Adenovirus-mediated E2F-1 Gene Transfer Inhibits MDM2 Expression and Efficiently Induces Apoptosis in MDM2-overexpressing Tumor Cells¹

MDM2, the protein product of mdm2 oncogene, is frequently overexpressed in many human tumors, including sarcomas, gliomas, melanomas, and breast and esophageal cancers (1). Overexpression of MDM2 has been correlated with a more aggressive tumor phenotype in soft tissue sarcomas and osteosarcomas (2, 3). A study of 211 adult soft tissue sarcomas, which revealed that 37% of tumors overexpressed MDM2, also found a significant correlation between increased MDM2 expression and both high tumor grade and decreased survival (4). Similarly, amplification of the mdm2 gene has been demonstrated in 33% of recurrent or metastatic osteosarcomas, but not in primary tumors (5). These observations suggest that MDM2 contributes significantly to tumorigenesis and tumor progression and that therapeutic strategies designed to abrogate the effects of MDM2 may be effective.

MDM2 is thought to exert its major effects by binding to and inactivating p53. MDM2 binds directly to the transactivation domain of p53, inhibits the transcriptional activation function of p53, promotes p53 degradation, and inhibits the ability of p53 to induce cell cycle arrest and apoptosis (6, 7, 8, 9, 10, 11, 12, 13, 14). p53 is inactivated in the majority of human cancers by deletion or mutation of the gene, or by binding to viral oncoproteins (15). MDM2 overexpression in tumor cells, therefore, represents an alternative mechanism of p53 inactivation. Recently, it has been demonstrated that the INK4a tumor suppressor gene product, p19^Arf, binds to MDM2 and impairs the ability of MDM2 to inhibit p53 function (16).

Several factors suggest that MDM2 may have oncogenic effects that are independent of p53. First, although most human tumors with an amplified mdm2 gene have a wild-type p53 gene, sarcomas containing both p53 mutation and mdm2 gene amplification are associated with a worse prognosis than those with mutation of either gene alone (4). This indicates that MDM2 overexpression may result in accelerated tumor growth even in the absence of functional p53. Second, MDM2 overexpression has been shown to transform cells in culture in the absence of a functioning p53 gene (17). Finally, targeted MDM2 overexpression in mammary glands of mice in vivo leads to abnormal cell cycle regulation with uncoupling of S phase from mitosis, abnormal cell proliferation, and mammary hypertrophy to a similar degree in both a p53+/+ and p53−/− background (18). In a background of either E2F-1 deletion or E2F-1 overexpression in vivo, MDM2 overexpression had similar effects on mammary glands, suggesting that the effects of MDM2 are not related to direct interaction with E2F-1 (19). These findings suggest, however, that MDM2 has important regulatory properties, in addition to p53 inhibition.

The molecular interactions of MDM2 provide clues to possible important oncogenic mechanisms, in addition to p53 inactivation. MDM2 associates with another tumor suppressor, the retinoblastoma gene product, pRB. MDM2 has been shown to bind to the COOH terminus of pRB, impair the ability of pRB to inhibit E2F transcriptional activity, and inhibit the capacity of pRB to induce G₁ arrest (20). pRB exerts its cell cycle regulatory function by binding and sequestering the E2F family of cell cycle regulatory transcription factors, of which the actions of E2F-1 are best understood (21). Furthermore, MDM2 has been shown to bind directly to E2F-1 and augment the transcriptional activity of E2F-1 (22). In this way, MDM2 may encourage tumor cell growth by inhibiting p53- and pRB-mediated cell cycle arrest and apoptosis, while at the same time stimulating cell cycle progression from G₁ to S phase by interaction with E2F-1. Therefore, it is possible that E2F-1 regulation represents an important component of MDM2 function.

E2F-1 itself has complex and sometimes paradoxical properties. Initial studies suggested that E2F-1 can function as an oncogene when overexpressed (23, 24, 25). As a transcription factor, E2F-1 promotes cell cycle progression by activation of a series of genes that are critical for G₁ to S phase transition (21, 26). In this regard, E2F-1 can act as a potent growth-promoting factor. Overexpression of E2F-1 prevents cell cycle arrest of fibroblasts after serum deprivation, and stimulates quiescent cells to enter into S phase (27, 28). Furthermore, transgenic mouse experiments demonstrated that E2F-1 overexpression under the control of a keratin 5 promoter resulted in epidermal hyperplasia and, in a background of activated v-Ha-ras, led to spontaneous development of benign skin papillomas, suggesting that E2F-1 may, under some circumstances, promote tumor formation in vivo (29). However, E2F-1 overexpression restricted to the testis in vivo led to testicular atrophy due to apoptosis (30).

