Olfactory Receptor Family 7 Subfamily C Member 1 Is a Novel Marker of Colon Cancer–Initiating Cells and Is a Potent Target of Immunotherapy

Colorectal cancer (CRC) is one of the most common malignancies throughout the world, with more than 140,000 new cases every year in United States (1). Recent progress in CRC treatments has improved the prognosis, but advanced disease with recurrence or distant metastasis is usually incurable and has an unfavorable prognosis (2, 3). Cancer stem–like cells/cancer-initiating cells (CSC/CIC) are defined by their ability of tumor initiation, self-renewal, and differentiation (4). CICs are resistant to standard cancer therapies including chemotherapy and radiotherapy, and they are therefore thought to be responsible for tumor recurrence and metastasis (5,6). CICs are a reasonable target for cancer treatment, and elimination of CICs is essential to cure cancer.

Cancer immunotherapy is a novel approach to target cancer and is expected to become the fourth most widely used cancer therapy after surgery, chemotherapy, and radiotherapy. Identification of the first human tumor-associated antigens (TAAs) has enabled cancer immunotherapy to be performed using antigenic peptides derived from TAAs (7–10). Clinical trials using antigenic peptides derived from TAAs are now being conducted in several facilities, and some trials have shown fascinating and promising results; however, the results of most studies using antigenic peptide vaccination have been disappointing (11). Recently, melanoma-initiating cells (melanoma stem cells) have been isolated as CD271⁺ cells, and CD271⁺ melanoma cells lack the expression of melanoma antigens including TYR, MART1, and MAGEs. Low expression levels of melanoma antigens in melanoma-initiating cells are thought to be the reason for resistance to peptide vaccination therapies using melanoma antigens (12). And recent studies indicate that CICs may contribute to make immunosuppressive microenvironment (13). We therefore hypothesized that CIC-targeting immunotherapy might improve the efficiency of cancer immunotherapy. To address the feasibility of CIC-targeting immunotherapy, we isolated CRC CICs as side population (SP) cells by Hoechst 33342 staining and found that SP cells are susceptible to CTLs as are non-CIC main population (MP) cells (14). We also reported that CICs of malignant fibrous histiocytoma could be recognized by autologous CTLs derived from tumor-infiltrating lymphocytes (15). Thus, CTLs can recognize treatment-resistant CICs.

In this study, we identified a novel CIC antigen in CRC, olfactory receptor family 7 subfamily C member 1(OR7C1). A gene knockdown study using siRNAs and an OR7C1 overexpression model showed that OR7C1 has a role in the maintenance of CICs of CRC. A high expression level of OR7C1 was shown to be a poor prognostic factor in the immunohistochemical staining of OR7C1. An OR7C1-specific CTL clone showed significantly stronger antitumor effect than that of a shared antigen-specific CTL clone in an in vivo CTL adoptive transfer model. These results indicate that the CIC antigen OR7C1 is a novel marker for CRC CICs and is a promising target for CIC-targeting immunotherapy.

Colon adenocarcinoma cell lines SW480 (HLA-A*0201/2402, B*0702/1508, Cw*0702/0704), HT29 (HLA-A1/24), HCT15 (HLA-A*0201/2402, B*08/3501, Cw*07/04) were kind gifts from Dr. K. Imai (Sapporo Medical University, Sapporo, Japan) at 1999. All cells was cultured in DMEM (Sigma Chemical Co.) supplemented with 10% FBS (Life Technologies, Inc.). HCT15-B2M, a stable transfectant of HCT15 cells with β2 microglobulin cDNA, was cultured in DMEM supplemented with 10% FBS and 10 μg/mL puromycin (Sigma). SW480, HT29, and HCT15 cells were authenticated by us using the PCR-SSCP (Single Strand Conformation Polymorphism) method as described previously (16). Erythroleukemia cell line K562 was purchased from ATCC and was cultured in RPMI1640 (Sigma) supplemented with 10% FBS. HLA-A*2402 stably transfected TAP-deficient cell line T2 A-A*2402 (T2-A24) cell was a kind gift from Dr. K.Kuzushima (Aichi Cancer Center Research Institute, Nagoya, Japan) and was cultured in RPMI1640 supplemented with 10% FBS and 0.8 mg/mL G418 (Life Technologies, Inc.). No specific authentication of K562 cells and T2-A24 cells were performed.

