Adenovirus E1A Does Not Induce the Ewing Tumor-associated Gene Fusion EWS-FLI1¹ Free

EFTs,³a group of largely undifferentiated tumors of bone and soft tissue,arise from an as yet uncharacterized progenitor cell in childhood or adolescence. This group of highly aggressive neoplasms is characterized by the expression of a chimeric transcription factor as the result of a gene fusion between EWS on chromosome 22q12 and a member of the ets oncogene family (for review, see Ref.1). In 85% of cases, EWS is rearranged with FLI1 on chromosome 11q24 (2). Genomic rearrangement sites are scattered over 6 and 40 kb in EWSand FLI1, respectively, with no recombination hot spots(3). Breakpoint diversity results in the production of variably sized, patient-specific fusion products. This variability has been reported to impact on the treatment outcome for patients with localized disease (4, 5), possibly because of the variable transactivating potential of the differently sized EWS-FLI1 chimeric transcription factors (6). Sequencing of 113 individual interchromosomal junctions did not provide any clues to the involvement of a specific recombinase in the gene rearrangement but rather suggested a complex sequence of independent strand breaks, local reconstruction, and interchromosomal joining as the source of the tumor-specific aberration (7). The mechanism underlying this illegitimate recombination process remains unknown. Recently,based on the morphological similarity of EFT cells and Ad-transformed cells only, Sanchez-Prieto et al. (8) tested for the presence of EWS-FLI1 chimeric products in the stably Ad5-E1-transformed human embryonal kidney cell line HEK293 and in transiently E1A-expressing HeLa cervix carcinoma cells, human fibroblasts, and keratinocytes with positive results. The Ad5 E1A gene encodes two major proteins, 289R and 243R, that interact with a variety of cellular proteins to mediate oncogenic and antioncogenic functions(for a recent review, see Ref. 9). The finding that E1A may elicit the generation of oncogenic gene fusions would have far-reaching consequences (10), specifically because the EWS-FLI1 gene fusion is representative of a whole class of similar tumor-associated gene rearrangements between EWS or close relatives and other transcription factor genes(reviewed in Ref. 1). Because EWS-FLI1-transfected rodent cells give rise to tumors with small round cell morphology(11), as does E1A, it has been speculated that E1A may exert some of its oncogenic functions via the induction of chromosomal alterations (10). One of the issues of concern arising from such a finding is the safety aspect of possible use of wild-type virus-contaminated adenoviral vectors in gene therapy. We therefore sought to verify the previously published data regarding an association between E1A expression and the generation of the EFT-specific chromosomal aberration. However, we were unable to confirm the presence of an EWS gene rearrangement and expression of the corresponding fusion product not only in HEK293 cells but also in other stably and transiently E1A-expressing cell lines of different origin.

Ewing tumor cell lines used in this study were raised at the Children’s Cancer Research Institute (Vienna, Austria; STA-ET series;Ref. 12) or kindly provided by F. van Valen (Department of Pediatrics, University of Münster, Germany; VH64, WE68, and WE-M2–68), J. Biedler (Memorial Sloan Kettering Cancer Center, New York, New York; SK-N-MC), G. Hamilton (Department of Surgery,University of Vienna, Austria; SAL2 and SMB), G. M. Lenoir (IARC,Lyon, France; IARC-EW2 and IARC-EW7), and T. Triche (Department of Pathology, Children’s Hospital, Los Angeles, CA; TC252). The EFT cell lines A673, RDES, and SK-ES1, as well as the neuroblastoma cell line LAN1, the ALL cell line MOLT, the human fibroblast cell line Hs68,and the Ad5-E1-transformed human embryonal kidney cell line HEK293(ATCC CRL-1573; Ref. 13), were obtained from the American Type Culture Collection (Manassas, VA). EcR293, a subclone of HEK293 expressing an ecdyson receptor transgene, was purchased from Invitrogen (Groningen, the Netherlands). Neuroblastoma cell lines SJ-NB-7 and Vi856 were kindly provided by T. Look (Dana Farber Cancer Institute, Boston, MA) and O. Majdic (Institute of Immunology,University of Vienna, Austria), respectively. Human embryonic retinoblast cell lines derived by stable transfection of plasmids encoding the early region 1 (E1) of Ad type 5 (HER911 and PER.c6; Ref.14) and Ad type 12 (HER-RIC.c7 and HER-RIC.c4; Ref.15), respectively, had been isolated at the Department of Molecular Cell Biology (Leiden University Medical Centre, Leiden, the Netherlands).

