Reports of the ability of estrogenic agents such as 17β-estradiol (E2), estriol (E3) and bisphenol A (BPA) to induce micronuclei (MN) in MCF-7 breast cancer cells have prompted us to investigate whether these effects are linked to activation of the estrogen receptor (ER) α. Coadministration of tamoxifen and the pure ER antagonist ICI 182 780 to cells treated with E2 and E3 did not lead to significant reductions in micronucleus frequencies. Since these antiestrogens interfere with the transcriptional activity of the ER and block promotion of ER-dependent gene expression, it appears that this process is not involved in micronucleus formation. However, ER activation also triggers rapid signaling via the Src/Raf/extracellular signal-regulated kinase (Erk) pathway. When MCF-7 cells were exposed to E2 and BPA in combination with the specific kinase inhibitors pyrazolopyrimidine and 2′-amino-3′-methoxyflavone, reductions in micronucleus frequencies occurred. These findings suggest that the Src/Raf/Erk pathway plays a role in micronucleus formation by estrogenic agents. Enhanced activation of the Src/Raf/Erk cascade disturbs the localization of Aurora B kinase to kinetochores, leading to a defective spindle checkpoint with chromosome malsegregation. Using antikinetochore CREST antibody staining, a high proportion of micronucleus containing kinetochores was observed, indicating that such processes are relevant to the induction of MN by estrogens. Our results suggest that estrogens induce MN by causing improper chromosome segregation, possibly by interfering with kinase signaling that controls the spindle checkpoint, or by inducing centrosome amplification. Our findings may have some relevance in explaining the effects of estrogens in the later stages of breast carcinogenesis.
It is generally accepted that cumulative exposure of the mammary gland to the sex hormone 17β-estradiol (E2) is a strong risk factor for the development of breast cancer in women ( 1 , 2 ). The role of E2 in carcinogenesis has traditionally been explained in terms of non-genotoxic processes, such as increased proliferation of hormone-responsive cells ( 3 ). With the observation that DNA-reactive intermediates can emerge from E2 came the realization that the hormone may also have a genotoxic effect ( 4 , 5 ). However, reports that steroidal estrogens do not induce mutations in commonly used point mutation assays ( 6 ) have motivated investigations of the hormone’s ability to induce genomic instability and changes in chromosome numbers (aneuploidy) ( 6 ).
In cell lines lacking the estrogen receptor (ER), aneuploidy has been observed at very high concentrations of E2 (>10 μM) ( 7 , 8 ). At similar non-physiological levels, the hormone can also produce micronuclei (MN), small bodies of nuclear material ejected from the cell nucleus into the cytoplasm of the affected cells ( 9 ). The frequency of MN in cells is commonly taken as an indication of genomic instability and aneuploidy ( 10 , 11 ). However, the physiological relevance of these observations was in doubt until Fischer et al. ( 12 ) demonstrated that E2 causes MN at much lower, nanomolar concentrations in ER-positive human breast cancer (MCF-7) cells. Similar observations were made with a number of E2 metabolites and derivatives, including estrone and estriol (E3) ( 13 ). ER-positive human ovary BG-1 cells also show micronucleus formation at nanomolar levels of E2 ( 14 ), suggesting that the hormone’s ability to induce this effect may be related to the presence of the ER.
This idea has been further investigated by analyzing associations between cell proliferation and the extent of micronucleus formation and by exploring the influence of antiestrogens on E2-mediated micronucleus induction. In MCF-7 cells, administration of E2 leads to an upsurge in cell proliferation ( 15 , 16 ), and this rise has been associated with increases in micronucleus frequency ( 12 ). In addition, coadministration of hydroxy tamoxifen with the hormone has been shown to almost completely suppress MN in MCF-7 and BG-1 cells, supporting the idea of an involvement of ER activation in this effect ( 12 , 14 ). Taken together, these observations have led Fischer et al. ( 12 ) to argue that the stimulation of MCF-7 cells by E2 leads to shorter cell cycles with a concomitant overriding of cell cycle checkpoints, less time for proper DNA repair and a resultant increase in MN. However, the proposed association between cell proliferation and micronucleus induction in MCF-7 cells is in conflict with reports by Yared et al. ( 13 ). In their hands, estrone, E2 and E3 all stimulated cell division, but only estrone and E2 caused MN, although differences in experimental protocol may explain some of these divergent findings.
