CRC Molecular and Cellular Pharmacology Research Group, School of Biological Sciences, The University of Manchester, G.38, Stopford Building, Oxford Road, Manchester, M13 9PT, UK

^aAuthor for correspondence

The Mdm2 protein is frequently overexpressed in human non-seminomatous germ cell tumours and transitional carcinoma of the bladder where it may contribute to tolerance of wtp53. Mdm2 forms an autoregulatory feedback loop with p53; the Mdm2 gene is responsive to transactivation by p53 and once synthesized the Mdm2 protein terminates the p53 response. We show here that the topoisomerase poison etoposide, like ultra violet irradiation, inhibits Mdm2 synthesis. Cytotoxic concentrations of etoposide (IC90 for >3 h) result in inhibition of Mdm2 induction at both the RNA and protein level. Rapid apoptosis ensues. Global transcription is not inhibited: p21^waf-1/cip-1 and GADD45 expression increase in a dose dependent manner. Inhibition of Mdm2 synthesis depends on the continuous presence of etoposide, suggesting the DNA damage may prevent transcription. Downregulation of Mdm2 transcript occurs in cells expressing HPV16-E6 suggesting that inhibition of Mdm2 transcription is p53-independent. When cells are treated with a pulse (1 h) of etoposide and reincubated in drug free medium, Mdm2 synthesis commences immediately after damage is repaired (3 h) and the p53 response is attenuated. Induction of apoptosis and loss of clonogenicity are 3 - 5-fold lower under pulse treatment conditions. This is the first observation of inhibition of Mdm2 transcription following treatment with topoisomerase (topo II) poisons, a feature that may be useful in tumour types where p53 is tolerated by overexpression of Mdm2.

testicular neoplasms; p53; Mdm2; etoposide; apoptosis

SSB, single-strand breaks; FBA, filter binding assay; ECL, enhanced chemiluminescence; IC, inhibitory concentration - derived from clonogenic assays; SST, sequence specific transcriptional activity; TGCT, testicular germ cell tumour; TdT, terminal deoxy nucleotide transferase; RE, rad equivalent

Introduction

Mdm2 overexpression provides an alternative mechanism to inactivate p53 in tumours where there are no mutations in the p53 gene. Mdm2 has been implicated in the pathogenesis of over 40% of human tumours (for review see Haines, 1997). The mdm2 proto-oncogene was originally identified as an amplified gene present on episomes (mouse double minute chromosome) in a spontaneously transformed mouse cell line (Fakharzadeh et al., 1991). Mdm2 was shown to be transforming by enforced overexpression in immortalized NIH3T3 cells, where it induced tumorigenicity (Fakharzadeh et al., 1991). In human tumours, overexpression of Mdm2 results from either gene amplification (e.g. in soft tissue sarcomas and breast; Oliner et al., 1992) or enhanced transcription and translation (e.g. in testicular germ cell tumours and transitional carcinoma of the bladder; Barbareschi et al., 1995; Fleischhacker et al., 1994; Landers et al., 1997; Lianes et al., 1994; Riou et al., 1995).

The transforming properties of Mdm2 can be attributed to its interactions with key proteins involved in regulation of cell cycle progression and apoptosis. In common with the DNA tumour viruses, adenovirus and papilloma virus, Mdm2 inactivates both the p53 and Rb tumor suppressor proteins (Levine, 1990; Momand et al., 1992; Oliner et al., 1993; Weinberg, 1993; Xiao et al., 1995). Mdm2 regulates p53 function through two mechanisms. Firstly, it binds to the transactivation domain of p53 and inhibits the ability of p53 to transactivate target genes (Chen et al., 1995; Momand et al., 1992; Oliner et al., 1993). Secondly, it targets p53 for proteosome- dependent degradation regulating the extent and duration of the p53 response (Haupt et al., 1997; Kubbutat et al., 1997). Following induction of p53 by UV irradiation the p53-dependent upregulation of Mdm2 is delayed until activation of p53-dependent target genes such as p21^waf-1/cip-1 and GADD45 are complete. Mdm2 then terminates the p53 response (Chen et al., 1994; Reinke and Lozano, 1997; Wu and Levine 1997). Through these activities Mdm2 inhibits both the growth arrest and apoptosis inducing activities of p53 (Chen et al., 1996; Finlay 1993; Haupt et al., 1996; Kondo et al., 1995; Momand et al., 1992; Oliner et al., 1993). Mdm2 also regulates Rb by protein-protein interactions. Mdm2 can bind to and inactivate the growth inhibitory function of Rb even when superactive pocket mutants (insensitive to kinase inactivation) are employed (Xiao et al., 1995).

During normal development the Mdm2/p53 interaction is critical and deletion of the mdm2 gene causes lethality in mice carrying a functional p53 gene (Jones et al., 1995; Montes-de-Oca-Luna et al., 1995). However, the role of Mdm2 in tumour development probably also depends on the p53-independent functions of Mdm2. Patients with sarcomas overexpressing both Mdm2 and p53 have a worse prognosis than those overexpressing p53 alone (Cordon-Cardo et al., 1994). Mdm2 can abrogate Rb and p107-induced growth arrest and provide a growth advantage in cells lacking p53 (Dubs-Poterszman et al., 1995). Moreover, Mdm2 interacts directly with the S-phase promoting transcription factor E2F1 (Martin et al., 1995; Xiao et al., 1995).

In an earlier study we demonstrated that the p53 of testicular germ cell tumours (TGCT) and transitional cell carcinoma (TCC) of the bladder was transcriptionally active after etoposide treatment (Chresta et al., 1996). In the clinic these two urological tumours frequently overexpress Mdm2 and this could block p53-mediated apoptosis and cell cycle arrest (Barbareschi et al., 1995; Fleischhacker et al., 1994; Guillou et al., 1996; Riou et al., 1995). We therefore investigated the p53/Mdm2 autoregulatory feedback loop under control conditions and after etoposide treatment. We report here that Mdm2 was overexpressed in the absence of gene amplification in both TGCT and TCC. Importantly we found that etoposide was able to perturb the p53/Mdm2 feedback loop. Etoposide inhibited Mdm2 expression whilst inducing the expression of other p53 target genes GADD45 and p21^waf-1/cip-1. Defining the pathways of Mdm2 overexpression and analysis of the p53/Mdm2 autoregulatory feedback loop provides information which can be used to facilitate restoration of p53 function in human tumours.

