 120 IU/mL) 24 h later and fed with AIM-V medium containing 5% AS and IL-2 (30 IU/mL) on day 5, 8, and 11. On day 14, expanded T cells were harvested to examine specific response to epitope peptide by IFN-γ enzyme-linked immunosorbent assay (ELISA).  Both CTL clones significantly produced IFN-γ responding to COS7 cells expressing HLA-A * 2402 and corresponding gene. Similar results were obtained in three independent experiments. Independently induced other CTL clones also produced significant amount of IFN-γ when exposed with COS7 cell expressing both HLA-A * 2402 and respective gene (data not shown). R/S ratio, responder cell (CTL clone)/stimulator cell (COS7 cell) ratio.

Cytotoxicity
Assay. Specific cytotoxic activity of induced CTL clones was tested by a 4 h 51 Cr release assay as previously described [25]. Data are represented as the mean ± SD of triplicate samples. The expression of HLA-A * 2402 and oncogene-derived protein was confirmed by Western blotting using anti-Myc (Upstate Biotechnology, Lake Placid, NY) or anti-FLAG antibody (Sigma-Aldrich). Two days after transfection, the transfected cells were harvested with versene (Invitrogen) and used to stimulate peptide-specific CTL clones. IFN-γ production by CTLs was examined by IFN-γ-specific ELISA.

Induction of CTL Responding to CDH3-or KIF20A-Derived Peptide Restricted with HLA-A * 2402.
Based on the analysis with the binding prediction software "BIMAS," we synthesized 21 CDH3-derived epitope-peptides and 28 KIF20A-derived epitope-peptides that were expected to have high affinity to HLA-A * 2402 molecule and activate CTLs (Table 1).

Recognition of Cells Endogenously Expressing Both HLA-A * 2402 and Respective Protein by Peptide-Specific CTL Clones.
We then examined that the established peptide-specific CTL clones can recognize cells that express HLA-A * 2402 and the target proteins. COS7 cells were transfected with plasmid designed to express HLA-A * 2402 molecule and/or that to express the full length protein of CDH3 or KIF20A. We confirmed expression of these proteins by western blotting (data not shown). CDH3-10-807 peptide responding CTL clone substantially produced IFN-γ when exposed to COS7 cells that expressing both HLA-A * 2402 and CDH3, but not COS7 cells that expressing either HLA-A * 2402 or CDH3 (Figure 3(a)). Similarly, KIF20A-10-66 peptide responding CTL clone produced significant amount of IFN-γ when exposed to COS7 cells that expressing both HLA-A * 2402 and KIF20A, but not COS7 cells that expressing either HLA-A * 2402 or KIF20A (Figure 3(b)). Both CTL clones also produced IFN-γ responding to COS7 cells, which transfected with pIRES-vector containing both HLA-A * 2402 and respective oncogene (data not shown). On the other hand, CTL clones responding to KIF20A-9-305 peptide or KIF20A-10-304 peptide did not produce IFN-γ when exposed to COS7 cells expressing both HLA-A * 2402 and KIF20A (data not shown). Only CDH3-10-807 peptide and KIF20A-10-66 peptide, but not other candidate peptides, were able to induce CTL responding to COS7 cells expressing HLA-A * 2402 and CDH3 or KIF20A, albeit we have tried several times using PBMC derived from different healthy donors (data not shown).

Cytotoxic Activity of CTLs.
We also examined cytotoxic activity of CTL clones. CDH3-10-807 or KIF20A-10-66 peptide-specific CTL clone demonstrated cytotoxic activity against HLA-A * 2402-positive TISI cells when respective peptide was pulsed, but not when HIV-A24 peptide was pulsed or peptide was not pulsed (Figures 5(a) and 5(b)). These results suggested that CDH3-10-807 or KIF20A-10-66 peptide-specific CTL clone specifically exerted cytotoxic activity responding to respective epitope peptide binding to HLA-A * 2402 on cells.
No homologous sequence to CDH3-10-807 peptide or KIF20A-10-66 peptide was demonstrated by the homology research using the BLAST algorithm http://blast.ncbi.nlm .nih.gov/Blast.cgi (data not shown), suggesting that these peptide would be the unique epitope peptide presented on HLA-A * 2402 of CDH3 or KIF20A-expressing cells.

Discussion
Pancreatic cancer is one of the most malignant cancers, since 5-year survival rate is only 5% and the therapeutic modalities are very limited [26,27]. Both CDH3 and KIF20A were upregulated in the majority of pancreatic cancers and have oncogenic functions [15,16]. Thus, CDH3 and KIF20A would be promising target molecules to develop novel therapeutic strategies for pancreatic cancer. Hence, we identified HLA-A * 0201-restricted peptides derived from CDH3 and KIF20A [28,29].
Predicted binding score of KIF20A-10-66 peptide to HLA-A * 2402 was relatively low when compared with that of CDH3-10-807 peptide. We previously reported epitope peptides derived from RNF43 and IMP-3, and those peptides also have low affinity to HLA molecule [22,24]. Interestingly, both peptides have been already applied for clinical trials as peptide-based immunotherapy and CTL were obtained in many cancer patients [11,12]. These results suggested that some peptides possibly induce CTLs albeit binding score was low by BIMAS prediction and KIF20A-10-66 peptide, as well as CDH3-10-807 peptide, possibly induces CTL in cancer patients.
Recent improvement and development of cancer therapies, including combined treatments of standard therapies (chemotherapy, radiotherapy, and surgical resection), substantially improved the survival of advanced cancer patients [27]. However, unfavorable adverse events are still often observed. On the other hand, immune therapies inducing cancer-cell-specific CTLs are now developed to improve the efficacy against cancers and the quality of life of patients. Ongoing several clinical trials using epitope peptides derived from TAA have been proving the evidence that CTL-inducing therapies are much less harmful to the patients [10][11][12]. However, efficacy of some vaccine therapy trials is still limited mainly due to the development of escaping variant cancer cells that lost targeted TAA expression during the treatment [30]. Therefore, it is generally thought that the therapeutic efficacy would be improved when the origin of vaccinated peptide is functionary essential molecule for cancer cell survival, proliferation, and/or motility. We have been screening epitope peptides derived from cancer-specific genes and reported several epitope peptides, which can elicit specific CTL responses [21][22][23][24]. Some of these peptides have been already applied for translational researches of multipeptide vaccine to treat esophageal cancer and colorectal cancer [11,12]. Moreover, multiple-antigen vaccine therapy was suggested to more effectively hinder escape mechanisms in the guidance from Food and Drug Administration (Guidance for Industry: Clinical Considerations for Therapeutic Cancer Vaccines). Thus, we believe that identification of CTL-inducible epitope peptides derived from several molecules that play critical roles in various types of cancer is important to develop multipeptide cocktail, and that resulted in the improvement of efficacy of CTL-inducing cancer therapies.
Presented results demonstrated that CDH3-10-807 peptide and KIF20A-10-66 peptide pulsed DCs induced specific CTL to possibly exert antitumor effect. The immunogenicity of CDH3-10-807 peptide and KIF20A-10-66 peptide should be examined in patients bearing these genes-expressing cancers, and we are now going to conduct clinical trials.