Int. J. Cancer: 70, 291–296 (1997) r 1997 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer DIFFERENT POINTS OF ACTION OF RETINOIDS AND ANTI-ESTROGENS IN G1 PHASE IDENTIFIED IN SYNCHRONIZED T-47D BREAST CANCER CELLS Nicholas R.C. WILCKEN1, Elizabeth A. MUSGROVE1 and Robert L. SUTHERLAND1* 1Cancer Research Program, Garvan Institute of Medical Research, St. Vincent’s Hospital, Darlinghurst, Sydney, Australia Both retinoids and anti-estrogens inhibit breast cancer cell proliferation with accumulation of cells in the G1 phase of the cell cycle, but the effect of retinoids is delayed compared to that of anti-estrogens. To determine whether this temporal difference is due to a simple delay in the action of retinoids on a common site or to different sites of action within the G1 phase, we studied the cell cycle effects of retinoic acid (RA) and the anti-estrogen ICI 164384 (ICI) in T-47D cells partially synchronized by mevalonic acid rescue of lovastatin-induced cell cycle arrest. We found that cells entering the cell cycle semi-synchronously after mevalonic acid rescue of lovastatin treatment were immediately susceptible to ICI but not RA. This suggests that RA may act at a point up-stream and ICI at a point down-stream of lovastatin action. Consistent with this, cells recommencing cell cycle progression after RA treatment were susceptible to the effects of lovastatin, while cells pre-treated with ICI then rescued with estradiol were not. In addition, cells rescued from cell cycle arrest induced by either RA, ICI or lovastatin entered S phase with the same kinetics. Our findings suggest, first, that within G1, RA acts before and ICI acts after the point of lovastatin action and, second, that despite these differences in the initiation of cell cycle arrest, the final nature of the cell cycle arrest is similar. Hence, retinoids and anti-estrogens may be expected to target different cell cycle–regulatory molecules to initiate cell cycle arrest, while overcoming this arrest may be accomplished by the activation of a common molecular pathway. Int. J. Cancer, 70:291–296, 1997. r 1997 Wiley-Liss, Inc. Tamoxifen has been in clinical use for the treatment of breast cancer for 2 decades and is currently being evaluated as a preventive agent. Retinoids have also shown promise as potential therapeutic and preventive agents for breast and other cancers, though strong evidence from animal studies has yet to be translated into success in clinical trials in breast cancer. While the cell cycle effects of anti-estrogens (both steroidal and non-steroidal) on ER1 breast cancer cells are well defined, less is known about retinoid action in these cells, yet potential differences and similarities between the 2 agents may be relevant to both their molecular mechanisms of action and their clinical use. The potential importance of preventive agents has increased recently with the identification of genes (BRCA1 and BRCA2) which, when mutated, cause inherited breast cancer. With the almost inevitable introduction of genetic screening for selected groups will come the identification of women at very high risk of breast cancer, for whom at present very little can be done. In the experimental setting, the combination of retinoids and antiestrogens is more effective than either agent alone in preventing mammary tumors (McCormick and Moon, 1986; Ratko et al., 1989; Anzano et al., 1994), suggesting an attractive strategy for breast cancer prevention in women at high risk. Delineating separate mechanisms of action for retinoids and anti-estrogens may lead to the ability to predict differential response rates, thus refining prevention strategies. As well, if common steps are required to overcome the cell cycle effects of retinoids and anti-estrogens, the molecular characterization of such steps may point to mechanisms of resistance to these treatments. Anti-estrogens were first shown to inhibit cell cycle progression in the G1 phase, and subsequently a series of experiments using mitotically selected cells showed that breast cancer cells were susceptible to anti-estrogens from early to mid-G1 (Sutherland et al., 1983; Taylor et al., 1983). Molecular studies have consequently focused on G1 events and the way in which anti-estrogens interact with the proteins that control cell cycle progression in G1 (Watts et al., 1995). While the growth-inhibitory effects of retinoids in