Identification of Novel HLA-A*24:02-Restricted Epitope Derived from a Homeobox Protein Expressed in Hematological Malignancies

The homeobox protein, PEPP2 (RHOXF2), has been suggested as a cancer/testis (CT) antigen based on its expression pattern. However, the peptide epitope of PEPP2 that is recognized by cytotoxic T cells (CTLs) is unknown. In this study, we revealed that PEPP2 gene was highly expressed in myeloid leukemia cells and some other hematological malignancies. This gene was also expressed in leukemic stem-like cells. We next identified the first reported epitope peptide (PEPP2271-279). The CTLs induced by PEPP2271-279 recognized PEPP2-positive target cells in an HLA-A*24:02-restricted manner. We also found that a demethylating agent, 5-aza-2’-deoxycytidine, could enhance PEPP2 expression in leukemia cells but not in blood mononuclear cells from healthy donors. The cytotoxic activity of anti-PEPP2 CTL against leukemic cells treated with 5-aza-2’-deoxycytidine was higher than that directed against untreated cells. These results suggest a clinical rationale that combined treatment with this novel antigen-specific immunotherapy together with demethylating agents might be effective in therapy-resistant myeloid leukemia patients.


Introduction
The development of chemotherapy regimens and targeted therapies has improved the survival of patients with leukemia. However, this disease continues to recur following conventional therapies in some patients, leading to poor prognosis [1,2]. Therefore, eradication of residual disease by additional treatment is necessary for these therapy-resistant patients. By this rationale, immunotherapy using antigen specifically expressed by leukemic cells might be an attractive strategy to cure leukemia patients [3,4,5]. committee of the Keio University School of Medicine (13-67-5) and Faculty of Pharmacy (111004-1). Written informed consent was obtained from all patients.

Sorting of LSCs
Leukemia cells were washed with phosphate-buffered saline (PBS; Sigma-Aldrich) and stained with R-phycoerythrin (PE)-conjugated monoclonal antibody against CD34 and fluorescein isothiocyanate (FITC)-conjugated monoclonal antibody against CD38 (Miltenyi Biotec, Auburn, CA) for 30 minutes on ice. After washing with PBS, cells were analyzed and separated using a MoFlo cell sorter system (Beckman Coulter Inc., Brea, CA).
Treatment with demethylating agent PL21 and HL60 cells and PBMNCs from HD were cultured in AIM-V medium (Life Technologies) supplemented with 10% human serum (Lonza, Walkersville, MD) in 6-well culture plates and 200 nM of 5-aza-2 0 -deoxycytidine (Sigma-Aldrich) was added every 24 hours for 72 hours.

Immunohistochemical staining
Immunohistochemical staining was performed as follows: Cells were attached to a slide using a Cytospin™ 4 cytocentrifuge (Thermo Fisher Scientific, Waltham, MA), then fixed with 2% paraformaldehyde and permeabilized with Triton X followed by blocking with 1% bovine serum albumin (BSA) in PBS. The sample was stained with monoclonal anti-PEPP2 antibody (Sigma-Aldrich) followed by Alexa Fluor488 1 goat-anti-mouse IgG (Molecular Probes, Life Technologies)

Peptide binding assay and peptide stability assay
We assessed the binding ability of candidate peptides to HLA-A24 as described previously [14]. CIR-A24 cells were incubated with 50 μg/mL peptides and 5μg/ mL of β2-microglobulin in serum-free media for 24 hours. After washing with PBS, the expression of HLA-A24 was measured by staining the cells with FITC-conjugated anti-HLA-A24 antibody (Mitsubishi-Kagaku, Tokyo, Japan). We also checked the stability of candidate peptides bound to the HLA-A24 molecule. CIR-A24 cells were incubated with 50 μg/mL peptides and 5 μg/mL of β2-microglobulin in serum-free media for 24 hours, followed by addition of 50 μg/ mL of brefeldin A (Sigma-Aldrich) for 4 hours. The expression of HLA-A24 was detected by staining the cells with FITC-conjugated anti-HLA-A24 antibody (Mitsubishi). MFI (mean fluorescence intensity) ratio was calculated as follows; MFI of CIR-A24 pulsed with each candidate peptide / MFI of CIR-A24 without peptide.

