A New Melanoma Antigen Fatty Acid–Binding Protein 7, Involved in Proliferation and Invasion, Is a Potential Target for Immunotherapy and Molecular Target Therapy

The identification of molecules that are preferentially expressed in cancer cells, particularly those that are involved in the formation of malignant phenotypes, is important not only for understanding cancer biology but also for the development of new diagnostic and therapeutic methods (1–3). For a long time, we have been attempting to identify human melanoma antigens to understand the immune responses to melanoma and to develop an effective immunotherapy (1, 4, 5). Among the melanoma antigens identified previously by us and other groups, melanocyte-specific, melanoma-overexpressed, and cancer-testis antigens can be identified using various systematic gene expression analyses, including DNA chip/microarray, serial analysis of gene expression (SAGE), etc., or various cDNA subtraction techniques, including representational differential analysis, etc. We have reported previously the identification of new melanocyte/melanoma-specific antigens, FCRL/FREB and PAX3, using DNA chip and SAGE, respectively (6, 7).

Most immunotherapies to date have used melanoma antigens that are not necessary for the survival and proliferation of melanoma cells, including melanocyte-specific antigens, gp100, MART-1/Melan-A, and tyrosinase, and cancer-testis antigens, MAGEs and NY-ESO-1 (8, 9). One of the problems has been tumor escape from immune attack through loss of antigens (10–12). It is preferable to use tumor antigens involved in the proliferation and survival of cancer cells because of decreased likelihood to develop antigen-loss variants during immunotherapy. In fact, mutated antigens possibly involved in melanoma formation, including β-catenin, cyclin-dependent kinase 4, MART-2 (mutated Ski), and MUM-3 (mutated RNA helicase), were isolated from patients with a favorable prognosis after immunotherapy (13). A high proportion of MUM-3-specific CTL was observed in the peripheral blood from the patient (14). However, these immunogenic mutated antigens are often quite uncommon or not presented by common HLA types, resulting in difficulty in their use as immunotherapy for a broad population of patients. Thus, the identification of additional antigens that are frequently overexpressed and involved in the formation of malignant phenotypes is required for the development of effective immunotherapy.

In this study, we attempted to identify molecules that are overexpressed in melanoma by comparing the expression profile of melanoma with those of various normal tissues using GeneChip and have identified fatty acid–binding protein 7 (FABP7), a member of the FABP family, which is likely involved in lipid metabolism (15). Human FABP7 was reported to be expressed in developing neuroglial cells and gliomas (16, 17) and involved in differentiation and migration of developing glial cells (18). We have shown that FABP7 is also expressed in melanoma cells and is involved in the enhanced proliferation and invasion of some melanoma cell lines through experiments that used both FABP7 down-regulation and FABP7 up-regulation. In addition, we showed that FABP7 is an immunogenic antigen in melanoma patients by showing the frequent presence of specific IgG antibody in sera from melanoma patients. These results indicated that FABP7 may be a potential target for the development of diagnostic and therapeutic methods, including immunotherapy and molecular target therapy.

Cell lines and tissues. Cell lines and tissues used were as follows: skin fibroblast, TIL1362 (T cell), 888 EBV-B cell, SKmel23, 888mel, A375mel, 586mel, 501Amel, 526mel, 397mel, 1362mel, 1363mel, 928mel, 624mel, MMG1, SMKT1, WM266mel (melanoma), U-87MG, GI-1, U-251, A172 (glioma), Daudi, K562 (lymphoma and leukemia), K1S (lung cancer), TE10 (esophageal cancer), PK8 (pancreas cancer), KU1920 (renal cancer), KU-7 (bladder cancer), PC-3 (prostate cancer), MDA231 (breast cancer), RD-cell (myosarcoma), and human umbilical vascular endothelial cell (HUVEC). These cancer cell lines were either obtained from various researchers [Dr. Steven A. Rosenberg (Surgery Branch, National Cancer Institute, Bethesda, MD), Dr. Makoto Sunamura (Division of Gastroenterological Surgery, Tohoku University Graduate School of Medical Science, Sendai, Japan), Dr. Masaru Murai (Department of Urology, Keio University, Tokyo, Japan), and Dr. Tsutomu Honjou (Morinaga Institute of Biological Science, Tokyo, Japan)] or purchased from the American Type Culture Collection (Manassas, VA) and cultured as described previously (7). Normal tissues, including brain, kidney, spleen, muscle, lung, placenta, and testis, were purchased from BD Biosciences Clontech (Palo Alto, CA). Tumor specimens were obtained from patients who had undergone surgical resection at Shinshu University Hospital (Nagano, Japan), Keio University Hospital, and John Wayne Cancer Institute (Santa Monica, CA) with informed consent according to the institutional guidelines and were stored at −80°C until use.

