Characterisation of VP-16-induced DNA cleavage in oestrogen-stimulated human breast cancer cells.

Cycling cells are recognised to be more susceptible than quiescent cells to the cytotoxic action of many commonly used cancer chemotherapeutic agents. We have found that oestrogen stimulation of T-47D human breast cancer cells is accompanied by a two-fold increase in VP-16-induced DNA cleavage as measured by alkaline DNA unwinding, and that this increase in DNA cleavage is accompanied by a corresponding enhancement of drug-induced cytostasis. The enhancement of VP-16-induced DNA cleavage seen with oestrogen exposure is antagonised both by antioestrogen treatment and by cycloheximide, an inhibitor of protein synthesis, but not by the DNA synthesis inhibitor aphidicolin. Increased c-myc protein synthesis is detectable within an hour of oestrogen exposure, while increased VP-16-induced DNA cleavage is detectable within 4h and increased DNA synthesis within 16h. Only small changes in cell-cycle distribution occur with oestrogen stimulation. In the absence of VP-16, oestrogen does not reduce DNA double-strandedness, nor does it induce changes in chromatin structure as measured by alterations in DNA superhelicity or chromatin accessibility. These findings suggest that oestrogen enhances VP-16-induced DNA damage in asynchronously growing G1-phase cells and that this enhancement may be dependent at some point upon de novo protein synthesis. Oestrogen pre-treatment of T-47D human breast cancer cells improves the therapeutic index of VP-16 without the need for cell synchronisation or highly precise drug scheduling.

Oestrogen stimulation potentiates the cytotoxicity of Sphase-active drugs in synchronised human breast cancer cells in vitro (Weichselbaum et al., 1978). DNA cleavage induced by non-phase-specific drugs such as m-AMSA has also been reported to be enhanced by oestrogen 'priming' of unsynchronised human breast cancer cell cultures (Zwelling et al., 1983), but the extent to which this DNA cleavage is reponsible for cytotoxicity remains uncertain. Moreover, it is not known whether such enhancement of DNA cleavage reflects changes in chromatin structure induced by oestrogen, such as changes in accessibility of chromatin to drug interaction (Kuo, 1981) or changes in DNA superhelicity (Lipetz et al., (1982).
Clinical trials using tamoxifen synchronisation and subsequent oestrogenic 'recruitment' prior to cytotoxic therapy of breast cancer have been undertaken (Eisenhauer et al., 1984;Lippman et al., 1984;Paridaens et al., 1985;Conte et al., 1987) but have so far failed to demonstrate any significant overall survival benefit (Davidson & Lippman, 1987). Part of the difficulty in applying this strategy successfully in vivo may relate to optimal scheduling of cytotoxic therapy (especially S-phase-specific drugs) and reversal of tamoxifeninduced cytostasis (Furr & Jordan, 1984). Here we report our findings using oestrogen priming of unsynchronised T-47D human breast cancer cells prior to VP-16 treatment, and show that this results in enhanced DNA cleavage which occurs independently of enhanced DNA synthesis or measurable chromatin modification, and which is also accompanied by enhanced toxicity.

Materials and methods
Cell culture, hormone stimulation and cell growth T-47D cells were obtained from the American Type Culture Collection (Rockville, MD) in their 84th passage. Cell stocks were maintained as monolayer cultures in RPMI plus 10% foetal calf serum, glutamine and antibiotics (subsequently referred to as complete medium) and incubated at 37°C in 5% CO2 in air. For three weeks prior to experiments involving hormone stimulation, cells were cultured in medium supplemented with 5% dextran-charcoal-stripped srum (Reddel et al., 1984) unless stated otherwise; cells were able to be passaged in this oestrogen-deprived medium for over six months without reduction in plating efficiency. 17-floestradiol was added to cultures at a final concentration of 10-8 M in 0.1% ethanol 24 h prior to experiments unless specified otherwise, while control samples received ethanol alone. 4-hydroxytamoxifen (a gift of Dr A.H. Todd, ICI plc Macclesfield, UK) was added simultaneously with oestrogen at a concentration of 10-7 M. For growth curves, cells were detached using trypsin/EDTA and counted in triplicate using a Coulter counter. Cytotoxicity was measured using the growth delay method validated by Leonessa et al. (1986) in which cells are grown for five days in complete medium following experimental treatment.