Generation of PEPP2 peptide-specific CTLs from human PBMCs
CTLs were generated by in vitro stimulation with peptide-pulsed autologous dendritic cells and phytohemagglutinin (PHA) blasts. Briefly, PBMNCs were isolated from whole blood of HLA-A Ã 24:02-positive HD by Ficoll density gradient centrifugation. Dendritic cells (DC) were generated by isolating CD14 + cells using a MACS separation system (Miltenyi Biotec, Bergisch Gladbach, Germany). The CD14 + cells were cultured in AIM-V medium supplemented with 10% human serum, 100 ng/mL of IL-4 (R&D Systems) and 100 ng/mL of GM-CSF (R&D Systems) for 5 days, then 20 ng/mL of TNF-α (R&D Systems) was added to generate DC. PHA blasts were derived from CD14cells by culturing in AIM-V medium supplemented with 10% human serum, 100 units of IL-2 and 1μ g/mL of PHA for 2 days. The DC or PHA blasts were pulsed with 50μg/L of peptide at room temperature for 3 hours and then irradiated (mediXtec, Chiba, Japan). On day 0, CD14cells were stimulated with irradiated peptide-pulsed DC in AIM-V medium with 10% human serum supplemented with 10 ng/mL of IL-7 (R&D Systems), then 50 ng/mL of IL-2 was added every 2 to 3 days. The cells were stimulated weekly with irradiated peptide-pulsed autologous PHA blasts, then CD3 + CD8 + cells were purified with immunomagnetic beads (Miltenyi Biotec).

Enzyme-linked immunospot (ELISpot) assay
IFN-γ secretion of CTLs in response to target cells was detected using ELISpot assay. Briefly, effector cells were incubated with target cells in plates coated with anti-IFNγ antibody (1-D1K; Mabtech Inc., Cincinnati, OH). After incubation for 20 hours, biotinylated antibody specific for IFN γ (7-B6-1; Mabtech) was added for 2 hours at room temperature and then streptavidin-alkaline phosphatase (Mabtech) was added for 1 hour. To develop spots, nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate (Biorad) were added and the color development was stopped by rinsing with distilled water. The resulting spots were counted using a CTL-ImmunoSpot 1 analyzer (Cellular Technology Ltd., Shaker Heights, OH)

Cytotoxicity assay
The cytotoxic activity of the CTLs was measured using a standard 51 Cr release assay. Target cells were labeled with 50 μCi of 51 Cr (PerkinElmer Inc., Waltham, MA) for 60 min at 37°C. CIR-A24 cells were then incubated with 5 μg/mL of peptide for 1 hour. For the blocking assay, KMS21 cells were treated with mouse anti-HLA class I antibody (w6/32; eBioscience, San Diego, CA) or mouse IgG isotype antibody (eBioscience). The target cells were mixed with effector cells. After 4 hours, supernatants were transferred to LumaPlates (PerkinElmer) and allowed to air dry in a hood overnight. Plates were sealed and counted in MicroBeta 1 Trilux instrument (PerkinElmer). Percent specific lysis for 51 Cr release was calculated as [(experimental 51 Cr release-spontaneous 51 Cr release) / (maximum 51 Cr release-spontaneous 51 Cr release)] x 100.