Gene expression analysis using GeneChip. GeneChip analysis of a highly pigmented human melanoma cell line SKmel23, primary cultured melanocytes, keratinocytes, HUVEC, and liver and stomach tissues were done using oligonucleotide arrays (GeneChip HuGene FL array and Hu35K subarrays A, B, C, and D, containing ∼5,600 human genes and 35,000 human expressed sequence tags; Affymetrix, Santa Clara, CA) as described previously (6, 19). Briefly, biotin-labeled cRNAs generated from these cells were hybridized to the DNA chips, and they were scanned with a GeneArray Scanner (Hewlett-Packard, Santa Clara, CA). The scanned images were processed using the GeneChip software 3.1.

Reverse transcription-PCR and real-time PCR. cDNAs were synthesized with an oligo(dT) primer from total RNAs. Thirty cycles of reverse transcription-PCR (RT-PCR) were done at 95°C, 56°C, and 72°C for 30 seconds for FABP7-specific primers 5′-CTGGAAGCTGACCAACAGTC-3′ and 5′-CTCATAGTGGCGAACAGCAA-3′. Real-time PCR analysis of FABP7 expression in various melanocytic lesions was done with the iCycler iQ Real-time PCR Detection System (Bio-Rad Laboratories, Hercules, CA) using 250 ng of total RNA. The PCR reaction was subjected to 40 cycles at 95°C, 56°C, and 72°C for 60 seconds for FABP7 and 40 cycles at 95°C, 55°C, and 72°C for 60 seconds for glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The mRNA copy number was calculated using the Real-time Detection System software. Specific plasmid controls for FABP7 and GAPDH were synthesized as described previously (20), and standard curves for each marker were generated with the threshold cycle of six serial dilutions of plasmid templates (10⁶-10¹⁰). The quantitative expression of FABP7 in cells transfected with either small interfering RNA (siRNA) or FABP7 cDNA was evaluated with real-time PCR using Taqman Gene Expression Assays with the FABP7-specific probe according to the manufacturer's protocol (Applied Biosystems, Foster City, CA). GAPDH was used as an internal control.

Western blot analysis and phage-plaque assays. The coding region of FABP7 prepared by PCR was inserted into pET16b (Novagen, Madison, WI), ZAP Express vector (Stratagene, La Jolla, CA), or pcDNA3.1(+) (Invitrogen, Carlsbad, CA) to generate proteins produced by bacterial, phage, or mammalian cell, respectively, as described previously (21). Briefly, the proteins were produced in Escherichia coli BL21(DE3)pLysS transformed with pET16b/FABP7. After purification using His-Tag columns, SDS-PAGE was done, and the proteins were blotted onto nitrocellulose membranes and reacted with sera (1:100 dilution). After reaction with goat anti-human IgG(Fc) antibody, positive bands were detected by reaction with alkaline phosphatase–conjugated antibody, nitroblue tetrazolium, and 5-bromo-4-chloro-3-indolyl phosphate.

The phage-plaque assays were done as described previously (22). Briefly, the recombinant FABP7 bacteriophages and phages with no insert were mixed at a 2:3 ratio and infected into E. coli XL-1blueMRF', and the induced protein was transferred to nitrocellulose membranes. They were incubated with sera (1:100 dilution), and positive plaques were detected by enzymatic detection. The presence of serum IgG antibodies specific for the recombinant FABP7 protein was evaluated by comparing plaques produced from the FABP7 recombinant phage and control phages with no insert.

Immunohistochemical study. Polyclonal rabbit anti-FABP7 antibodies were generated by two independent immunizations of two rabbits each with the bacterial recombinant FABP7 protein after purification by Ni⁺ columns. Immunohistochemistry was done on formalin-fixed, paraffin-embedded sections from tissues. After deparaffinization, the sections to be reacted with the FABP7 polyclonal antibody were boiled for 10 minutes in 10 mmol/L citrate buffer. The sections for HMB45 were not pretreated. For immunohistochemical detection, DAKO EnVision (DAKO, Carpinteria, CA) was used with streptavidin-conjugated horseradish peroxidase and diaminobenzidine. Nuclei were stained with Giemsa. H&E staining was done in all sections.