Drug treatment VP-16 was stored as a light-protected 34mM stock at room temperature. Bleomycin sulphate was a gift of Lundbeck Ltd (Luton, Beds UK). Following pre-treatment of samples with either oestrogen or ethanol, drugs were diluted in water and added to samples at various concentrations for 1 h at 370C. Following removal of treated medium, monolayers were washed twice in PBS and then either frozen (using 10mM Tris-HCl, 100mM NaCl, 10mM EDTA, 1 mgml-1 bovine serum albumin, pH 8.0; for alkaline DNA unwinding assays) or re-fed with complete medium (for growth curves). Colcemid (60ngml-1) was used for stathmokinetic experiments. For experiments involving cycloheximide, oestrogen was added to samples for 6h only prior to VP-16 exposure; designated samples received cycloheximide 10 gml-1 at the same time as oestrogen administration. Aphidicolin (0.25ygml-1) was added to samples at the same time as oestrogen for 24h prior to VP-16 exposure, and flow cytometry confirmed the induction of G1-S arrest in aphidicolin-treated samples.
DNA damage assays (a) Alkaline unwinding This technique quantifies DNA cleavage as a time-dependent function of DNA unwinding in alkali, since the rate of DNA unwinding is known to vary linearly with the amount of pre-existing DNA strand-breakage (Ahnstrom & Erixon, 1973). DNA unwinding was determined by fluorometry using a bisbenzamide dye (Latt & Stetten, 1976). As described previously (Smith et al., 1986), treated cell monolayers were frozen and then detached by rapid thawing. Following resuspension in ice-cold buffer, 0.5ml aliquots of the cell suspension were analysed using the method developed by Kanter & Schwartz (1982). Briefly, quadruplicate samples were exposed to 0.1 N NaOH for 60 min at 4°C and neutralised with 0.1 N HCI. Detergent buffer (0.16% sodium lauroyl sarcosinate, 0.2M KH2PO4, 0.04M disodium EDTA, pH 7.4) containing Hoechst 33342 at a final concentration of 0.25 M was then added, and the samples homogenised using a 150W ultrasonic disintegrator. Fluorescence was measured 24h later using a Perkin-Elmer MPF-4 spectrofluorimeter. A second set of samples was processed in parallel except that the alkaline lysates were subjected to an additional sonication to accelerate DNA unwinding and thereby permit calculation of background fluorescence associated with minimal residual DNA doublestrandedness; while a third set of samples was not exposed to alkali at all, permitting calculation of fluorescence associated with maximal DNA doublestrandedness. Treatment-induced enhancement of DNA unwinding was determined by the expression f= -100 log (Dx/Dc), where Dx and Dc represent the percentage of doublestranded DNA in treated and control samples respectively. DNA damage induced by one Gy X-irradiation was consistently detectable, and use of the algorithm yielded a linear damage-induced response up to at least 16Gy. X-ray calibration of the assay was used to determine breaks/109 daltons DNA assuming 0.5 breaks/109 daltons mol. wt./Gy (Kohn et al., 1976).
(b) Nucleoid sedimentation As originally described by Cook & Brazell (1975) and modified by Farzaneh et al. (1982), cells were detached and resuspended in cold PBS. Fifty pl of suspension was then deposited onto 150 y1 lysis buffer (final concentration 2mm EDTA, 0.5% Triton X-100, 100mM Tris-HCI, 2 M NaCl, pH 8.0) over 3.8 ml 15-30% sucrose gradients containing 1mM EDTA, 10mM Tris-HCI, 2M NaCl and 1pM Hoechst-33342 at pH 8.0. Cells were then lysed for 30 min at room temperature prior to centrifugation for 30 min at 50,000g. For titration experiments designed to measure the superhelicity of these histone-depleted nucleoid structures (Lipetz et al., 1982), ethidium bromide was incorporated into both the lysis buffer and the sucrose gradient at the specified concentration in lieu of Hoechst-33342. Transcriptionally active chromatin is known to exhibit increased sensitivity to nuclease-induced DNA nicking (Gazit & Cedar, 1980). To determine whether oestrogen stimulation of T-47D cells was associated with any gross change in chromatin accessibility to nuclease, cells were permeabilised and detached by freeze-thawing, then resuspended in prewarmed buffer (10mM Tris-HCl, 10mM EDTA, 100mM NaCl, 1 mg ml -1 bovine serum albumin, pH 7.75) and incubated in a 37°C water-bath for 15 min with varying concentrations of DNase II (20 Kunitz units/ml; bovine spleen type; Sigma). The reaction was stopped by dilution with ice-cold buffer at pH 8.0 in preparation for the alkaline unwinding assay.