These observations have prompted us to investigate further a possible role of ER activation in micronucleus induction. In addition to the classical genomic pathways (activation and translocation of ER to the nucleus), steroids can induce rapid non-genomic effects via secondary signaling pathways, such as the mitogen-activated protein kinase (MAPK) cascade ( 17 ). At the cytoplasmic level, the activated ER promotes stimulation of extracellular signal-regulated kinases (Erks), as well as Src and Ras, which are important upstream mediators ( 18 , 19 ). It has also been shown that environmental estrogens such as bisphenol A (BPA) are able to trigger similar non-genomic actions, such as phosphorylation of Erk1/Erk2 ( 20 ), but the consequences of these signaling events for genomic instability in mammary gland cells have not yet been explored.
Saavedra et al. ( 21 ) observed that induced overexpression of Ras can increase micronucleus frequencies in (ER negative) NIH 3T3 cells. Increased levels of v-ras precipitated the acquisition of one or more centromeres (centrosome overamplification), leading to the formation of mitotic bridges and malsegregation of chromosomes. All these changes gave rise to increased micronucleus frequencies, but could be suppressed by inhibition of Erk1/Erk2 ( 21 ). These results point to the involvement of MAPKs in unblocking the mitosis checkpoint, where cell cycle progression is delayed until all chromosomes are correctly aligned along the spindle equator. Regulatory proteins that override this checkpoint, thereby triggering cell cycle progression before correct chromosome alignment is completed, are a phosphorylation target of the Ras/Erk1/Erk2 pathway ( 22 ).
In the light of these findings, we hypothesize that Src and Erk1/Erk2 phosphorylation due to ER activation may play a role in micronucleus formation by estrogenic agents. To verify this hypothesis, we studied whether the induction of this effect by E2 and other estrogenic agents could be suppressed by administration of selective inhibitors of Erk1/Erk2 and Src. On the basis of our observations, we argue that it is the overriding of the spindle checkpoint due to Src/Ras/Erk activation, and not a shortening of the cell cycle due to the action of ‘mitogenic’ E2 ( 12 , 14 ), that plays a role in micronucleus induction by steroidal estrogens and other ER agonists.
Materials and methods
Chemicals and reagents
E2 (99% purity), E3 (98% purity), BPA (>99% purity), TAM (99% purity) and epidermal growth factor (EGF) were purchased from Sigma–Aldrich (Dorset, UK). Benzo[ a ]pyrene (96.5% purity) was purchased from Acros Organics (Fisher Scientific, Loughborough, UK). 2′-Amino-3′-methoxyflavone (PD 98059) was obtained from Promega (Southampton, UK), pyrazolopyrimidine (PP2) from Merck Biosciences Ltd (Nottingham, UK) and ICI 182 780 was a generous gift from Zeneca (Cheshire, UK).
Primary antikinetochore CREST antibody was purchased from Antibodies (Davis, CA) and fluorescein isothiocyanate (FITC)-conjugated goat polyvalent anti-human antibody from Sigma–Aldrich. DPX mounting media was from BDH (Leicestershire, UK) and Vectashield antifade solution containing 4′,6-diamidino-2-phenylindole from Vector Laboratories, Ltd (Peterborough, UK). All other chemicals and reagents were obtained from Sigma–Aldrich and Invitrogen (Paisley, UK).
E2, E3 and BPA were dissolved in ultrapure high-performance liquid chromatography grade ethanol (EtOH) as 1 mM stocks. EGF was prepared as a 100 μg/ml stock in reconstitution solution (0.2 μm-filtered 10 mM acetic acid containing 0.1% bovine serum albumin). PD 98059, PP2, TAM and B[ a ]P were all dissolved in dimethyl sulfoxide (DMSO) to produce stock solutions of 20, 3.3, 67.2 and 1 mM, respectively. Stock solutions and subsequent dilutions were stored at –20°C.
