The cellular oncogene EWS/activating transcription factor 1 is ... - Nature

© 2000 Nature America, Inc. 0929-1903/00/$15.00/⫹0 www.nature.com/cgt

The cellular oncogene EWS/activating transcription factor 1 is unable to activate adenovirus-borne promoters: Implications for cytotoxic prodrug therapy of malignant melanoma of soft parts Raymond W. M. Lung and Kevin A. W. Lee Department of Biology, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, Special Administrative Region, China. The cellular oncoprotein Ewing’s sarcoma oncogene (EWS)/activating transcription factor 1 (ATF1) is a highly specific marker for malignant melanoma of soft parts (MMSP) and is a potent activator of several cAMP-inducible promoters, including the somatostatin promoter. Here we explored the potential for using the somatostatin promoter to direct toxic gene expression in MMSP cells. When introduced into MMSP cells, a somatostatin-herpes simplex virus thymidine kinase fusion gene confers strong and cell-specific sensitivity to the cytotoxic prodrug ganciclovir. Ganciclovir sensitivity requires the ATF-binding site present in the somatostatin promoter, indicating that toxic gene expression is caused by EWS/ATF1. We also tested the efficacy of recombinant adenoviruses adenoviruses for gene delivery and expression in two MMSP cell lines (DTC1 and Su-ccs-1). Surprisingly, several promoters (including somatostatin) that are strongly activated by EWS/ATF1 in transient assays are not activated in DTC1 and Su-ccs-1 cells when present in an adenovirus vector. In summary, our findings demonstrate the potential for using the somatostatin promoter for cytotoxic prodrug therapy for MMSP. However, first-generation adenovirus vectors cannot be used as promoter delivery vehicles for toxic gene expression in MMSP cells. Cancer Gene Therapy (2000) 7, 396 – 406

Key words: EWS/ATF1 oncogene; malignant melanoma of soft parts; adenovirus; gene therapy.

C

hromosomal translocations of the N-terminal region of the Ewing’s sarcoma oncogene (EWS) to a variety of cellular transcription factors produces dominant oncogenes (EWS fusion proteins (EFPs)) that cause distinct sarcomas.1,2 In malignant melanoma of soft parts (MMSP), EWS is fused to activating transcription factor 1 (ATF1);3 in Ewing’s sarcoma, EWS is fused to the ETS domain family members;4 – 8 and in desmoplastic small round cell tumors, EWS is fused to the Wilm’s tumor oncogene, WT1.9 Because EFPs are strong transcriptional activators, it is thought that EFPinduced malignancies arise via aberrant gene activation, with the tumor type being specified by the EWS fusion partner. MMSP (also known as clear cell sarcoma) is typically associated with tendons and aponeuroses and is thought to be of neuroectodermal origin.10,11 MMSP is a very aggressive, early onset tumor that initially grows slowly, resulting in 20% of patients having metastases at the time of diagnosis.12 Current treatment involves surgery Received April 30, 1999; accepted July 8, 1999. adenovirusdress correspondence and reprint requests to Dr. Kevin A. W. Lee, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, S. A. R. China. E-mail address: bokaw@ usthk.ust.hk

396

combined with chemotherapy and radiotherapy,10,13 but recurrence is high and mortality is ⬃45% due to metastatic disease. The need for more effective therapy of MMSP and other EFP-related tumors is apparent and emerging gene therapy techniques may offer new approaches. In the case of Ewing’s sarcoma5 and desmoplastic small round cell tumors14 the oncogenic properties of EFPs have been demonstrated directly and deregulated target genes have been identified.15–20 Furthermore, in Ewing’s sarcoma, agents that antagonize EWS-Fli1 fusion proteins can also inhibit cellular proliferation,22–24 indicating the potential for the development of therapeutic agents that target EWS/Fli. In contrast to EWS/ Fli, the role that EWS/ATF1 plays in tumorigenesis is poorly characterized. Because ATF1 is a cAMP-inducible activator,25,26 it is predicted that EWS/ATF1 acts via constitutive activation of cAMP-inducible genes. However, the critical target genes have not yet been identified. In addition, a role for EWS/ATF1 in tumor maintenance has not yet been established. Consequently, it is not known whether therapeutic approaches aimed at antagonizing EWS/ATF1 will be applicable to MMSP. An alternative therapeutic approach (cytotoxic prodrug therapy) for MMSP involves the exploitation of tumor-specific promoters to target tumor cells for the

Cancer Gene Therapy, Vol 7, No 3, 2000: pp 396 – 406

397

LUNG AND LEE: GENE THERAPY FOR MMSP

action of cytotoxic agents.27–29 Previous studies have characterized EWS/ATF1 as a tumor-specific activator of promoters containing ATF-binding sites.30 –32 Transcriptional activation is dependent upon the EWS activation domain (EAD)30,31 and the DNA-binding domain of ATF1.30 –32 Most significantly, promoters that can be activated by EWS/ATF1 (e.g., the somatostatin promoter) are constitutively active when transiently introduced into tumor-derived cell lines (DTC1 and Su-ccs-1, hereafter referred to as MMSP cells) containing endogenous EWS/ATF1.30 In addition to a cell-specific targeting agent (such as a tumor-specific promoter), cytotoxic prodrug treatment requires a safe and efficient vector to introduce the therapeutic gene and an effective prodrug. Among many available systems, recombinant adenoviruses (adenoviruss)33,34 and the herpes simplex virus thymidine kinase (hsvtk) gene have been well studied. The combined use of hsvtk as a toxic gene and adenovirus as a gene delivery vehicle has proven effective for prodrug sensitization of other tumor cell lines.35–37 In the current study, we tested the use of a somatostatin promoter-hsvtk fusion gene and adenovirus vectors for cytotoxic prodrug treatment of MMSP cell lines. Our results demonstrate the potential of the somatostatin promoter and hsvtk as cytotoxic agents for MMSP cells. Surprisingly however, several promoters (including somatostatin) that are strongly activated in MMSP cells using plasmid-based transient assays are not activated when present in an adenovirus vector. Thus, first-generation adenovirus vectors cannot be used as promoter delivery vehicles for EWS/ATF1-dependent toxic gene expression in MMSP cells. MATERIALS AND METHODS