cell culture are well documented in a variety of cell types (e.g., Lotan, 1979), as are chemopreventive effects in rodent models of mammary cancer (e.g., Moon et al., 1992), there is less information and less consensus about the precise cell cycle effects seen in culture. This is partly due to the known cell type-specific actions of retinoids, making it difficult to extrapolate from one study to another. Nevertheless, in the majority of studies involving breast cancer cells, DNA synthesis is reduced, and, as with anti-estrogen treatment, cells accumulate in G1 phase (Ueda et al., 1980; Marth et al., 1985; Rubin et al., 1994; Zhao et al., 1995). For both anti-estrogens and retinoids, inhibition of cell cycle progression is closely correlated with reduced cell proliferation, but the effects of retinoids and anti-estrogens on G1 progression have different kinetics (Wilcken et al., 1996). We therefore sought to define more closely the relationship between retinoids and anti-estrogens with respect to their cell cycle effects by extending our previously published data on asynchronously proliferating T-47D breast cancer cells to populations of synchronous cells. Cell synchronization was achieved with the mevalonic acid rescue of lovastatin-treated cells (described in Keyomarsi et al., 1991). Lovastatin and related drugs inhibit HMG-CoA reductase, a rate-limiting enzyme in the cholesterol synthesis pathway, an action which in cell culture results in cell cycle arrest early in the G1 phase (Jakobisiak et al., 1991). This can then be reversed by addition to the culture medium of mevalonic acid, the immediate product of HMG-CoA reductase, and as a result cells recommence cell cycle progression in a synchronous manner. Our data using this cell culture system demonstrate that the cell cycle effects of retinoic acid (RA) and the pure steroidal antiestrogen ICI 164384 are different, despite the fact that both agents cause cells to accumulate in G1 phase and that differences in the molecular pathways by which they control cell proliferation, therefore, are to be expected. MATERIAL AND METHODS Material T-47D cells were supplied by E.G. and G. Mason Research Institute (Worcester, MA) for the National Cancer Institute Breast Cancer Program Cell Culture Bank, and MCF-7 cells were from the Michigan Cancer Foundation (Detroit, MI). RPMI 1640 medium containing HEPES (N-2-hydroxyethylpiperazine-N8-ethanesulfonic acid) (20 mM), sodium bicarbonate (14 mM), L-glutamine (6 mM) and 20 µg/ml gentamicin was used throughout. Stock cultures were maintained in medium supplemented with 10 µg/ml human Contract grant sponsor: The National Health and Medical Research Council of Australia. *Correspondence to: Garvan Institute of Medical Research, St. Vincent’s Hospital, Darlinghurst, NSW 2010, Australia. Fax: 61 (2) 9295 8321. Received 16 July 1996; revised 1 October 1996. WILCKEN ET AL. 292 insulin (Actrapid; CSL-Novo, North Rocks, Australia) and 10% FCS, and cell growth experiments were carried out in medium supplemented with 5% FCS. To maintain maximal growth rates and therefore observe maximal growth inhibition, serum was not charcoal-stripped. RA was purchased from Sigma (St. Louis, MO) and Ro 13-7410 (P-[(E)-2-(5,6,7,8,-tetrahydro-5,5,8,8,-tetramethyl2-naphthalenyl)]-1- propenyl-benzoate) was obtained from Dr. M. Klaus (F. Hoffman-La Roche, Basel, Switzerland). ICI 164384 (N-n-butyl-N-methyl-[3,17b-dihydroxyestra-1,3,5 (10)-triene-7ayl] undecanamide) was obtained from Dr. A. Wakeling (Zeneca, Macclesfield, UK). The anti-estrogen and retinoids were stored as 1,000-fold concentrated stock solutions at 270°C in either absolute ethanol (ICI 164384 and RA) or DMSO (Ro 13-7410), and the retinoids were sealed with argon and protected from light. Retinoid stock solutions were discarded after 1 week. Lovastatin was purchased from Merck Sharp and Dohme (Rahway, NJ) and converted from its lactone pro-drug form to the open acid form by dissolving of the compound in 1 M NaOH, neutralizing with HCl to a pH of 7.2 and making up to a volume with distilled water to produce a 10 mM stock solution that was stored at 220°C. Mevalonic acid lactone was purchased from Sigma. Cell growth experiments For growth-inhibition assays, 105 cells were plated into 25 cm2 flasks. One to 2 days later, 5 µl aliquots of drug (ICI 164384, RA or Ro 13-7410) were added directly to the culture medium. Cells