Expression of HLA class I, HLA-DR, and PD-L1
Untreated cells or cells treated with 5-aza-2 0 -deoxycytidine were washed with PBS and stained with PE-conjugated monoclonal antibody against HLA class I (eBioscience), PE-conjugated monoclonal antibody against HLA-DR (Gen-Probe, San Diego, CA), PE-conjugated monoclonal antibody against PD-L1 (eBioscience) and mouse isotype IgG 1 antibody (Beckman Coulter) for 30 minutes on ice. After washing with PBS, cells were analyzed using a FACSCalibur flow cytometer equipped with CellQuest software (BD Biosciences, San Jose, CA).
Expression of the PD-L1 gene was detected by quantitative PCR. Extraction and cDNA synthesis from PL21 cells were conducted as described in 'Detection of PEPP2 expression with RT-PCR' section. Quantitative PCR was performed using KOD SYBR qPCR Mix (TOYOBO, Osaka, Japan) with PD-L1-specific primers (Forward; 5-gacctatatgtggtagagtatgg tagc-3 0 , Reverse; 5-tgccattaatgccctctgatg-3 0 ). Quantitative PCR of GAPDH gene was performed as described above. The relative expression of PD-L1 was calculated by dividing results of PD-L1 by that of GAPDH.

Statistical analysis
Results are presented as mean ± s.e. Groups were compared using Student's t-test. Differences were considered significant when p < 0.05.

PEPP2 gene expression in cancer cells
We evaluated PEPP2 gene expression in various types of cancer cell lines by quantitative RT-PCR (Fig 1A and 1B). Especially myeloid leukemia cells such as K562 (CML), NB4 (APL), KU812 (CML), and also U937 (histiocytic lymphoma) expressed the gene at high levels, which was compatible with the results of DNA microarray (A of S1 We also examined PEPP2 gene expression in leukemic patients' samples. High expression was observed in 11 of 19 BM samples from APL ( Fig 1C) and 8 of 20 CML patients (Fig 1D). On the other hand, PBMNC and BMMNC isolated from healthy donors expressed little to no levels of the gene (Fig 1E). In addition, the PEPP2 gene expression pattern did not fully correlate with that of the WT1 gene in APL samples. For example, APL sample 3 and 5 were PEPP2positive and WT1-negative (Fig 1F).

PEPP2 expression in leukemic stem cells
We then examined if the PEPP2 gene was expressed in LSCs. We sorted the CD34 + CD38cells from the AML cell line KG1a (Fig 2A, S2 Fig) and the CML cell line KU812 (Fig 2D, S2 Fig), as LSCs of myelogenous leukemia are considered to be enriched in this fraction, which was also previously reported in leukemia cell lines including KG1a [20,21]. The sorted CD34 + 38cells and CD34cells isolated from both cell lines expressed the PEPP2 gene (Fig 2B and 2E). Expression of PEPP2 protein was also confirmed by immunohistochemistry (Fig 2C and 2F).

PEPP2-derived peptide screening
We synthesized 24 PEPP2-derived candidate peptides based on scores predicted by the SYF-PEITHI and BIMAS programs (Table 1). We then performed peptide binding and stabilization assays. We first selected 2 peptides, PEPP2 271-279 (EFGPFPFVI), PEPP2 253-261 (SLPLPLMLL) which showed high score in stability assay, since peptides recognized by CTL should not only bind to HLA molecule but also stay there for longer time. We also chose PEPP2 140-149 (AFT-PIQLQEL), which showed the best binding ability. The three PEPP2 peptides were used for further screening for induction of CTL. Induction of anti-PEPP2 CTL from healthy donors Using the PEPP2 271-279 peptide, we could induce PEPP2-specific CTL from HLA-A Ã 24:02 positive lymphocytes isolated from healthy donors (Fig 3A). We could not induce PEPP2-specific CTL with the PEPP2 253-261 or PEPP2 140-149 peptides (data not shown).
The CTLs could lyse C1R-A24 cells only when the target cells were pulsed with PEPP2 271-279 peptide (Fig 3B). The recognition by CTLs was peptide dose-dependent (Fig 3C). We also examined their cytotoxicity against several cancer cell lines. The CTLs lysed only HLA-A Ã 24:02-positive and PEPP2-positive KMS21 cells (Fig 3D) and failed to recognize HLA-A Ã 24:02-positive PBMNC from HD which expressed low levels of PEPP2 expression (HD7 in Fig 1E). This cytotoxicity was suppressed by addition of anti-HLA class I antibody (Fig 3E).
Up-regulation of PEPP2 expression by 5-aza-2 0 -deoxycytidine Gene expression of many CTAs is controlled by methylation of their promoter regions. Therefore, we checked whether PEPP2 expression is up-regulated by 5-aza-2p-deoxycytidine in leukemia cell lines HL60 and PL21, which don't express PEPP2 gene. As shown in Fig 4A and 4B, PEPP2 gene expression was significantly up-regulated after treatment with 5-aza-2 0 -deoxycytidine. PEPP2 protein expression was also increased in treated cells compared to untreated cells (Fig 4C). In contrast, gene and protein expression did not change after treatment of PBMCs from healthy donor with 5-aza-2 0 -deoxycytidine.