Plasmids p⌬(⫺71)SomCAT contains the somatostatin promoter at position ⫺71, fused to the chloramphenicol acetyl transferase (CAT) coding sequences.38 p⌬(⫺71)SomCAT contains a single ATF1-binding site, which is required for activation by both cAMP38 and EWS/ATF1.30 p⌬(⫺42)SomCAT is the same as p⌬(⫺71)SomCAT, except that sequences from position ⫺71 to ⫺42 of the somatostatin promoter are deleted.30 Therefore, p⌬(⫺42)SomCAT lacks the ATF1-binding site of the somatostatin promoter.30 p⌬(⫺42)SomTK and p⌬(⫺71)SomTK are identical with p⌬(⫺42)SomCAT and p⌬(⫺71)SomCAT, respectively, except that the CAT sequence is replaced by the cDNA sequence of hsvtk. pSVEA contains the complete EWS/ ATF1 coding sequence with 123 of 5 untranslated sequences from EWS and can express EWS/ATF1 under the control of the simian virus 40 (SV40) early promoter. pRSVCAT contains the Rous sarcoma virus (RSV) long terminal repeat linked to CAT,39 and pCAT-TM-control (Promega, Madison, Wis) contains the SV40 enhancer and early promoter linked to CAT. pVIP25CAT40 contains the vasoactive intestinal peptide (VIP) cAMP-responsive element containing two ATF-binding sites, fused to the RSV promoter from positions ⫺50 to ⫹39 and to CAT. pVIP4CAT contains 240 bp of the VIP promoter, including the cAMP-responsive element linked to the reporter CAT gene.40 pCAT-BstN1 (Ref. 41, referred to as pE3CAT in Fig 2) contains the adenovirus E3 promoter to position ⫺105 fused to CAT. p⌬E1(⫺71A) (used to construct AD71) was obtained by

Cancer Gene Therapy, Vol 7, No 3, 2000

inserting a BamHI/AatII fragment from p⌬(⫺71A)SomCAT and an oligonucleotide to reconstruct the somatostatin promoter to position ⫺71 into the multiple cloning site of p⌬E1SP1A. p⌬E1(⫺42A) (used to construct AD42) was obtained by insertion of an AatII/BamHI blunt fragment from p⌬(⫺71)SomCAT into the EcoRI/BamHI blunt sites of the multiple cloning region of p⌬E1sp1A. p⌬E1(25A) (used to construct AD25V) was obtained by inserting a BamHI/AflIII blunt fragment from pVIP25CAT into the EcoRI/BamHI blunt sites of p⌬E1sp1A. p⌬E1(VIPB) (used to construct ADVIP) was obtained by inserting a SacI/BamHI blunt fragment from the pVIP4CAT sequence into the EcoRI/BamHI blunt sites of p⌬E1sp1B.

Construction and propagation of recombinant adenoviruss Recombinant adenoviruss (adenovirus type 5) were constructed by in vivo recombination of plasmids in 293 cells using adenovirus vector construction kit C (Microbix Biosystems, Toronto, Canada). p⌬E1SP1A and p⌬E1SP1B contain the left-hand end of adenovirus type 5 (from base pair 22 (0 map units) (m.u.) to base pair 5790 (16.1 m.u.)), with a deletion of E1 sequences from base pair 342 to base pair 3523 (1.0 –9.8 m.u.), but containing the entire packaging signal. A multicloning site is present in the E1 region, allowing for insertion of the foreign gene. p⌬E1SP1A and p⌬E1SP1B are identical except for the orientation of the multicloning site between ClaI and BglII. The pBHG10 plasmid contains the rest of the adenovirus type 5 genome and a small overlapping region with both p⌬E1sp1A and p⌬E1sp1B to allow recombination after cotransfection into 293 cells.42 pBHG10 is noninfectious due to lack of a packaging signal needed for encapsidation of viral DNA;43 in addition, to allow a bigger insert, the nonessential E3 region (78.3– 85.8m.u.) is deleted. Low-passage 293 cells (⬍40 passages) in a 60-mm dish were set up at 50% confluence. A total of 10 ␮g of p⌬E1(⫺71A), p⌬E1(⫺42A), p⌬E1(25A), or p⌬E1(VIPB) and 10 ␮g of pBHG10 were then cotransfected into the cells by calcium phosphate precipitation. Medium was renewed the following day and on reaching confluence, the cells were trypsinized and diluted. On release from the cells, recombinant viruses were confirmed by polymerase chain reaction analysis of isolated viral DNA and titrated on 293 cells. Purification of viral DNA was performed as follows. Infected 293 cells were harvested and lysed by the addition of 400 ␮L of lysis buffer (20 mM of Tris-Cl (pH 7.5), 0.5% Nonidet P-40, 500 mM NaCl, and 1 mM dithiothreitol) on ice for 5 minutes. After centrifugation for 5 minutes in a microfuge, the supernatant was adjusted by the addition of 15 ␮L of 20% sodium dodecyl sulfate and 150 ␮g of proteinase K and incubated at 55°C for 30 minutes. The sample was then phenol-extracted, ethanol-precipitated, and finally treated with 20 ␮g/mL ribonuclease A for 30 minutes to remove the RNA.

Cell culture and adenovirus infections The JEG3,44 Su-ccs-1,11 and DTC130 cells lines were maintained as monolayers in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal calf serum. PC12 cells were maintained as monolayers in DMEM containing 10% fetal calf serum and 5% horse serum. RAT1 is a rat embryo fibroblast cell line and MEL28 is a human melanoma cell line. Human embryo kidney 293 cells (Microbix Biosystems) contain and express the early region 1 (E1) of adenovirus type 5. Clones 711, 714, and 716 (collectively called D71TK cells) are independent clones of DTC1 cell lines containing ⌬(⫺71)Som/ hsvtk fusion gene in the genome. Clones 422, 423, and 424

398 (collectively called D42TK cells) are independent clones of DTC1 cell lines containing ⌬(⫺42)Som/hsvtk fusion gene. Polymerase chain reaction analysis of genomic DNA confirmed the presence of the SomTK fusion gene in each clone (data not shown). Conditions for adenovirus infections were as follows: Cells in a 60-mm culture dish at ⬃50% confluence were infected at a multiplicity of infection (MOI) as indicated in the figure legend. adenovirussorption of the virus was carried out in 0.5 mL of DMEM buffered with 20 mM N-2-hydroxyethylpiperazine-N⬘-2-ethanesulfonic acid (pH 7.2) for 45 minutes, followed by dilution with fresh growth media.