were then harvested and counted under phase contrast microscopy, and DNA flow cytometry was performed as previously described (Musgrove et al., 1989). Briefly, trypsinized cells were resuspended in medium at a concentration of 2 to 3 3 105 cells/ml. Ethidium bromide (Sigma) and mithramycin (Pfizer, West Ryde, Australia) were added at final concentrations of 40 µg/ml and 12.5 µg/ml, respectively, in the presence of 0.2% Triton X-100 (Sigma) and, for mithramycin, at a final concentration of 7.5 mM MgCl2. Stained samples were stored either at 4°C overnight or at room temperature for 4 hr. DNA flow cytometry was performed on a FACStar laser-based flow cytometer (Becton Dickinson, Mountain View, CA), and data were acquired and analyzed using the accompanying software (Becton Dickinson). DNA histograms contained 30,000 events and had a coefficient of variation of 2–4%. retinoids and anti-estrogens invariably leads to a delay in the stereotypical cell cycle action of the retinoids compared with the anti-estrogens (Wilcken et al., 1996). Figure 1 shows the mean effect of equipotent concentrations of RA and the pure steroidal anti-estrogen ICI 164384 on S phase entry in T-47D cells and demonstrates that the same effect is also seen in MCF-7 breast cancer cells. Since MCF-7 cells are both more sensitive to anti-estrogens and less sensitive to retinoids than T-47D cells, however, lower concentrations of anti-estrogen and higher concentrations of retinoid are required to achieve equipotency. As shown in Figure 1, MCF-7 cells, therefore, have been treated with the retinoid analogue Ro 13-7410 to match the added sensitivity of MCF-7 cells to anti-estrogens. In both cell lines, the retinoids tested had little effect on the percent of cells in S phase (%S phase) after 16 hr of treatment, while the anti-estrogen had a near maximal effect at this time, yet by 40 hr the effects on %S phase were broadly similar for both classes of agent. This delay in retinoid cell cycle action is not seen in genomic actions of retinoids, such as the transcriptional regulation of androgen and progesterone receptor gene expression in T-47D cells (Clarke et al., 1991; Hall et al., 1992; Wilcken et al., 1996), implying that access to receptors and DNA is not a limiting factor in retinoid action. Thus, in both breast cancer cell lines there is a difference between retinoid and anti-estrogen effects which represents either different mechanistic pathways or a simple delay in the onset of retinoid action on the cell cycle. Subsequent experiments were confined to T-47D cells and aimed at distinguishing between these 2 possibilities by comparing the effects of RA and ICI 164384 in synchronized cells. Inducing cell synchrony with lovastatin Initial experiments were carried out to determine the concentration and duration of lovastatin treatment that gave the greatest extent of G1 arrest and subsequent synchronous progression through the cell cycle after mevalonic acid rescue in T-47D cells (data not shown). The optimized procedure was as follows: Cell synchronization Cells (105) were plated into 25 cm2 flasks, and after 1–2 days aliquots of 10 mM lovastatin were added to make a final concentration of 40 µM. Twenty-four hours later, medium was removed and fresh medium added together with 2 M mevalonic acid to yield a final concentration of 5 mM. In some experiments, RA, ICI 164384 or vehicle (0.1% ethanol) were also added at this time. Reversal experiments Two types of experiment were carried out. First, 105 cells were plated into 25 cm2 flasks; after 1 day, RA was added for 48 hr, or ICI 164384 or lovastatin for 24 hr, at which time the medium was replaced in all flasks. Estradiol was then added to the cells previously treated with RA or ICI 164384 and mevalonic acid to the cells previously treated with lovastatin, and cells were then harvested at intervals for DNA flow cytometry. Second, cells were treated with either RA for 48 hr or ICI 164384 for 24 hr, after which medium was replaced and estradiol added to all flasks. Then, either lovastatin or H2O was added, and cells were harvested at intervals for DNA flow cytometry. RESULTS AND DISCUSSION Differential effects of retinoids and anti-estrogens on asynchronous cells We have previously shown that, while the final effect on cell cycle phase distribution is the same for several retinoids and anti-estrogens in T-47D cells (an accumulation