Published papers

We applied a strategy that utilized a combination of systematic gene expression analysis with various tissues and immunological detection with sera from melanoma patients to identify melanoma antigens expressed preferentially in melanoma and melanocytes. We selected 101 genes by comparing cDNA profiles obtained by GeneChip analysis of a highly pigmented melanoma cell line, SKmel23, primary cultured melanocytes, HUVECs cultured endothelial cells, keratinocytes, liver and stomach. After the additional selection with criterion of high registered frequency of each cDNA in melanocyte-related cDNA libraries in the NCBI database, 15 genes including 12 known melanocyte specific genes were identified. One of the remaining 3 genes, FCRL/FREB, encoding a member of the Fc receptor family that was previously reported to express in germinal center B cells, was found to express preferentially in melanocytes and melanoma tissues by RT-PCR and Northern blot analysis. The FCRL/FREB protein was detected in the cytoplasm of melanoma cells by staining with the murine polyclonal antibody and by transfection with GFP-fused FCRL/FREB cDNA. The bacterial recombinant protein was recognized by serum IgG antibody obtained from some patients with melanoma. These results suggest that FCRL/FREB may function in melanocytes and melanoma and may be useful for development of diagnostic methods for various pigment disorders and immunotherapy of melanoma.

To clarify the correlation of genomic alterations with clinical and histological features, we performed metaphase comparative genomic hybridization analysis on 20 primary cutaneous melanomas, which were obtained by laser capture or manual microdissection, and 16 melanoma cell lines. There were no differences in the average number of aberrations between acral melanomas (AM) and non-AM, although gains of 5q and 11q13 were more frequent (P = 0.05) and 10q loss was less frequent (P = 0.01) in AM than in non-AM. Although tumor thickness is considered a measurable estimate of clinical expression, there were no differences in the average number of aberrations among 4 groups, classified by thickness of the tumor. While the majority of aberrations were equally distributed among the 4 groups, 6p gains were found only in the thickest tumors. Patients with 6p or 1q gains had a lower overall survival rate than those without them (P = 0.0002 or P = 0.013). While gains of 1q, 2q, 3p, 3q, 7q, 20p, and 20q were more frequent in the cell lines than in the primary tumors (P < 0.01), losses of 6q, 9p, 10p, and 10q were equally found in both cell lines and primary tumors. The present study showed that chromosomal aberrations had already occurred in the thinner tumors, and that 6p and 1q gains may be a prognostic factor.

Objectives. To examine whether autoantibodies against bone morphogenetic protein receptor-II (BMPR-II) or activin receptor-like kinase 1 (ALK-1) are associated with pulmonary arterial hypertension (PAH) in patients with mixed connective tissue disease (MCTD). Methods. We studied sera from 37 MCTD patients with or without PAH, six patients with idiopathic PAH, and 30 healthy controls. Circulating anti-BMPR-II and anti-ALK-1 antibodies were detected using immunoprecipitation of recombinant antigens generated by in vitro transcription/translation and indirect immunofluorescence of cultured cells that were induced to express these antigens by gene transfer. Anti-BMPR-II antibodies were further examined by immunoprecipitation and immunoblotting using a recombinant fragment of the extracellular domain of BMPR-II. Results. Serum anti-BMPR-II and anti-ALK-1 autoantibodies were not detected in MCTD patients irrespective of the presence or absence of PAH, or in patients with idiopathic PAH. Conclusions. Our finding does not support the hypothesis that autoantibody-mediated dysregulation of signals through BMPR-II or ALK-1 contributes to the development of PAH in patients with connective tissue diseases.

Antiphospholipid syndrome (APS) is an autoimmune prothrombotic disorder in association with autoantibodies to phospholipid (PL)-binding plasms proteins, such as β2-glycoprotein I (β2GPI). We have recently found that CD4+ T cells autoreactive to β 2GPI in patients with APS preferentially recognize a cryptic peptide encompassing amino acid residues 276-290 (p276-290), which contains the major PL-binding site, in the context of DR53. However, it is not clear how previously cryptic p276-290 becomes visible to the immune system and elicits a pathogenics autoimmune response to β2GPI. Here we show that presentation of a disease-relevant cryptic T-cell determinant in β2GPI is induced as a direct consequence of antigen processing from β2GPI bound to anionic PL. Dendritic cells or macrophages pulsed with PL-bound β2GPI induced a response of p276-290-specific CD4+ T-cell lines generated from the patients in an HLA-DR-restricted and antigen-processing-dependent manner but those with β2GPI or PL alone did not. In addition, the p276-290-reactive T-cell response was primed by stimulating peripheral blood T cells from DR53-carrying healthy individuals with dendritic cells bearing PL-bound β2GPI in vitro. Our finding is the first demonstration of an in vitro mechanism eliciting pathogenic autoreactive T-cell responses to β2GPI and should be useful in clarifying the pathogenesis of APS.

