Synovial sarcomas (SSs) account for approximately 7% to 10% of all human soft tissue sarcomas and predominantly affect the extremities but can occur in any part of the body, including the head and neck [20, 49]. Despite its name, SS arises from tissues other than normal synovial lining. Ultrastructural studies of epithelium-like cells in biphasic synovial sarcomas revealed characteristics of a true glandular epithelium, which were not observed in either superficial or deeper cells of synovium [21]. Further, immunohistochemical studies have demonstrated keratins and epithelial membrane antigen-positive epithelium-like cells in biphasic synovial sarcomas and unlike those found in the lining cells of synovium [43]. Therefore, the name “synovial sarcoma” is believed inappropriate for this class of tumor, and the cell of origin of SS is still unknown [35].

Genetically, SSs are characterized as tumors harboring the reciprocal chromosomal translocation t(18;X) producing an SYT-SSX fusion gene [31]. The myf5 gene is an important transcription factor regulating the differentiation of cells along the myogenic lineage [5]. A recent report using a conditional knock-in mouse model using the promoter of the myf5 gene as a driver of the SYT-SSX gene demonstrated sarcomatous tumors strongly resembling human SS [22]. This suggested the precursor of SS was cells of the myogenic lineage. The global gene expression profile of the SS-like tumors that developed in these knock-in mice, however, differed considerably from that reported in human SS [20], possibly as a result of the difference in cell origin.

We previously performed gene expression profiling of 47 soft tissue sarcomas using a genome-wide cDNA microarray and reported SS was clustered into the same group with malignant peripheral nerve sheath tumor (MPNST), the precursor cells of which are neural crest-derived cells [36]. Among the products of the 26 genes whose expression was upregulated in SS, collagen Type IX is one of the major components of cartilage [17], neurofilament heavy polypeptide (NFH) has important roles in the assembly of filaments and in the radial growth of axons [28], and endothelin 3 is an important signal for melanocytes and enteric neurons [7]. In addition, the expression of the chondrogenic transcription activator SOX9 [9] is upregulated in SSs [2]. These results suggest the cellular origin of SS is neural crest-derived cells capable of differentiating in multiple directions. Neural crest-derived cells are precursors of neuroblastomas [41, 42], malignant melanomas [18], and the Ewing family of tumors [46], which are sarcomas harboring a tumor-specific fusion oncoprotein (EWS-Fli1 or EWS- ERG) [34]. SS is known to develop at unusual anatomic sites such as the pleurae [6], visceral organs [10], bone marrow [24], and peripheral nerves [14]. Molecular diagnoses using the SYT-SSX fusion transcript as a marker have proved some renal tumors previously diagnosed as different diseases were in fact SSs [4]. This diverse distribution may favor the notion that SS originate from neural crest-derived cells, which migrate into many types of tissues.

The goal of this study was to demonstrate the neural origin of SS using primary tumors and cell lines and also to apply the differentiation induction therapy for the treatment of SS in combination with a growth signal inhibitor.

Using SS tumor tissues and cell lines, the hypothesis of neural origin of SS was investigated by mRNA and protein expression analyses of a set of neural tissue genes such as nestin, which is expressed in undifferentiated cells [42]. Also, several neural inducers were used to induce the neural phenotype in cell lines, and the therapeutic application was examined by in vitro and in vivo growth inhibition assay.

Tumor tissues were obtained at either biopsy or resection surgery and kept at −80°C. Informed consent was obtained from each patient and tumor samples were approved for analysis by the Ethics Committee of the Faculty of Medicine, Kyoto University. Five human SS cell lines (YaFuSS, HS-SY-II, SYO-1, Fuji, 1273/99) were used in this study. YaFuSS and HS-SY-II cells have the SYT-SSX1 fusion gene, and the others have the SYT-SSX2 fusion gene (data not shown). The properties of YaFuSS were described previously [26]. HS-SY-II was a gift from H. Sonobe (Kochi Medical School, Japan) [44], SYO-1 from A. Kawai (National Cancer Center, Japan) [29], Fuji from S. Tanaka (Hokkaido University, Japan) [37], and 1273/99 from O. Larsson (Karolinska Institute, Stockholm, Sweden). NMS-2 (MPNST) was a gift from A. Ogose (Niigata University, Japan) [25]. Saos2 (osteosarcoma), HT1080 (fibrosarcoma), COLO205 (colon cancer), SW480 (colon cancer), and NT2/D1 (neurogenic clone of pluripotent embryonal carcinoma [3]) were purchased from the American Type Culture Collection (Manassas, VA). Adult human mesenchymal stem cells were obtained from BioWhittaker (Walkersville, MD) and maintained in MSCGM™ medium (BioWhittaker). Tumor cell lines Fuji, SW480, and COLO205 were maintained in RPMI 1640 medium (Sigma-Aldrich, St Louis, MO) with 10% fetal bovine serum (FBS; HyClone, Logan, UT). Other tumor cell lines were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM; Sigma) with 10% FBS. OPTI-MEM^® I, which contains insulin and transferrin as protein supplements (Invitrogen, Carlsberg, CA), was used in serum-free cultures.