Transfections, CAT assays, and RNA analysis All transfections were carried out by calcium phosphate coprecipitation, and CAT assays were performed as described previously.45 Precipitates contained 5 ␮g of reporter plasmid, the indicated amount of activator plasmid(s), and 20 ␮g of total DNA made up with pSP64 as carrier. One-third of the precipitate was added to ⬃50% confluent JEG3 cells in a 60-mm culture dish. For quantitation of results, the percentage of conversion of unacetylated to acetylated 14C-chloramphenicol under linear assay conditions was determined by excision of spots from the thin-layer chromatography plate and quantitation of radioactivity using a liquid scintillation counter. Preparation of total cellular RNA and detection of specific transcripts by primer extension was as described previously.46

Production of DTC1 transfectants and ganciclovir (GCV) sensitivity assays A 50% confluent monolayer of DTC1 cells (⬃2.5 ⫻ 106 cells) was transfected by calcium phosphate coprecipitation39 with 5 ␮g of the desired plasmid with or without 1 ␮g of polyneo plasmid, conferring neomycin resistance. DTC1 cells were used because Su-ccs-1 cells do not yield viable colonies due to a very low transfection efficiency and/or cytotoxicity at low cell density (K.A.W.L., unpublished observation). Transfected cells were washed with phosphate-buffered saline, and old medium was replaced by fresh growth medium at 1 day posttransfection. Cells were then split to 25% confluence into medium with 400 ␮g/mL G418 (Life Technologies, Gaithersburg, Md; stored as 40 mg/mL in sterile-filtered 100 mM N-2-hydroxyethylpiperazine-N⬘-2-ethanesulfonic acid (pH 7.2)). Selection medium was changed daily for the first 10 days and every 2 days afterward. Cells were set up in duplicate at 30% confluence in 24-well plates the day before the addition of GCV (Cymevene, 500 mg GCV/546 mg from Ben Venue Laboratories, Bedford, Ohio, under license from Syntex Laboratories, Palo Alto, Calif) to the growth medium. GCV was dissolved in autoclaved H2O at a concentration of 0.5 mg/mL and stored at 4°C. Appropriate drug concentrations were made by diluting the stock solution with fresh medium immediately before each experiment. The same volume of medium was added for the control experiment, medium was changed every 2 days, and viable cell counts were taken using trypan blue exclusion and a hemocytometer.

RESULTS

Cytotoxic prodrug treatment of DTC1 cells in vitro A truncated version of the somatostatin promoter (hereafter simply referred to as the somatostatin promoter) containing promoter sequences to ⫺71 including a single ATF-binding site, is highly active when transiently intro-


duced into both MMSP cell lines (DTC1 and Su-ccs1).30 We wanted to test the ability of a somatostatin promoter-hsvtk fusion gene to express hsvtk in the above cells and thereby confer sensitivity to the cytotoxic prodrug GCV. To establish a test system for GCV sensitivity, we produced DTC1 transfectants containing a stably integrated somatostatin promoter linked to CAT as a reporter. Initially, we used CAT instead of hsvtk as a simpler test for somatostatin promoter activity under the desired conditions. Two pools of DTC1derived stable transfectants (D42 and D71) were obtained after G418 selection. D42 represents a pool of DTC1 cells containing the ⌬(⫺42)SomCAT fusion gene (lacking the ATF-binding site in the somatostatin promoter); D71 represents a pool of cells containing the ⌬(⫺71)SomCAT fusion gene (containing the ATF site). Somatostatin promoter activity was assessed by CAT assay (Fig 1A). Similar to results obtained in transient assays,30 there was no detection of somatostatin promoter activity in D42 cells (or parental DTC1 cells), whereas strong activity was obtained in D71 cells. PC12 cells containing a stably integrated cAMP-inducible VIP promoter linked to CAT serve as a comparison for activated levels of transcription. In the presence of 20 ␮M forskolin (which increases intracellular cAMP levels), the VIP promoter was activated in PC12 cells to a level that was similar to that seen for the somatostatin promoter in D71 cells. The above results show that (on average within a pool of transfectants) chromosomal copies of the somatostatin promoter are active in DTC1 cells. Promoter activity is dependent upon the ATFbinding site, consistent with the promoter being activated by endogenous EWS/ATF1. We produced stable transfectants from DTC1 cells containing somatostatin promoter-hsvtk cDNA fusions and tested their sensitivity to GCV. GCV is nontoxic and is converted into a cytotoxic DNA synthesis chain terminator by hsvtk. D71TK and D42TK represent pools of transfectants similar to D71 and D42 cells, except that the CAT reporter is replaced by hsvtk cDNA. In the absence of GCV, D71TK cells exhibit the same growth kinetics as parental DTC1 cells (Fig 1B). adenovirusding GCV (5 ␮g/mL) to the cell culture medium has no effect on the survival of parental DTC1 cells but severely inhibits the growth of D71TK cells, starting at day 4 and ending in cell death by day 12. Individual G418-resistant clones were isolated and tested for sensitivity to GCV (Fig 1C). Three independent clones (711, 714, and 716) isolated from D71TK cells and three independent clones (422, 423, and 424) isolated from D42TK cells were examined. After treatment with 0.1 ␮g/mL GCV for 10 days, the survival rates of clone 716 (48%), clone 714 (38%), and clone 711 (35%) were lower than the survival rates of all three clones (422, 423, and 424, average survival of 97%) from D42TK cells. At 0.3 ␮g/mL GCV, the survival rates at 10 days for 716 (8% survival), 714 (5%), and 711 (5%) were quite low, whereas the survival of 422, 423, and 424 was only slightly affected (average of 95%). Thus, D71TK cells are sensitive to GCV at concentrations previously reported to be toxic for other