Three pairs of frozen colorectal adenocarcinoma and corresponding non-neoplastic colorectal tissues, which were used for RT-PCR, were obtained from surgically resected tissues removed at Kushiro City General Hospital. Sixty-seven specimens, used for immunohistochemical staining of OR7C1, were obtained from surgical resection tissues at Department of Gastroenterological Surgery I, Hokkaido University Graduate School of Medicine (Sapporo, Japan). Normal tissues used for immunohistochemical staining of OR7C1 were obtained from pathologic dissection at the Department of Pathology, Sapporo Medical University School of Medicine (Sapporo, Japan). To obtain CTL clones, we used blood from three healthy volunteer donors and three colon cancer patients at Higashi Sapporo Hospital (Sapporo, Japan). We obtained informed consent from all patients and volunteer donors according to the guidelines of the Declaration of Helsinki.

SP analysis and flow cytometry analysis of OR7C1 were performed as described previously and as described in the Supplementary Materials and Methods (14,17).

Xenograft transplantation model was performed as described previously (18). Cancer stem cell frequency was calculated by web program Extreme Limiting Dilution Analysis (ELDA; http://bioinf.wehi.edu.au/software/elda/) software (19).

Total RNAs (3 μg) derived from SW480 SP cells and SW480 MP cells were reverse transcribed to generate double-stranded cDNA and labeled using Cy3-cytidine or Cy5-cytidine. Then the labeled cRNAs were hybridized to a oligonucleotide microarray (Panorama Human Micro Array, Sigma) and scanned by Genepix 4000B Microarray Scanner. Microarray raw data and processed data have been deposited in the Array Express database (E-MTAB-2605).

RT-PCR analysis was performed as described previously and as described in Supplementary Materials and Methods (14,20).

To assay sphere formation efficiency, 10³ cells were plated in ultra-low attachment 6-well plates (Corning Incorporated Life Sciences) and cultured in DMEM/F12 (Life Technologies, Inc.,) supplemented with 20 ng/mL EGF and 20 ng/mL basic fibroblast growth factor (R&D Systems). Cells were incubated in a 5% CO₂ incubator for 2 weeks, and the number of spheres was counted. The data was shown in (number of spheres) × 100/(numbers of seeded cells)%.

An OR7C1 gene knockdown experiment was performed using siRNA. OR7C1 siRNA duplexes were designed and synthesized using the BLOCK-it RNAi designer system (Life Technologies, Inc.). The oligonucleotide encoding OR7C1 siRNA1 was 5′-TATACTTTGCAACTGGCGTCCTGGG-3′, siRNA2 was 5′-CCCAAGAATTTCTCCTCCTGGGATT-3′, siRNA3 was 5′-CAGAGATTCAGTTCATTCTCTTTGG-3′. Cells were seeded at 50% confluence and transfections were carried out using Lipofectamine RNAi max (Life Technologies, Inc.) in Opti-MEM according to the manufacturer's instructions.

Five peptides, OR7C1_34(10) (MYLVTFTGNL), OR7C1_93(10) (TYAGCLSQIF), OR7C1_277(9) (MYTMVTPML), OR7C1_251(10) (FYGTGFGVYL) and OR7C1_131(9) (HYTVIMNPQL), were designed from the amino acid sequence of OR7C1 based on the HLA-A24–binding motifs. All of the peptides were synthesized at and purchased from Life Technologies, Inc. Peptides were dissolved in DMSO and stored at −80°C before use.

Peptide binding affinity to an HLA-A24 was assessed by HLA-A24 stabilization assay as described previously (21).

PBMCs were isolated from three healthy volunteers and three colon cancer patients as described previously and as described in Supplementary Materials and Methods (21, 22).

CTL induction was performed as described previously (23). To obtain CTL clones, standard limiting dilution was performed as described previously (22). The lytic activity of CTLs was tested by a ⁵¹Cr release assay and IFN-γ ELISpot assay as described previously (21, 22).

NOD/SCID mice were inoculated subcutaneously in the back with 1 × 10³ SW480 SP cells. Three weeks later, when the tumor had started to become palpable, 5 × 10⁴ OR7C1_ 93(10)-specific CTL clone cells or 5 × 10⁴ Cep55/c10orf3_ 193(10)-specific CTL clone cells or PBS were injected intravenously. The same adoptive transfer procedure was performed 4 weeks after inoculation with SP cells. Tumor size was assessed weekly.