Tritiated thymidine incorporation 3H-thymidine (specific activity 6.7Cimmol-1; New England Nuclear, Boston MA) was added to samples at a concentration of 0.5pCiml-1 together with either oestrogen or ethanol for the required period. Cell monolayers were washed three times in PBS supplemented with magnesium and calcium, lysed using 4-aminosalicylic acid (6% w/v; BDH, Poole, Dorset), triisopropylnaphthalenesulphonic acid (1% w/v; Eastman Kodak, Rochester, NY) and butan-2-ol (6% v/v), then removed with a rubber policeman, mixed with trichloroacetic acid (10% w/v) and the acid-insoluble radioactivity isolated on glass fibre filters. The filters were mixed with 10mls scintillant before being counted with a single-channel ratio programme and converted to disintegrations per minute using a chloroform quench curve.
C-myc protein synthesis The ELISA assay developed by Moore & Evan (1987) was used. Briefly, pan-myc (antibody was adsorbed to microtitre wells. Cell monolayers were washed with PBS containing 0.02% EDTA and 0.1% sodium azide and then removed using a rubber policeman. Cells were centrifuged, resuspended at a density of 5 x 107 cellsml-1, and lysed. Lysates were prepared by boiling cells in SDS and 50mM dithiothreitol, alkylating with 100mM iodoacetamide, shearing the DNA by repeated passage through a 26-gauge needle, and diluting in Nonidet P40. Samples were then incubated in microtitre plates. A second anti-myc mono- I r IT f'% clonal antibody conjugated to alkaline phosphatase then recognised captured human myc protein, and bound alkaline phosphatase was detected colorimetrically. The reaction was stopped with acid and optical density determined at 494nm. Flow cytometry Following trypsin/EDTA detachment, cells were stained with ethidium bromide 50 ugml-1 plus 0.125% Triton X-100 (Taylor & Milthorpe, 1980) and ribonuclease 0.5mgml-1 for 10min prior to analysis. Samples were monitored using a flow cytometer incorporating an Innova 70-5 argon laser (Coherent, Palo Alto, CA) tuned to 488 nm at 200 mV. DNA fluorescence distributions were analysed by computer using a cell-cycle phase-fitting programme which assumes normal distributions for GI aand G2M phase populations and which calculates a probability function for the S phase distribution based upon the means and standard deviations of the GI and G2M phase (Watson et al., 1987). In Figure 2, the oestrogen concentration which optimally stimulates cell growth (represented in Figure 2a as minimal doubling time) corresponds to the optimal concentration for enhancement of VP-16-induced DNA cleavage ( Figure 2b) and VP-16-induced cell growth retardation (Figure 2c). These data suggest a relationship between the growthstimulatory effect of oestrogen and the observed enhancement of VP-16-induced DNA damage.
(ii) Effect of inhibitors on oestrogen-induced DNA damage enhancement Inhibition of the growth-stimulatory effect of oestrogen using a potent antioestrogen, 4-hydroxytamoxifen, results in antagonism of the observed enhancement of VP-16-induced DNA cleavage (Figure 3a). Simultaneous exposure of cells to oestrogen and the protein synthesis inhibitor, cycloheximide, also leads to antagonism of this effect (Figure 3b). However, inhibition of DNA synthesis using the DNA polymerase inhibitor aphidicolin does not affect the enhancement of DNA cleavage seen in oestrogen-primed cells (Figure 3c). These findings suggest that the enhancing effect of oestrogen on VP-16-induced DNA cleavage is mediated at some point by new protein synthesis but not by entry of cells into DNA synthesis.
(iii) Influence of oestrogen of DNA doublestrandedness and chromatin structure Oestrogen stimulation of up to 72h duration failed to cause any change in DNA scission as measured by nucleoid sedimentation (Figure 4a), and similar results were observed in cells simultaneously treated with the poly(ADP-ribosyl) transferase inhibitor 3-aminobenzamide (data not shown), suggesting that oestrogen stimulation alone may not be associated with either long-lived or transient DNA strandbreak induction. DNA cleavage was also measured by alkaline unwinding in cells treated with oestrogen for 24h, and no effect was seen over a 3 h unwinding period ( Figure  4a). This implies that pre-existing DNA cleavage in samples not exposed to drug does not contribute to the observed Effect of oestrogen concentration on percentage enhancement of DNA cleavage induced in samples exposed to 5 1M VP-16 for 1 h at 37°C when compared to ethanol control. fx, DNA cleavage in oestrogen-treated samples (see Materials and methods); fc, DNA cleavage in control samples; (c) Effect of oestrogen concentration on growth delay induced by 1 h exposure to 5pM VP-16. Oestrogen-treated samples were counted in triplicate and expressed as a percentage of control values following 5 days' growth in complete medium subsequent to VP-16 treatment. In samples not exposed to VP-16, the total number of cell doublings following the 5-day incubation in complete medium consistently differed by <5% between samples initially pre-treated with either oestrogen or ethanol. Nx, number of cells in samples incubated in complete medium for five days following VP-16 exposure; Nc, number of cells in samples following 5 days' growth in complete medium without VP-16 exposure. 0, control cells; 0, oestrogen-treated cells.