Anti-neuron-specific class III β-tubulin antibody (Tuj-1) and recombinant human bone morphogenetic protein 4 (rhBMP4) were purchased from R&D Systems (Minneapolis, MN). Recombinant human fibroblast growth factor 2 (rhFGF2) was obtained from Oncogene Research Products (San Diego, CA).

The expression of neural tissue-related genes was analyzed by reverse transcription (RT)-PCR in 29 sarcoma tumor samples, including 18 SSs and also 12 tumor cell lines including five SS cell lines. Total RNA was extracted using TRIzol^® reagent (Invitrogen) following the manufacturer’s instructions. After treatment with DNase I (Nippon Gene, Osaka, Japan), 1 mg of total RNA was reverse-transcribed into single-stranded cDNA using the oligo(dT) primer and SuperScript™ II reverse transcriptase (Invitrogen). PCR was performed using 1 mL of RT product in a final volume of 25 mL containing 20 pmol each of the sense and antisense primers, 2.5 mmol/L MgCl₂, 0.2 mmol/L of each dNTP, and 1 unit of rTaq polymerase (Toyobo, Osaka, Japan). All PCRs were performed using GeneAmp^® 9700 (PE Applied Biosystem, Foster City, CA). Information on the primers is available on request.

To investigate whether SS cell lines are able to differentiate into the neural cell lineage, cells were treated with ATRA, rhFGF2, or rhBMP4, all known as inducers of neural differentiation [12, 27, 33]. Cytoplasmic staining of Tuj-1 and the mRNA expression of MASH1 were used as markers for neural induction by these treatments. For neurogenic induction, cells (2.5 × 10³) in DMEM with 10% FBS were seeded in an eight-well chamber slide (Permanox^® type; Nalgene Nunc International, Rochester, NY) and cultured overnight. Then the medium was replaced with serum-free OPTI-MEM^® I medium with vehicle (0.1% dimethyl sulfoxide), all-trans-retinoic acid (ATRA) (1 mmol/L), rhFGF2 (100 ng/mL), or rhBMP4 (50 ng/mL) and cultured for 14 days (for rhBMP4) or 7 days (for the others). The cells were fixed with 4% paraformaldehyde for 20 minutes at room temperature, permeabilized, blocked with phosphate-buffered saline (PBS) containing 0.1% Triton^® X-100 (Sigma) and 0.5% bovine serum albumin for 1 hour, and incubated with Tuj-1 (1:500) followed by tetramethylrhodamine isothiocyanate-conjugated antimouse immunoglobulin (1:100; DakoCytomation, Glostrup, Denmark). After three washes with PBS, the cells were incubated with 4′,6-diamidino-2-phenylindole. To check the expression of mammalian achaete-scute homolog 1 (MASH1), cells were treated with 1 mmol/L ATRA (for 7 days), 100 ng/mL rhFGF2 (for 48 hours), or 50 ng/mL rhBMP4 (for 48 hours) and total RNA was extracted as described previously.

We have reported the FGF signaling pathway is important for growth in SS cells [26]. Treatment with a FGF receptor (FGFR) induced growth arrest but failed to induce cell death, either necrotic or apoptotic [26]. For in vitro treatment with ATRA and an FGFR inhibitor, cells (1.0 × 10⁴) were seeded on collagen-coated 60-mm dishes, and after their adhesion to the dishes, DMEM with 10% FBS was exchanged for DMEM with 10% FBS containing ATRA (1 or 10 μmol/L) and/or a FGFR-specific inhibitor, PD166866 (100 or 500 nmol/L) provided by Pfizer Global Research and Development (Groton, CT). In the control, DMEM with 10% FBS was changed every 2 to 3 days. The counting of cells was performed once every 2 days until Day 17.

To investigate whether these growth-inhibitory effects occur in vivo, treatment with ATRA of human SS xenograft in nude mice was performed using SYO-1 cells. For in vivo treatment with ATRA, the procedures were performed as previously described [38]. Briefly, SYO-1 (5 × 10⁶) cells suspended in 100 mL of PBS were subcutaneously injected into the hind flank of male BALB/c nu/nu athymic mice at 5 weeks of age (Japan SLC, Hamamatsu, Japan). Mice were randomly assigned to ATRA-treated, vehicle-treated (sesame oil), and untreated groups (two mice per group). Starting on Day 2, the day after the inoculation of SYO-1 cells (Day 1), ATRA (20 mg/kg) was given to the mice by mouth through a gastric tube in 0.1 mL of purified sesame oil (Sigma) six times a week. Tumor size was measured with Vernier calipers, and the volume was calculated as π/6 x length x width x height on Days 3, 7, 10, 14, and 17.