Discussion
Many CTAs have been reported, however, the number of antigens that can elicit strong immunity and eradicate cancer cells in patients is limited [22].
In this study, we tried to identify the first epitope peptide derived from a homeobox protein, PEPP2. Homeobox genes encode 60-amino acid homeodomains, which are considered to be transcription factors that regulate the developmental process, and sometimes carcinogenesis [23]. Although the target gene or precise function of human PEPP2 is still unknown, Hu et al. [24] reported that Rhox5, a mouse homologue of human PEPP2, could suppress Unc5c, or netrin-1 receptor, which is a tumor-suppressor gene with pro-apoptotic function. Li et al. [25] showed that knockdown of Rhox5 in a colon cancer cell line suppressed cell proliferation and migration in mice. There is also a report that transducion of PEPP2 into murine monocytic leukemia cell line enhanced leukemogenesis in irradiated mice [26]. These results indicate an important role of PEPP2 in the development of leukemia and suggest that targeting PEPP2 might be effective in the eradication of leukemia cells.
We first clarified that the PEPP2 gene is highly expressed in cell lines derived from hematological malignancies and found that especially myeloid leukemia such as APL and CML cells expressed high amount of PEPP2. Then, we also observed its high expression in several BM samples from APL and CML patients, some of which were negative for WT1, suggesting that using both antigens as CTL epitopes would broaden the number of patients who can receive immunotherapy.
On the other hand, PEPP2 expression in PBMNC or BMMNC from healthy donors was not observed or was extremely low compared to leukemia cells, which suggests that PEPP2 is an ideal target for therapy of leukemia. We didn't observe high expression of the PEPP2 gene in other types of cancer cells including brain tumor, lung cancer, and pancreatic cancer, although it has been reported that this gene is expressed in colon cancer and gastric cancer [26], which were not included in our sample set.
Since target antigens expressed by cancer stem cells are better candidates for therapeutic targets, we then examined whether the PEPP2 gene is expressed in LSCs. We found that the PEPP2 gene was expressed in both the CD34 + CD38fraction and the CD34fraction in KG1a and KU812 cells. Although its expression level was not restricted to the CD34 + CD38fraction, immunotherapy targeting PEPP2 might contribute to eradicating residual LSCs in myeloid leukemia patients.
We also identified a HLA-A Ã 24:02-restricted epitope (PEPP2 271-279 ) by screening 24 PEPP2-derived candidate peptides. This epitope is located in the C-terminal domain of the 288 amino acid sequence of PEPP2 and does not overlap with the homeodomain, indicating that this epitope is PEPP2-specific. Anti-PEPP2 CTLs could be consistently induced from PBMNCs of HLA-A Ã 24:02-positive healthy donors. These CTLs recognized epitope peptide-pulsed target cells as well as PEPP2-positive cancer cells in an HLA-A Ã 24:02-restricted manner, indicating that PEPP2 271-279 is an HLA-A Ã 24:02-restricted epitope specific to PEPP2.
Antigen-specific immunotherapy cannot be applied for patients without antigen expression by their cancer cells. Therefore, induction of target antigens in antigen-negative cancer cells is useful for the treatment of wide variety of cancer patients. Demethylating agents including 5-aza-2 0 -deoxycytidine is known to increase expression of several CTAs, such as MAGE and NY-ESO1 by epigenetically modulating their gene expression [27]. It has been shown that the expression level of the Rhox5 gene is up-regulated by epigenetic drugs in mouse cancer cell lines via demethylation of the gene promoter region [28]. Here we evaluated PEPP2 expression in human leukemic cells and normal blood cells after treatment with physiologically tolerable doses of 5-aza-2 0 -deoxycytidine. Expression level of both the PEPP2 gene and protein were significantly enhanced in AML cell lines that originally didn't express PEPP2.