enhancement of VP-16-induced DNA cleavage as assayed by alkaline DNA unwinding. That oestrogen also does not alter the overall superhelicity of DNA in T-47D cells is suggested by Figure 4b, which shows that a similar amount of ethidium bromide is required to 'untwist' negatively supercoiled DNA (thereby resulting in minimum sedimentation) in both oestrogen-treated and control cells. Chromatin accessibility to drug treatment appears to be similarly unaffected by oestrogen in either viable (Figure 4c) or freeze-thaw-permeabilised (Figure 4d) cells, since the induction of strand-breaks by either bleomycin or DNase II respectively is unchanged in cells stimulated for 24h. (iv) Time-course of events following oestrogen stimulation Synthesis of c-myc protein peaks after one hour of oestrogen stimulation, and is maintained at this level for 24 h ( Figure  Sa). This final level of c-myc protein in stimulated cells represents approximately twice that measured in control cells. The amount of DNA cleavage induced by 5pM VP-16 also reaches double control levels in cells stimulated for 24 h, and -50% of this increase is achieved within 8 h ( Figure   5b). In contrast, no change in tritiated thymidine incorporation (Figure 5c), fraction of cells involved in DNA synthesis (Figure 5d), or rate of cell-cycle traverse ( Figure  5e) is measurable after 8 h oestrogen stimulation (Figure 5c). This sequence of events suggests that oestrogen-induced mitogenesis per se is preceded by very early changes in protein (such as c-myc) synthesis, and that the observed enhancement of VP-16-induced DNA cleavage is related to an intermediate phase of cell activation which precedes stimulation of DNA synthesis.

Discussion
The finding that oestrogen-stimulated T-47D cells sustain higher levels of VP-16-induced DNA cleavage ( Figure 2b) and VP-16-induced cytostasis (Figure 2c) than do oestrogendeprived cells extends the observation of Sullivan et al. (1986) that synchronously proliferating cells incur greater VP-16-induced DNA cleavage and cytotoxicity than do quiescent cells. Although not conclusive, these findings are consistent with a causative relationship between the assayed DNA cleavage and the observed toxicity of VP-16. Indirect evidence favouring this relationship has also been provided by the work of other groups interested in the mechanism of action of VP-16 Estey et al., 1987). This is an important point if such observations are to serve as the basis of new strategies for improving the therapeutic index of clinical cancer therapy. The effects of cycloheximide and aphidicolin on VP-16-induced DNA cleavage in oestrogen-primed cells are in strong agreement with those reported by Chow & Ross (1987) for the effects of these inhibitors on enhanced VP-16-induced cleavage seen in synchronised cultures released from quiescence. The inhibitory effect of cycloheximide documented in both reports indicates that the enhanced drug-induced DNA cleavage witnessed in activated cell cultures is dependent at some point on new protein synthesis. This is perhaps not surprising given the very early protein synthetic events detectable immediately following stimulation ( Figure Sa), events which presumably play an important role in the sequential cascade of mitogenic activation. On the other hand, the lack of antagonism induced by aphidicolin in both studies suggests that enhancement of VP-16-induced DNA cleavage occurs predominantly, if not exclusively, within activated GI-phase cells. This scenario is clearly more attractive for the design of clinical trials based on target cell stimulation than is a strategy dependent upon highly schedule-specific administration of S-phase-active drugs.
That the enhancement of VP-16-induced DNA cleavage does indeed occur in Gl-phase cells is further supported by the data presented in Figure 5, which shows that VP-16induced DNA cleavage is enhanced several hours prior to enhancement of DNA synthesis. Since c-myc protein levels are increased only two-fold by oestrogen stimulation of this cell systemwhereas numerous studies of synchronised cell systems confirm a ten-fold increment in the expression of this gene following release of cells from quiescence (Kelly et al., 1983;Dean et al., 1986) -the possibility exists that only a fraction of the T-47D cell population is activated by oestrogen exposure. This hypothesis is consistent with the relatively small increase in S-phase cells documented after 24h oestrogen exposure (Figure 5d). Furthermore, if only a subpopulation of GI-phase cells is in fact activated by oestrogen, the observed overall two-fold enhancement of VP-16-induced DNA cleavage could represent a gross underestimate of true DNA cleavage enhancement in the putative oestrogen-activated cell subset. Flow cytometric studies are now underway to clarify this issue by examining the effect of VP-16-induced DNA damage at the single-cell level.