RNA interference assay. dsRNAs homologous to regions of the FABP7 open reading frame were synthesized by Invitrogen (Stealth RNAi). Stealth RNAi Negative Control duplexes (Medium GC Duplex, Invitrogen) were used as control. BlOCK-iT Fluorescent Oligo (Invitrogen) was used to evaluate transfection efficiency. The target sequences of the siRNAs for FABP7 are as follows: 6 (5′-GACAAAGUGGUCAUCAGGACUCUCA-3′), 7 (5′-GGGAGAAGAGUUUGAUGAAACCACU-3′), and 8 (5′-GCCUGGAUGGAGACAAACUUGUUCA-3′). Cells were transfected with dsRNAs using LipofectAMINE 2000 (Invitrogen) or siLentFect (Bio-Rad Laboratories).

Optimal transfection conditions were determined initially. Cells were seeded at 20% to 50% confluence in 24-well plates (Nunc, Roskilde, Denmark). The concentrations of the siRNA solutions were determined using BlOCK-iT Fluorescent Oligo at 10 to 50 nmol/L, and the transfection reagents were used at 1 μL/well. The final volume in each well in all experiments was 600 μL. Transfections with LipofectAMINE 2000 were done under serum-free conditions using Opti-MEM-1 (Invitrogen) as the diluting medium. The condition that resulted in >80% of cells being transduced with BlOCK-iT Fluorescent Oligo was used for further experiments.

Cell proliferation assay. The WST-1 Cell Proliferation Assay System (Takara, Kyoto, Japan) was used to evaluate cell proliferation in addition to the standard cell counting with trypan blue staining. Premix WST-1 solutions (10 μL) were added to the wells containing cells in 0.1 mL medium at the 2nd day of the transfection of 3 × 10³ to 6 × 10³ cells with siRNAs. After 1-hour incubation at 37°C, the absorbance at 450 nm was measured against a reference wavelength of 655 nm.

Matrigel invasion assay. Cells were plated in BioCoat Matrigel invasion chambers (BD Biosciences, San Jose, CA) at a cell density of 2.5 × 10⁴ to 10 × 10⁴ per chamber in RPMI 1640 supplemented with (outer chamber) or without (inner chamber) 10% fetal bovine serum. After 22-hour incubation at 37°C and 5% CO₂, noninvading cells remaining on the top surface of the chamber were removed by scrubbing with a cotton-tipped swab, and the invaded cells that had adhered to the bottom surface of the chamber membranes were fixed and counted after cell staining with Diff-Quik stain (Sysmex, Kobe, Japan) according to the manufacturer's instructions.

Statistical analysis. Statistical analyses were done using the unpaired Student's t test. When mRNA copy number was compared during nevi and primary and metastatic melanoma, the Mann-Whitney U test was used.

The identification of molecules overexpressed in melanoma cells using GeneChip. To identify molecules that were overexpressed in melanoma cells compared with melanocytes and other normal cells, cDNA expression profiles obtained using GeneChip (U35; containing 42,000 probes) from a highly pigmented melanoma cell line SKmel23, primary cultured melanocytes, HUVEC, keratinocytes, and fresh tissues from liver and stomach were compared. Eight genes whose average difference scores (19, 23) in SKmel23 were >100- and >8-fold compared with those in other normal cells were identified (Table 1). These genes represented five previously identified melanoma antigens, including cancer-testis antigens MAGE-A2, MAGE-A3, MAGE-A12, and PRAME, a melanocyte-specific antigen MATP (AIM1), and three other molecules, including FABP7, PEG10, and FLJ16517. Further studies were done on the latter three molecules.