399

Figure 1. Cytotoxicity of a somatostatin-hsvtk fusion gene in DTC1 cells. A: Somatostatin promoter activity in stably transfected DTC1 cells. Cell extracts were prepared from parental DTC1 cells (DTC1) or from pools of G418-selected cells containing the somatostatin promoter linked to CAT as a reporter. D42 is a pool of cells containing p⌬(⫺42)SomCAT (which lacks the ATF-binding site in the somatostatin promoter); D71 is a pool of cells containing p⌬(⫺71)SomCAT (including the ATF-binding site). PC12 represents a stable transfectant containing a chromosomal cAMP-inducible VIP-CAT reporter. A representative autoradiogram for CAT assays (conversion of chloramphenicol (c) to acetylated chloramphenicol (Ac)) is shown. B: GCV sensitivity of DTC1 cells containing a somatostatin-hsvtk fusion gene. Growth curves are shown for parental DTC1 cells or for a pool of stable DTC1 transfectants (D71TK) containing a stably integrated somatostatin-hsvtk fusion gene. GCV was included in the cell culture medium at a concentration of 5 ␮g/mL. C: GCV sensitivity of individual transfectants. Cells were treated with the indicated concentrations of GCV, and cell survival was determined by viable cell counting after 10 days of GCV treatment. Three independent clones (711, 714, and 716) isolated from D71TK cells and three independent clones (422, 423, and 424) isolated from D42TK cells were examined. Survival rates for the three control (EWS/ATF1-negative) cell lines (MEL28, JEG3, and RAT1) are also shown.


400


Figure 2. A: Transcriptional activation of selected promoters by EWS/ATF1 in JEG3 cells. Cells were transfected with 5 ␮g of reporter plasmid containing the test promoter (shown above) in the absence or presence of pSVEWS/ATF1 (5 ␮g) (shown below) made up to a total of 20 ␮g DNA with pSP64. CAT assays were performed at 40 hours posttransfection. Reporter plasmids are described in Materials and Methods. B: Structure of recombinant adenoviruss. The open box at the top represents the entire 36-kb adenovirus genome. ITR is the inverted terminal repeat sequence, the E3 gene is deleted (⌬E3), and foreign promoterCAT inserts into the E1 region to produce different recombinant viruses (AD42, AD71, AD25V, and ADVIP) are shown below. The transcription orientation for the CAT reporter gene is indicated by an arrow, and the test promoter sequences are indicated in black boxes with coordinates relative to the transcription start site (⫹1) indicated.

hsvtk-expressing tumor cells.35,47 As an additional test of specificity, we examined the effect of GCV (after introduction of the SomTK fusion gene) in three other mammalian cell lines (MEL28, JEG3, and RAT1) that do not express EWS/ATF1 (Fig 1C). Similar to clones 422, 423, and 424, MEL28 (90% survival), JEG3 (87%), and RAT1 (85%) were all only slightly affected by GCV treatment. In summary, the above results show that a somatostatin promoter-hsvtk fusion can confer GCVdependent cytotoxicity to DTC1 cells. This effect is cell-specific and (as indicated by the requirement for the ATF site in the somatostatin promoter) most likely results from somatostatin promoter activation by endogenous EWS/ATF1.

Design of recombinant adenoviruss for toxic gene delivery In a transient cotransfection assay EWS/ATF1 can activate several promoters that contain ATF-binding sites.30 Trans-activation of four such promoters by exogenous

EWS/ATF1 in JEG3 cells is shown (Fig 2A). VIP25CAT (containing the cAMP-response-element from the VIP promoter linked to the RSV TATA box), VIP4CAT (containing the intact VIP promoter to position ⫺94), and ⌬(⫺71)SomCAT are all strongly trans-activated (⬃100-fold) by EWS/ATF1. In addition, the adenovirus E3 promoter, which contains an ATF site required for trans-activation by cAMP,48 –50 is strongly activated by EWS/ATF1 in this assay (Fig 2A). We constructed recombinant adenoviruss containing the above promoters (and a control virus derived from ⌬(⫺42)SomCAT) linked to CAT (Fig 2B). Again, we initially used CAT instead of hsvtk as a simpler test for promoter activity, and we refer to the above promoters (AD71, AD42, AD25V, and ADVIP) as virus-borne ATF-dependent promoters (VAPs).

VAP activity during adenovirus infection DTC1 and Su-ccs-1 were infected with AD71 and AD42 at different MOIs, and CAT assays performed at 48



401

Figure 3. A: Viral promoter activity during infection of MMSP cells. The same samples of recombinant adenoviruss (shown in Fig 2) were used to infect DTC1, Su-ccs-1, and 293 cells at different MOIs. CAT assays were performed at 48 hpi unless otherwise indicated, and representative autoradiograms of CAT assays are shown. B: Transactivation using DTC1 cells in a transient assay. Test promoters were assayed after the transfection of DTC1 and JEG3 cells. Transfection conditions were as described in the legend to Figure 2, and the same precipitate for each test promoter was evenly divided and added to either DTC1 and JEG3 cells. CAT assays were performed at 40 hours posttransfection.

hours postinfection (hpi) (Fig 3A). At MOIs up to 100 (or 1000, data not shown), both AD71 and AD42 showed no detectable promoter activity in DTC1 or Su-ccs-1 cells. Using the transient assay described previously,30 we tested the ability of the DTC1 cells used in the above experiment to support activation of the somatostatin promoter (Fig 3B). As shown previously,30 the somatostatin promoter has high activity in DTC1 cells (comparable or greater than the activity of the SV40 and RSV promoters) compared with the near background levels observed in JEG3 cells (that lack endogenous EWS/ATF1). Thus, the lack of VAP activity observed in DTC1 cells is not due to loss of trans-activation capacity of the batch of DTC1 cells used. To confirm the integrity of the recombinant viruses, to test their infectivity, and to test whether activators other than EWS/ATF1 can activate VAPs, we infected 293 cells with AD42 and


AD71 and analyzed VAP activity by CAT assay (Fig 3A). 293 cells are useful for the above purposes because they express an endogenous adenovirus E1A gene that is an activator of many ATF-dependent promoters.46,48 For AD71, promoter activity is readily detectable at low MOIs and reaches a maximum at an MOI of between 20 and 100 (14.7% conversion in the CAT assay at an MOI of 100). For AD42, promoter activity was also detectable and saturated at an MOI of between 100 and 400 (1.4% conversion at an MOI of 400). The maximal promoter activity for AD71 is therefore ⬃10-fold higher than AD42. The results show that AD42 and AD71 are infectious in 293 cells, and that the somatostatin promoter can be activated in these cells either by E1A or by other viral activators produced during a viral infection. Similar to AD42 and AD71, AD25V and ADVIP exhibited no VAP activity in either DTC1 or Su-ccs-1 cells,