Results

Overexpression of Mdm2 in testicular and bladder tumour cells tolerating wtp53

Figure 1a shows that Mdm2 protein is overexpressed in all testis tumour lines (TGCT) and in a bladder tumour cell lines which has wtp53 (RT4). The Mdm2 levels are 3 - 13-fold higher in TGCT and RT4 compared to either normal testis or bladder tumour lines with non-functional p53 (RT112 and HT1376). India ink staining of the immunoblot shows that the absence of Mdm2 in the normal testis and RT112 and HT1376 is not a product of under loading (there is slight overloading in the normal sample and 833K). The full length 5.5 kb Mdm2 transcript is also relatively highly expressed in the tumour cell lines tolerating wtp53 (TGCT and RT4) compared to those with mutant p53 (RT112 and HT1376) (Figure 1b). Analysis of Mdm2 gene copy number by Southern blotting showed the Mdm2 gene is not amplified, suggesting upregulation of Mdm2 is at the level of transcription/translation (Figure 1c).

Regulation of the p53/Mdm2 feedback loop after etoposide treatment

Since overexpression of Mdm2 is known to inhibit the transactivation activity of p53 (see Introduction) and since we have previously shown p53 in TCC and TGCT is transcriptionally active after etoposide treatment it was of interest to determine the effect of etoposide on Mdm2 expression. The two cell lines studied were the TGCT cell line 833K and the TCC cell line RT4 both of which have wtp53 by sequence analysis (deJong et al., 1997; Rieger et al., 1995). Equitoxic concentrations of etoposide (derived from clonogenic assays) were used in both cell types. In the TGCT cell line the etoposide concentration ranged from 3.3 M (99% cells viable) to 52 M (99% cells dead), whereas in the etoposide resistant cell line, RT4, the concentrations ranged from 80 (10% dead) to 240 M (90% dead).

In 833K testicular tumour cells following continuous treatment with etoposide, p53 protein levels increase by approximately fivefold at all doses (4 h incubation), whereas stimulation of Mdm2 transcript levels is maximal at a concentration of etoposide that kills only 10% of cells (IC₁₀) and is inhibited above an IC₅₀ (Figure 2a,b). The activity of p53 is not inhibited at these high doses as other p53-dependent genes p21^waf-1/cip-1 and GADD45 show a dose dependent increase (Figure 2c,d). The expression of the 1 kb Bax transcript and GAPDH is unchanged by drug treatment demonstrating there is not a general down regulation of transcription or degradation of RNA (Figure 2c). Identical results have been found in another germ cell tumour line (GH).

We extended these studies to include a transitional carcinoma of the bladder cell line in which we have shown p21 upregulation is p53-dependent (Figure 3). The concentrations of etoposide required to induce DNA damage and inhibit the colony formation of the chemosresistant bladder tumour cells (RT4) are much higher than those required to result in cell death in the TGCT cell lines (Chresta et al., 1996). However, as can be seen in Figure 2e the effects of etoposide on Mdm2 and other p53-dependent genes are similar. At the lowest dose of etoposide employed, IC₁₀ (82 M), Mdm2 upregulation is completely inhibited, whereas p21 and GADD45 are increased 3 - 4-fold. Treatment of RT4 with IC₅₀ or above of etoposide resulted in downregulation of Mdm2 relative to control and reduced the upregulation of p21 and GADD45 (Figure 2e). High dose etoposide (IC₅₀ or greater) had no effect on GAPDH levels in RT4 suggesting the effects are not a result of general RNA degradation (see Figure 3b).

Inhibition of Mdm2 expression by etoposide is p53-independent

High dose UV-irradiation (10 and 20 J/m²) similarly results in differential regulation of Mdm2 and p21; Mdm2 synthesis is delayed until damage is repaired (Reinke and Lozano, 1997; Wu and Levine 1997). Downregulation of Mdm2 transcript levels in response to high dose UV-irradiation has been attributed to a p53-independent UV responsive repressor of Mdm2 transcription (Wu and Levine 1997). To investigate if the etoposide effects are also p53-independent, we examined the inhibition of Mdm2 expression in RT4 cells engineered to express human papilloma virus E6 protein. The E6 expressing cells have undetectable levels of p53 protein and fail to upregulate p21 RNA or protein in response to etoposide treatment (Figure 3a,b). However, as can be seen in Figure 3b down regulation of Mdm2 by an IC₅₀ concentration of etoposide was unaffected by the presence of E6, demonstrating that the inhibitory effects of etoposide are p53-independent.

Inhibition of Mdm2 transcription is reversible following repair of etoposide-induced DNA damage

Inhibition of Mdm2 by etoposide and UV-irradiation could result from activation of an etoposide/UV-responsive repressor of Mdm2 transcription (Wu and Levine 1997). Alternatively, etoposide (and possibly UV) may act directly to inhibit transcription. Etoposide is a topoisomerase II (topo II) poison which inhibits the resealing activity of topo II, resulting in a bulky DNA protein complex. The presence of etoposide and UV-induced lesions in DNA can directly impair transcription (Ljungman and Zhang, 1996; Riou et al., 1993). If the Mdm2 transcription is inhibited by etoposide damage then removal of etoposide, which results in reversal of etoposide topo II complex, should allow transcription to proceed. Figure 4 demonstrates that both the etoposide-induced single strand breaks and the repression of Mdm2 synthesis are readily reversed after drug removal. The levels of Mdm2 RNA and protein are much lower in response to continuous treatment with an IC₁₀ or IC₉₀ of etoposide compared to when repair is allowed after pulse treatment (Figure 4b, top panels). Drug removal also results in a decrease in p53 levels, probably as a result of Mdm2 upregulation (Figure 4b, lower panel). The suppression of GADD45 and p21 synthesis seen at high concentrations of etoposide in RT4 is also relieved by drug removal (results not shown). Results shown are for the TGCT 833K cell line, identical findings have been seen in another TGCT cell line (GH).