Expression of the FABP7 transcript in melanoma cell lines and tissues. Using 30 cycles of RT-PCR analysis, FABP7 was detected in brain and weakly detected in both skeletal muscle in normal fresh tissues and melanocytes among cultured noncancer cells. Among three nevus tissues [1, dysplastic nevus; 2 and 3, acquired melanocytic nevi (AMN)], FABP7 was detected in one AMN (3) tissue. Among the cancer cell lines, FABP7 was detected in all 15 melanoma cell lines tested and in 2 of 4 glioma cell lines but not in other types of cancer cell lines (Fig. 1A). Among fresh cancer tissues, FABP7 was detected in 10 of 11 melanoma cell lines and in all 4 glioma samples but not in 7 samples from other cancers. PEG10 and FLJ16517 were detected in various normal and cancer cell lines and tissues (data not shown). Thus, further studies were not attempted on these two genes. Using real-time PCR analysis to a maximum of 40 cycles, 6 of 11 (54.5%) nevi, 54 of 65 (83.1%) primary melanoma tissues, and 31 of 48 (64.6%) metastatic melanoma tissues were positive for the FABP7 expression (Fig. 1B). Lower levels of the FABP7 mRNA were observed in metastatic melanoma compared with primary melanoma (P < 0.001). Similarly, by Northern blot analysis, FABP7 expression was detected in three melanoma cell lines, in one of the brain tumor cell lines, and faintly in adult brain tissue (Fig. 1C).

Expression of the FABP7 protein in melanoma cell lines and tissues. To evaluate the expression of the FABP7 protein, a rabbit polyclonal antibody specific for FABP7 was generated by immunization with the bacterial recombinant FABP7. As shown in Fig. 2A, Western blot analysis with this polyclonal antibody detected a 15-kDa band in 293T cells that were transfected with the FABP7 cDNA but not in untransfected 293T cells, indicating that this antibody was specific for FABP7. The antibody detected similar-sized bands in all eight melanoma cell lines, in one of three brain tumor cell lines, and in one primary (2) and three metastatic (1, 3, and 4) melanoma tissues, but not in other cell lines and tissues, including primary cultured melanocytes (Fig. 2A), showing the expression of the FABP7 protein in melanoma and brain tumors. Further immunohistochemical examination was then done using this FABP7-specific antibody. The cytoplasm of three melanoma cell lines, but not primary cultured melanocytes, was stained, whereas a control rabbit preimmune serum did not stain any of these cultured cells (Fig. 2B). Among 15 primary melanoma tissues tested, 11 were found to express the FABP7 protein as shown in a representative result (Fig. 2C). Therefore, the FABP7 protein was frequently expressed in melanoma.

FABP7 is involved in the cell proliferation and Matrigel invasion of melanoma cell lines. We then investigated the functional role of FABP7 in melanoma. We first examined the effects of FABP7 down-regulation with specific siRNAs on malignant phenotypes of melanoma, including cell proliferation, survival, invasion, and production of the angiogenic factor vascular endothelial growth factor (VEGF) and the immunosuppressive cytokine interleukin-10 (IL-10). Six melanoma cell lines that express FABP7 (888mel, WM266mel, 526mel, SKmel23, SKmel28, and A375mel) were evaluated for the effects of the FABP7 down-regulation using three siRNAs (6, 7, and 8). siRNA transfection was done under conditions leading to transfection of >80% of the cells with the FITC siRNA.

The FABP7 mRNA detected by Taqman real-time PCR was significantly decreased in 888mel at day 3 after transfection of siRNAs 6, 7, or 8 (Fig. 3A). Cell numbers counted by trypan blue assay at day 3 (Fig. 3B) and cells detected by the WST-1 assay (Fig. 3C) at day 2 were decreased in the 888mel melanoma cell line transfected with siRNAs 6, 7, or 8 compared with the cells transfected with the control siRNA (Stealth RNAi Negative Control duplexes) without significant increase of apoptotic cells (data not shown). Similarly, a significant inhibition by the siRNA of cell proliferation of WM266 was observed (Supplementary Fig. S1A-C). These results indicated that FABP7 is involved in cell proliferation of the 888mel and WM266 melanoma cell lines.

The effect of FABP7 down-regulation by siRNAs on the invasion ability of the melanoma cell lines was then evaluated using Matrigel invasion assay. When invasion ability was measured by the number of cells migrated to the outer membranes in the chambers after 22-hour incubation, the invasion ability of 888mel transfected with siRNAs 6, 7, or 8 was decreased compared with the cells transfected with the control RNA as shown in a representative result of three independent experiments (Fig. 3D). Similar inhibition of the invasion ability of WM266mel was decreased (Supplementary Fig. S1D). Cell proliferation and invasion ability of SKmel23, 526mel, SKmel28, or A375mel transfected with the FABP7 siRNAs were not significantly decreased (data not shown). Although FABP5 (epidermal FABP), which was isolated from keratinocytes of psoriasis tissues, was reported to be involved in the enhanced production of VEGF in PC-3M prostate cancer cell line (24), no inhibition of VEGF and IL-10 production was observed in 888mel (data not shown).

