We have analyzed the transcriptional effects of SYT-SSX1 expression in bone marrow derived human mesenchymal stem cells isolated from pediatric or adolescent patients. These cells may provide an appropriate model to study synovial sarcoma development, based on the generally recognized notion that SS originates from as yet unidentified pluripotent stem cells capable of mesenchymal and neuroectodermal differentiation. The only available transgenic model of SS 
thus far, suggests that development of SS is linked to the expression of the early myogenic marker Myf5. Co-expression of early markers for different tissue lineages has been observed in MSCs 
, without necessarily being associated with loss of plasticity. Human MSC expressed Myf5
with up to a 30-fold population-dependent transcript level variation (data not shown). However, we could not establish whether the observed differences in expression were due to varying enrichment of a specific sub-population or to homogeneous, donor specific, traits.
display a transcriptional profile with significant similarity to the gene expression signature of synovial sarcoma, supporting the notion that these cells could have features common to the pluripotent mesenchymal cell of origin of SS. Analysis of the transcriptional profile of hMSCSYT-SSX1
revealed overexpression of genes related to nervous system development, suggesting that SYT-SSX1 may exert some degree of pressure toward neuronal differentiation in mesenchymal stem cells. Several reports 
, including a recent proteomics-based study 
, have shown a similar gene expression profile among clear cell sarcoma, synovial sarcoma, and MPNST supporting the assumption that these three tumors may be derived from, or differentiate toward, neuroectodermal cells. Rare cases of synovial sarcoma, identified by SYT-SSX expression, have in fact been reported to express neural immunomarkers 
Single population analysis according to Gene Ontology (GO) annotation revealed remarkable variation among batches. The observed variation involved single genes whose overexpression has been associated with synovial sarcoma, including BCL2 and IGF2 as well as clusters of genes implicated in cell trafficking and differentiation. Thus, some populations of MSC appeared to be more permissive than others for SYT-SSX-induced changes in the expression of genes relevant to fundamental requirements for normal and cancer stem cell biology. Similarly, single population analysis revealed greater similarity of some MSCSYT-SSX1 population transcriptomes than others to SS gene expression signatures, supporting the hypothesis that features which distinguish independent hMSC isolates and contribute to their heterogeneity may be key for SYT-SSX function. Our present observations suggest that the nature of these putative features may, at least in part, be epigenetically determined.
Transcriptome analysis of hMSCSYT-SSX1 showed that a major effect of SYT-SSX in hMSCs involves changes in the expression of epigenetically regulated genes, including imprinted genes, genes that contain CpG island in their TSS and chromatin related genes. Epigenetic de-regulation has been suggested to be a central effect of the aberrant expression of SYT-SSX and a possible mechanism underlying synovial sarcoma formation. The present transcriptome analysis of hMSC expressing SYT-SSX strongly supports this notion.
Consistent with the variability of MSCSYT-SSX1 transcriptome relatedness to SS signatures and that of GO term overrepresentation, single population analysis limited to datasets of epigenetically regulated genes showed marked qualitative and quantitative differences among the four hMSC isolates, the most striking being the divergent effect of SYT-SSX on the expression of imprinted genes. It is therefore conceivable that epigenetic features displayed only by some hMSC populations permit SYT-SSX to affect expression of genes implicated in biological functions relevant to stem cells and SS. We therefore sought divergent epigenetic characteristics among the MSC populations that may explain the significant variations observed in the transcriptional effect of SYT-SSX.
Assessment of the H19/IGF2
cluster provided support for our hypothesis. IGF2
is considered to be one of the signature genes of SS and is part of one of the best characterized imprinted clusters. Deregulation of its expression has been suggested to play a role in the development of several types of cancer. Real time PCR experiments revealed that different hMSC isolates display highly variable levels of IGF2
transcripts. Although a complex network of long range interactions and multiple looping are emerging as newly recognized regulators of H19
and IGF2 
, the methylation status at the H19
imprinting control region (ICR) remains a basic regulatory factor according to the shared enhancer model. Bisulfite transformation analysis revealed a highly divergent methylation pattern among hMSC populations both at the H19
ICR and in a second region downstream of the H19
gene. In those populations that were found to be informative, the methylation pattern at the H19
ICR was shown to be compatible with maintenance or loss of imprinting and to correlate accordingly with a mono or bi-allelic IGF2
expression that could explain, at least in part, the different level of mRNA measured by RT-PCR. Thus different hMSC populations displayed a different imprinting status at the H19/IGF2
ICR. Epigenetic variation, which could be a function of numerous factors, including age and environmental conditions, has been shown to characterize stem cells and to play an important role in determining cell commitment and plasticity 
. Consistent with this notion, the cells used in the present study were derived from different donors whose age variation, although in the pediatric/adolescent range, could conceivably explain, at least in part, their distinct epigenetic features.