Data for samples in the xenograft model, cytotoxicity assay and CTL adoptive transfer model were analyzed by Student t test, with P < 0.05 conferring statistical significance. Statistical analyses were performed with SPSS (version 11 for Windows; SPSS Inc), and GraphPad Prism (version 4.0 for Windows; GraphPad Software Inc) was used for plotting Kaplan–Meier curves. Pearson χ² square tests were used to determine the significance of associations between characteristic variables. Survival rates were calculated using Kaplan–Meier method, and differences between groups were tested using the log-rank test.

We previously reported that side population (SP) cells derived from SW480, HT29. and HCT15 CRC cells showed high tumor-initiating ability, high expression levels of stem cell markers including SOX2, POU5F1, and Nanog, and a high level of resistance against chemotherapeutic agents compared with those of MP cells (14). In this study, we isolated SP cells from SW480, HT29, and HCT15 cells as sources of CRC CICs to screen for novel CIC-related antigens. SP cells derived from SW480, HT29, and HCT15 cells showed higher expression levels of stem cell markers including SOX2, POU5F1, and LGR5, higher tumor-initiating ability higher levels of tumor-initiating cells and higher sphere formation ability than those of MP cells (Supplementary Fig. S1A–S1D; Table 1). CRC CIC–related genes were screened by a cDNA microarray using mRNAs of SP cells and MP cells derived from SW480 cells. Approximately 600 genes showed more than 2-fold higher expression levels than those in MP cells (Supplementary Table S1). We further investigated the expression profiles of candidate genes by RT-PCR. Most of the genes showed ubiquitous expression in normal organs; however, one SP-specific gene, olfactory receptor, family 7, subfamily C, member 1 (OR7C1), showed expression only in the testis among adult normal tissues and fetal tissues (Fig. 1A). OR7C1 mRNA was preferentially expressed in SP cells derived from SW480, HT29, and HCT15 cells rather than in MP cells (Fig. 1B). Therefore, OR7C1 is a novel cancer/testis (CT) antigen and is also a novel CIC antigen. The expression of OR7C1 mRNA in human primary carcinoma tissues was examined by RT-PCR. OR7C1 mRNA was expressed in all primary CRC tissues, but its expression was not detected in adjacent normal colon mucosa tissues. Stem cell markers (SOX2, POU5F1, LGR5) were also detectable in only CRC tissues (Fig. 1C). We then evaluated the expression level of OR7C1 protein using a polyclonal anti-OR7C1. The specificity of the OR7C1 antibody was confirmed by Western blotting using 293T cells transiently transfected with OR7C1 DNA (Supplementary Fig. S1E). OR7C1 protein was preferentially expressed in SP cells derived from SW480, HT29, and HCT15 cells rather than in MP cells as shown by Western blot analysis (Fig. 1D). Immunohistochemical staining revealed that OR7C1-positive cells were present in primary spermatocytes, secondary spermatocytes, spermatid and sperm of the testis. OR7C1 was not detected in normal colon tissues (Fig. 1E).

To investigate the expression of OR7C1 protein in human CRC tissues, we stained 67 human CRC tissues using anti-OR7C1 antibody and 56 cases showed OR7C1-positive staining (positivity: 83.5%). The clinicopathologic status of each of the patients is summarized in Supplementary Table S2. No significant correlations with OR7C1 staining were found for median age of the patients, gender, clinical stage, site of the lesion and pathologic grade (Table S2). We classified the OR7C1 staining into four groups according to the staining intensity of cancer cells (intensity score (IS): IS 0: negative, IS 1+: weakly positive, IS 2+: moderately positive, IS 3+: strongly positive), and IS 0 and 1+ were classified as low OR7C1 IS and IS 2+ and 3+ were classified as high OR7C1 IS (Fig. 1F). The 10-year survival rates of low OR7C1 IS and high OR7C1 IS were 100% and 63.1%, respectively. The high OR7C1 IS group showed remarkably poorer prognosis than that of the low OR7C1 IS group (Fig. 1G, P <0.001). Strikingly, all low OR7C1 IS patients were alive at the time of investigation. We wanted to determine whether OR7C1-intensity would be a prognosis factor, but we were not able to perform the univariate analysis because all of the low IS group patients were alive.