Rapid activation of the Mdm2 autoregulatory feedback loop after drug removal

The results in Figure 4b demonstrate that when Mdm2 protein is upregulated (following pulse treatment) there is subsequent p53 downregulation. To analyse this further we determined the kinetics of Mdm2 induction and p53 decline following a short pulse treatment with etoposide. In 833K cells p53 is elevated at the 2 h timepoint (immunoblot in Figure 5a). However, by 4 h, p53 protein levels decrease, coincident with Mdm2 synthesis. Downregulation of p53 at 4 h just precedes downregulation of the p21^waf-1/cip-1 transcript (Figure 5b). This suggests the Mdm2 autoregulatory feedback loop rapidly terminates the p53 response after etoposide damage is repaired. As can be seen in Figure 5c the levels of p21 RNA and p53 protein are reduced when cells are pulse treated allowing repair. The transient decrease in Mdm2 protein (but not RNA) at 7 h results from proteolysis by caspases (Erhardt et al., 1997) (Figure 5a). This timepoint is co-incident with a peak in apoptosis in 833K cells under these conditions. Activation of caspases was confirmed by an increase in the cleavage product of polyADPribose polymerase (Figure 5a). The slight decrease in Mdm2 protein at 2 h after drug treatment occurs consistently in 833K and RT4 cells; it probably results from temporary inhibition of Mdm2 synthesis during the 1 h pulse treatment.

Continuous etoposide treatment prolongs the p53 response and results in rapid apoptosis

DNA damaging agents can activate p53-mediated apoptosis whilst Mdm2 can protect against p53-mediated apoptosis (Chen et al., 1996; Haupt et al., 1996; Kondo et al., 1995). Since continuous treatment with etoposide both depletes Mdm2 and increases the extent and duration of the p53 response (Figure 5c), we have determined if this treatment schedule also affects the activation of apoptosis. As can be seen in Figure 6a there is up to 65% increase in apoptosis produced by a 4 h continuous incubation compared to a 1 h short pulse with etoposide. This is not just an alteration in the kinetics of apoptosis as there is also a significant increase in the loss of clonogenicity (long-term survival) (Figure 6b). Although prolonged etoposide treatment would be expected to kill more cells, the dramatic effect of such a small difference in treatment time is unusual. For example, in the human leukemia cell lines MOLT-4, where p53 is inactivated by mutation, there is only a 3% increase in apoptotic cells following continuous treament with 50 M etoposide for 4 h compared to a 1 h pulse treatment (Figure 6c).

Mdm-2 expression is differentially affected by etoposide and the radiomimetic drug bleomycin

Etoposide (and UV) can induce the p53 of ataxia telangiectasia cells (AT) despite the fact it is resistant to induction by other DNA damaging agents such as -irradiation and bleomycin (Canman et al., 1994). Potentially, downregulation of Mdm2 by etoposide (this study) and UV may augment p53 upregulation under conditions when the usual signal transduction mechanisms are defective. We therefore determined the effects of Bleomycin on Mdm2 expression in 833K (TGCT) and RT4 (TCC) cells. We demonstrate that Bleomycin (like etoposide) produced DNA strand breaks that were rapidly reversed. Alkaline elution demonstrated that 0.5 g/ml bleomycin produced 370 and 400 rad equivalent single strand breaks in 833K and RT4 respectively (data not shown). However, in contrast to the findings with etoposide, continuous treatment with an IC90 dose of bleomycin upregulates both p21 and Mdm2 (Figure 7a). Although bleomycin is an effective inhibitor of colony formation in 833K and RT4 it was found to be a weak inducer of apoptosis (Figure 7b). Whether the weak activation of apoptosis results from Mdm2 upregulation has yet to be investigated. In contrast to results with etoposide, a 1 or 4 h treatment with bleomycin produced an equivalent level of cell killing measured by colony formation (Figure 7c).

Discussion

We have previously demonstrated that human testicular (TGCT) and bladder (TCC) tumour cell lines express wtp53 that is activated following etoposide treatment (Chresta et al., 1996). Tolerance of wtp53 in tumours frequently results from overexpression of Mdm2 (see Introduction; Haines, 1997). The results presented here demonstrate that Mdm2 is overexpressed in both testicular and bladder carcinoma cells. However, we show that the Mdm2-p53 autoregulatory feedback loop can be inhibited in dose-dependent manner by etoposide treatment.

Mdm2 RNA and protein were found to be overexpressed in the full length 5.5 kb and 97 kDa forms in all TGCT and in TCC tolerating wtp53 (Figure 1). There was no evidence for amplification of the Mdm2 gene suggesting overexpression of Mdm2 results from upregulation at either the transcriptional or translational level. Mdm2 can be expressed from two distinct promoters, P1 which is p53-independent and P2 which is p53-dependent (Barak et al., 1994). In the TCC cell line RT4 the endogenous Mdm2 levels were regulated p53-independently (demonstrated using HPV-E6 to target p53 for degradation - Figure 3b). This is of interest as expression from the p53-independent promoter has been demonstrated to be regulated by growth factors (Shaulian et al., 1997). This suggests the possibility that in tumours such as TGCT and TCC aberrant survival signalling pathways could induce overexpression of Mdm2 and hence contribute to tumorigenesis.