Other evidence suggests that E2F-1 also functions as a tumor suppressor. Homozygous E2F-1 null (knockout) mice demonstrate increased cell proliferation and tumor formation in several tissues (31, 32, 33, 34). In addition, overexpression of E2F-1 has been shown to induce apoptosis, which involves cooperation with p53 in several systems (28, 35, 36, 37) and is independent of p53 in others (38, 39, 40). This indicates that E2F-1 plays a role not only in stimulating cellular proliferation, but in coordinating programmed cell death.

MDM2-overexpressing cells have been shown to be resistant to gene therapy strategies that have been shown to be effective in other cell types. MDM2-overexpressing cells are resistant to the growth-suppressing effects of adenovirus-mediated p53 gene transfer (41). Adenovirus-mediated p21 overexpression resulted in growth arrest of MDM2-overexpressing cells, but apoptotic cell death was not evident (41). Recently, adenovirus-mediated E2F-1 expression has been investigated as a potential gene therapy strategy in glioma, breast, and ovarian cancers (42, 43, 44, 45). E2F-1 overexpression was found to induce cancer cell apoptosis, which did not require a functioning p53 gene. We hypothesized E2F-1 overexpression might be an effective gene strategy in MDM2-overexpressing tumors by exploiting the interaction of E2F-1 with MDM2. However, Kowalik et al. (46) recently demonstrated that adenovirus-mediated MDM2 overexpression inhibited E2F-1-induced apoptosis in REF52 fibroblasts. For this reason, we surmised that MDM2-overexpressing tumor cells might be resistant to adenovirus-mediated E2F-1 overexpression, as well. Results of the present studies indicate that, in tumors that overexpress MDM2, E2F-1 overexpression results in efficient induction of apoptosis, which is associated with inhibition of MDM2 expression.

OsACL, U2OS, and NIH3T3 cell lines were purchased from American Type Culture Collection (Manassas, VA). 3T3DM cells were the generous gift of Dr. Donna George (University of Pennsylvania Medical Center, Philadelphia, PA). OsACL and U2OS are human osteosarcoma cell lines that overexpress MDM2 (1, 47). The 3T3DM cell line (mouse fibrosarcoma) overexpresses MDM2 as a result of mdm2 gene amplification (48). NIH3T3 cells (mouse fibroblast) express normal levels of MDM2. All cell lines contain wild-type p53. OsACL cells were cultured in RPMI-1640 medium, and U2OS cells were cultured in McCoy’s 5A medium. NIH3T3 and 3T3DM cells were cultured in DMEM. All cell culture reagents were obtained from Life Technologies, Inc. (Bethesda, MD). All media were supplemented with 10% heat-inactivated fetal bovine serum and penicillin/streptomycin (100 units/ml). Cells were cultured in a 5% CO₂ incubator at 37°C, and the medium was changed every 3 days.

Two replication-defective recombinant adenoviral vectors were used. The Ad5CMV-E2F-1 vector has been deleted in the adenoviral E1 subunit and contains the transgene E2F-1 under the control of the CMV⁴ promoter (42). Ad5-CMV-nls-LacZ (referred to herein as Ad5CMV-LacZ and generously provided by Dr. B. French, University of Virginia) was used as a control vector that expresses nuclear-localized β-galactosidase under the same promoter (49). Both vectors were propagated in the 293 cell line and titered using standard plaque assays (50). For infections, 1 × 10⁶ cells were plated in 10-cm tissue culture plates. The following day, the media was removed and cells were infected by adding the adenoviral vectors in 1-ml α-MEM at the MOI of 100 plaque-forming units/cell. Mock infection was performed by treatment of cells with vehicle (media) only. One hour after incubation at 37°C, the medium was removed and 10 ml of fresh α-MEM with 5% FBS was added. Cells were harvested at specific time points for analysis.