RNA extraction and RT-PCR for EWS-FLI1 and germ-line EWS expression were performed as described previously (16). Negative controls and sensitivity controls were included in every individual determination as described previously (17). Amplification of the 5′-end of germ-line FLI1 cDNA was achieved using primers TATTAAGGAGGCTCTGTCG and ATGCTCCTGTGTCCACAG. Expression of Ad5 E1A isoforms was monitored after DNase I treatment of RNA by RT-PCR using primers GTATGATTTAGACGTGACGG and GATAGCAGGCGCCATTTTAG.

Northern blotting of total RNA was performed according to standard procedures using a 783-bp EcoRI/BamHI fragment and a 663-bp EcoRI/HindII fragment from cloned EWS-FLI1 type 1 cDNA as EWS- and FLI1-specific probes, respectively.

EcoRI-digested genomic DNA was separated on a 0.8% agarose gel, transferred to a nylon membrane (Amersham Hybond-N), and hybridized to probe HP.5 (3) for visualization of EWS-derived fragments.

For the detection of germ-line FLI1 and EWS-FLI1 protein, whole cell extracts were separated on a 10% Laemmli gel, blotted onto a nitrocellulose membrane (Amersham), and reacted with monoclonal antibody 7.3, which was raised against an 82-amino acid COOH-terminal FLI1 peptide (18).

HeLa, Hs68, and SJ-NB-7 cells were transiently transfected using LipofectAMINE Plus reagent (Life Technologies, Inc., Paisley,UK) according to the manufacturer’s recommendations.

One of the surprising findings in the previously published study about induction of the EFT-specific translocation by E1A was that in individual transient transfections, only single EWS-FLI1 fusion types were detectable, rather than a polyclonal rearrangement. This result may be explained if the viral gene product promotes EWS-FLI1 recombination only at a very low frequency. To study abundance and heterogeneity of this gene rearrangement after transfer of Ad5 E1A, we transiently expressed the viral oncogene from a cytomegalovirus promoter (19) after lipofection in HeLa cervix carcinoma cells, the human fibroblast cell line Hs68, and the human neuroblastoma cell line SJ-NB-7. Transfection efficiencies, as evidenced by cotransfection of a LacZ reporter gene, were 2%, 4%, and >80%,respectively, corresponding to a calculated number of 20,000, 40,000,and >800,000 E1A-transfected cells. Expression of EWS-FLI1 was monitored 70 h posttransfection using a highly sensitive one-tube nested PCR approach capable of identifying target transcripts from as few as one EWS-FLI1-expressing cell per million (Refs. 16and 17; Fig. 1,A) and using the same primer combination and RT-PCR conditions applied by Sanchez-Prieto et al. (8;data not shown). Integrity of RNA and cDNA preparations was confirmed by amplification of germ-line EWS (Fig. 1,B). Although transcripts corresponding to the 289R and 243R isoforms of E1A could be amplified from all transfectants (Fig. 1,C), no EWS-FLI1 chimeric product was obtained with any of these methods. Consequently,we estimated the frequency of E1A-induced gene rearrangements, if they occur at all, to be lower than 1 in 10⁴ to 1 in 10⁶. We then sought to confirm EWS-FLI1 expression in HEK293 cells, which have been reported to express a type 1 fusion (8). However, in contrast to EFT cell lines, no PCR product was obtained from HEK293 cells (Fig. 1,A). To confirm that we were using authentic HEK293 cells, we also tested EcR293 cells, a putative HEK293-derived subclone obtained from a different commercial source. A common origin and the purity of these two cell lines were confirmed by microsatellite analysis (data not shown). Again, no EWS-FLI1 transcript was detectable with both RT-PCR approaches. As a complementary piece of evidence, we performed Northern blot analysis. A probe corresponding to the 3′ terminus of FLI1 recognized a band of approximately 4.5 kb not only in EFT cells expressing EWS-FLI1 (SK-N-MC) and in the ALL cell line MOLT, which is known to express germ-line FLI1, but also in HEK293 cells, as reported by Sanchez-Prieto (Fig. 2,A). However, when probed with a DNA fragment from the EWS 5′portion present not only in the germ-line EWS transcript but also in all EFT-derived chimeric RNA, a signal was obtained for SK-N-MC only(Fig. 2,B). Because germ-line FLI1 and EWS-FLI1 transcripts are of similar size, we supposed that HEK293 cells express germ-line FLI1. This assumption was confirmed by RT-PCR using primers specific for the very 5′-end of FLI1 that is absent from chimeric EWS-FLI1 (Fig. 1,D). However, when HEK293 and EcR293 were compared with three different EFT cell lines (VH64 and IARC-EW2 expressing an EWS exon7/FLI1 exon 5 fusion and STA-ET2.2 carrying an EWS exon 9/FLI1 exon 4 rearrangement) and with MOLT cells on the Western blot, the monoclonal FLI1-specific antibody 7.3 (18) recognized EWS-FLI1 as bands of about M_r90,000 in the EFT cells exclusively and recognized FLI1 as a band of about M_r 60,000 in MOLT only(Fig. 3). No corresponding protein was detectable in the HEK293 and EcR293 cells, despite the presence of a germ-line FLI1 transcript at low abundance. In addition, we studied a series of stably Ad5-E1 (HER911 and PER.c6)- and Ad 12-E1 (HER-RIC.c4 and HER-RIC.c7)- transformed human embryonic retinoblasts. As in HEK293 cells, no EWS-FLI1 was detectable, but germ-line FLI1 transcripts were identified in these cell lines (Fig. 1). To account for a low incidence of the EWS-FLI1 rearrangement in E1A-expressing cells, we compared short- and long-term cultures of PER.c6 cells. The results were identically negative for passage 9 and passage 122. The presence of PCR-detectable germ-line FLI1 transcripts in Ad-transformed cell lines may be due to the expression of the 289R form of E1A that possesses a potent transcriptional activation function and which, by binding to several components of the basic transcriptional machinery, may alter gene expression profiles (9). However, when we tested HeLa,Hs68, and SJ-NB-7 cells before and after E1A transfection, no dependence of FLI1 expression on the presence of E1A was seen: HeLa was negative; and Hs68 as well as SJ-NB-7 was positive (Fig. 1 D). Germ-line FLI1 expression also varied in EFT cells(compare VH64 with SK-N-MC).