402


Figure 4. adenovirus replication and gene expression in MMSP cells. A: Cells were infected with WT or dl312 adenoviruss as described in Materials and Methods. At different times after infection, culture medium and remaining cells were freeze/thawed three times, the debris was removed by centrifugation, and viral titers were determined by plaque assay using 293 cells. B: E3 gene expression during adenovirus infection of MMSP cells. DTC1 and Succs-1 cells were infected with WT adenovirus (unless otherwise indicated) at different MOIs or for different times (as indicated above). Total RNA was extracted, and correctly initiated E3 transcripts (indicated to the side) were detected by primer extension using 32Plabeled primers as described in Materials and Methods. Experiments were repeated twice. C: Trans-activation of the somatostatin promoter by WT adenovirus. DTC1 or 293 cells were infected with recombinant AD71 virus (MOI ⫽ 100) in the absence or presence of WT adenovirus (MOI ⫽ 50). CAT assays were performed at 24 hpi, and a representative CAT assay is shown. Experiments were repeated twice.

whereas VAP activity was readily detected after infection of 293 cells (Fig 3A). Thus, the inability of endogenous EWS/ATF1 to activate VAPs in MMSP cells does not reflect a general repression of VAPs in infected cells.

Viral replication and gene expression in MMSP cells The low activity of VAPs in MMSP cells could be caused either by inefficient delivery of adenovirus to the nucleus of MMSP cells or the inability of EWS/ATF1 to activate transcription in the context of viral chromatin. To distinguish between these possibilities, we examined viral replication and gene expression in MMSP cells (Fig 4). Replication of wild-type (WT) adenovirus serves as a positive control for normal (E1A-dependent) replication, and an E1A-negative virus (dl312) is very similar to the recombinant viruses (AD42, AD71, ADVIP, and AD25V) used in our gene transfer experiments. Cells

were infected for virus replication experiments, and the virus produced was quantitated by plaque assay at different times after infection (Fig 4A). As a control for viral infections and plaque assays, 293 cells were included. WT virus propagates to high titer in 293 cells, and due to the presence of endogenous E1A in these cells, the replication of an E1A mutant virus (dl312) is similar to WT virus. Replication in 293 cells confirms the presence of infectious viruses for the experiments in MMSP cells. For both DTC1 and Su-ccs-1 cells, WT virus replicates to high titer (comparable with 293 cells), although the kinetics of replication are delayed relative to 293 cells. Maximal levels of virus production take 4 days for both DTC1 and Su-ccs-1 cells but only 2 days for 293 cells. As expected, dl312 showed limited ability to replicate in both DTC1 and Su-ccs-1 cells, and cell viability was not obviously affected even after 14 days of infection with dl312. The inability of dl312 to replicate in



MMSP cells indicates that the E1A-negative recombinant viruses used in our experiments do not replicate in MMSP cells. Consistent with the efficient replication of adenovirus in MMSP cells, the viral E1A protein is expressed in the nucleus (data not shown); in addition, we have shown previously that two adenovirus early promoters (E2 and E3) are active during a WT infection of MMSP cells, depending upon the viral E1A protein (Ref. 51, Fig 4B). To further characterize the behavior of adenovirus during infection of MMSP cells, we examined E3 promoter activity in more detail. Consistent with the delayed kinetics of viral replication in MMSP cells, the E3 promoter is also activated later than in 293 cells. In 293 cells, the E3 promoter is typically activated at ⬃6 hpi, whereas in MMSP cells the promoter has only low activity at 20 hpi and is strongly activated only at later times (between 20 and 40 hpi). Titration experiments using different MOIs show that E3 gene expression is saturated at a relatively low MOI (saturation is reached at an MOI of ⬍20, Fig 4B), indicating that MMSP cells take up the virus efficiently. The late onset of E3 gene expression is therefore probably not due to inefficient uptake of virus, but more likely reflects delayed kinetics for early viral gene expression. Because the E3 promoter contains an ATF1-binding site and is strongly activated by EWS/ATF1 in the transient assay (see Fig 2A), it is also of significance that, like other VAPs, the E3 promoter is not activated by endogenous EWS/ATF1 in the context of a viral chromatin. Taken together, the above data indicate that VAPs are effectively delivered to the nucleus of MMSP cells but that endogenous EWS/ATF1 is unable to activate these promoters. To test whether VAPs are generally repressed or whether they are specifically refractory to activation by EWS/ATF1, we asked whether E1A (which can activate VAPs in 293 cells (Fig 3A)) can activate VAPs in MMSP cells (Fig 4C). DTC1 cells were coinfected with WT adenovirus (containing E1A) and AD71, and CAT assays were performed at 24 hpi. Infection of 293 cells with AD71 was performed as a positive control for viral infection and CAT assays. As shown earlier, AD71 infection alone produces no detectable promoter activity in DTC1 cells, whereas coinfection with WT virus results in significant promoter activation. Thus, E1A protein (or possibly some other viral proteins) can activate VAPs in DTC1 cells. Activation during viral infection of DTC1 cells is not as great as that observed in 293 cells. This might reflect a lower ability of E1A to activate in DTC1 cells or the delayed kinetics for VAP activation as shown for the E3 promoter (Fig 4B). In summary, the above results show that the inability of EWS/ATF1 to activate VAPs is not due to a general repression mechanism (because E1A can activate VAPs), but rather is specific for EWS/ATF1 and/or MMSP cells. DISCUSSION We have evaluated the use of tumor-specific promoters and an adenovirus vector for cytotoxic prodrug treat-