The cell lines infected with Ad5CMV-LacZ were assayed for β-galactosidase expression by the X-gal staining method, as described previously (51). Briefly, 48 h after infection, the cells were washed with PBS and fixed in 2% (v/v) formaldehyde and 0.2% (v/v) glutaraldehyde in PBS (pH 7.4) for 5 min at 4°C. The cells were then washed and stained with X-gal solution [1 mg/ml 5-bromo-4chloro-3-indolyl-b-galactopyranoside, 5 mm K₄Fe(CN)₆, 5 mm K₃Fe(CN)₆, and 2 mm MgCl₂ in PBS (pH 6.5)] for 12–18 h at 37°C. Blue staining of cell nuclei identified transduced cells. Mock-infected cells and cells transduced with other adenoviral vectors served as controls.

Cells were treated by mock infection or infection with Ad5CMV-E2F-1 or Ad5CMV-LacZ at the MOI of 100. Cells were harvested at selected time points and lysed in radioimmunoprecipitation assay lysis buffer (50 mm Tris-HCl, 150 mm NaCl, 1% NP40, 0.5% sodium deoxycholate, and 0.1% SDS) with a protease inhibitor mixture [4-(2-aminoethyl)-benzenesulfonyl fluoride, pepstatin A, transepoxysuccinyl-l-leucylamido(4-guanidino)butane(E-64), bestatin, leupeptin, and aprotinin; 10 μl/10⁶ cells; Sigma Chemical Co., St. Louis, MO] for 30 min. Cell lysates were centrifuged, and protein concentration was determined by BIO-RAD DC protein assay (Bio-Rad, Hercules, CA). Equal amounts of cellular protein were electrophoresed in 8% (E2F-1, PARP, MDM2), or 12% (CPP32, Bax, Bcl-2, p53) SDS-polyacrylamide gels and transferred to a Hybond-PVDP membrane (Amersham Corp., Arlington Heights, IL). The membrane was first incubated with the following primary antibodies: mouse antihuman E2F-1 mAb (Santa Cruz Biotechnology, Santa Cruz, CA); mouse-antihuman MDM2 mAb (Calbiochem, Oncogene Research Products, Cambridge, MA); mouse-antihuman-CPP32 mAb (Transduction Laboratory, Lexington, KY); mouse-anti-p53 mAb (Calbiochem, Oncogene Research Products); rabbit-antihuman-Bax pAb and rabbit-antihuman-Bcl-2 pAb (PharMingen, San Diego, CA); and then with antimouse immunoglobulin or antirabbit immunoglobulin, peroxidase-linked, species-specific whole antibody (Amersham Corp.). Enhanced chemiluminescence reagents were used to detect the signals, according to the manufacturer’s instructions (Amersham Corp.).

Cell proliferation was assessed 24, 48, and 72 h after infection by measuring the conversion of the trazolium salt WST-1 to formazan, according to the manufacturer’s instructions (Boehringer Mannheim, Indianapolis, IN). Briefly, cells were plated into 12-well plates and infected with the adenoviral vectors 24 h later. At each time point, 50 μl of WST-1 were added to each well and cultured at 37°C for 1.5 h. The supernatant from each plate was collected for measurement of absorbance at 415 nm and 650 nm (the latter as a reference wavelength). Under the experimental conditions of the present studies, there is a direct correlation between the absorbance at 415 nm and the cell numbers. The results were expressed as the percentage of the absorbance of control (uninfected) cells. Cell viability was assessed by the trypan blue exclusion. Cells were stained with 0.4% trypan blue for 5 min and counted using a hemacytometer.

Both adherent and nonadherent cells were harvested, washed once with PBS, and fixed in 70% (v/v) ethanol at −20°C for up to 1 week. Cells were pelleted, washed once with PBS, and resuspended in propidium iodide solution [50 μg/ml propidium iodide and 0.5 mg/ml RNase in PBS (pH 7.4)] for 30 min in the dark. Flow cytometric analysis was performed on a FACScan Flow Cytometer (Becton Dickinson, San Jose, CA). The data from 10,000 cells were collected and analyzed using CellFIT Cell-Cycle Analysis Software (Version 2.01.2). The subdiploid population was calculated as an estimate of the apoptotic cell population.