To confirm the enhancing effects of FABP7 on cell proliferation and invasion, we also examined the effects of the FABP7 overexpression (Fig. 4A) on the FABP7-negative embryonic kidney cell line 293T by transient transfection of the FABP7 cDNA. Cell proliferation, measured by WST-1 (Fig. 4B) and Matrigel invasion assays (Fig. 4C), of the 293T cells with the overexpressed FABP7 was increased significantly. Together with the results from the siRNA experiments on melanoma cell lines described above, FABP7 seems to be involved in the enhanced cell proliferation and invasion of certain types of cell lines, including melanoma.

FABP7 is an immunogenic antigen that induces IgG antibody in patients with melanoma. Because five of eight selected molecules in this GeneChip screening were previously identified melanoma antigens recognized by either T cells or antibody, the immunogenicity of FABP7 was evaluated by detecting IgG antibody specific for FABP7 in sera from patients with various cancers and healthy individuals. IgG antibodies specific for the bacterial recombinant FABP7 protein were detected in sera from 8 of 31 (26%) melanoma patients as shown in a representative Western blot experiment (Fig. 5A,, lanes 6 and 9) but not in any sera from 11 brain tumor patients (Fig. 5A,, lanes 17-20) or from 30 healthy individuals (Fig. 5A,, lanes 12-15). The serum FABP7 IgG antibody was also detected against phage plaques expressing the recombinant FABP7 protein. In this phage assay, the FABP7-specific IgG was detected in sera from 14 of 25 (56%) melanoma patients and from 1 of 21 (4.8%) brain tumor patients but not in any sera from a total of 81 patients with various cancers and from 38 healthy individuals (Fig. 5B). The different positive rates between the Western blot analysis under denaturing conditions and the phage-plaque assay may be explained by the recognition of conformational epitopes or different sensitivity of these assays. Therefore, FABP7 is an immunogenic antigen that frequently induces an IgG antibody response in melanoma patients.

The ideal tumor antigens for immunotherapy are frequently overexpressed molecules that are involved in the formation of malignant phenotypes, including proliferation, survival, and invasion. In the present study, using GeneChip analysis, we identified FABP7, for the first time, as an immunogenic melanoma antigen that is frequently expressed and involved in the enhanced cell proliferation and extracellular matrix invasion of melanoma cells.

FABPs are intracellular proteins with a low molecular weight (14-16 kDa) and are involved in lipid metabolism (15). Human FABP7 (B-FABP) was first isolated from a fetal brain cDNA library, and a transcript was detected in adult human brain and skeletal muscle but not in other normal adult tissues (25). The murine FABP7 is specifically expressed in certain cells, including radial glial cells and immature astrocytes in the developing brain (26–28). In adult mice, some neuroglial cells, including cells in the olfactory bulb, and radial glial cells were reported to express FABP7 (16). These spatial and temporal expression indicated the functional role of FABP7 in the differentiation and migration of neuroglial cells (18).

In this study, FABP7 was shown to be expressed in both primary and metastatic melanoma. de Wit et al. (29) recently reported that one rarely metastasizing melanoma cell line, IF6, but not a frequently metastasizing melanoma cell line, Mel57, expressed the FABP7 protein and that the FABP7 mRNA was expressed in all 12 nevi and 5 of 6 primary melanoma and 1 of 10 metastatic melanoma by quantitative PCR analysis, suggesting the down-regulation of FABP7 in the progression of melanoma. Although we have observed the lower expression of the FABP7 mRNA in metastatic melanoma compared with primary melanoma, we were able to detect the expression of the FABP7 transcripts and proteins in many metastatic melanoma cell lines and tissues. The reason for this disparity between these results remains unclear but may be due to the sensitivity of the PCR assays. In addition, it remained to be further investigated why FABP7, possibly involved in the enhanced proliferation and invasion, was rather expressed at a lower level in metastatic melanoma.