403 ment of MMSP tumor cells in vitro. Using an integrated somatostatin promoter linked to hsvtk, GCV sensitivity can be conferred to DTC1 cells. This cytotoxic effect is specific for DTC1 cells and requires the ATF-binding site in the somatostatin promoter, indicating that cytotoxicity is caused by EWS/ATF1. The finding that the somatostatin promoter can be activated in MMSP cells (presumably by EWS/ATF1) when present in a chromosomal context is significant, because D71TK cells provide a cell system for the further evaluation of the cytotoxic prodrug approach for MMSP therapy. For example, in the future, D71TK cells can be used for tumor production in nude mice and subsequent testing of GCV sensitivity of tumors in intact animals. The above findings indicate that the somatostatin-hsvtk fusion tested (or hsvtk fusions to other promoters that are strongly activated by EWS/ATF1) has great potential for cytotoxic prodrug treatment of MMSP tumor cells in vivo. In contrast to previous studies that exploited adenovirus as a vector for cellular promoters,35,36,52–55 promoters that can be activated by EWS/ATF1 in transient assays are not activated in the context of adenovirus chromatin. VAPs are efficiently delivered to the nucleus of MMSP cells but cannot be activated by EWS/ATF1. Therefore, our findings indicate that first-generation adenovirus vectors (such as the one used in our study) may have limited potential as a gene delivery vehicle for cytotoxic prodrug treatment of MMSP cells. The reason for the inability of EWS/ATF1 to activate VAPs is unclear. A variety of VAPs are refractory to EWS/ATF1 and (as far as our studies go) this block is not restricted to position or orientation within the viral genome. In addition, the inability of EWS/ATF1 to activate resident viral promoters (E3 and probably E2) indicates that the lack of response of cellular promoters is not due to the absence of putative cis elements that are required for transcription in adenovirus chromatin. Efficient replication of adenovirus together with E3 promoter activation in MMSP cells demonstrates that these cells can support E1A-dependent transcription of VAPs. Therefore, our results are consistent with an inability of EWS/ATF1 to activate VAPs. In infected cells, viral chromatin is thought to be a distinct nucleoprotein complex (the adenovirus-core) containing the viral core proteins VII and V.56 In vitro transcription from adenovirus-core templates is repressed and requires a protein, template-activating factor-1, which has been suggested to modify the adenovirus-core to allow functional interactions with both replication and transcription factors.57 Therefore, it is possible that, in the absence of E1A, template-activating factor-1 (or other similar factors) is unable to modify the adenovirus-core, resulting in a block to activation by EWS/ATF1. However, such an effect would be relatively specific for EWS/ATF1 and/or MMSP cells, because many other promoters can be activated by their corresponding cellular activators (independently of E1A) when present in the adenovirus-core.35,36,52–55 Future in vitro studies

404 will be required to test the effect of adenovirus-core templates on trans-activation by EWS/ATF1. Irrespective of the protein factors involved, it is likely that cis-acting viral DNA sequences play a role (indirectly, due to their implied role in viral chromatin assembly) in preventing VAPs being activated by EWS/ ATF1. Therefore, it is possible that vector modification could overcome the effect. The use of larger regions of EWS/ATF1-dependent promoter sequences instead of the small truncated promoters used in our study is one option. Second-generation adenovirus vectors58,59 containing less of the viral genome enable this possibility to be tested by allowing larger foreign DNA inserts. Another option would be to test so called “gutless vectors” lacking all adenovirus genes.60,61 If the adenovirus-core proteins involved in viral chromatin assembly are indeed responsible for inhibiting EWS/ATF1, then the use of gutless vectors should overcome this problem. Other novel therapies being developed for EWSrelated sarcomas focus on inhibiting the function of EFPs inside the tumor cells. For Ewing’s sarcoma, expression of antisense RNA to EWS/Fli1 transcripts can successfully lower tumorigenicity.21,23,24 In addition, the development of small molecule inhibitors that target the EAD is an attractive long-term option. The EAD has a highly repetitive primary structure4,32 and makes distinctive contacts with the transcriptional machinery,62,63 indicating that the EAD is a likely target for small molecule inhibitors. However, because it is currently not known whether EWS/ATF1 is required for MMSP tumor maintenance, the potential that EAD inhibitors would hold for MMSP therapy is unclear. In view of this, the properties that we have described for tumor-specific EWS/ATF1-dependent promoters suggest that approaches to cytotoxic prodrug therapy for MMSP should be pursued. Currently, the main advantage of this approach is that it relies simply on the well-characterized tumor specificity and potent trans-activation properties of EWS/ATF1 without regard for the role of EWS/ATF1 in tumorigenesis. The use of second-generation Ad vectors or other methods29,34 for toxic gene delivery to MMSP cells remains a priority. ACKNOWLEDGMENTS We thank Dr. Helen Hurst (Imperial Cancer Research Fund, Hammersmith, UK) for supplying the E3CAT reporter plasmid and Dr. Chris Ring (Glaxo Wellcome, Stevenage, UK) for advice on recombinant adenoviruses. GCV was kindly supplied by the Department of Clinical Pathology, Tuen Mun Hospital (Hong Kong, Special Administrative Region, China). This work was supported by Hong Kong Government Research Grants Council Grant HKUST 645/96 M (to K.A.W.L).

REFERENCES 1. Rabbitts TH. Chromosomal translocations in human cancer. Nature. 1994;372:143–149. 2. Ladanyi M. The emerging molecular genetics of sarcoma translocations. Diagn Mol Pathol.. 1994;4:162–173.