To identify new melanoma antigens using systematic gene expression analysis combined with rapid screening of patient sera for immunogenicity, a serial analysis of gene expression (SAGE) method was applied to profile transcripts in a highly pigmented melanoma cell line SKmel23. 25,997 SAGE tags consisting of 10,382 unique transcripts were sequenced. This melanoma SAGE library was compared with a testis SAGE library and the colon SAGE database, and to the cDNA database obtained by random sequencing of a melanocyte cDNA library. Among the 15 tags finally selected with criteria of preferential expression on melanoma and melanocytes at relatively high frequency, two tags were further analyzed for their structure and immunogenicity. One was identified as PAX3, and its isoform, PAX3d, was found to be dominantly expressed in melanoma and melanocytes. The other was derived from a novel gene and its full-length cDNA clone was isolated. Preferential expression of these genes in melanoma and melanocytes was confirmed by RT-PCR and Northern blot analysis. The recombinant bacterial PAX3d protein was recognized by serum IgG from some patients with melanoma and Vogt-Koyanagi-Harada (VKH) disease but not from healthy individuals, indicating that PAX3d is a new melanocyte-specific antigen immunogenic in patients with melanoma or VKH disease. The authors report two melanocyte/melanoma-specific molecules, which may be useful for development of diagnosis or treatment of these pigment disorders. In addition, a system using SAGE and immunoscreening with patients' sera is shown to be an efficient method for the systematic identification of tumor antigens.

Increase of Skp-2, which is involved in the degradation of cell cycle regulators including p27Kip1, p21 and c-myc, is one of the important mechanisms for dysregulation of cell cycles in various cancers. We applied RNA interference (RNAi) for Skp-2 by using HIV-lentiviral or adenoviral vectors for a human small-cell lung carcinoma cell line with increased Skp-2 to evaluate RNAi strategy for cancer gene therapy. HIV-lentivirus-mediated RNAi for Skp-2 resulted in efficient inhibition of the in vitro cell growth of cancer cells with increased Skp-2 through the increase of p27Kip1 and p21, but no significant effect on the growth of cells without high Skp-2 expression. Furthermore, intratumoral administration of adenovirus siRNA vector for Skp-2 efficiently inhibited growth of established subcutaneous tumor on NOD/SCID mice. These results indicate that the Skp-2 RNAi may be a useful strategy for gene therapy of cancers with high Skp-2 expression.

Identification of human melanoma antigens by various molecular biological and immunological techniques and evaluation of tumor reactive T cells in patients with the identified tumor antigen and HLA tetramer technology, not only provided us more profound understanding of anti-tumor immune responses in human, but also led to reveal basic problems in each step towards immunological tumor rejection, including systemic suppressive mechanisms such as regulatory T cell induction and local inhibitory environment in tumors. Based on these results obtained from the basic and clinical researches, various improvements have been applied for immunotherapy, including active immunization with modified antigenic peptides and recombinant virus, T cell adoptive transfer with lymphodepletive pretreatment, and administration of anti-CTLA-4 Ab, although further improvement is necessary. The translational research performed on melanoma, would facilitate development of immunotherapy for other cancers.

By screening a human testis cDNA library with glioma patients' sera, we isolated a transcriptional factor, SOX6. Here, we analyzed SOX6 expression in gliomas having a range of malignancy grades using immunostaining. Murine Sox6 is a transcriptional factor that is specifically expressed in the developing central nervous system and in the early stages of chondrogenesis in mouse embryos. The reverse transcription-polymerase chain reaction (RT-PCR) revealed that the SOX6 gene was more highly expressed in glioma tissues and fetal brain than in normal adult brain and other cancer cells, except melanoma cells. Immunohistochemical analysis with the anti-SOX6 antibody showed that all the glioma tissues analyzed (14 glioblastomas, 14 anaplastic astrocytomas, 3 anaplastic oligoastrocytomas, 5 diffuse astrocytomas, 1 oligodendroglioma, and 1 pilocytic astrocytoma) expressed SOX6 in tumor cells, but only a few SOX6-positive cells were detected in nonneoplastic tissues from the cerebral cortex. These results indicate that the developmentally regulated transcription factor SOX6 may be a potential diagnostic marker for gliomas.