in G1 phase and a decline in S phase), the time course of these effects is different. Thus, treatment of T-47D cells with a series of structurally different FIGURE 1 – Effects of retinoids and anti-estrogens on %S phase in asynchronous T-47D and MCF-7 cells. T-47D or MCF-7 cells proliferating in medium supplemented with 5% FCS were treated with retinoid, anti-estrogen or vehicle (0.1% ethanol) and harvested after 16 and 40 hr for determination of cell cycle phase distribution by flow cytometry. Data are presented relative to the %S phase of vehicletreated cells. T-47D cells were treated with RA (1026 M, filled histograms: mean 6 SEM) or the anti-estrogen ICI 164384 (1027 M, striped histograms: mean 6 range). MCF-7 cells were treated with the retinoid Ro 13-7410 (1025 M, filled histograms: mean 6 range) or the anti-estrogen ICI 164384 (3 3 1029 M, striped histograms: mean 6 range). RETINOIDS, ANTI-ESTROGENS AND BREAST CANCER exponentially growing cells were incubated with 40 µM lovastatin for 24 hr, at which time the medium was replaced and 5 mM mevalonic acid added. Cells were then harvested at intervals over the subsequent 56 hr and cell cycle phase distribution measured. Figure 2 shows that cell synchrony was achieved, with a synchronous cohort of cells entering S phase 8 hr after treatment with mevalonic acid. A maximal 4-fold increase in %S phase to approximately 40% was observed between 20 and 24 hr. This initial increase in %S phase was followed 8–10 hr later by an increase in the number of cells in G2 1 M phase of a similar proportional magnitude to the increase in %S phase (i.e., about 4-fold). In addition, there was a second, smaller peak of cells in S phase at about 40 hr, indicating that the next generation of cells had entered a second cell cycle in partial synchrony. Therefore, the results are consistent with studies in other cell types (Keyomarsi et al., 1991): a synchronous population of cells entered the cell cycle after rescue from lovastatin arrest, progressed through S phase and on to mitosis and thence to a second partially synchronous cell cycle. Although this method does not yield a perfectly synchronous population, the degree of enrichment of cells in S phase seen was sufficient to allow meaningful interpretation of the effects of anti-estrogens and retinoids on cell cycle progression. The technique also represents a good compromise between the desirable effect of achieving complete synchrony and the undesirable effect of using synchronizing agents that may themselves have confounding cell cycle or metabolic effects. For example, the standard method for obtaining a population of synchronized cells is mitotic selection since this technique selects cells that are at the beginning of the cell cycle but does not affect underlying RNA and protein synthesis. However, there is cell line-dependent variability in the purity of the selected population, and T-47D cells produce a smaller proportion of synchronized cells than do MCF-7 cells. Nevertheless, mitotic selection has been performed on T-47D cells (Musgrove, 1992), and the resultant progression of cells through G1 phase and into S phase is very similar to that shown in Figure 2, with little change seen in %S phase over the first 9 hr and a steep rise then occurring to a peak of about 50% after about 21 hr. In addition, lovastatin treatment does not affect overall RNA and protein synthesis (Jakobisiak et al., 1991; Keyomarsi et al., 1991) and, therefore, is likely to be superior to other methods of synchronization, such as growth factor withdrawal or treatment FIGURE 2 – Synchronization of T-47D cells. T-47D cells proliferating in medium supplemented with 5% FCS were treated with 40 µM lovastatin for 24 hr, after which medium was replaced and 5 mM mevalonic acid added. Cells were then harvested at intervals over 56 hr for determination of cell cycle phase distribution by flow cytometry. S phase data are the mean 6 SEM of 3 independent experiments. G2 1 M data are from a representative experiment. 