Moreover, PBMNCs from healthy donors failed to express the PEPP2 gene after treatment with 5-aza-2 0 -deoxycytidine. In a clinical setting, this phenomenon is extremely important to avoid side effects caused by target expression in normal cells. Coral et al [29] have reported that expression of the MAGE gene family is also up-regulated only in tumor tissue in mouse, however, the precise mechanism causing this difference between cancer cells and normal cells remains to be investigated.
We also revealed that 5-aza-2 0 -deoxycytidine-treated leukemia cells became sensitive to killing by anti-PEPP2 CTLs. This killing was not a direct effect of the drug on leukemia cells, since viability of leukemia cells was still more than 90% after treatment with the drug (data not shown). In addition, the CTL recognition was not observed with healthy donor's PBMNCs treated with 5-aza-2 0 -deoxycytidine, which didn't express PEPP2, suggesting that this CTL recognition was PEPP2-specific. This finding is consistent with reports using other antigens, such as NY-ESO1 and PRAME [30,31]. In addition, the expression level of HLA class I in leukemia cells was up-regulated after treatment with 5-aza-2 0 -deoxycytidine, which could facilitate killing of leukemia cells by CTLs. These epigenetic drugs have been approved in many countries and widely used for the treatment of patients with myelodysplastic syndrome. Toor et al. [32] recently reported that administration of azacitidine and lenalidomide followed by autologous lymphocyte infusion could enhance expression of CTAs including MAGEA3 in myeloma patients and antigen-specific CTLs were observed in some subjects. Therefore, the combination of PEPP2-specific immunotherapy with these epigenetic drugs might be an attractive strategy for treatment of leukemia patients. However, we also found that PD-L1, an immune checkpoint molecule, was slightly elevated in 5-aza-2i-deoxycytidine-treated leukemia cells, which might have reduced CTL recognition. This up-regulation of PD-L1 has also been reported in other cancer, such as non-small cell lung cancer or MDS, although the precise mechanism is remained to be determined [33,34]. Therefore, addition of anti-PD-L1 blockade, which is now widely gaining attention for its immune-modifying ability in cancer treatment, might be necessary for further enhancement of cytotoxic activity against the target cells [35].
In conclusion, we have demonstrated that a homeobox protein, PEPP2, is highly expressed in myeloid leukemia cells and its expression was up-regulated by demethylating agent in a leukemia-specific manner. We also identified a novel HLA-A Ã 24:02-binding epitope, PEPP2 271-279 , which might represent a good target antigen for immunotherapy in therapy-resistant myeloid leukemia patients, especially in combination with a demethylating agent.
Supporting Information S1 Fig. PEPP2 gene expression in various normal tissues and cancer cell lines. The expression pattern of the PEPP2 gene was detected by DNA microarray (A), and PEPP2 was highly expressed in testis and myelogenous leukemia cell lines (APL, CML). PEPP2 expression level was compared between normal testis and K562 by standard PCR (B) and quantitative PCR (C). Water was used for template as negative control (NC). (TIF)