There is now much evidence suggesting that most VP-16induced DNA cleavage revealed by alkalior proteinasebased DNA damage assays represents stabilised 'cleavable complexes' of DNA and the intranuclear enzyme topoisomerase II (Chen et al., 1984;Ross et al., 1984). The importance of chromatin structure (addressed in Figure 4) in mediating this type of DNA lesion has been highlighted by the work of Udvardy & Schedl (1986). Absolute levels of topoisomerase II have, moreover, recently been recognised to DNase II concentration (Ku ml-1) Figure 4 Influence of oestrogen on DNA and chromatin structure in cells not exposed to VP-16: (a) Effect of oestrogen on DNA double-strandedness as measured by either nucleoid sedimentation or alkaline DNA unwinding. Up to 72 h oestrogen exposure (0) leads to no change in nucleoid sedimentation compared to control (0). Samples pre-treated with oestrogen for 24 h (-) show no change in the rate of alkaline DNA unwinding when compared with ethanol-treated controls (A) over a 3 h unwinding period; (b) Influence of 24 h oestrogen pretreatment on DNA superhelicity as measured by ethidium bromide titration of histone-depleted nucleoids. The amount of ethidium bromide required to induce minimum sedimentation (i.e. minimal DNA supercoiling) is not distinguishably different in oestrogen-treated (0) and control (0) samples. Datum points are based on one of two similar experiments; (c) Bleomycin-induced DNA scission as an index of chromatin accessibility in oestrogen-treated (0) and control (0) cells; (d) DNase II nicking as an index of chromatin accessibility in freeze-thaw-permeabilised cells. 0, control cells; *, oestrogen-treated cells.
be higher in proliferating than in quiescent cells (Heck & Earnshaw, 1986). However, the report that enzyme levels (as determined by immunoblotting) rise only in S-and G2-phase cells following stimulationeven though enhancement of VP-16-induced DNA cleavage is unaffected by inhibition of DNA synthesis using aphidicolin (Chow & Ross, 1987)suggests that enzyme activation within GI-phase cells may accompany cellular activation, a model supported by the data presented here. Indeed, in vitro data have already implicated phosphorylation (Ackerman et al., 1985), poly(ADP-ribosyl)ation (Darby et al., 1985) and calciummediated pathways (Osheroff & Zechiedrich, 1987) as candidate mechanisms for topoisomerase II activation. Hence the inhibition of VP-16-induced cleavage seen with cycloheximide exposure could signify inhibited synthesis of a molecule contributing to the process of enzyme activation, rather than necessarily implicating inhibition of topoisomerase II synthesis per se. We are currently exploring these questions further by correlating other parameters of topoisomerase II availability with VP-16-induced DNA cleavage.
The findings presented here suggest that oestrogen stimulation of asynchronously growing human breast cancer Increase in 3H-thymidine incorporation during oestrogen stimulation. Standard errors are based on triplicate measurements; (d) Oestrogen-stimulated increase in rate of cell-cycle traverse. GI-phase emptying was determined for samples exposed to either oestrogen or ethanol alone for the required period by simultaneously treating samples with colcemid 60ngml-1. Since this manoeuvre prevents mitotic cells re-entering GI, the rate at which cells exit from Gl can be used as an index of the rate of cell-cycle traverse. Control values at each time-point were subtracted from those of oestrogen-treated samples to yield a net percentage of Gl-phase cells exiting from Gl as a consequence of oestrogen stimulation alone; (e) Effect of oestrogen stimulation on cell-cycle redistribution as measured by absolute increase in S-phase fraction relative to control. The percentage of cells in S-phase was calculated by flow cytometric analysis. 0, control cells; 0, oestrogen-treated cells.
cells is a clinically relevant strategy for improving the therapeutic index of VP-16. Since drugs more commonly used to palliate breast cancer-such as mitoxantroneare also known to interact with topoisomerase II (Crespi et al., 1986), this strategy could be rationally combined with stateof-the-art management regimens without recourse to potentially confounding cell synchronisation protocols. A clinical trial is now being designed to determine the empirical value of this approach in patients with disseminated breast cancer.
The authors would like to thank Dr John Moore for his assistance in performing the c-myc protein estimations, and Dr Gerard Evan for providing the c-myc antibody. R.J.E. was supported by the Sir Robert Menzies Memorial Trust and in part by the Royal Australasian College of Physicians.