293 with hydroxyurea, but similar in efficacy to mitotic selection in T-47D cells. Effects of ICI 164384 and RA on synchronized cells The next experiment was designed to test whether cells synchronized by this method are susceptible to the growth-inhibitory effect of the pure steroidal anti-estrogen ICI 164384. A model placing lovastatin action early in G1 phase and ICI 164384 action somewhat later predicts that ICI 164384 would suppress some or all of the S phase entry seen with lovastatin–mevalonic acid synchronisation, and this was observed. As Figure 3 demonstrates, cells rescued with mevalonic acid reached a peak %S phase of 45% at 24 hr, while cells treated with 1027 M ICI 164384 at the same time as mevalonic acid was added reached a peak of only 30%. This suppression of S phase entry was followed by a decline in entry into G2 1 M, demonstrating that fewer cells entered and completed the cell cycle in the presence of ICI 164384. Furthermore, the second S phase peak seen in the synchronized cells was completely ablated by ICI 164384, showing that ICI 164384 was active during the second round of replication. Therefore, sensitivity to the cell cycle effects of ICI 164384 lies down-stream of the point at which lovastatin arrests cells. To test whether the relationship between RA and lovastatin effects was the same as that for ICI 164384, the same experiment was undertaken with RA. In this case (Fig. 4), RA did not appreciably affect the synchronous entry of cells into S phase at 24 hr but did reduce the second, smaller wave of S phase entry, as had ICI 164384 in the previous experiment. The proportion of cells entering G2 1 M phase was also unaffected at 24 hr and reduced at 48 hr, consistent with the S phase data. The results of these experiments demonstrate that the cell cycle actions of RA and ICI 164384 are different since cells rescued from FIGURE 3 – Effect of ICI 164384 on cells synchronized by lovastatin– mevalonic acid. T-47D cells proliferating in medium supplemented with 5% FCS were treated with 40 µM lovastatin for 24 hr, after which medium was replaced and 5 mM mevalonic acid added together with either 1027 M ICI 164384 or vehicle (0.1% ethanol). Cells were then harvested at intervals over 56 hr for determination of cell cycle phase distribution by DNA flow cytometry. Similar data were obtained in a second independent experiment. 294 WILCKEN ET AL. FIGURE 4 – Effect of RA on cells synchronized by lovastatin– mevalonic acid. T-47D cells proliferating in medium supplemented with 5% FCS were treated with 40 µM lovastatin for 24 hr, after which medium was replaced and 5 mM mevalonic acid added together with either 1026 M RA or vehicle (0.1% ethanol). Cells were then harvested at intervals over 56 hr for determination of cell cycle phase distribution by flow cytometry. Similar data were obtained in a second independent experiment. RA-treated cells, however, no specific agonist was available; therefore, the medium was changed to remove as much retinoid as possible. Estradiol was also added as a control, though it has previously been demonstrated that estradiol in this setting does not have an effect on RA-treated cells (Wilcken et al., 1996). Thus, the effects of both ICI 164384 and lovastatin were specifically reversed (by estradiol and mevalonic acid, respectively), while RA was simply removed from the medium. Figure 5 shows a pooled time course of lovastatin-rescued cells synchronously entering S phase, as seen in Figures 2–4, with a peak value of 40–45% at 24 hr. Estradiol reversal of ICI 164384 treatment also led to synchronous re-entry into S phase but with 35% of cells in S phase at 24 hr, while removal of RA resulted in a more modest reversal, with a peak S phase value of 28–30% at 24 hr. Therefore, the quantitative effect of ‘‘reversing’’ the 3 treatments differed, and this effect was independent of the starting %S phases, all of which were similar. However, this is most likely a reflection of the varying degrees with which the respective treatments were reversed. Both lovastatin and ICI 164384 treatments were specifically reversed (with mevalonic acid and estradiol, respectively), while the effect of RA was ‘‘reversed’’ simply by the addition of fresh medium, there being no direct way to antagonize RA. Therefore, it might be expected that there would be less S phase entry with RA ‘‘reversal’’. The same could apply to ICI 164384, where a ligand-receptor-DNA binding process must be reversed by estradiol, while mevalonic acid addition restores a single biochemical pathway. However, despite the difference in the magnitude of S phase entry, the timing of entry into S phase after reversal of cell cycle arrest was very similar for all 3 treatments. Since the data presented above suggest that the 