We previously demonstrated that the developmentally regulated gene, SOX6, is strongly expressed in glioma cells and in the fetal brain, but only faintly in the normal adult brain. Recent studies have indicated that brain tumor cells may share antigens, signaling systems, and behavior with neural stem/progenitor cells. To test the validity of this proposition, we analyzed the expression of SOX6 in various human central nervous system (CNS) tumors. Immunohistochemical analysis revealed that astrocytic and oligodendroglial tumors expressed SOX6; neuronal-glial cell tumors (central neurocytoma) and embryonal tumors (medulloblastoma), which arise from multipotential stem cell precursors, also showed a high intensity of SOX6 staining. In contrast, ependymal tumors (ependymoma and subependymoma), meningioma, and schwannoma, which are all well differentiated tumors, showed either no staining or only faint staining for SOX6. These results suggest that SOX6 may be expressed in bipotential or multipotential cells capable of neuronal and glial differentiation, but not in fully differentiated cells. SOX6 may be a useful marker for the diagnosis of tumors arising from immature bipotential cells that may differentiate into neuronal and glial cells.

Objective. To detect and characterize the autoreactive CD8+ T cells to major histocompatibility complex class 1 chain-related gene A (MICA), a stress-inducible antigen preferentially expressed on the epithelium and endothelium, in patients with Behçet's disease (BD). Methods. A candidate for the antigenic MICA peptide was selected based on its predicted binding affinity for HLA-B51 and proteasomal cleavage sites. Peripheral blood T cells from 14 patients with BD and 15 healthy controls were repeatedly stimulated with the MICA peptide, and the specific T cell response was measured by peptide-induced interferon-γ. Cytotoxic T lymphocyte activity was examined by chromium-51 release from an BLA-151-transfected B cell line in the presence of the MICA peptide. Results. A 9-mer peptide AAAAAIFVI (termed MICA transmembrane [MICA-TM]) was selected as a candidate for the antigenic peptide presented by HLA-B51. A specific T cell response to MICA-TM was detected in 4 patients with BD (29%) but in none of the 15 healthy donors. All 4 responders had HLA-B51 and active disease, and the specific T cell response was lost after the BD-related symptoms disappeared. The MICA-induced T cell response was specifically inhibited by anti-HLA class I antibody or by CD8+ cell depletion. MICA-reactive T cells recognized an HLA-B51-transfected B cell line pulsed with MICA-TM or a B cell line transfected with both HLA-B51 and MICA in the absence of exogenous peptides. Finally, MICA-stimulated T cell lines lysed the HLA-B51-expressing B cell line in the presence of MICA-TM. Conclusion. HLA-B51-restricted cytotoxic T lymphocytes autoreactive to MICA may be involved in the pathogenesis of BD.

Recent identification of human tumor antigens recognized by T cells made it possible to analyze anti-tumor immune responses quantitatively and qualitatively in human and to develop immunotherapy in a more scientific way. The identification methods have been being improved during the past 10 years. When tumor reactive T-cells are available, DNA expression cloning or direct peptide sequencing from HPLC fractions extracted from MHC class I/peptide complex can be performed. However, it is difficult to establish tumor reactive T cells in many tumors. In those cases, we can apply the reverse immunology approach, in which tumor antigen candidates are first identified by various genetic or immunological methods, including cDNA subtraction between cancers and various tissues with cDNA profiles obtained through systematic gene analysis such as DNA chip/microarrays, SAGE, or public gene databases, or cDNA expression cloning with patients' sera (SEREX). Then antigenicity of the identified candidates is confirmed by demonstration of T cell induction against candidates through in vitro T cell induction or immunization of HLA transgenic mice. We have made various modifications on these techniques. Identification of human tumor antigens in various cancers will lead to development of more effective immunotherapy in the future.

