To the editor—Breakpoint variation and the resulting fusion type heterogeneity are characteristic of the Ewing tumor-associated EWS–FLI1 gene rearrangement. Sanchez-Prieto et al. report high efficiency induction of EWS–FLI1recombination in a variety of cell types by adenovirus E1A (ref 1). To our surprise, only a single fusion transcript type (type 1 or type 2) was found in individual E1A transfections, instead of the expected polyclonal rearrangement. To characterize the E1A-induced gene fusions in more detail, we sought to repeat the experiments described by Sanchez-Prieto et al.

Adenovirus 5 E1A was expressed from a CMV promoter in HeLa cervix carcinoma cells, a human fibroblast cell line and a neuroblastoma cell line after lipofection. Transfection efficiencies were approximately 2%, 4% and 80%, respectively. We used a one-tube nested PCR approach capable of identifying one EWS–FLI1-expressing cell in a total of 106 cells6 to screen for expression of the Ewing tumor specific gene rearrangement 70 hours after transfection. No EWS–FLI1 transcript was detectable in any of the transiently E1A-transfected cells, although E1A expression was confirmed.

Next, we studied HEK293 human embryonal kidney cells immortalized with adenovirus 5 DNA and a subclone of this cell line EcR293 obtained from two different commercial sources (ATCC and Invitrogen), for which a common origin could be confirmed by microsatellite analysis. Although Sanchez-Prieto et al. reported that HEK293 cells carry and express an EWS–FLI1 type 1 rearrangement, we were unable to detect the presence of the gene rearrangement on Southern blot analysis of genomic DNA (a piece of evidence lacking from the study by Sanchez-Prieto et al.); nor could we identify expression of an EWS–FLI1 chimeric RNA by northern blot analysis, by RT–PCR using the conditions applied by the authors, or by single-tube nested PCR. Furthermore, the EWS–FLI1 protein was undetectable on western blots probed with an FLI1-specific polyclonal antibody7. In contrast, all Ewing tumor cell lines studied in parallel, tested positive for EWS–FLI1 using all the methods described herein.

Finally, we tested 27 Ewing tumor cell lines and 19 primary Ewing tumors with established EWS–FLI1 expression for the presence of adenovirus 5 E1A DNA/RNA by RT–PCR. None of the samples produced positive results.

In summary, we were unable to confirm an association between adenovirus E1A expression and the oncogenic EWS–FLI1 gene rearrangement, either in Ewing tumors or in any other cell type tested. Thus, there is no evidence for a major role of viral genes in the generation of Ewing tumors.