Routine cell culture
Human mammary carcinoma MCF-7 cells were a generous gift from M.Dufresne, University of Windsor, Ontario, Canada. Cells were routinely maintained in 75 cm 2 cell culture flasks (Greiner, Gloucestershire, UK) in alpha minimal essential medium supplemented with heat-inactivated 5% fetal bovine serum, here referred to as ‘complete medium’. The medium was replaced every other day and cells were subcultured once a week, when 70–80% of confluence was reached. When harvested, cells were disaggregated with a solution of 0.05% trypsin and 0.02% ethylenediaminetetraacetic acid to form a single cell suspension. Cells were cultured at 37°C in a humidified atmosphere of 5% CO 2 in air, over a maximum of 10 passages.
Assay for cytotoxicity
The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay was carried out as described by Mosmann ( 23 ) with minor changes. Briefly, MCF-7 cells suspended in complete medium were seeded in flat-bottomed 96-well plates (NUNC, VWR International, Leicestershire, UK) at a density of 5000 cells per well. Cells were left to attach for 24 h after which the media was replaced with 100 μl of assay medium, which consisted of Dulbecco’s modified Eagle’s medium without phenol red, supplemented with 5% charcoal-dextran-treated fetal bovine serum, containing the test compounds at different concentrations. Cells were then incubated for 24 h under normal growing conditions. After this period, the medium was removed and replaced with a combination of 100 μl assay media and 20 μl of MTT solution [5 mg/ml in phosphate-buffered saline (PBS)]. Cells were left for 4 h in this solution, allowing viable cells to reduce the yellow MTT into dark blue crystals of formazan. Crystals were dissolved in 150 μl/well MTT solubilizing solution [20% wt/vol sodium dodecyl sulfate, 40% N , N -dimethyl formamide, 2% vol/vol glacial acetic acid and 1% vol/vol hydrochloric acid (from a 2 M stock) in distilled water, pH 4.7] overnight. The optical density of solubilized formazan product was photometrically quantitated at a wavelength of 560 nm using a plate reader (Labsystems Multiscan, VWR International)
The cytokinesis-block micronucleus assay
MCF-7 cells were disaggregated, resuspended in complete medium and seeded (3 ml) at a density of 1 × 10 4 cells/ml in 30 mm petri dishes containing two 18 × 18 mm glass cover slips. After 24 h, this medium was replaced with assay medium containing the test compounds. The cells were typically exposed for 24 h, and the final concentrations of the test agents were as follows: 1 nM E2, 10 nM E3, 5 μM BPA and 100 ng/ml EGF. The volume of the dilutions added to the cells was calculated so that the final EtOH concentration did not exceed 0.1% (vol/vol). The solvent control contained 0.1% (vol/vol) EtOH. B[ a ]P was added at a concentration of 10 μM ensuring that the final concentration of DMSO did not exceed 0.5% (vol/vol). Here, the solvent control contained 0.5% DMSO.
For cotreatment with inhibitors, these were added to the cells jointly with the test agents at the concentrations: PD 98059 25 μM, PP2 25 μM, ICI 182 780 1 μM and TAM 10 μM. A mixture of 0.5% DMSO and 0.1% EtOH was used as a solvent control.
Following the 24 h treatment period, assay medium containing 3 μg/ml cytokinesis inhibitor cytochalasin B was added and left for an additional 18 h. The medium was then removed and cells were washed briefly with 1% Hanks’ balanced salt solution and fixed in 70% cold EtOH for 20 min. Cells were stained in 5% Giemsa at room temperature. Subsequently, each cover slip was carefully lifted from the petri dishes with forceps and dipped in distilled H 2 O to wash off all staining. All cover slips were air dried before mounting on glass microscopic slides using DPX mounting media. MN were assessed in binucleated (BN) cells that had completed nuclear division following exposure to the test chemicals. For each slide, a minimum of 1000 BN cells were examined, and the frequency of BN cells containing MN was determined as number of MN per 1000 BN cells. In solvent-treated controls, the number of BN cells with MN typically ranged from 96 to 101, with a SEM of <10%. In some cases, the number of MN in BN cells was recorded to obtain information about the frequency of cells containing one, two or more micronucleus. All slides were coded to ensure that the researcher scoring MN frequencies was blinded toward treatment regimens. Typically, a minimum of three independent experiments were conducted following organisation for economic development and cooperation guideline 487.