3 agents tested have different points of action, this result implies that the final growth- lovastatin arrest were immediately susceptible to the effects of ICI 164384 but not RA. An explanation compatible with these observations is that ICI 164384 acts at a point in G1 phase after the action of lovastatin, while cells are susceptible to RA action only before the point at which lovastatin acts since cells released from lovastatin were not susceptible to RA until the next cell cycle. However, a major factor confounding such an interpretation is the delay in the onset of RA action on the cell cycle, as demonstrated in Figure 1. Thus, these results are also compatible with RA and ICI 164384 acting at the same point in G1 phase but with the delay in RA action explaining the differential effects on synchronized cells. Effects of lovastatin on cells recovering from ICI 164384 or RA treatment Two further sets of experiments were carried out to address this question. First, it could be postulated that if drugs which arrested cell cycle progression at different points within G1 phase were removed or their effects reversed, cells might enter S phase at different times. Therefore, the time taken to recover from cell cycle arrest could provide some indication as to where in the cell cycle arrest had occurred. This question was examined by treating cells with RA (48 hr), ICI 164384 (24 hr) or lovastatin (24 hr) and then reversing the effects of these treatments as much as possible, to allow the recommencement of cell cycle progression. In the case of lovastatin-treated cells, this was straightforward, and Figures 2–4 demonstrate how effectively fresh medium with added mevalonic acid overcomes the effect of lovastatin. Maximal reversal of anti-estrogen was achieved by changing the medium and adding estradiol to the ICI 164384-treated cells. For the FIGURE 5 – Effect of reversing the cell cycle arrest caused by lovastatin, ICI 164384 or RA. T-47D cells proliferating in medium supplemented with 5% FCS were treated with 40 µM lovastatin for 24 hr (L, h) or 1027 M ICI 164384 for 24 hr (ICI, n) or 1026 M RA for 48 hr (RA, s), and at time zero these treatments were reversed. The medium in all flasks was changed; 5 mM mevalonic acid was added to lovastatin-treated cells and 1027 M estradiol was added to RA- and ICI 164384-treated cells. Cells were then harvested at intervals over 28 hr for determination of cell cycle phase distribution by flow cytometry. Data points represent the mean 6 SEM of 3 independent experiments. RETINOIDS, ANTI-ESTROGENS AND BREAST CANCER arrested state is similar for all 3 agents. An alternative explanation is that rescued cells, having been arrested at different points in the cell cycle, proceed through G1 phase at different rates, thus arriving at S phase at the same time. Finally, if the point of action of RA is indeed earlier in G1 phase than the point of lovastatin action, one would predict that cells released from RA-induced arrest would be immediately susceptible to lovastatin, while cells released from ICI-induced arrest would not. If RA acted after lovastatin, such cells would need to progress through the cell cycle to a round of replication before being affected by lovastatin. Therefore, exponentially growing cells were treated with either RA for 48 hr or ICI 164384 for 24 hr, then, the resulting block in cell cycle progression induced by each drug was relieved by changing to drug-free medium with estradiol added, as in the previous experiment. At the same time, lovastatin (or H2O as a control) was added, and cells were harvested at intervals over a period of 28 hr. The results of this experiment are shown in Figure 6. Cells rescued from RA-induced arrest re-initiated cell cycle progression, with %S phase more than doubling from 12% after 48 hr of RA treatment to 28% 24 hr after this treatment was reversed, as seen in Figure 5. However, in the presence of lovastatin, cells remained in an arrested state (Fig. 6). In contrast, while cells rescued from ICI 164384-induced arrest re-entered S phase (Figs. 5, 6), lovastatin added at the time of reversal had no major effect on this cell cycle progression (Fig. 6). Consistent with the lack of effect lovastatin had on recovery from anti-estrogen treatment, shown in Figure 6, Bonapace et al. (1996) have reported that estradiol can induce cell cycle progression in MCF-7 cells treated