We also studied integrity of the EWS gene by genomic Southern blotting (Fig. 4) and by means of molecular cytogenetics (data not shown). On Southern blots of EcoRI-digested chromosomal DNA, only the germ-line configuration of EWS was detectable in HEK293 and EcR293 cells, whereas type 1 EWS-FLI1-expressing EFT cell lines A673, SK-N-MC,and STA-ET-1 all showed an additional signal corresponding to the rearranged allele. The absence of an EWS-FLI1 gene rearrangement in the Ad5-E1-transformed human embryonal kidney cells was further supported by interphase cytogenetics using a combination of cosmid probes flanking the EWS breakpoint region on chromosome 22 (20). No splitting of signals was observed,confirming the integrity of the EWS gene (data not shown). Similarly, chromosome painting of PER.c6 cells did not provide any evidence for a rearrangement between chromosomes 11 and 22 in continuously E1A-expressing cells (data not shown).

Finally, we screened genomic DNA of 27 EFT cell lines (Fig. 1 E) and cDNA from 19 primary tumor samples (data not shown)for the presence of sequences from the adenoviral E1A gene,with negative results.

Although Ad infection can cause disease in animals, no oncogenic role has been known for any Ad serotype in humans. Therefore,replication-defective, E1-deleted adenoviral vectors have been used as efficient gene transfer vehicles for gene therapy. However, at low frequencies, these vectors may be converted to replication-competent infectious virus during propagation on E1-expressing helper cells, when they reacquire the E1 region by homologous recombination. The conditionally replicating vectors that contain the Ad E1Agene are studied in clinical trials (21). Therefore, the postulated potential of E1A to generate oncogenic gene fusions in human cells is of significant concern. However, we were unable to confirm an association between E1A expression and the presence or the generation of an EWS-FLI1 gene rearrangement. We have studied a series of cell types with identical negative results. In the transient transfections, we considered the possibility that the EWS-FLI1 recombination may be a rare event by using a highly sensitive nested RT-PCR approach. We also accounted for a low incidence of E1A-induced gene rearrangements when we studied stably E1A-expressing cells not only at low passage numbers but at high passage numbers as well. By applying DNA-, RNA-, and protein-based methods, we were able to unambiguously exclude a correlation between E1A expression and the presence of an EFT-specific gene fusion even in HEK293 cells that have previously been reported by Sanchez-Prieto et al. (8) to carry and express an EWS-FLI1 type 1 gene fusion. The basis for the contradictory nature of our results and the results of Sanchez-Prieto et al. (8) remains obscure. Besides the inherent RT-PCR risk of cross contamination, our finding of germ-line FLI1 expression in several cell lines including HEK293 may be part of an explanation. The use of ill-defined commercial antibodies may have also contributed to a false positive interpretation of the previously published results. Finally, because E1A was absent from all EFT material studied, there is no evidence that Ad E1A may contribute to EFT pathogenesis. Our data imply that there is no reason to abstain from the use of adenoviral vectors for gene therapy in general and from the use of E1A-containing vectors for cancer gene therapy in particular.

We thank O. Delattre (Institut Curie, Paris, France) for kindly providing monoclonal antibody 7.3, probe HP.5, and cosmids G9,F10, F7, and E4.