Cells were seeded and treated in the same way as for the cytokinesis-block micronucleus assay. Cells were then fixed in 100% methanol for 1 h at −20°C, washed three times for 5 min in PBS/0.1% Tween 20 and blocked with goat serum for 1 h at 37°C/5% CO 2 . Cells were then incubated with the primary antikinetochore antibody CREST (diluted 1:10 in PBS) overnight at 37°C/5% CO 2 . After washing three times in PBS/0.1% Tween 20, cells were incubated with a secondary FITC-conjugated goat polyvalent anti-human antibody (1:300 in PBS) for 1 h at 37°C. Slides were washed several times with PBS/0.1% Tween 20 and mounted in a Vectashield antifade solution containing 4′,6-diamidino-2-phenylindole.
Using the fluorescence of nuclear material stained with 4′,6-diamidino-2-phenylindole (excitation at 345 nm), 400–700 BN cells per slide were examined for the presence of kinetochores by monitoring CREST-associated FITC fluorescence (excitation 490 nm). FITC staining of MN is diagnostic for the presence of kinetochores and suggests that MN staining FITC positive contain entire chromosomes. MN were scored as CREST positive and CREST negative. Microscopy was carried out with a Leitz (Wetzlar, Germany) Dialux 20 fluorescent microscope equipped with a ×63 objective.
Micronucleus frequencies are derived from count values and, therefore, were not normally distributed, which prohibited application of parametric significance tests. For this reason, effects different from controls were estimated by the Mann–Whitney test. In order to address the multiple testing problems deriving from the comparison of more than one dose group to the same controls, the individual raw P values were adjusted by the step-up method of Hochberg ( 26 ) and used to assess statistical significance (α = 5%, two sided).
Cytotoxicity of test agents
Based on the dose–response relationships recorded in the E-Screen ( 27 ), concentrations of each chemical that produced a maximal mitogenic effect were selected to investigate the formation of MN.
The MTT assay was then employed to evaluate the cytotoxicity of the chosen agents in MCF-7 cells. No significant cytotoxic effects were observed for any of the concentrations tested.
E2- and BPA-induced micronucleus formation in MCF-7 cells
Exposure of MCF-7 cells to E2 and BPA led to increases in the frequency of MN, when compared with solvent controls ( Figure 1 ). At the concentrations chosen (1 nM E2 and 5 μM BPA), the number of cells showing one micronucleus was twice as high as in solvent controls. B[ a ]P, the well-known genotoxic carcinogen and clastogen, which was used here as a positive control for micronucleus formation, induced a 3-fold increase of cells with one micronucleus relative to solvent controls. The formation of multiple MN in single BN cells was negligible with E2 and BPA, but occurred more frequently with B[ a ]P. The micronucleus frequencies reported here for E2 are in good quantitative agreement with the observations by Yared et al. ( 13 ).
ER antagonists TAM and ICI 182 780 failed to suppress MN induction by estrogens
Fischer et al. ( 12 ) argued that micronucleus formation is linked to ER activation. Therefore, we reasoned that hormone treatment of MCF-7 cells in the presence of ER antagonists should lead to reduced micronucleus frequencies. To test this idea, we initially conducted a series of experiments with the antiestrogen TAM. As shown in Figure 2 , we were not able to confirm the observations by Fischer et al. ( 12 ) of an inhibitory effect of TAM in combination with E2. A lack of reduction in micronucleus frequencies was also observed with the steroidal estrogen E3. However, in contrast to Fischer’s findings, but in agreement with the results communicated by Crofton-Sleigh et al. ( 28 ), TAM on its own caused micronucleus frequencies significantly higher than those in solvent controls.
In a second series of experiments, we explored the effects of the pure ER antagonist ICI 182 780. On its own, the antagonist induced elevated levels of MN. In combination with E2, there was very little, if any, suppression of micronucleus formation by ICI 182 780 ( Figure 2 ). In the experiments with E3, there was even a marked increase in micronucleus frequency. To ensure that this absence of effect was not due to a lack of biological activity, both antiestrogens were tested in the E-Screen in combination with E2 and shown to decrease the effects of the hormone in a concentration-dependent manner (data not shown).