with the closely related simvastatin, demonstrating that estradiol-mediated pathways can override the effects of HMG-CoA reductase inhibitors. These data are again consistent with the hypothesis that ICI 164384 and RA act at different points in the cell cycle. While cells recovering from treatment with lovastatin, ICI 164384 or RA enter S phase at the same time (albeit with some differences in magnitude), cells recovering from treatment with ICI 164384 or RA do not respond to treatment with lovastatin in the same way. Thus, the hypothesis generated from the data presented in Figures 3 and 4 (that within the cell cycle RA acts before and ICI 164384 after lovastatin) is supported. Since in breast cancer cells both anti-estrogens and retinoids inhibit cell cycle progression in the G1 phase of the cell cycle, both agents must act, either directly or indirectly, on molecules that are necessary for transit through G1. Progression through G1 phase is regulated by the sequential activation of cyclin-CDK complexes, which in turn phosphorylate down-stream targets, the best defined of which is pRB, the gene product of the tumor-suppressor gene RB-1. Hyperphosphorylation of pRB is necessary for entry into S phase, and there is, thus, a strong link between G1 cyclin induction, CDK activation, pRB phosphorylation and transit through G1 phase (Sherr, 1994). While the major event that activates CDKs is the binding of cyclin, both phosphorylation and dephosphorylation of certain residues on the CDK itself are also required. Anti-estrogen treatment of breast cancer cells is accompanied by decreases in cyclin D1 mRNA and protein levels, CDK4 activity and pRB phosphorylation, followed by G1 cell cycle arrest (Watts et al., 1995; Wilcken et al., 1996). The time sequence of these events strongly suggests a causative relationship. Conversely, RA caused a fall in CDK4 activity and pRB phosphorylation without affecting cyclin D1 levels (Wilcken et al., 1996). While such findings suggest that RA may act by inducing an inhibitor of CDK4 activity, this remains to be demonstrated. Moreover, the need to induce the synthesis of such a molecule to levels where CDK activity is inhibited could explain the delay in RA action on asynchronous cells seen in Figure 1. Another untested possibility is that RA indirectly affects the necessary phosphorylation or dephosphorylation of CDK4. 295 FIGURE 6 – Effect of lovastatin on cells re-entering the cell cycle after reversal of RA or ICI 164384 treatment. T-47D cells proliferating in medium supplemented with 5% FCS were treated with 1026 M RA for 48 hr or 1027 M ICI 164384 for 24 hr, and at time zero these treatments were reversed. The medium was changed and 1027 M estradiol added to all flasks plus either 40 µM lovastatin (open symbols) or H2O (closed symbols). Cells were then harvested at intervals over 28 hr for determination of cell cycle phase distribution by flow cytometry. Data points represent the mean 6 SEM of 3 independent experiments. The cell cycle studies reported here thus support the notion that the molecular targets of anti-estrogens and retinoids are likely to differ and that initial targets of RA action might be expected to regulate progression through an early part of G1 phase. However, we have also shown that reversal of cell cycle arrest induced by either RA or ICI 164384 (or lovastatin) resulted in entry into S phase at the same time, suggesting that the maintenance of a growth-arrested state in both cases may be overcome by the 296 WILCKEN ET AL. activation of common steps. Molecular studies point to the importance of CDK4 function, and the regulation of this molecule (albeit by different mechanisms) may be a critical factor in the action of both retinoids and anti-estrogens. Finally, the demonstration of a difference in cell cycle action and the implication of different molecular mechanisms of action lend weight to animal data that consistently show synergism between anti-estrogens and retinoids when these are used as anticarcinogenic agents. Collectively, these observations strongly suggest that a combination of these agents may be a more effective mode of cancer protection in those at high risk than the use of one or other agent alone. ACKNOWLEDGEMENTS This study was supported by the National Health and Medical Research Council of Australia (NHMRC). N.R.C.W. is the recipient of an NHMRC Medical Postgraduate