Our study was directed toward identifying a primary cell that is permissive for EWS-FLI-1-mediated transformation. Following the observation that EWS-FLI-1 transforms primary mouse mesenchymal progenitor cells (MPC) to form Ewing's sarcoma–like tumors (1), we tested these tumors for CD99 expression as one among several features of the Ewing's sarcoma phenotype.
We found that one of the most widely used anti-human CD99 antibodies recognized MPCEWS-FLI-1 but not MPCBcl2 tumor cells. As positive controls, we used mouse pancreas sections based on the described expression of CD99 in pancreatic islet cells. The antibody did indeed stain pancreatic islet cells, and Western blot analysis of MPCEWS-FLI-1 cell lysates using the antibody revealed a 32-kDa band similar to the one expressed in human Ewing's family tumor (EFT) cells. MPCs displayed a weak band recognized by the antibody, raising the possibility that they may well express a low baseline level of a mouse CD99 homologue that is undetectable by conventional immunohistochemistry. We confirmed the Western blot results using a polyclonal antibody (ab27271, Abcam, Cambridge, United Kingdom). Thus, the human anti-CD99 antibody seems to recognize a related mouse molecule or homologue. Whether this corresponds to the product of the recently isolated mouse cDNA or yet another homologue or isoform remains to be determined.
Importantly, we based the notion that EFTs originate in MPCs on the following observations: (a) EWS-FLI-1 expression transformed primary MPCs, establishing that MPCs display the permissiveness that EWS-FLI-1 oncogenicity requires. (b) MPCEWS-FLI-1 tumors display a transcriptome characterized by induction and repression of numerous transcripts found to display the same movement in various EWS-FLI-1-transfected cells and EFTs. (c) MPCEWS-FLI-1 tumors up-regulate and are dependent on insulin-like growth factor-I, a major functional feature of EFTs. (d) MPCEWS-FLI-1 tumors display a morphology reminiscent of human EFT and neural differentiation markers, in addition to anti-CD99 antibody reactivity, as do their human counterparts.
Our observations are supported by two recent studies by other groups: First, Torchia et al. (2) showed that EWS-FLI-1 can block differentiation of MPCs, which is consistent with the poorly differentiated appearance of EFTs. Second, Robert Ilaria's group (3) showed that expression of EWS-FLI-1 in unsorted primary bone marrow–derived cells resulted in the formation of tumors that resemble human EFT. Their tumor cells expressed Sca-1, a key mesenchymal stem cell marker that characterized the MPCs used in our study.
In summary, we feel that our observations provide strong evidence that MPCs are a candidate cell type from which EFTs originate.