Usually, overexpressed Mdm2 blocks p53-activity. This inhibition can be overcome by three mechanisms. Firstly, DNA damaging agents can induce post-translational modification of p53 which disrupts the Mdm2/p53 interaction (Shieh et al., 1997; Siliciano et al., 1997). Secondly, inhibition of Mdm2 synthesis by antisense oligonucleotides can activate p53 (Chen et al., 1998). Thirdly, the p53/Mdm2 complex can be disrupted by physical/steric means using antibodies or mini-proteins (Blaydes et al., 1997; Bottger et al., 1997).

How does etoposide activate p53 in Mdm2 overexpressing cells? Firstly, topoisomerase poisons, such as etoposide, activate at least two protein kinases which phosphorylate the amino-terminal of p53, casein kinase 1 and DNA protein kinase (Knippschild et al., 1996, 1997; Shieh et al., 1997; Siliciano et al., 1997). Phosphorylation in this region has been demonstrated to prevent Mdm2 binding to p53, which should enhance p53 activity (Shieh et al., 1997). Secondly, inhibition of Mdm2 synthesis by etoposide could synergise with a post-translational activation mechanism to enhance activation of p53. In cells such as TGCT and TCC, that overexpress Mdm2, the post-translational modification activation mechanism may not be able to result in a significant p53-response. Recent studies by Chen et al. (1998) have shown that the DNA damaging agent camptothecin cannot achieve maximal activation of p53 in Mdm2 overexpressing JAR cells, unless Mdm2 synthesis is also inhibited. Our results similarly suggest that etoposide treatment conditions which reduce Mdm2 levels (continuous 4 h) result in higher p53 levels and transcriptional activity (Figure 5c). Presumably etoposide-induced inhibition of Mdm2 synthesis prevents Mdm2 mediated ubiquitination and proteolysis of p53. Studies comparing UV and -irradiation demonstrated a prolonged p53 response in response to UV but not -irradiation. Interestingly, UV but not -irradiation, decreases both Mdm2 protein levels and the ubiquitination of p53 (Maki and Howley 1997; Wu and Levine 1997).

Studies in mouse testicular germ cell tumours (F9) suggest that both post-translational modification and inhibition of Mdm2 synthesis could contribute to activation of p53 by etoposide. Reich et al. (1982) showed that cycloheximide results in stabilization of p53, as a result of downregulation of proteins that control p53 stability (presumably Mdm2). However, the transcriptional activity of p53 was not increased unless the cells were also treated with etoposide (Lutzker and Levine, 1996). Further studies with antisense Mdm2 will allow us to determine whether downregulation of Mdm2 facilitates activation of p53 by DNA damaging agents. If Mdm2 antisense can, like etoposide, promote apoptosis in TGCT and TCC cells, then these tumours would be good candidates for treatment using inhibitory mini proteins which disrupt the p53/Mdm2 interaction (Bottger et al., 1997).

The downregulation of Mdm2 by etoposide was unexpected and is in contrast to earlier reports of induction of Mdm2 by the DNA strand breaking agents -irradiation and bleomycin (Blaydes et al., 1997; Chen et al., 1994; Perry et al., 1993). However, differential activation of p21^waf-1/cip-1 and Mdm2 have been reported previously in response to high dose UV-irradiation (Blaydes et al., 1997; Lu et al., 1996; Lu and Levine, 1993; Perry et al., 1993; Reinke and Lozano, 1997; Wu and Levine 1997). UV results in sequential activation of p21^waf-1/cip-1 and Mdm2; this induces cell cycle arrest and allows potential repair synthesis before Mdm2 terminates the p53 response (Lu et al., 1996; Perry et al., 1993; Reinke et al., 1997; Wu and Levine 1997). Several hypotheses have been put forward to explain the delayed activation of Mdm2 by high dose UV-irradiation. Firstly, Reinke and Lozano (1997) suggest that the higher affinity of p53 for the p21^waf-1/cip-1 promoter may result in preferential activation of p21^waf-1/cip-1 before Mdm2 following UV-irradiation (10 J/m²). However, we find that low doses of etoposide (IC₁₀ - 7 M) can result in coincident synthesis of p21^waf-1/cip-1 and Mdm2 RNA suggesting this is not the case for etoposide activated p53 (Figure 2). Secondly, Lu et al. (1996) proposed that high dose UV may inhibit p53 activity, either directly or indirectly. However, we have shown that p53 activates p21^waf-1/cip-1 and GADD45 under conditions where Mdm2 RNA synthesis is inhibited (Figures 2 and 3). Therefore an inhibitor of p53 transcriptional activity would have to specifically affect the activity of p53 on the Mdm2 promoter. Finally, Wu and Levine (1997) suggest high dose UV-irradiation may induce a p53-independent repressor which is rapidly activated by DNA damage and rapidly inactivated following repair. Our data is consistent with this hypothesis, however, several observations from our work and that of others lead us to suggest an alternative hypothesis; that damage to the DNA template per se (by UV and etoposide) inhibits Mdm2 synthesis.

We suggest that when DNA damage is above a certain threshold Mdm2 synthesis is prevented by bulky adduct damage to the DNA template. This would effectively prevent Mdm2 synthesis until repair had occurred. The specificity for inhibition of Mdm2 probably occurs because the promoter has altered chromatin conformation. The Mdm2 P1 and P2 promoters are nucleosome free and display constitutive DNAse 1 sensitivity (Xiao et al., 1998). In addition Mdm2 is a short-half life protein which is very actively transcribed so will be the site of enhanced topoisomerase activity. Similar findings of etoposide hypersensitivity were previously described for DNAse 1 sensitive region in the c-myc gene (Riou et al., 1993). It is of note that c-myc and mdm2 both form episomes and are frequently amplified in human tumour's suggesting their chromatin conformation may be unusual.