Next, we investigated overall survival according to OR7C1-positive levels. OR7C1-positive rate showed a moderate positive correlation with OR7C1 staining intensity (Supplementary Fig. S1F, R = 0.576, P = 0.0002). The cases were classified into four groups according to the OR7C1-positive proportion [proportion score (PS): PS: negative, PS 1: <10%, PS 2: 10%–50%, PS3: >50%). The 10-year survival rates of patients in the OR7C1 PS1, PS2, and PS3 groups were 86.5%, 89.0%, and 60.0%, respectively, and the OR7C1PS 3 group showed poorer prognosis than that in the OR7C1 PS1 or PS2 group (Supplementary Fig. S1G). Univariate analysis showed that OR7C1 PS3 colorectal carcinoma is a risk factor for poor prognosis after surgical resection. (Table 2; relative risk: 5.08, P = 0.013).

A recent study showed that increased expression of ALDH1 was associated with poor clinical outcome in colorectal cancer cases (24). We thus investigated the expression of ALDH1 and classified the expression into IS0-IS3. The positive frequencies of OR7C1 and ALDH1 showed a statistically significant correlation (Supplementary Fig. S1H, R = 0.426, P < 0.01). We found that the prognosis of ALDH1 strong intensity (IS2 and IS3) showed a tendency to be related to poorer prognosis but did not reach statistical significance (Fig. 1H). Univariate analysis showed that ALDH1 highly positive (PS3) colorectal carcinoma is a risk factor for poor prognosis after surgical resection (Table 2; relative risk: 4.07, P = 0.029). Finally, we performed the multivariate analysis, and only OR7C1 high positivity (PS3) was an independent risk factor for poor prognosis (Table 2).

To evaluate the functions of OR7C1 in colon cancer cells, we generated OR7C1-overexpressed SW480 cells (SW480-OR7C1 cells). The expression of OR7C1 mRNA was confirmed by quantitative reverse transcription PCR (qRT-PCR; Fig. 2A). The expression levels of stem cell markers including SOX2, LGR5, and POU1F5 were higher in SW480-OR7C1 cells than in SW480 cells (Fig. 2A). The ratio of SP cells in SW480-OR7C1 cells was higher than that in SW480 cells (Fig. 2B). SW480-OR7C1 cells showed higher sphere-forming ability than that of SW480 cells (Fig. 2C). The tumorigenicity of SW480-OR7C1 cells in vivo was examined by injecting cells into NOD/SCID mice. Tumor formation was initiated with as few as 10³ SW480-OR7C1 cells. On the other hand, SW480 cells did not initiate the formation of a tumor even with injection of 10⁴ cells (Fig. 2D; Table 1). To generalize these observations, the same experimental procedure was performed by using two additional CRC cell lines (HCT15 cells and HT29 cells), and similar results were obtained (Table 1; Supplementary Fig. S2). The frequency of CICs was significantly higher in SW480-OR7C1 cells than in SW480 cells (Table 1). These results indicate that CRC cells with overexpression of the OR7C1 gene have properties of CIC.

To confirm the functions of OR7C1 in CRC cells, we performed OR7C1 gene knockdown experiments using SW480 SP cells. We designed three different siRNAs and confirmed the suppression of OR7C1 mRNA by siRNAs (Fig. 2E). Gene knockdown of OR7C1 decreased the expression levels of stem cell markers including SOX2, LGR5, and POU1F5 (Fig. 2E) and the percentage of SP cells in SW480 cells (Fig. 2F). Gene knockdown of OR7C1 significantly decreased sphere-forming ability of SW480 SP cells (Fig. 2G). We evaluated the tumorigenicity of SW480 SP cells in which OR7C1 mRNA was knocked down by siRNAs. Control siRNA-transfected SW480 cells initiated tumor formation with as few as 10² cells, whereas OR7C1 siRNA-transfected SW480 cells needed 10³ cells to initiate tumor formation (Table 1). The tumor was significantly smaller with OR7C1 siRNA-transfected SW480 SP cells than with control siRNA-transfected cells (Fig. 2H). Similar results were obtained by SP cells derived from two additional CRC cell lines, HCT15 and HT29 (Table 1 and Fig. S3). These results indicate that OR7C1 has an essential role in the maintenance of CRC CICs.