Several methods were used to confirm apoptotic cell death. First, cellular morphology was assessed by harvesting both adherent and nonadherent cells, which were then washed once with PBS. Cytospins were air-dried and stained with the standard Wright and Giemsa method (52). Second, in situ TUNEL assay identifies internucleosomal DNA strand breaks characteristic of apoptosis. Cells were fixed in 4% formaldehyde in PBS (pH 7.4) for 15 min at room temperature. After centrifugation, cells were resuspended in 80% ethanol and stored at 4°C for up to 1 week. A TdT-FragEL DNA fragmentation detection kit (Calbiochem, Oncogene Research Products) was used to detect apoptosis, according to instructions provided by the manufacturer. Third, PARP cleavage assay has been shown to be a sensitive method for detection of apoptosis (53). PARP cleavage assays were performed using a monoclonal mouse anti-PARP antibody (Calbiochem, Oncogene Research Products) at a dilution of 1:100.

To estimate the transduction efficiency of adenoviral vectors, cell lines U2OS and OsACL were infected with Ad5CMV-LacZ at the MOI of 100. Transduction efficiency was determined by measuring the percentage of blue cells following cytochemical staining with X-gal 48 h after infection. Greater than 95% transduction efficiency was detected in these cell lines (data not shown). No significant cytotoxic effect was noted at this viral concentration.

Infection of OsACL and U2OS cells with Ad5CMV-E2F-1 resulted in substantial overexpression of E2F-1 at 24 and 48 h (Fig. 1). In contrast, the baseline expression of E2F-1 was undetectable in mock-infected and Ad5CMV-LacZ-infected control groups.

Proliferation assays were performed to evaluate the effect of adenovirus-mediated E2F-1 expression on cell growth in vitro. After mock infection or infection with Ad5CMV-LacZ or Ad5CMV-E2F-1, WST-1 assay was performed. At 24 h after infection, no growth inhibition was observed in Ad5CMV-E2F-1-infected cells when compared with control groups (Fig. 2). By 48 h, E2F-1-overexpressing cells exhibited a 2-fold growth inhibition compared with controls, and many of the cells detached from the tissue culture plates and appeared shrunken. By 72 h, <15% of OsACL cells and almost no U2OS cells were adherent in the Ad5CMV-E2F-1-infected groups, and substantial growth inhibition was evident. In contrast, Ad5CMV-LacZ- and mock-infected cells demonstrated increasing growth and remained adherent at 72 h.

At 24 h after infection, nearly 100% of cells in all groups were viable (Fig. 3). By 72 h, however, <15% of E2F-1-overexpressing cells were viable. Little toxicity was seen with Ad5CMV-LacZ infection. Rapid loss of cell viability was also observed in 3T3DM and NIH3T3 cells after infection with Ad5CMVE2F-1 (data not shown). These data demonstrate the E2F-1-mediated inhibition of cellular proliferation is related to widespread loss of cell viability.

To investigate the mechanism of E2F-1-induced sarcoma cell death, cell cycle analysis was performed. As shown in Fig. 4 and Table 1, E2F-1 overexpression in U2OS cells resulted in a substantial increase in the S phase population by 24 h, which continued to increase up until 48 h. A subdiploid population of U2OS cells, characteristic of apoptotic cell death, is evident at 24 h after Ad5CMV-E2F-1 infection, which increased up to 72 h. In the OsACL cell line, S phase accumulation was evident by 48 h after Ad5CMV-E2F-1 infection. By 96 h, however, a substantial subdiploid population was present, as well as a large S phase population. Neither mock-infected nor Ad5CMV-LacZ-infected cells demonstrated substantial changes in cell cycle profiles. These data suggest that overexpression of E2F-1 promotes premature and unscheduled entry into S phase, followed by apoptotic cell death.

Several experiments were performed to verify that E2F-1-induced cell death in the MDM2-overexpressing osteosarcoma cell lines was the result of apoptosis. First, cell morphology demonstrated typical changes characteristic of apoptotic cell death, including cell shrinkage, cytoplasmic blebbing, chromatin condensation, nuclear fragmentation, and formation of apoptotic bodies (data not shown). These changes were not seen in mock-infected or Ad5CMV-LacZ-infected cells.