The model of a UV-induced repressor of Mdm2 synthesis is attractive because it suggests temporal control in the activation of p53-dependent genes with different function. Delayed activation of Mdm2 allowing p21 to halt cell cycle progression until DNA damage is repaired. However, several pieces of evidence are not consistent with this model. Firstly, when the level of DNA damage produced by either UV-irradiation (50 J/m²) or etoposide (>90 M) is high, the specificity for Mdm2 is lost and both GADD45 and p21 expression are inhibited (Figure 2e) (Lu et al., 1996; Reinke and Lozano, 1997). Secondly, Reinke and Lozano (1997) have shown that the p53 from extracts of UV-treated cells (50 J/m²) can still bind to an exogenously added p21 promoter, demonstrating p53 DNA binding activity is not repressed. Thirdly, when we studied the effects of Bleomycin, which produces single and double strand scission by free radical damage, we found that Mdm2 was upregulated without delay. Thus, Mdm2 is induced despite the fact that repair has not taken place and 90% of these cells are destined to die (Figure 7). This suggests that the bulky adducts produced by etoposide and UV-damage inhibit Mdm2 synthesis, rather than a repressor activated by high levels of unrepairable damage.

The p53-dependent activation of Mdm2 usually limits the extent and duration of the p53 response after DNA damage (Chen et al., 1994; Haupt et al., 1997; Kubbutat et al., 1997). This can be seen in Figure 5, where downregulation of p53 is coincident with upregulation of Mdm2 4 h after drug treatment. Under these pulse treatment conditions activation of p21^waf-1/cip-1 occurs in the short time window between activation of p53 and further upregulation of Mdm2 (Figure 5b). Thus, Mdm2 effectively and rapidly terminates the p53 response. In contrast, under conditions of continuous treatment with etoposide Mdm2 upregulation is repressed and the levels of p53 and p21^waf-1/cip-1 remain at maximal levels (Figure 5c). Significantly these conditions also result in a dramatic increase in the activation of apoptosis and long-term loss of viability (Figure 6a,b). Downregulation of Mdm2 by antisense has similarly been demonstrated to enhance camptothecin and cisplatin-induced apoptosis (Chen et al., 1998; Kondo et al., 1995).

This is the first observation of Mdm2 regulation by a topoisomerase II poison. The results are strikingly similar to the effects of high dose UV-irradiation and it is possible that both UV and etoposide activate a repressor of Mdm2 synthesis. However, the rapid reversibility of the etoposide effect suggests Mdm2 expression could be inhibited by bulky adduct damage to the template which effectively maintains the p53 response until repair has taken place.

Materials and methods

Cell culture and cytotoxicity assays

All the cell lines were grown routinely in 25 cm² flasks in RPMI 1640 medium with 10% (v/v) heat inactivated fetal calf serum, 2 mM glutamine, at 37°C in a humidified atmosphere of 5% CO₂ in air. Each cell line was used over a maximum of 12 passages to minimize changes that occur during prolonged culture. All cell lines were routinely tested and found to be mycoplasma free. Drug sensitivity was measured by clonogenic assays. One thousand cells were plated per well in 6-well plates (2 cm²) and allowed to adhere overnight. They were then treated for indicated times with etoposide or vehicle control, dimethylsulfoxide. The drug was then washed from the cells with three changes of PBS and the cells reincubated in fresh medium for 10 - 14 days. Colonies were stained with methylene blue and colonies of greater than 50 cells were counted.

Western blotting

Samples were collected 4 h after either continuous or 1 h pulse drug treatment. Following 4 h continuous drug treatment, apoptotic cells detached from the monolayer. These cells were collected by centrifugation and combined with the monolayer cells which had been detached by scraping into versene (GIBCO BRL, Paisley, Scotland, UK). Cells were rinsed in ice cold PBS containing the following protease inhibitors (PI), aprotinin (1%), benzamidine (1 mM), phenylmethylsulphonylfluoride (1 mM), trypsin/chymotrypsin inhibitor (10 g/ml) (Sigma, Poole, UK) then resuspended in 300 l of PBS with PI and sonicated using a Soniprep 150 (MSE) for 10 s at 12 microns, allowed to cool, then sonicated again for 10 s at 16 microns. Protein concentration was determined using a BioRad protein assay kit and 15 g of protein used per sample. Samples were boiled for 5 min in SDS sample buffer and then electrophoresed on 0.75 mm polyacrylamide slab gels. Proteins were transferred to Immobilon-P membrane (Millipore, Harrow, UK) at 80 V for 1.5 h.

Southern blotting

Cells (1´10⁶) were washed in PBS then lysed in 250 l of sarcosyl lysis solution (0.2% sarcosyl, 2 M NaCl, 0.04 M EDTA, pH 10) containing proteinase K at 0.5 mg/ml, the lysate was incubated overnight at 45°C, then diluted with one volume of TE buffer and DNA precipitated by addition of two volumes of 95% ethanol (-20°C). Precipitates were allowed to form at -20°C for 30 min and then collected by centrifugation at 4°C. The pellets were washed in 70% ethanol then 95% ethanol, dried and resuspended in TE and restricted overnight with EcoRI prior to electrophoresis. Gels were blotted and probed as described by Maniatis under high stringency conditions (Sambrook et al., 1989). The probe used for Mdm2 is a HindIII fragment of the human cDNA clone FL4. This region spans nucleotides 578 - 1462 of the human Mdm2 cDNA.

Northern blotting

Cells were collected as for Western blotting, rinsed in PBS then total RNA was isolated using RNAzol B as described by the manufacturers (Biogenesis). Fifteen g of total RNA was loaded per lane onto formaldehyde gels and separated overnight. Samples were prepared, electrophoresed and blotted as described by Maniatis. Filters were washed 3 - 5 times for 15 min with 2´SSC; 0.1% SDS at room temperature and for 1 - 10 min at 65°C with 0.2´SSC; 0.1% SDS depending upon the probes used.

DNA fragmentation filter binding assay (FBA)²

Apoptosis was quantified using the filter binding assay (Bertrand et al., 1991). Percentage DNA fragmentation assayed using the FBA corresponded well with frequency of cells demonstrating apoptotic morphology by Hoechst 33258 staining.