OR7C1 overexpression and knockdown experiments indicated that OR7C1 has a role in CRC CICs, and we therefore characterized OR7C1-positive (OR7C1⁺) cells derived from CRC cells. The polyclonal antibody we used in this study is specific for the putative extracellular domain of OR7C1, and we evaluated staining of CRC cells by flow cytometry. OR7C1-overexpressed SW480 cells were stained as a positive control and analyzed by flow cytometry. The specificity of OR7C1 staining was also confirmed by flow cytometry in 293T cells transiently transfected with OR7C1 (data not shown), indicating that the polyclonal antibody could detect OR7C1 protein by flow cytometry analysis. A cell population with positive OR7C1 staining was detected in OR7C1-overexpressed SW480 cells, and the percentage of OR7C1⁺ cells was 95.0% (Fig. 3A). We further analyzed CRC cells and found that approximately 2.0% OR7C1⁺ cells were present in SW480, HCT15, and HT29 cells (Fig. 3B). A higher level of OR7C1 gene expression in OR7C1⁺ cells was confirmed by qRT-PCR (Fig. 3C). The expression of stem cell markers in OR7C1⁺ cells was analyzed by qRT-PCR, and OR7C1⁺ cells showed higher expression levels of stem cell markers including SOX2, LGR5, and POU1F5 than the levels in OR7C1-negative (OR7C1⁻) cells (Fig. 3C). We then investigated whether the OR7C1⁺ cells have properties of CICs. Xenograft transplantation of 10³ OR7C1⁺ cells and 10³ OR7C1⁻ cells derived from SW480, HCT15, and HT29 cells revealed that the tumor derived from OR7C1⁺ cells was significantly larger than that derived from OR7C1⁻ cells (Fig. 3D). Xenograft transplantation of as few as 10¹ SW480 OR7C1⁺ cells initiated tumor formation in 1 of 5 mice, whereas SW480 OR7C1⁻ cells did not initiate tumor formation even with injection of 10³ cells, and the percentages of CICs were higher in OR7C1⁺ cells than that in OR7C1⁻ cells (Table 1). HCT15 OR7C1⁺ cells initiated tumor formation in 1 of 5 mice with injection of 10¹ cells, whereas HCT15 OR7C1⁻ cells needed 10² cells to initiate tumor formation. On the other hand, no difference was observed in initiating tumor formation by HT29 OR7C1⁺ cells and HT29 OR7C1⁻ cells. OR7C1⁺ cells showed higher proliferation ability in vitro (Fig. 3E) and the higher sphere-forming ability (Fig. 3F) than those of OR7C1⁻ cells. These results indicate that OR7C1⁺ colorectal cancer cells have characteristics of CICs and that OR7C1 is a novel marker for CRC CICs.

OR7C1 is a novel CIC antigen and it has a role in the maintenance of CRC CICs. Furthermore, OR7C1 is a novel cancer/testis (CT) antigen, thus OR7C1 is a promising target of CIC-targeting cancer immunotherapy. We induced OR7C1-specific CTLs derived from HLA-A*2402–positive colon cancer patients and healthy volunteer donors to evaluate the immunogenicity of OR7C1. We designed and synthesized OR7C1 candidate peptides carrying HLA-A*2402, a frequent HLA allele in Asian and other areas, binding motifs, as described in a previous report (21). HLA-A24-binding ability was assessed by an HLA-A24 binding assay, and all of tested peptides showed higher binding affinity than that of the negative control peptide (Table S3).