Apoptosis is associated with internucleosomal degradation of genomic DNA, which can be assessed by in situ TUNEL staining. Using this approach, it is possible to confirm that DNA cleavage has occurred and free 3′-OH groups are generated by cellular endonucleases (54). At 72 h after infection, Ad5CMV-E2F-1-infected cells demonstrated abundant TUNEL staining (Fig. 5). In contrast, neither mock-infected nor Ad5CMV-E2F-1-infected cells showed significant evidence of internucleosomal DNA fragmentation.

Little is known about the mechanism by which E2F-1 induces apoptosis. In some cells, it has been suggested that increased levels of p53 may mediate the effects of E2F-1 overexpression (28, 35, 36, 37). In OsACL and U2OS cells, which overexpress MDM2, p53 levels are normally undetectable, and remained undetectable after infection with Ad5CMV-E2F-1 (Fig. 6,A). p53 expression is detectable after UV irradiation (Fig. 6 B). Therefore, it seems that effects of E2F-1 overexpression in these MDM2-overexpressing cells are not mediated by significant increases in p53 levels.

E2F-1 overexpression in OsACL and U2OS cells resulted in a marked decrease in MDM2 expression (Fig. 7,A). Apoptosis can also result from a relative increase in the proapoptotic protein Bax or decrease in the antiapoptotic protein Bcl-2. Bcl-2 levels were undetectable at all time points in OsACL and U2OS cell lines after infection (data not shown). In OsACL cells, at 48 h a faster migrating Bax-reactive band was evident, without change in the intensity of the Bax-reactive p21 band, but by 72 h the faster migrating band predominated (Fig. 7 B). In U2OS cells, Bax levels declined in Ad5CMV-E2F-1-infected cells at 24 h, recovered by 48 h, and by 72 h a faster migrating band is also seen.

Involvement of the caspase cascade in E2F-1-mediated apoptosis was suggested in a recent study of human glioma cells (30). To evaluate the role of caspase cascade activation in E2F-1-mediated apoptosis of MDM2-overexpressing osteosarcoma cells, Western blot analysis of caspase 3/CPP32 and PARP was performed. There is a decrease in the proenzyme CPP32 levels in both cell lines after E2F-1 overexpression (Fig. 8,A). An apoptosis-specific PARP cleavage fragment (85–90 kDa) was present in both cell lines infected with Ad5CMV-E2F-1, but not in controls (Fig. 8 B). Taken together, these data extend the observation that the caspase cascade is involved in E2F-1-induced apoptosis to osteosarcoma cells and suggest that this may be a common mechanism.

Overexpression of E2F-1 has been shown to induce apoptosis in several cell types, both normal and malignant (28, 35, 36, 37, 38, 39, 40, 42, 43, 44, 45). Although the mechanisms by which E2F-1 induces apoptosis are not completely understood, it has been suggested that apoptosis results from incompatible signals for proliferation and cell cycle arrest. One such set of conflicting signals is the concomitant stimulation of E2F-1 and p53 activity. Previous studies have shown that E2F-1 and p53 cooperate to mediate apoptosis (28, 35, 38, 46). For example, E2F-1-induced apoptosis in fibroblasts is potentiated by high levels of endogenous wild-type p53 (28, 35). There is also evidence that overexpression of E2F-1 induces the accumulation of p53, again implicating p53 in E2F-1-mediated apoptosis (46, 55). Although it was originally believed that p53 was essential for E2F-1-mediated apoptosis, it is now clear that p53 is not always required (38, 39, 40, 42, 43, 44, 45). In the present study, p53 levels remained undetectable by Western blot analysis after E2F-1 overexpression, suggesting that apoptosis in these cells is independent of p53 function, as well.

Recently, it has been shown that the apoptotic function of E2F-1 is separable from the ability to accelerate entry into DNA synthesis—COOH-terminal truncation mutants of E2F-1 lacking the transactivation domain efficiently induce apoptosis without stimulating DNA synthesis (38, 39). The exact contribution of these two functions of E2F-1 in promoting apoptosis in MDM2-overexpressing tumors remains unclear. For example, unscheduled entry into S phase initiated by E2F-1 overexpression may trigger a series of events that lead to apoptosis. Alternatively, E2F-1-mediated apoptosis in these tumors can be completely explained by the apoptotic function of E2F-1, with no contribution of the transactivation function of this protein. Although the apoptotic and DNA synthesis-promoting functions of E2F-1 can be dissociated, each function may contribute to apoptotic death in MDM2-overexpressing tumor cells. Additional studies are under way to characterize further the domains of E2F-1 that are essential to induce apoptosis in these cells.