Alkaline elution

DNA single-strand breaks (SSBs) were measured by alkaline elution (pH 12.1) as described by Kohn et al. (1976). Cells in early logarithmic phase growth (1 - 2´10⁵/ml) were labelled for 30 h with 0.015 Ci ¹⁴C-thymidine/ml (specific activity, 56 mCi/mmol; Amersham, UK), washed and reincubated in fresh medium for 2 h. The cells were then exposed to etoposide for 1 h. Following drug treatment cells were rinsed with ice cold PBS and then scraped into ice cold versene to minimize the possibility of reversal of damage. Cells were collected by centrifugation and resuspended in ice cold PBS. 8´10⁵ cells were loaded onto each filter and rinsed with 10 ml ice-cold PBS before lysis. For measurement of DNA SSBs cells were lysed using a sarcosyl lysis solution (0.2% sodium dodecyl sarcosine; 2 M NaCl; 0.04M EDTA pH 10) onto 2 m polycarbonate filters (Millipore). DNA was eluted for 15 h at 0.04 ml/min with tetraethylammonium hydroxide (pH 12.1) containing 0.1% SDS. DNA remaining on the filter and in the filter funnels was released as described (Kohn et al., 1976). The frequency of SSBs induced by etoposide or bleomycin were converted to rad equivalents using a calibration graph derived from the number of SSBs produced by a known X-ray dose.

Acknowledgements

We would like to thank John Hickman for helpful discussion and review of the manuscript. We would also like to thank the following people for the kind donations of probes and antibodies. Mdm2: Ab 2A10 a kind gift from Dr A Levine (Princeton, New Jersey, USA); Probes: p53 - BamHI insert of pC53-SN3; Mdm2, HindIII fragment of the human cDNA clone FL4 - both kindly provided by Dr B Vogelstein (Johns Hopkins Oncology Centre, Baltimore, MD, USA); 21^waf-1/cip-1 - p21^waf-1/cip-1 insert from pBABE, kindly provided by Dr S Picksley (University of Dundee, Scotland), GADD45 cDNA insert from PCRIII - kindly provided by Dr Campomenosi (University of Dundee, Scotland). Supported by project grant SP2234 from the Cancer Research Campaign.

Barak Y, Gottlieb E, Juven GT and Oren M. (1994). Genes Dev. 8, 1739-1749. MEDLINE

Barbareschi M, Girlando S, Fellin G, Graffer U, Luciani L and DallaPalma P. (1995). Urol. Res. 22, 349-352. MEDLINE

Bertrand R, Sarang M, Jenkin J, Kerrigan D and Pommier Y. (1991). Cancer Res. 51, 6280-6285. MEDLINE

Blaydes JP, Gire V, Rowson JM and Wynford-Thomas D. (1997). Oncogene 14, 1859-1868. MEDLINE

Bottger A, Bottger V, Sparks A, Liu WL, Howard SF and Lane DP. (1997). Curr. Biol. 7, 860-869. MEDLINE

Canman CE, Wolff AC, Chen CY, Fornace AJ and Kastan MB. (1994). Cancer Res. 54, 5054-5058. MEDLINE

Chen CY, Oliner JD, Zhan Q, Fornace AJ, Vogelstein B and Kastan MB. (1994). Proc. Natl. Acad. Sci. USA 91, 2684-2688. MEDLINE

Chen JD, Lin JY and Levine AJ. (1995). Mol. Med. 1, 141-142.

Chen JD, Wu X, Lin JY and Levine AJ. (1996). Mol. Cell. Biol. 16, 2445-2452. MEDLINE

Chen L, Agrawal S, Zhou W, Zhang R and Chen J. (1998). Proc. Natl. Acad. Sci. USA 95, 195-200. MEDLINE

Chresta CM, Masters JRW and Hickman JA. (1996). Cancer Res. 56, 1834-1841. MEDLINE

Cordon-Cardo C, Latres E, Drobnjak M, Oliva MR, Pollack D, Woodruff JM, Marechal V, Chen J, Brennan MF and Levine AJ. (1994). Cancer Res. 54, 794-799. MEDLINE

deJong S, Dam WA, deVries EGE and Muldner NH. (1997). Proc. AACR 38, 638.

Dubs-Poterszman MC, Tocque B and Wasylyk B. (1995). Oncogene 11, 2445-2449. MEDLINE

Erhardt P, Tomaselli KJ and Cooper GM. (1997). J. Biol. Chem. 272, 15049-15052. Article MEDLINE

Fakharzadeh SS, Trusko SP and George DL. (1991). EMBO J. 10, 1565-1569. MEDLINE

Finlay CA. (1993). Mol. Cell. Biol. 13, 301-306. MEDLINE

Fleischhacker M, Strohmeyer T, Imai Y, Slamon DJ and Koeffler HP. (1994). Mod. Pathol. 7, 435-439. MEDLINE

Guillou L, Estreicher A, Chaubert P, Hurlimann J, Kurt AM, Metthez G, Iggo R, Gray AC, Jichlinski P, Leisinger HJ and Benhattar J. (1996). Am. J. Path. 149, 1221-1228. MEDLINE

Haines DS. (1997). Leukemia and Lymphoma 26, 227-238. MEDLINE

Haupt Y, Barak Y and Oren M. (1996). EMBO J. 15, 1596-1606. MEDLINE

Haupt Y, Maya R, Kazaz A and Oren M. (1997). Nature 387, 296-299. MEDLINE

Jones SN, Roe AE, Donehower LA and Bradley A. (1995). Nature 378, 206-208. MEDLINE

Knippschild U, Milne DM, Campbell LE, DeMaggio AJ, Christenson E, Hoekstra MF and Meek DW. (1997). Oncogene 15, 1727-1736. MEDLINE