PBMCs of 5 CRC patients and 3 healthy donors were stimulated with mixture of 5 peptides, and reactivity for each peptide was evaluated by an IFN-γ ELISpot assay and a ⁵¹Cr release assay. The reactivity for each peptide is shown in Supplementary Fig. S4A and S4B, and OR7C1_277(9), OR7C1_34(10), OR7C1_251(10), and OR7C1_93(10) peptides were shown to be immunogenic epitopes. Among the tested peptides, OR7C1_93(10) showed highest CTL induction rates, inducing specific CTLs from 6 of 8 cases as detected by ELISpot (patients: 4, healthy donors: 2) and 5 of 8 cases by ⁵¹Cr release assay (patients: 3, healthy donors: 2; Supplementary Fig. S4A and S4B). Therefore, OR7C1_93(10) peptide might be the most immunogenic epitope of OR7C1, and we thus generated an OR7C1_93(10)-specific CTL clone for further analysis (clone: #23). The OR7C1_93(10)-specific CTL clone exhibited cytotoxic activity for T2-A24 cells pulsed with OR7C1_93(10) peptide but not for T2-A24 cells without the peptide or K562 cells (Fig. 4A). To confirm that the OR7C1_93(10) peptide is endogenously expressed in CRC CICs, we performed a cytotoxicity assay using SP and MP cells derived from SW480, HCT15-B2M, and HT29 cells. The OR7C1_93(10)-specific CTL clone showed greater cytotoxicity for SP cells derived from SW480 cells, HCT15-B2M cells, and HT29 cells than for MP cells derived from SW480 cells, HCT15-B2M cells, and HT29 cells and for HLA-A*2402⁻OR7C1⁻ K562 cells (Fig. 4B). Moreover, the CTL clone showed significantly greater cytotoxicity for OR7C1⁺ cells derived from SW480 cells, HCT15-B2M cells, and HT29 cells than for OR7C1⁻ cells derived from SW480 cells, HCT15-B2M cells, and HT29 cells (Fig. S4C). The cytotoxicities for SP cells and OR7C1⁺ cells derived from HCT15-B2M cells were blocked by an anti-HLA class I mAb (Fig. 4C and Supplementary Fig. S4D). These results indicate that OR7C1_34(10), OR7C1_93(10), OR7C1_251(10), and OR7C1_277(9) peptides are immunogenic and that OR7C1_93(10) peptide is endogenously processed and presented on the cell surface of colon CICs.

The OR7C1-specific CTL clone efficiently recognized isolated CICs. We then examined whether the CTL clone can recognize CICs specifically in CIC and non-CIC mixed conditions. OR7C1-specific CTL clone #23 was added to SW480 colon cancer cell culture twice (Fig. 4D), and the ratios of SP cells were determined 1 week after the second CTL mixture. SP cells were observed in control the CD8⁺ T-cell mixture, but the frequency of SP cells was reduced in the OR7C1-specific CTL clone mixture (Fig. 4E), indicating that the OR7C1-specific CTL clone specifically recognizes SP cells.

We then investigated the potency of CIC-related antigens. We previously identified a novel TAA, CEP55 (22). CEP55 is expressed in both SP cells and MP cells derived from colon carcinoma cells, and a CTL clone specific for CEP55 peptide could recognize SP cells and MP cells at similar levels (14). In a previous review article, we proposed that TAAs that are recognized by CTLs can be divided into 3 categories according to the expression pattern in CICs and non-CICs: (i) CIC antigens, which are expressed preferentially in CICs, (ii) shared antigens, which are expressed in both CICs and non-CICs, and (iii) non-CIC antigens, which are expressed preferentially in non-CICs (25). In this study, we investigated the expression profiles of several known TAAs in SP cells and MP cells by RT-PCR. OR7C1, MAGEA3, and MAGEA4 were preferentially expressed in SP cells (CIC antigens), CEP55, BIRC5, and AURKA were expressed in both SP cells and MP cells (shared antigens), and EGLN3 was preferentially expressed in MP cells (non-CIC antigen; Fig. 4F). As both CIC antigens and shared antigens are expressed in CICs, we compared the potency of a CIC antigen and that of a shared antigen using CTL clones specific for OR7C1 and CEP55, respectively. The OR7C1-specific CTL clone showed greater cytotoxicity for SP cells than for MP cells, whereas the CEP55-specific CTL clone showed the same levels of cytotoxicity for both SP cells and MP cells (Fig. 4G). The antitumor effects in vivo of the OR7C1-specific CTL clone and the CEP55-specific CTL clone were compared by using a CTL therapeutic adoptive transfer model (Fig. 4H). Both OR7C1-specific and CEP55-specific CTL clones showed significant antitumor effects compared with the effects in the negative control group (Fig. 4I). Interestingly, the OR7C1-specific CTL clone showed a significantly greater antitumor effect than that of the CEP55-specific CTL clone (Fig. 4I). Similar results could be observed in two independent experiments. These results indicate that immunotherapy targeting only CICs using a CIC antigen-specific CTL is more potent than immunotherapy targeting all cancer cells using shared antigens.