Among the six melanoma cell lines that exhibited increased FABP7 expression, cell proliferation and Matrigel invasion of two melanoma cell lines, 888mel and WM266mel, but not other four melanoma cell lines, were significantly inhibited when FABP7 was down-regulated by the specific RNA interference. The reasons for the different involvement of FABP7 in the malignant phenotypes among melanoma cell lines remain unclear. Additional signals may be more important in the proliferation and invasion of the latter four melanoma cell lines. Although FABP7 may function through enhanced lipid metabolism via lipid receptors, such as peroxisome proliferator-activated receptor (PPAR), the exact functional mechanisms of the role of FABP7 in the enhanced cell proliferation and invasion remain to be investigated.

Some members of the FABPs have been reported to be overexpressed in cancers. FABP5 (epidermal or cutaneous FABP) is overexpressed in prostate, bladder, skin, and breast cancers (24, 30), and FABP4 (adipocyte FABP) is overexpressed in bladder cancer (31). FABP7 was also reported to be expressed in some human glioma cell lines and tissues (17), and its transcription may be regulated by phosphorylation of nuclear factor I (32). In a recent study of gene expression analysis using glioblastoma multiforme specimens, the expression of many genes that are normally expressed in neuroglial progenitor cells, including FABP7 and SOX, was observed in glioma cells, indicating that glioma cells exhibit the nature of progenitor cells (33). We have previously isolated SOX6 as a frequently expressed glioma antigen (34). FABP7 seems to be mainly expressed and functions at certain stages of developing neuroglial cells. The current finding of the FABP7 overexpression in melanoma may indicate that FABP7 is also expressed at a certain stage of developing melanocytes.

In the same gene expression analysis, FABP7 was also shown in a group of survival-related genes, for which expression correlated with poor survival of glioma patients. Overexpression of FABP7 in the low-expressing FABP7 glioma cell line by transfection with FABP7 cDNA facilitated the migrating ability of the glioma cells similar to our results obtained in melanoma. In contrast, a decrease or loss of FABP7 was reported in breast cancer (35). The functional role of FABP7 in cancer cells may differ among tissues through different use of downstream signals, including the PPAR isotypes (36).

In this study, FABP7 was shown to be an antigen immunogenic in many melanoma patients. It may be useful for immunotherapy as an antigen capable of inducing CD4⁺ helper T cells, particularly in those patients with positive IgG responses. Although FABP7 may be detected in some normal tissues, including developing glial cells and lactating mammary epithelial cells (37), the use of FABP7 as a target for CD8⁺ cytotoxic T cells in immunotherapy may not be excluded because of different susceptibility of normal and tumor tissues to T cells. Furthermore, the involvement of FABP7 in malignant phenotypes, including cell proliferation and invasion in some melanoma cells, suggests that FABP7 is an attractive antigen possibly due to the decreased likelihood of development of antigen loss variants in the immunotherapy for some patients with melanoma and possibly glioma.

The preferential expression of FABP7 in melanoma and glioma also supports the use of FABP7 in development of diagnostic methods, including serum diagnosis as a tumor marker at the mRNA and protein levels and immunohistochemical diagnosis of melanoma and glioma. The FABP7 protein is found in culture medium (18), suggesting that it might be secreted into the blood of patients. In addition, the serum IgG antibody may also be useful for prognostic diagnosis of patients with positive antibody because we have frequently observed the disappearance of IgG antibody specific for tumor antigens after successful treatment (21, 38). As shown in the glioma patients, it should be investigated whether FABP7 can be a prognostic marker for melanoma patients through enhanced cell proliferation and invasion ability.

In summary, by comparing expression profiles of melanoma with those of various normal tissues using GeneChip, we have identified a novel melanoma antigen, FABP7, which is frequently expressed in melanoma. In addition to its possible use in diagnosis, the involvement of FABP7 in cell proliferation and invasion of some melanoma cells and its immunogenic nature in many melanoma patients indicate that FABP7 could be a potential target for immunotherapy and molecular therapy in some patients with FABP7-expressing melanoma.

Grant support: Grant-in-Aid for Cancer Research from the Ministry of Health, Labor and Welfare; Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan; Sagawa Foundation for Promotion of Cancer Research; Japanese Foundation for Multidisciplinary Treatment of Cancer; Promotion and Mutual Aid Cooperation for Private Schools for Japan; Keio University School of Medicine Special Grant-in-Aid for Innovative Collaborative Research Projects; and Keio Gijuku Academic Development Funds.

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.