Knippschild U, Milne DM, Campbell LE and Meek DW. (1996). Oncogene 13, 1387-1393. MEDLINE

Kohn KW, Erickson LC, Ewig R and Friedman CA. (1976). Biochemistry 15, 4629-4637. MEDLINE

Kondo S, Barnett GH, Hara H, Morimura T and Takeuchi J. (1995). Oncogene 10, 2001-2006. MEDLINE

Kubbutat MHG, Jones SN and Vousden KH. (1997). Nature 387, 299-303. MEDLINE

Landers JE, Cassei SL and George DL. (1997). Cancer Res. 57, 3562-3568. MEDLINE

Levine AJ. (1990). Virology 177, 419-426. MEDLINE

Lianes P, Orlow I, Zhang ZF, Olvia MR, Sarkins AS, Reuter VE and Cordon-Cardo C. (1994). J. Natl. Cancer Inst. 86, 1325-1330. MEDLINE

Ljungman M and Zhang F. (1996). Oncogene 13, 823-831. MEDLINE

Lu X, Burbridge SA, Griffin S and Smith HM. (1996). Oncogene 13, 413-418. MEDLINE

Lu X and Lane DP. (1993). Cell 75, 765-778. MEDLINE

Lutzker SG and Levine AJ. (1996). Nature Med. 2, 804-810. MEDLINE

Maki CG and Howley PM. (1997). Mol. Cell. Biol. 17, 355-363. MEDLINE

Martin K, Trouche D, Hagemeier C, Sorensen TS, La TN and Kouzarides T. (1995). Nature 375, 691-694. MEDLINE

Momand J, Zambetti GP, Olson DC, George D and Levine AJ. (1992). Cell 69, 1237-1245. MEDLINE

Montes-de-Oca-Luna R, Wagner DS and Lozano G. (1995). Nature 378, 203-206. MEDLINE

Oliner JD, Kinzler KW, Meltzer PS, George DL and Vogelstein B. (1992). Nature 358, 80-83. MEDLINE

Oliner JD, Pietenpol JA, Thiagalingam S, Gyuris J, Kinzler KW and Vogelstein B. (1993). Nature 362, 857-860. MEDLINE

Perry ME, Piette J, Zawadzki JA, Harvey D and Levine AJ. (1993). Proc. Natl. Acad. Sci. USA 90, 11623-11627. MEDLINE

Reich NC, Oren M and Levine AJ. (1983). Mol. Cell. Biol. 3, 2143-2150. MEDLINE

Reinke V and Lozano G. (1997). Radiation Res. 148, 115-122.

Rieger KM, Little AF, Swart JM, Kastrinakis WV, Fitzgerald JM, Hess DT, Libertino JA and Summerhayes IC. (1995). Br. J. Cancer 72, 683-690. MEDLINE

Riou G, Barrios M, Prost S, Terrier MJ, Theodore C and Levine AJ. (1995). Mol. Carcinog. 12, 124-131. MEDLINE

Riou JF, Gabillot M and Riou G. (1993). FEBS Lett. 334, 369-372. MEDLINE

Sambrook J, Fritsch EF and Maniatis T. (1989). Molecular cloning: A laboratory manual (CSH, 1989). .

Shaulian E, Resnitzky D, Shifman O, Blandino G, Amsterdam A, Yayon A and Oren M. (1997). Oncogene 15, 2717-2725. MEDLINE

Shieh SY, Ikeda M, Taya Y and Prives C. (1997). Cell 91, 325-334. MEDLINE

Sigalas I, Calvert AH, Anderson JJ, Neal DE and Lunec J. (1996). Nature Med. 2, 912-917. MEDLINE

Siliciano JD, Canman CE, Taya Y, Sakaguchi K, Appella E and Kastan MB. (1997). Genes Dev. 11, 3471-3481. MEDLINE

Weinberg R. (1993). Neuron 11, 191-196. MEDLINE

Wu L and Levine AJ. (1997). Molec. Med. 3, 441-451.

Xiao G, White D and Bargonetti J. (1998). Oncogene 16, 1171-1181. MEDLINE

Xiao ZX, Chen J, Levine AJ, Modjtahedi N, Xing J, Sellers WR and Livingston DM. (1995). Nature 375, 694-697. MEDLINE

Figure 1 Overexpression of Mdm2 in the absence of gene amplification in TGCT and TCC tolerating wtp53. (a) Endogenous Mdm2 and p53 protein levels were determined by Western blotting using 15 g of whole cell extract, blots were probed with monoclonal Ab D0-1 for p53 and MoAb 2A10 for Mdm2. (b) Mdm2 and GAPDH transcript levels were analysed by Northern blotting using 15 g of total RNA. Mdm2 was detected using a HindIII fragment of the Mdm2 cDNA which detects nucleotides 578 - 1462 of the 2.2 kb cDNA (the HindIII fragment was used to avoid c-terminal Alu sequences). A single transcript at 5.5 kb was detected using this probe. However, we cannot rule out the presence of alternative transcripts which have only N- or C-terminal regions of the molecule which would not be detected with this probe (Sigalas et al., 1996). (c) Southern blotting of 10 g genomic DNA restricted with EcoRI and probed with the HindIII fragment of Mdm2 cDNA. Loading was found to be equal by ethidium staining. Only two bands were detected due to use of the HindIII fragment probe, these were of the same size and intensity in all lines. All experiments were repeated three times and results are representative