3. Zucman J, Delattre O, Desmaze C, et al. EWS and ATF1 gene fusion induced by t(12;22) translocation in malignant melanoma of soft parts. Nat Genet. 1993;4:341–345. 4. Delattre O, Zucman J, Plougastel B, et al. Gene fusion with an ETS DNA-binding domain caused by chromosome translocation in human tumours. Nature. 1992;359:162– 165. 5. May WA, Gishizky ML, Lessnick SL, et al. Ewings sarcoma 11–22 translocation produces a chimeric transcription factor that requires the DNA-binding domain encoded by fli1 for transformation. Proc Natl Acad Sci USA. 1993;90:5752–5756. 6. May WA, Lessnick SL, Braun BS, et al. The Ewing’s sarcoma EWS/FLI-1 fusion gene encodes a more potent transcriptional activator and is a more powerful transforming gene than FLI-1. Mol Cell Biol. 1993;13:7393–7398. 7. Zucman J, Melot T, Desmaze C, et al. Combinatorial generation of variable fusion proteins in the Ewing family of tumours. EMBO J. 1993;12:4481– 4487. 8. Sorenson PHB, Lessnick SL, Lopez-Terrada D, Liu XF, Triche TJ, Denny CT. A second Ewing’s sarcoma translocation, t(21;22), fuses the EWS gene to another ETSfamily transcription factor, ERG. Nat Genet. 1994;6:146 – 151. 9. Ladanyi M, Gerald W. Fusion of the EWS and WT1 genes in the desmoplastic small round cell tumour. Cancer Res. 1994;54:2837–2840. 10. Chung EB, Enzinger FM. Malignant melanoma of soft parts. Am J Surg Pathol. 1983;7:405– 413. 11. Epstein AL, Martin AO, Kempson R. Use of a newly established human cell line (SU-CCS-1) to demonstrate the relationship between clear cell sarcoma to malignant melanoma. Cancer Res. 1984;44:1265–1274. 12. Hicks MJ, Saldivar VA, Chintagumpala MM, Cooley LD. Malignant melanoma of soft parts involving the head and neck region: review of literature and case report. Ultrastruct Pathol. 1995;19:395– 400. 13. Caraway NP, Fanning CV, Wojcik EM, Staerkel GA, Benjamin RS, Ordonez NG. Cytology of malignant melanoma of soft parts: fine-needle aspirates and exfoliative specimens Diagn Cytopathol. 1993;9:632– 638. 14. Kim J, Lee KAW, Pelletier J. The desmoplastic small round cell tumor t(11;22) translocation produces EWS/ WT1 isoforms with differing oncogenic properties. Oncogene. 1998;16:1973–1979. 15. Braun BS, Frieden R, Lessnick SL, May WA, Denny CT. Identification of target genes for the Ewing’s sarcoma EWS/FLI fusion protein by representational difference analysis. Mol Cell Biol. 1995;15:4623– 4630. 16. Thompson AD, Braun BS, Arvand A, et al. EAT-2 is a novel SH2 domain-containing protein that is up-regulated by Ewing’s sarcoma EWS/FLI1 fusion gene. Oncogene. 1996;13:2649 –2658. 17. Karnieli E, Werner H, Rauscher FJ III, Benjamin LE, LeRoith D. The IGF-1 receptor gene promoter is a molecular target for the Ewing’s sarcoma-Wilms’ tumor 1 fusion protein. J Biol Chem. 1996;271:19304 –19309. 18. May WA, Arvand A, Thompson AD, Braun BS, Wright M, Denny CT. EWS/FLI1-induced manic fringe renders NIH 3T3 cells tumorigenic. Nat Genet. 1997:17:495– 497. 19. Watson DK, Robinson L, Hodge DR, Kola I, Papas TS, Seth A. FLI1 and EWS-FLI1 function as ternary complex factors and ELK1 and SAP1a function as ternary and quaternary complex factors on the Egr1 promoter serum response elements. Oncogene. 1997;14:213–221. 20. Lee SB, Kolquist KA, Nichols K, et al. The EWS-WT1


405


21.

22. 23.

24.

25. 26. 27.

28. 29. 30.

31.

32.

33. 34. 35.

36.

37.

38.

translocation product induces PDGFA in desmoplastic small round-cell tumour. Nat Genet. 1997;17:309 –313. Ouchida M, Ohno T, Fujimura Y, Rao VN, Reddy ESP. Loss of tumorigenicity of Ewing’s sarcoma cells expressing antisense RNA to EWS-fusion transcripts. Oncogene. 1995;11:1049 –1054. Kovar H, Aryee DNT, Jug G, et al. EWS/FLI-1 antagonists induce growth inhibition of Ewing tumor cells in vitro. Cell Growth and Differ. 1996;7:429 – 437. Yi H-K, Fujimura Y, Ouchida M, Prasad DDK, Rao VN, Reddy ESP. Inhibition of apoptosis by normal and aberrant Fli-1 and erg proteins involved in human solid tumors and leukemias. Oncogene. 1997;14:1259 –1268. Tanaka K, Iwakuma T, Harimaya K, Sato H, Iwamoto Y. EWS-Fli1 antisense oligodeoxynucleotide inhibits proliferation of human Ewing’s sarcoma and primitive neuroectodermal tumor cells. J Clin Invest. 1997;99:239 –247. Lee KAW, Masson N. Transcriptional regulation by CREB and its relatives. Biochim Biophys Acta. 1993;1174: 221–233. Ribeiro A, Brown AD, Lee KAW. An in vivo assay for members of the CREB family of transcription factors. J Biol Chem. 1994;269:31124 –31128. Huber BE, Richards CA, Krenitsky TA. Retroviral gene therapy for the treatment of hepatocellular carcinoma: an innovative approach for cancer therapy. Proc Natl Acad Sci USA. 1991;88:8039 – 8043. Harris JD, Gutierrez AA, Hurst HC, Sikora K, Lemoine NR. Gene therapy for cancer using tumour-specific prodrug activation. Gene Ther. 1994;1:170 –175. Harris JD, Lemoine NR. Strategies for targeted gene therapy. Trends Genet. 1996;12:400 – 405. Brown AD, Lopez-Terrada D, Denny CT, Lee KAW. Promoters containing ATF-binding sites are de-regulated in cells that express the EWS/ATF1 oncogene. Oncogene. 1995;10:1749 –1756. Fujimura Y, Ohno T, Siddique H, Lee L, Rao VN, Reddy ESP. The EWS-ATF-1 gene involved in malignant melanoma of soft parts with t(12;22) chromosome translocation encodes a constitutive transcriptional activator. Oncogene. 1996;12:159 –167. Pan S, Koh YM, Dunn TA, Li KKC, Lee KAW. The EWS/ATF1 fusion protein contains a dispersed activation domain that functions directly. Oncogene. 1998;16:1625– 1631. Yeh P, Perricaudet M. Advances in adenoviral vectors: from genetic engineering to their biology. FASEB J. 1997; 11:615– 623. Anderson WF. Human gene therapy. Nature. 1998;392:25– 30. Kaneko S, Hallenbeck P, Kotani T, et al. Adenovirusmediated gene therapy of hepatocellular carcinoma using cancer-specific gene expression. Cancer Res. 1995;55:5283– 5287. Wills KN, Huang W-M, Harris MP, Machemer T, Maneval DC, Gregory RJ. Gene therapy for hepatocellular carcinoma: chemosensitivity conferred by adenovirus-mediated transfer of the HSV-1 thymidine kinase gene. Cancer Gene Ther. 1995;2:191–197. Hall SJ, Mutchnik SE, Chen SH, Woo SLC, Thompson TC. Adenovirus-mediated herpes simplex virus thymidine kinase gene and ganciclovir therapy leads to systemic activity against spontaneous and induced metastasis in an orthotopic mouse model of prostate cancer. Int J Cancer. 1997;70:183–187. Montminy MR, Sevarino KA, Wagner JA, Mandel G,


39.