Several methods have been reported for isolation of CICs including the use of cell surface markers, use of SP cells, ALDEFLUOR assay and tumor sphere formation (6). In this study and our previous study, we confirmed that SP cells derived from CRC cells have higher tumor-initiating ability, higher sphere-forming ability, and higher expression levels of stem cell markers than those of MP cells derived from CRC cells. Thus, SP cells used in this study are enriched with CICs and are a reasonable source of CICs. In the microarray screening, we found an olfactory receptor family gene, OR7C1. OR7C1 is a member of the olfactory receptor family, characterized by G protein–coupled seven transmembrane receptors, and is expressed in the tongue (26,27). The olfactory receptor family includes about 350 genes and they are usually expressed in the olfactory epithelium. Several kinds of olfactory receptors have been identified in tissues other than olfactory tissues, including the testis, tongue, and placenta; however, functions of most olfactory receptors expressed in nonolfactory tissues are still elusive (28). A previous study showed that an olfactory receptor, OR51E2, is expressed in prostate cancer cells and that activation of OR51E2 by its ligand, β-ionone and inhibited proliferation of prostate cancer cells (29). Another olfactory receptor family member, OR51E1, was reported to be a potential novel biomarker for small intestine neuroendocrine carcinoma (30). Expression of the olfactory receptor OR13C4 was shown to be significantly associated with risk of pancreatic cancer by pathway analysis in a genome-wide association study (31). These findings suggest that ectopically expressed olfactory receptors in cancer cells may have some roles; however, there has been no report in which the relationship between an olfactory receptor and CICs was described. In the current study, we analyzed the functions of OR7C1 by gene overexpression and gene knockdown by siRNAs, and we found that OR7C1 has roles in the expression of stem cell genes including SOX2, POU5F1 and LGR5, in the frequency of SP cells, in sphere-forming ability in a floating condition and in tumor-initiating ability in vivo. These results indicate that OR7C1 has roles in the maintenance of CRC CICs, which might be controlled by the expression of stem cell molecules including SOX2, POU5F1, and LGR5. In the olfactory bulb, olfactory receptors receiving signals usually stimulate the G protein (Golf)-coupled cascade and then induce an increase of cyclic AMP (cAMP) synthesis by activation of adenylate cyclase. The increase in cAMP synthesis results in a change in the intracellular Ca²⁺ concentration by the opening of some channels and in depolarization of the cell membrane (28). It has been reported that the human sperm chemotaxis is mediated via this pathway (32). As OR7C1 is expressed in primary spermatocytes, secondary spermatocytes, spermatid, and sperm in the testis, it may have a role in sperm chemotaxis. Inactivation of OR51E2 could reduce growth of prostate cancer cells via this pathway and then phosphorylation of p38 and stress-activated protein kinase/c-Jun NH₂-terminal kinase (SAPK/JNK) MAPKs (29). The signaling pathway that induces the expression of stem cell–related genes via OR7C1 is still elusive, but activation of adenylate cyclase by Golf might be a possible pathway, and our preliminary data suggested that activation of Akt might also be involved in OR7C1 signal transduction (data not shown).

Previous work by Ben-Porath and colleagues revealed that the embryonic stem cell–like gene expression signature in cancer is related to a higher grade in breast cancers, glioblastomas, and bladder cancers (33). They found that activation targets of NANOG, POU5F1, SOX2, and c-MYC are overexpressed in poorly differentiated tumors. These findings indicate that ES-like gene expression might play a role in the stem-like phenotypes of tumors that have a biologically high-grade malignant potential. ES genes are related to cancer stem cell/cancer-initiating cell phenotype. SOX2 has roles in the maintenance of CICs of breast cancer, glioma, and lung cancer (18,34,35), and the expression of SOX2 is related to poorer prognosis in several malignancies (36–39). Relationships of other ES genes including OCT4, NANOG, and LIN28 with CSCs/CICs were described in other reports (40–43). On the other hand, there are controversial reports regarding the intestinal stem cell marker LGR5. Ziskin and colleagues reported that LGR5 did not have prognostic significance in CRC patients (44). On the other hand, Saigusa and colleagues reported that LGR5 expression and CD44 expression in cancer stroma may be coordinately associated with tumor relapse in locally advanced rectal cancer after preoperative chemoradiotherapy and that they are related to poorer overall survival (45). Moreover, combined expression levels of Aldh1 which is one of the CIC markers, Survivin, and EpCAM are strong independent prognostic factors, with high HRs, for survival and tumor recurrence in colon cancer patients and therefore reflect tumor aggressiveness (24). In this study, we investigated whether OR7C1 influences clinical prognosis using immunohistochemical staining of CRC tissues. We confirmed that patients in the highly positive and intense OR7C1 group have a low rate of overall survival and that only OR7C1 high positivity was an independent risk factor of poor prognosis. The frequencies of OR7C1-positive cells, which are thought to have upregulation of the stem cell genes SOX2, POU5F1, and LGR5, might reflect the total number of high-grade CICs and might be related to poorer prognosis.