Figure 2 Inhibition of Mdm2 transcription by etoposide. (a) 833K cells were treated continuously with etoposide for 3 h for RNA studies and 4 h for protein studies. Total RNA was isolated and 15 g separated on denaturing gels. Mdm2 was detected using the HindIII fragment of FL-4 as described in legend to Figure 1. C=control (DMSO - drug vehicle), and subsequent lanes are 3 h plus etoposide, 3.3, 6.6 (IC₁₀), 10.5(IC₅₀), 25.7(IC₉₀), 51.4 M. The GAPDH probed filter is shown for comparison. (IC=inhibitory concentration). (b) Quantitation of the levels of Mdm2 5.5 kb transcript and p53 protein following continuous treatment of 833K cells with the indicated concentrations of etoposide. Mdm2 and GAPDH transcript levels were determined by phospho-image analysis of the blot and Mdm2 results normalized to GAPDH. p53 levels were determined by scanning densitometry. (c) Northern analysis of Mdm2, Bax, p21^waf-1/cip-1 and GADD45 transcripts following 3 h continuous treatment of 833K cells with the indicated concentrations of etoposide. The GAPDH result is shown for comparison. (d) Quantitation of transcript levels of blots shown in (c) by phospho-image analysis. All results are normalized to GAPDH. (e) Induction of p21^waf-1/cip-1 and GADD45 in RT4 cells by 3 h treatment with the indicated concentrations of etoposide. Transcript levels were quantified by phospho-image analysis. All results are normalized to GAPDH. Results shown are representative of three individual experiments in RT4 and five experiments in 833K

Figure 3 Downregulation of Mdm2 is p53-independent. (a) Immunoblot analysis of 15 g of total cell extracts from RT4 cells stabily transfected with HPV16-E6 or Neo vector control. p53 protein is degraded in the E6 transfects and p21 induction by 4 h treatment with an IC₅₀ (96 M) of etoposide is inhibited. (b) Analysis of the effects of loss of p53 function on p21 and Mdm2 transcript levels in control and etoposide treated cells (4 h with 96 M). Transcript levels of Mdm2, p21 and GAPDH were determined as described in legend to Figure 2

Figure 4 Reversal of Mdm2 inhibition following repair of DNA damage. (a) Quantitation of etoposide-induced DNA damage and repair in 833K cells by alkaline elution, either immediately following a 60 min treatment with 2 M etoposide (no repair) or 1 h after the treatment. Cells were reincubated in drug free medium at 37°C for the 60 min (60 min repair). Damage is expressed as rad equivalents (RE). (b) Dose response of p53 (protein) and Mdm2 (RNA and protein) induction under continuous treatment conditions or under conditions where DNA damage is repaired (pulse treatment). Cells were treated continuously or for a 1 h pulse plus repair period. RNA samples were prepared at 3 h and protein samples at 4 h after drug treatment. The IC₁₀, ₅₀ and ₉₀ for 833K cells (TGCT) are 6.6, 10.5, and 25.7 M respectively. GAPDH is shown for comparison. Note, 4 h after drug treatment p53 levels are maintained at a higher levels in cells where continuous etoposide treatment prevents Mdm2 induction. Results are representative of three individual experiments

Figure 5 Analysis of the kinetics of the p53-Mdm2 autoregulatory feedback loop following drug removal. (a) Timecourse of induction of p53, p21 and Mdm2 following etoposide treatment (IC₅₀) in the testicular germ cell tumour line 833K. Samples were treated for 1 h with IC₅₀ (10.5 M) etoposide, washed and reincubated in drug free medium and samples of total RNA and protein prepared at the indicated times. Immunoblots and Northerns were analysed for Mdm2, p53, p21 and GAPDH as described in legend to Figure 2. PARP was detected using MoAb MCA 1522 from Serotec, UK. (b) Transcript levels of blots shown in (a) were quantified by phospho-image analysis and data is expressed relative to the untreated control. All samples are normalized for GAPDH to correct for loading. (c) Effect of pulse or continuous etoposide treatment (IC₁₀ - 7 M) on p53 protein and p21^waf-1/cip-1 transcript induction in 833K cells. Samples of total RNA or protein were prepared at 3 or 4 h respectively following 1 h pulse or continuous etoposide treatment. p53 protein levels were determined by immunoblotting and scanning densitometry. p21 and GAPDH transcript levels were determined by phospho-image analysis, data presented is normalized to GAPDH

Figure 6 Enhanced apoptosis in cells where the Mdm2 / p53 autoregulatory feedback loop is inhibited. (a) Activation of apoptosis in TGCT cells under pulse (1 h+3 h) or continuous (4 h) treatment conditions with the indicated concentrations of etoposide. Apoptosis was measured at 4 h by the filter binding assay and results confirmed by analysis of morphology using Hoechst staining. (b) Comparison of loss of clonogenicity following 1 or 4 h treatment with the indicated concentrations of etoposide. (c) Activation of apoptosis in MOLT-4 human lymphoblastoid cells under pulse (1 h+3 h) or continuous (4 h) treatment conditions with 50 M etoposide. Samples were taken at 4 h and apoptosis analysed by the filter binding assay. Note the dramatic 65% difference in apoptosis with 50 M etoposide pulse or continuous treatments at 4 h in TGCT (5a) compared to 3% difference in MOLT-4 under the same conditions (5c). All results are the mean±s.e. of three independent experiments

Figure 7 Comparison of the effect of Bleomycin and etoposide on Mdm2 levels, apoptosis and long-term survival in 833K TGCT cells. (a) Northern blot of p21, Mdm2 and GAPDH RNA levels in RT4 cells treated for 3 h continuously with drug vehicle only (PBS) - C; IC₉₀ of Bleomycin (10 g/ml) and IC₉₀ of etoposide (240 M). The right hand panel shows phospho-image analysis of the results shown in the left hand panel for RT4 and for 833K cells treated for 4 h with IC₉₀ of Bleomycin (1 g/ml). Results are normalized to GAPDH. (b) Activation of apoptosis in 833K and RT4 at indicated timepoints following 1 h treatment with 0.5 g/ml Bleomycin. Apoptosis was assayed by counting cells with condensed chromatin identified by staining with Hoechst 33258. (c) Loss of viability measured by clonogenic assays. Cells were treated for 1 or 4 h with indicated concentrations of Bleomycin and then reincubated in drug free medium for approximately 10 days to allow colony formation. Results are the mean±s.e. of three independent experiments