Research Scholarship. REFERENCES ANZANO, M.A., BYERS, S.W., SMITH, J.M., PEER, C.W., MULLEN, L.T., BROWN, C.C., ROBERTS, A.B. and SPORN, M.B., Prevention of breast cancer in the rat with 9-cis-retinoic acid as a single agent and in combination with tamoxifen. Cancer Res., 54, 4614–4617 (1994). BONAPACE, I.M., ADDEO, R., ALTUCCI, L., CICATIELLO, L., BIFULCO, M., LAEZZA, C., SALZANO, S., SICA, V., BRESCIANI, F. and WEISZ, A., 17bEstradiol overcomes a G1 block induced by HMG-CoA reductase inhibitors and fosters cell cycle progression without inducing ERK-1 and -2 MAP kinases. Oncogene, 12, 753–763 (1996). CLARKE, C.L., GRAHAM, J., ROMAN, S.D. and SUTHERLAND, R.L., Direct transcriptional regulation of the progesterone receptor by retinoic acid diminishes progestin responsiveness in the cancer cell line T-47D. J. biol. Chem., 266, 18969–18975 (1991). HALL, R.E., TILLEY, W.D., MCPHAUL, M.J. and SUTHERLAND, R.L., Regulation of androgen receptor gene expression by steroids and retinoic acid in human breast-cancer cells. Int. J. Cancer, 52, 778–784 (1992). JAKOBISIAK, M., BRUNO, S., SKIERSKI, J.S. and DARZYNKIEWICZ, Z., Cell cycle-specific effects of lovastatin. Proc. nat. Acad. Sci. (Wash.), 88, 3628–3632 (1991). KEYOMARSI, K., SANDOVAL, L., BAND, V. and PARDEE, A.B., Synchronization of tumour and normal cells from G1 to multiple cell cycles by lovastatin. Cancer Res., 51, 3602–3609 (1991). LOTAN, R., Different susceptibilities of human melanoma and breast carcinoma cell lines to retinoic acid-induced growth inhibition. Cancer Res., 39, 1014–1019 (1979). MARTH, C., BOCK, G. and DAXENBICHLER, G., Effect of 4-hydroxyphenylretinamide and retinoic acid on the proliferation and cell cycle of cultured human breast cancer cells. J. nat. Cancer Inst., 75, 871–875 (1985). MCCORMICK, D.L. and MOON, R.C., Retinoid–tamoxifen interaction in mammary cancer chemoprevention. Carcinogenesis, 7, 193–196 (1986). MOON, R.C., MEHTA, R.G. and DETRISAC, C.J., Retinoids as chemopreventive agents for breast cancer. Cancer Detection Prevention, 16, 73–79 (1992). MUSGROVE, E.A., Hormonal control of breast cancer cell cycle progression. PhD thesis, University of NSW, Sydney (1992). MUSGROVE, E.A., WAKELING, A.E. and SUTHERLAND, R.L., Points of action of estrogen antagonists and a calmodulin antagonist within the MCF-7 human breast cancer cell cycle. Cancer Res., 49, 2398–2404 (1989). RATKO, T.A., DETRISAC, C.J., DINGER, N.M., THOMAS, C.F., KELLOFF, G.J. and MOON, R.C., Chemopreventive efficacy of combined retinoid and tamoxifen treatment following surgical excision of a primary mammary cancer in female rats. Cancer Res., 49, 4472–4476 (1989). RUBIN, M., FENIG, E., ROSENAUER, A., MENENDEZ-BOTET, C., ACHKAR, C., BENTEL, J.M., YAHALOM, J., MENDELSOHN, J. and MILLER, W.H.J., 9-cisretinoic acid inhibits growth of breast cancer cells and down-regulates estrogen receptor RNA and protein. Cancer Res., 54, 6549–6556 (1994). SHERR, C.J., G1 phase progression: cycling on cue. Cell, 79, 551–555 (1994). SUTHERLAND, R.L., GREEN, M.D., HALL, R.E., REDDEL, R.R. and TAYLOR, I.W., Tamoxifen induces accumulation of MCF 7 human mammary carcinoma cells in the G0/G1 phase of the cell cycle. Europ. J. Cancer Clin. Oncol., 19, 615–621 (1983). TAYLOR, I.W., HODSON, P.J., GREEN, M.D. and SUTHERLAND, R.L., Effects of tamoxifen on cell cycle progression of synchronous MCF-7 human mammary carcinoma cells. Cancer Res., 43, 4007–4010 (1983). UEDA, H., TAKENAWA, T., MILLAN, J.C., GESELL, M.S. and BRANDES, D., The effects of retinoids on proliferative capacities and macromolecular synthesis in human breast cancer MCF-7 cells. Cancer, 46, 2203–2209 (1980). WATTS, C.K.W., BRADY, A., SARCEVIC, B., DEFAZIO, A., MUSGROVE, E.A. and SUTHERLAND, R.L., Antiestrogen inhibition of cell cycle progression in breast cancer cells is associated with inhibition of cyclin-dependent kinase activity and decreased retinoblastoma protein phosphorylation. Mol. Endocrinol., 9, 1804–1813 (1995). WILCKEN, N.R.C., SARCEVIC, B., MUSGROVE, E.A. and SUTHERLAND, R.L., Differential effects of retinoids and antiestrogens on cell cycle progression and cell cycle regulatory genes in human breast cancer cells. Cell Growth Differentiation, 7, 65–74 (1996). ZHAO, Z., ZHANG, Z.-P., SOPRANO, D.R. and SOPRANO, K.J., Effect of 9-cis-retinoic acid on growth and RXR expression in human breast cancer cells. Exp. Cell Res., 219, 555–561 (1995).