Comparative genomic hybridization (CGH) using 40 cell lines derived from malignant melanomas (MMs) revealed frequent amplification at 7q33-q34 containing BRAF gene, which often is mutated in MM. We found this gene to be amplified to a remarkable degree in the MM cell lines that exhibited high-level gains at 7q33-q34 in CGH. Among 40 cell lines, the eight lines that revealed neither BRAF nor NRAS mutations showed even higher levels of BRAF mRNA expression than the 32 mutated lines, although DNA amplification at 7q33-q34 was not detected in every lines overexpressing BRAF. MM cells that carried wild-type BRAF and NRAS showed constitutive over-expression of B-Raf protein and phosphorylation of extracellular signal-regulated kinase 1/2 (ERK1/2), even after serum starvation. Not only downregulation of the endogenously overexpressed wild-type B-Raf by antisense oligonucleotide but also a treatment with an inhibitor of mitogen-activated protein kinase kinase (MAPKK, MEK) reduced phosphorylated ERK1/2 and cell growth, whereas the exogenously expressed wild-type B-Raf promoted cell growth in MM cells. Our results provide the evidence that overexpression of wild-type B-Raf, in part but not always as a result of gene amplification, is one of the mechanisms underlying constitutive activation of the MAPK pathway that stimulates growth of MM cells.

Human tumor antigens recognized by T cells have been identified by means of various molecular biological and immunological methods, including cDNA expression cloning with patients' T cells and antibodies, cDNA subtraction using RDA and PCR differential display, systematic gene analysis such as DNA sequencing, CGH, DNA chip/microarray and SAGE, in vitro T cell induction and immunization of HLA transgenic mice. The identification of human tumor antigens has led to a better understanding of the nature of tumor antigens, anti-tumor immune responses in patients before and after immunotherapy, and tumor escape mechanisms. The information obtained from these researches has enabled us to develop and improve immunotherapy by attempting to overcome the identified problems, including intrinsically low immunogenicity of tumor antigens and several escape mechanisms, such as regulatory T cell induction. The existence of immunogenic unique antigens derived from genetic alterations in tumor cells, and the varied immunogenicity of shared tumor antigens among patients due to differing expression in tumor cells and immunoreactivity of patients, indicates that individualized immunotherapy should ideally be performed. The identified antigens will also be useful for development of diagnostic methods and molecular targeting therapy for cancer.

To identify the CpG islands differentially methylated in human glioma, we performed restriction landmark genomic scanning with a CpG methylation-sensitive enzyme. We found 12 spots, the intensity of which was entirely lost or decreased in both the human glioma tissues examined as compared with that in matched normal lymphocytes, indicating aberrant methylation of these CpG islands in gliomas. The expression of RFX1, one of the genes associated with the methylated CpG islands, was frequently decreased in human glioma cell lines and tissues. We also demonstrated that the isolated CpG island located in the seventh intron of the RFX1 gene had enhancer activity and was hypermethylated in all of the glioma tissues and cell lines analysed, but not in normal brains or lymphocytes. Treatment of glioma cells with a demethylating agent, 5-azacytidine, resulted in the expression of RFX1, indicating that the silencing of the RFX1 gene may be attributable to its methylation. RFX1 has been implicated in transcriptional downregulation of the proto-oncogene c-myc. By expression of the RFX1 gene, the cellular proliferative activity of glioma cells was suppressed. Taken together, these results suggest that the RFX1 gene may be epigenetically silenced in human gliomas and involved in glioma tumorigenesis.

Background Typical vascular features of systemic sclerosis include low capillary density and vascular obliteration. The formation and repair of blood vessels in adults involve vasculogenesis, which is mediated through the recruitment of bone-marrow-derived circulating endothelial precursors (CEP). We investigated whether vasculogenesis is impaired in patients with systemic sclerosis. Methods Peripheral blood was obtained from 11 patients with systemic sclerosis, 11 with rheumatoid arthritis, and 11 healthy controls. Factors potentially affecting the CEP number were matched among the three groups. CEP (identified as circulating cells positive for CD34, CD133, and the type 2 receptor for vascular endothelial growth factor) were quantified by cell sorting and three-colour flow cytometry. The circulating concentrations of angiogenic factors were measured by ELISA. The potential of CEP to differentiate into endothelial cells was assessed by the upregulation of von Willebrand factor after in-vitro maturation treatment. Findings The absolute number of CEP was much lower in patients with systemic sclerosis than in patients with rheumatoid arthritis or healthy controls (median 274 [IQR 178-395] vs 1154 [653-1524] and 1074 [713-1186] per 20 mL peripheral blood, respectively; p<0·0001 by Kruskal-Wallis test. Paradoxically, circulating concentrations of most angiogenic factors were significantly higher in patients with systemic sclerosis than in healthy controls. The proportion of CEP that differentiated into endothelial cells was significantly lower in patients with systemic sclerosis than in healthy controls (p<0·0001, Mann-Whitney test). Interpretation Insufficient vascular repair machinery due to defective vasculogenesis might contribute to vasculopathy in systemic sclerosis. Relevance to practice As well as providing an important insight into the pathogenesis of this disorder, these findings suggest that dysregulated vasculogenesis might be important in other disorders with abnormalities in vascular formation, including those with excessive formation of new vessels such as cancer and those with deficient vessel formation such as atherosclerosis. Circulating endothelial precursors could be a novel target for therapeutic strategies for ischaemic complications in patients with systemic sclerosis.