The data are shown as mean ± standard deviation. The one-factor analysis of variance (ANOVA) test was used to determine differences in the effect of ATRA in vitro. The repeated-measure ANOVA test was used to determine the difference in the effect of ATRA in vivo and in the effect of a combination of ATRA and PD166866 in vitro. All statistical analyses were performed using Statcel software (OMS Publishing Inc, Tokyo, Japan).

Synovial sarcoma is a rare sarcoma of unknown histologic origin. Despite its name, evidence suggests SS arises from tissues other than normal synovial lining. Based on the gene expression profile data in our previous study [36], we hypothesized SS stemmed from cells in the neural lineage, and the neural induction may be applied as a new therapeutic modality for SS.

Our study has a number of limitations. First, the mRNA expression analyses in this study were qualitative rather than quantitative, and tumors other than SS also expressed several neural tissue-related genes (Figs. 1, 2). The expression of particular genes such as p75NTR and CRABP1, however, was observed only in SS and MPNST, suggesting the common characteristics of these two tumors. There was a considerable heterogeneity with respect to the expression profile of neural tissue genes among SS, irrespective of morphologic or genetic classification (Figs. 1, 2). Quantitative analyses using a large number of tumor tissues are required to draw definite conclusions. Second, we used Tuj-1 as an indicator of neural differentiation, which was widely used as a neural marker, although it is not specific in cells in neural lineage. Additional experiments using other markers are now undertaken. Finally, the in vivo growth inhibition assay in this study was a preliminary one using only one cell, because SYO-1 was the only line producing tumors in nude mice constantly. Although no severe side effect was observed in this short-term assay, more intensive analyses using other cell lines will be required to confirm the efficacy of ATRA treatment.

Consistent with the results of our previous study [36], here we demonstrated further evidence of the neural origin of SS, particularly neurons, but not glia, and the differentiation induction by ATRA inhibited the growth of SS in vitro and in vivo. Growing evidence suggests the cellular origins of malignant tumors are tissue stem cells, making the neural crest stem cells a possible candidate for the precursor of SS [1, 11, 23, 30]. Neural crest stem cells are regarded as pluripotent cells capable of differentiating into multiple cell lineages, including chondrocytes, osteocytes, neuron, glia, and melanocytes [13, 32]. We have tried to induce the differentiation of SS cells into these lineages and found some SS cells could differentiate in more than one direction but not multiple directions (data not shown). It has been proposed that in between stem cells and terminally differentiated cells, some cells have the ability to differentiate into two or more lineages, some of which might be the precursor of SS cells [32]. Experiments using knock-in mice with a myf5-driven SYT-SSX suggested a myogenic origin of SS [22], but there are neural cells expressing the myf5 gene in muscle tissues [16]. Knock-in mice carrying the SYT-SSX gene driven by the promoter of an appropriate neural crest marker may clarify this point.

Redifferentiation therapy has been widely investigated as an alternative to standard chemotherapy. Retinoic acid has fundamental effects on cellular differentiation, which may accompany the growth arrest of each cell [45]. Retinoic acid and its derivatives are among the most intensively analyzed redifferentiating agents and are used for the treatment of hematologic and solid malignant tumors [38, 39, 50]. Most of the successes with ATRA therapy has been obtained in patients with acute promyelocytic leukemia characterized by the reciprocal chromosomal translocation t(15;17) generating the fusion oncoprotein PML-RARα [38]. More than 90% of patients with acute promyelocytic leukemia undergo complete disease remission when treated with ATRA plus chemotherapy [19]. The prominent effect of ATRA on this type of tumor, however, depended on the ability of ATRA to directly inhibit the function of the PML-RXRα oncoprotein and was not necessarily related to its redifferentiation property [19]. Experimental studies report growth-inhibitory effects on osteosarcoma [47], rhabdomyosarcoma [40], and AIDS-associated Kaposi sarcoma [15]. Approximately 30% of patients with AIDS-associated Kaposi sarcoma have a substantial clinical response to ATRA treatment [8], and one case report suggested ATRA treatment combined with interferon-α was effective against inoperable metastatic osteosarcoma [48]. Our data demonstrate SS cell lines were far more sensitive to treatment with ATRA than the other types of tumors analyzed, and NMS-2, a cell line derived from MPNST, which in clinical practice is resistant to all chemotherapeutic treatments, also showed marked sensitivity to treatment with ATRA. The growth-inhibitory effect of ATRA in vivo, however, was only partial because a high dose could not be administered as a result of toxic side effects (data not shown). Combined with the inhibition of the FGF signal, which is one of the essential growth signals in SS, ATRA seemed as effective at lower concentrations as at high concentrations, suggesting the possibility of a new therapeutic target for SS.