40.

41. 42.

43. 44. 45. 46.

47.

48.

49. 50.

51. 52.

53.

54.

55.

Goodman RH. Identification of a cyclic-AMP-responsive element within the rat somatostatin gene. Proc Natl Acad Sci USA. 1986;83:6682– 6686. Gorman CM, Merlino GT, Willingham MC, Pastan I, Howard BH. The Rous sarcoma virus long terminal repeat is a strong promoter when introduced into a variety of eukaryotic cells by DNA-mediated transfection. Proc Natl Acad Sci USA. 1982;79:6777– 6781. Tsukada T, Fink JS, Mandel G, Goodman RH. Identification of a region in the human vasoactive intestinal polypeptide gene responsible for regulation by cyclic AMP. J Biol Chem. 1987;262:8743– 8747. Weeks DL, Jones NC. Adenovirus E3-early promoter: sequences required for activation by E1A. Nucleic Acids Res. 1985;13:5389 –5402. Bett AJ, Haddara W, Prevec L, Graham FL. An efficient and flexible system for construction of adenovirus vectors with insertion or deletions in early regions 1 and 3. Proc Natl Acad Sci USA. 1994;91:8802– 8806. Hearing P, Shenk T. The adenovirus type 5 E1A transcriptional control region contains a duplicated enhancer element. Cell. 1983;33:695–703. Kohler PO, Bridson WE. Isolation of hormone-producing clonal lines of human choriocarcinoma. J Clin Endocrinol Metab. 1971;32:683– 687. Gorman CM, Moffat LF, Howard BH. Recombinant genomes which express chloramphenicol acetyltransferase in mammalian cells. Mol Cell Biol. 1982;2:1044 –1051. Lee KAW, Green M. A cellular transcription factor E4F1 interacts with an E1A-inducible enhancer and mediates constitutive enhancer function in vitro. EMBO J. 1987;6: 1345–1353. Ido A, Nakata K, Kato Y, et al. Gene therapy for hepatoma cells using a retrovirus vector carrying herpes simplex virus thymidine kinase gene under the control of human ␣-fetoprotein gene promoter. Cancer Res. 1995;55: 3105–3109. Lee KAW, Hai T-Y, Siva-Raman L, et al. A cellular protein, activating transcription factor, activates transcription of multiple E1A-inducible adenovirus early promoters. Proc Natl Acad Sci USA. 1987;84:8355– 8359. Sassone-Corsi P. Cyclic AMP induction of early adenovirus promoters involves sequences required for E1A transactivation. Proc Natl Acad Sci USA. 1988;85:7192–7196. Lee KAW, Fink JS, Goodman RH, Green MR. Distinguishable promoter elements are involved in transcriptional activation by E1A and cyclic AMP. Mol Cell Biol. 1989;9:4390 – 4397. Li KKC, Lee KAW. MMSP tumor cells expressing the EWS/ATF1 oncogene do not support cAMP-inducible transcription. Oncogene. 1998;16:1325–1331. Kanai F, Shiratori Y, Yoshida Y, et al. Gene therapy for ␣-fetoprotein-producing human hepatoma cells by adenovirus-mediated transfer of the herpes simplex virus thymidine kinase gene. Hepatology. 1996;23:1359 –1368. Osaki T, Tanio Y, Tachibana I, et al. Gene therapy for carcinoembryonic antigen-producing human lung cancer cells by cell type-specific expression of herpes simplex virus thymidine kinase gene. Cancer Res. 1994;54:5258 –5261. Lan KH, Kanai F, Shiratori Y, et al. Tumor-specific gene expression in carcinoembryonic antigen-producing gastric cancer cells using adenovirus vectors. Gastroenterology. 1996;111:1241–1251. Tanaka T, Kanai F, Okabe S, et al. Adenovirus-mediated prodrug gene therapy for carcinoembryonic antigen-pro-

406

56. 57.

58. 59. 60.

ducing human gastric carcinoma cells in vitro. Cancer Res. 1996;56:1341–1345. Chatterjee PK, Vayda ME, Flint SJ. Adenoviral protein VII packages viral DNA throughout the early phase of infection. EMBO J. 1986;5:1633–1644. Matsumoto K, Okuwaki M, Kawase H, Handa H, Hanaoka F, Nagata K. Stimulation of DNA transcription by the replication factor from the adenovirus genome in a chromatin-like structure. J Biol Chem. 1995;270:9645– 9650. Gao GP, Yang Y, Wilson JM. Biology of adenovirus vectors with E1 and E4 deletions for liver-directed gene therapy. J Virol. 1996;70:8934 – 8943. Dedieu JF, Vigne E, Torrent C, et al. Long-term gene delivery into the livers of immunocompetent mice with E1/E4-defective adenoviruses. J Virol. 1997;71:4626 – 4637. Clemens PR, Kochanek S, Sunada Y, et al. In vivo muscle


gene transfer of full-length dystrophin with an adenoviral vector that lacks all viral genes. Gene Ther. 1996;3:965– 972. 61. Chen HH, Mack LM, Kelly R, Ontell M, Kochanek S, Clemens PR. Persistence in muscle of an adenoviral vector that lacks all viral genes. Proc Natl Acad Sci USA. 1997; 94:1645–1650. 62. Bertolotti A, Melot T, Acker J, Vigneron M, Delattre O, Tora L. EWS, but not EWS-FLI-1, is associated with both TFIID and RNA polymerase II: interactions between two members of the TET family, EWS and hTAFII68, and subunits of TFIID and RNA polymerase II complexes. Mol Cell Biol. 1998;18:1489 –1497. 63. Petermann R, Mossier BM, Aryee DN, Khazak V, Golemis EA, Kovar H. Oncogenic EWS-Fli1 interacts with hsRPB7, a subunit of human RNA polymerase II. Oncogene. 1998;17:603– 610.