Src and Erk1/Erk2 inhibitors suppressed micronucleus induction by E2 and BPA
Activation of the ER by E2 and BPA leads to rapid phosphorylation of the kinases Erk1 and Erk2 ( 20 , 29 ). We investigated whether these phosphorylation events are linked to the induction of MN. Coincubation of E2-treated cells with the Src inhibitor PP2 led to a statistically significant decrease in micronucleus frequencies ( Figure 3 ). A similar effect was observed with the Erk1/Erk2 inhibitor PD 98059. Essentially, the same inhibition pattern was seen with BPA. Administered on their own, PP2 and PD 98059 led to slight, but statistically not significant changes in effect when compared with solvent controls.
The concentrations of inhibitors were chosen on the basis of previously published papers ( 30 , 31 ) and our own experiences of the activity of these compounds. By using western blotting analysis, we were able to ascertain that PD 98059 and PP2, at 25 μM, suppressed the phosphorylation of Erk1/Erk2 and Src, respectively (data not shown). A concern was whether the kinase inhibitors induced cytotoxicity at these chosen concentrations, both on their own or in combination with E2 or BPA. This was investigated in the MTT assay. PD 98059 (25 μM) caused a reduction in MCF-7 cell numbers to 90% relative to controls, when administered on its own. In combination with E2, cell numbers were 93% of control values. The cytotoxicity seen with the Src inhibitor PP2 was more pronounced. At 25 μM, PP2 induced a decrease in cell numbers to 68% of controls, and together with E2, 72% of control values were observed. A slight reversal of cell viability in the presence of E2 was notable, and this could be due to the ability of the hormone to suppress proapoptotic genes. BPA (5 μM) induced a drop in cell numbers to 75% relative to the control.
The induction of MN by EGF was suppressed by Src and Erk1/Erk2 inhibitors
Our data show that formation of MN by E2 and BPA was sensitive to disruption of Src/Raf/Erk signaling by selective inhibitors. This raised the possibility that activation of these pathways may lead to an increase in micronucleus frequency. To test this possibility, we administered the peptide EGF, a strong activator of the receptor tyrosine kinase EGF receptor, which in turn signals to Src and Erk1/Erk2 ( 32 ). EGF (100 ng/ml) revealed itself as capable of inducing the effect, with a frequency similar to that seen with B[ a ]P (10 μM). As expected, the formation of MN could be suppressed by coadministration of PP2 and PD 98059 ( Figure 4 ).
Micronucleus induction by B[a]P and effects of Src and Erk inhibitors
Next, we investigated whether the suppression of MN seen with PP2 and PD 98059 was limited to compounds able to bind and activate the ER or if it was a general feature related to micronucleus induction. To this end, we tested whether induction of the effect by the genotoxic carcinogen B[ a ]P would be influenced by PP2 and PD 98059. B[ a ]P itself is unable to bind or activate the ER ( 33 ). There was a small, albeit statistically not significant, effect of PP2 and PD 98059 on the MN frequency induced by B[ a ]P ( Figure 5 ), far less pronounced than the inhibition seen in the presence of E2 and BPA.
A large fraction of MN induced by estrogenic compounds contained kinetochores
Based on the data presented here and the results published by Saavedra et al. ( 21 ), we suspected that the formation of MN in cells treated with estrogens might result mainly from mitotic aberrations mediated by the Src/Raf/Erk signaling.
If this is the case, the increased activation of MAPKs by our test agents should lead to disruption of the mitotic spindle and improper segregation of chromosomes, with consequent increases in MN containing whole chromosomes, rather than MN with acentric chromosome fragments.
This hypothesis was tested by immunofluorescent staining of the centromeres using a CREST-specific antibody that distinguishes MN containing whole chromosomes or chromatids (CREST positive) from those containing only chromosomal fragments (CREST negative). As shown in Table I , the percentage of CREST-positive MN was elevated in cells treated with estrogenic chemicals and EGF, and this effect became more pronounced as the concentration of these agents increased, except for B[ a ]P and BPA. With BPA, it is possible that saturation was reached already at the lowest tested concentration of 5 μM, with no further increase in the number of CREST-positive MN at the higher concentrations (50 μM).
|Agent||Concentration||Experiment||Number of BN cells scored||Number of CREST-positive MN||Percent CREST-positive MN|
|B[ a ]P||10 μM||1||607||147||24|
|Agent||Concentration||Experiment||Number of BN cells scored||Number of CREST-positive MN||Percent CREST-positive MN|
|B[ a ]P||10 μM||1||607||147||24|
Data are from two independent experiments.