We have shown that adenovirus-mediated E2F-1 overexpression resulted in rapid and efficient apoptotic death of MDM2-overexpressing osteosarcoma and fibrosarcoma cells. This is of interest considering the results of Kowalik et al. (46), who found that adenovirus-mediated overexpression of MDM2 inhibited apoptosis induced by adenovirus-mediated E2F-1 gene transfer in the REF52 fibroblast cell line. In that study, however, E2F-1 overexpression was associated with a concomitant increase in p53 levels, and MDM2 overexpression blocked both p53 accumulation and E2F-1-mediated apoptosis. The effect of E2F-1 overexpression on levels of MDM2 after coinfection was not examined. In the present study, E2F-1 overexpression was not associated with p53 accumulation. However, in comparing our results with those of Kowalik et al. (46), it should be noted that there are substantial differences in the cell lines and experimental conditions, the MOI of adenovirus used, and, possibly, the relative levels of E2F-1 and MDM2 expression, which may account for the differences in results. Furthermore, it is likely that there is substantial variability among tumor types in p53-dependent and -independent mechanisms that mediate the apoptotic effects of E2F-1.

Fueyo et al. (43) found that E2F-1 overexpression in glioma cells resulted in a decrease in intracellular levels of the proapoptotic protein Bax. A similar finding was evident in U2OS cells at 24 h after infection with Ad5CMV-E2F-1, although this decline was transient. In OsACL cells, and to a lesser degree in U2OS cells, a lower molecular weight form of Bax was observed, coincident with the development of apoptosis. This could represent a Bax cleavage product related to caspase (56), calpain (57), or other proteinases. Furthermore, E2F-1-induced apoptosis was not related to changes in levels of the antiapoptotic protein Bcl-2. In human glioma cells, the caspase cascade has been implicated as a mediator of E2F-1-mediated apoptotic cell death (43). In OsACL and U2OS cells, CPP32 activation and PARP cleavage were evident after Ad5CMV-E2F-1 infection, indicating that the caspase cascade is involved in E2F-1-mediated apoptosis in MDM2-overexpressing osteosarcoma cells, as well. Caspase cascade activation may be a universal feature of E2F-1-mediated apoptosis.

In the present studies, adenovirus-mediated E2F-1 overexpression resulted in decreased levels of MDM2. This is a novel finding and suggests that E2F-1-mediated growth inhibition and apoptosis may result, at least in part, from inhibition of MDM2 function. Alternatively, MDM2 down-regulation may simply correlate with the induction of apoptosis, but play no functional role in promoting apoptotic cell death. Although the mechanisms underlying this decrease in MDM2 levels are unclear, MDM2 is known to be a caspase 3/CPP32 substrate (58, 59, 60). It is possible that the decrease in MDM2 levels is related to caspase cleavage of MDM2. Furthermore, MDM2 overexpression in some cell types is maintained by translational enhancement, and this may be affected by E2F-1 (1). Alterations of mdm2 transcription, translation, posttranslational modifications, or protein stability are also possible. Of note, however, is the fact that the decline in MDM2 levels was not accompanied by a detectable increase in p53 levels.

In conclusion, overexpression of E2F-1 by adenovirus-mediated gene transfer results in rapid and efficient apoptotic cell death of MDM2-overexpressing tumor cells, which does not seem to involve p53. E2F-1-mediated apoptosis was associated with diminished levels of MDM2 and activation of the caspase pathway. Because MDM2-overexpressing tumors are resistant to p53 gene therapy, adenovirus-mediated E2F-1 gene transfer may be a promising alternative. Further investigation into the mechanisms of E2F-1-mediated apoptosis is ongoing.

We are grateful to Dr. Donna George for providing the 3T3DM cells, Dr. Brent French for providing the Ad5CMV-LacZ vector, and Sherri Matthews for expert assistance with manuscript preparation.