Oncogenic mutations of molecules involved in the mitogen-activated protein kinase (MAPK) pathways provide signals mediating both tumor growth and invasion in various cancers including melanomas. BRAF somatic mutations, found in 66% of melanomas, have NIH3T3 transforming ability with the elevated kinase activity in vitro. We attempted to mediate RNA interference (RNAi) with HIV lentiviral vectors specific for either wild type or the most frequently mutated form of BMAF (V599E) in 10 melanoma cell lines, and found that RNAi inhibited the growth of most melanoma cell lines in vitro as well as in vivo, which was accompanied by decrease of both BRAF protein and ERK phosphorylation. Interestingly, the mutated BRAF (V599E)-specific siRNA inhibited the growth and MAPK activity of only melanoma cell lines with this mutation. Furthermore, BRAF RNAi inhibited matrigel invasion of melanoma cells accompanied with a decrease of matrix metalloproteinase activity and β1 integrin expression. These results clarify that the mutated BRAF (V599E) is essentially involved in malignant phenotype of melanoma cells through the MAPK activation and is an attractive molecular target for melanoma treatment. The lentivirus-mediated RNAi specific for oncogenic mutations may be a powerful technique for gene therapy of cancer.

Background. RNA interference (RNAi) is a phenomenon in which expression of an individual gene can be specifically silenced by introducing a double-stranded RNA, one complementary to the gene, into cells. This phenomenon can be observed in mammalian cells when small interfering RNAs (siRNAs) are used, and is receiving attention as the most powerful tool for reverse genetics in the post genome era. Several groups have developed vector-based siRNA-expression systems that can induce RNAi in living cells. Methods. We describe here a comparative analysis of various siRNA-expression systems, in which we examined the effects of stem length, loop sequence and insertion of mutation(s) and/or bulges in the stem sequence on silencing effects and on the stability of the vectors. Results. As a result of the comparative analysis, we determined the following optimized siRNA-expression system: U6 promoter-driven hairpin-type dsRNA with 21-nt stem length, three to four mutations in the sense strand only, and the optimized 9-nt loop sequence, derived from microRNA. Moreover, we demonstrate that the siRNA-expression system with a tetracycline-regulated U6 promoter(s) could have the potential to control RNAi in cells, and that the HTV vector-mediated transfer of an siRNA-expression cassette into cells resulted in efficient silencing of a target gene at a multiplicity of infection as low as five. Conclusion. The mutated hairpin siRNAs and their genetically stable coding vectors could be very useful for gene knockdown experiments, and could further benefit gene therapy using RNAi.

SEREX (serological identification of tumor antigens by recombinant cDNA expression cloning) is a powerful cDNA expression cloning technique for isolation of tumor antigens recognized by T cells using serum antibodies from cancer patients from whom it is difficult to raise tumor-reactive T cells. To efficiently identify tumor antigens, various modifications have been attempted for SEREX because of frequent isolation of antigens not associated with immune responses to tumor cells. These include use of a cDNA library made from testis, use of sera from patients who are immunized with autologous tumor-derived products who demonstrate tumor regression after treatment, and use of human IgG containing mice sera from immunodeficient mice implanted with human tumor tissues.

The MAGE-1 gene codes for a tumor-specific antigen, MZ2-E, that elicited a cytotoxic T-lymphocyte response in the melanoma patient from whom it was derived. We have developed a simplified method, using polymerase chain reaction amplification of exon 3 followed by restriction enzyme pattern analysis, to distinguish expression of the MAGE-1 gene from MAGE-2 and MAGE-3, other members of this gene family. MAGE-1 mRNA was expressed in 53% of 17 melanoma lines, two of seven Epstein-Barr virus-transformed B-cell lines, and 2 of 5 breast cell lines including a line established from normal breast epithelium. MAGE-1 is not likely to be the common melanoma antigen recognized by the other HLA-A1- or HLA-A2-restricted cytotoxic T-lymphocytes examined in this study, but the fact that it is expressed in about 50% of melanoma cell lines makes it a reasonable target for the immunotherapy of patients bearing HLA-A1.