We previously reported that CRC CICs isolated as SP cells were sensitive to CEP55-specific CTLs as well as non-CICs, although CICs were resistant to chemotherapy agents (14). Our recent mouse DNA vaccination models indicated that CIC antigens are more potent than shared antigens and that shared antigens are more potent than non-CIC antigens (17,46). The potency of CIC-targeting immunotherapy was shown for mouse melanoma and squamous cell carcinoma (47). However, there is no evidence that CIC antigens are the most potent also in the human CTL system, and there are no such known human CIC antigens. In this study, we identified a novel CIC antigen, OR7C1, and we confirmed the potency of a CIC targeting immunotherapy by using a mouse CTL adoptive transfer model. Our data indicate that targeting only CICs with a CIC antigen is a better strategy than targeting all cancer cells with a shared antigen. As shared antigens are also expressed in CICs, the exact mechanism by which CIC antigen is more potent than a shared antigen is still elusive. One possible explanation is the limited number of CTLs in vivo. There are huge numbers of cancer cells in apparent clinical tumors (a clinically diagnosable tumor of only 1 cm in diameter contains approximately 10⁹ cancer cells.). Therefore, targeting only CICs is more effective for eradicating CICs with limited numbers of CTLs and so might result in a significant antitumor effect. On the other hand, the effector/target ratio was relatively low with CTLs specific to a shared antigen as shared antigen-specific CTLs also recognize non-CICs, and CTLs may miss CICs. Thus, the antitumor effect of shared antigen-specific CTLs might be limited compared with that CIC antigen-specific CTLs. CIC-specific immunotherapy, however, misses non-CICs, which constitute the major population of cells in a tumor. As non-CICs have less tumor-initiating ability than do CICs, non-CICs may not be significant for tumor eradication.

In summary, we identified a novel CRC CIC marker, OR7C1. Expression of OR7C1 is a poor prognostic marker for CRCs. OR7C1_93(10) peptide is a promising candidate for CIC-targeting immunotherapy.

No potential conflicts of interest were disclosed.

Conception and design: R. Morita, Y. Hirohashi, T. Torigoe, S. Inoda, T. Tsukahara, M. Asaka, N. Sato

Development of methodology: R. Morita, Y. Hirohashi, S. Inoda, H. Asanuma

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Morita, S. Inoda, G. Kutomi, T. Terui, N. Minagawa, N. Takahashi, A. Taketomi, S. Todo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Morita, Y. Hirohashi, T. Torigoe, S. Inoda, T. Mariya, H. Asanuma

Writing, review, and/or revision of the manuscript: R. Morita, Y. Hirohashi, T. Torigoe

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Morita, T. Torigoe, A. Takahashi, H. Asanuma, Y. Tamura, T. Kubo, T. Terui, T. Kondo, A. Taketomi

Study supervision: T. Torigoe, Y. Tamura, T. Tsukahara, T. Kanaseki, T. Mizuguchi, K. Ishitani, S. Hashino, M. Asaka

The authors thank Drs. K. Imai and K. Kuzushima for providing cell lines, Dr. Y. Yoshida for clinical sample collection, and Ms. E. Yamamoto for technical assistance.

This study was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to N. Sato), program for developing the supporting system for upgrading education and research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to N. Sato), Health and Labour Sciences Research Grants, a grant-in-aid of Ono Cancer Research Fund (to T. Tsukahara), Sagawa Foundation for Promotion of Cancer Research (to Y. Hirohashi), Suharakinenzaidan Co., Ltd. (to Y. Hirohashi), and Kobayashi foundation for cancer research (to Y. Hirohashi). This study was also supported in part by Grants-in-Aid for Regional R&D Proposal-Based Program from Northern Advancement Center for Science & Technology of Hokkaido Japan (to Y. Hisrohashi and T. Tsukahara.).

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.