Expression of SYT-SSX in the four populations produced variable epigenetic effects.
Methylation analysis at the H19 ICR showed modest changes, hypermethylation on both alleles being induced by SYT-SSX in only one population whereas other populations displayed either no effect or opposite effects on the two alleles. The absence of methylation changes in population 3 can be reconciled with the observed absence of induction of IGF2 expression. Conversely in population 4 the hypermethylating effect of SYT-SSX at the 6th CTCF binding site may explain, in part, the induction of the IGF2 transcripts. The observed SYT-SSX-dependent switch from monoallelic to biallelic expression of IGF2 in this population, together with the methylation changes, is consistent with SYT-SSX-induced LOI according to the shared enhancer model. Hypermethylation of both alleles in these cells can also explain the observation that, in addition to the re-expression of the silent allele, expression of the active allele was increased.
In population 1, where a methylation pattern consistent with loss of imprinting and a corresponding baseline bi-allelic IGF2
expression were demonstrated, the effect of SYT-SSX at the 6th
CTCF binding site produced modest but opposite effects on the two alleles. SYT-SSX1-mediated enhancement of IGF2
expression in this population must therefore have been achieved by alternative mechanisms since it cannot be explained by the reactivation of a silent allele. The involvement of alternative and/or additional regulatory factors at the H19/IGF2
locus that may be directly or indirectly affected by SYT-SSX expression is suggested by several observations emerging from the present study. Concomitant induction of H19
observed in all cases is not compatible with the sole perturbation by SYT-SSX1 of ICR imprinting. Furthermore the similar activation of both P1 and P2–P4 IGF2
promoters is also suggestive of the existence of multiple regulatory mechanisms affected by the fusion protein since several independent observations suggest that not all IGF2
promoters are regulated exclusively by the imprinting control region. It has been reported that in hepatocytes and chondrocytes, IGF2
transcripts from promoter P1 are derived from both parental alleles, whereas transcripts from promoters P2, P3 and P4 are derived from a single parental allele 
. These observations suggest that P1 promoter activity could be at least partly independent of the ICR. It is noteworthy that the P1 transcript is reported to be expressed from both parental alleles in postnatal liver and fetal choroid plexus/leptomeninges 
, and that P1 promoter activity was observed not to be exclusively connected to IGF2
LOI in laryngeal squamous cell carcinoma 
Methylation analysis of regions outside the H19 ICR showed that SYT-SSX1 does not affect methylation specifically and exclusively at the H19 ICR but rather at different discrete regions with even opposite effects in adjacent segments and in different hMSC populations. The exact mechanism whereby SYT-SSX affects methylation and possibly the complex network of long range interactions and multiple looping that regulate the H19/IGF2 locus remains to be defined. Our data suggest that a specific epigenetic substrate, defined by a normal imprinting status and monoallelic expression of IGF2 are required for a strong effect of SYT-SSX on IGF2 expression and that changes in the baseline epigenetic status, can prevent SYT-SSX1 from exerting its effect on the H19 ICR. On the other hand our data also suggest that the effect of SYT-SSX is not limited to methylation changes at the H19 ICR but rather affects additional, hitherto undefined, regulatory mechanisms at the H19/IGF2 locus.
We have shown that introduction of SYT-SSX into different populations of hMSC has effects on epigenetic function that display cell-type specific qualitative and quantitative variation. We hypothesize that this variation could originate from the differences in the epigenetic context that the fusion protein encounters and that minor baseline epigenetic changes may have a relevant bearing on SYT-SSX function. It is possible that a highly specific epigenetic status is required for transformation of primary cells by SYT-SSX, which may explain, in part, the low frequency of SS. Such a permissive epigenetic status may be confined to cells at a specific stage of differentiation, as suggested by the recently reported transgenic mouse model.