We evaluated the ability of mitogens to form MN in a mammalian breast cancer cell line. In an effort to better understand the role of ER signaling in inducing genomic instability in breast cancer cells, our particular interest was in steroidal estrogens (E2 and E3) and xenoestrogens (BPA). Our observations are in agreement with previous reports of micronucleus formation by E2, E3 and BPA in MCF-7 cells ( 13 , 14 , 34 , 35 ).
However, our data do not agree with results by Fischer et al. ( 12 ), who reported a suppression of micronucleus formation in MCF-7 cells treated with E2 and the ER antagonist hydroxy tamoxifen. Although we observed a slight suppression of MN induction with TAM in combination with E2 or E3, this effect was not statistically significant. Similarly, the pure ER antiestrogen ICI 182 780 lacked a suppressive effect with E2 or E3. Instead, ICI 182 780 even caused a slight elevation of MN when combined with E3.
The discrepancies between our findings and those reported by Fischer et al. ( 12 ) could be due to differences in our respective experimental protocols. According to the procedure developed by Fenech ( 36 ), we incubated MCF-7 cells with TAM for one cell cycle (24 h), which enabled us to analyse MN exclusively in BN cells after one cell division. In contrast, Fischer et al. ( 12 ) exposed their cells for 96 h. According to Norppa et al. ( 37 ), incubations exceeding 24 h could result in the loss of MN, and this may explain why the frequencies reported by Fischer et al. ( 12 ) are somewhat lower than ours. 4-Hydroxy-TAM, employed by Fischer et al. , is a more potent ER antagonist than TAM used in our experiments ( 38 ). However, it is unlikely that this accounts for the absence of effects on MN in our hands because both TAM and ICI 182 780 induced a significant reduction in cell proliferation when tested in the E-Screen (data not shown).
Instead, our results indicate that the ‘classical’ genomic ER pathway with translocation of the receptor to the nucleus and induction of transcription does not mediate micronucleus formation. This led us to evaluate alternative mechanisms that could be at play in the induction of this effect by estrogenic compounds.
Previously published studies have shown that steroidal estrogens ( 39 , 40 ) and environmental estrogens, such as BPA ( 20 ), are able to rapidly and transiently activate a number of growth factor signaling events, including the Src/Raf/Erk pathway. These cellular signaling pathways are essential for good cell maintenance and are involved in the regulation of cell proliferation, differentiation and death. An interesting paper by Saavedra et al. ( 21 ) demonstrated that stimulation of this pathway by induced overexpression of Ras contributes to chromosomal instability and elevated micronucleus frequencies in embryonic mouse fibroblast cells.
These findings motivated us to investigate a possible role of the Src/Raf/Erk pathway in the formation of MN by estrogenic agents. Upon coadministration of inhibitors of the Src/Raf/Erk signaling module, statistically significant suppressions of E2- and BPA-induced MN were observed. In an alternate test of the importance of this pathway for micronucleus induction, we examined the effects of EGF, a strong stimulatory agent of Src/Raf/Erk signaling. For the first time, we show that EGF leads to an increase in MN. As with estrogens, the effect of EGF on micronucleus formation was sensitive to the presence of PD 98059 and PP2. These data echo Saavedra’s findings and lend strong support to the hypothesis that the Src/Raf/Erk cascade plays a role in micronucleus formation by ER agonists and other mitogenic agents.
Since MN cannot be formed without cells entering the cell cycle, it could be argued that the observed suppression of effect in combination treatments with PP2 or PD 98059 was simply the result of a blocking of cell cycle progression by these inhibitors. However, MN were scored only in cells that had undergone cell division, with cytokinesis halted by cytochalasin B. Thus, possible perturbations of micronucleus frequencies by alterations of cell cycle progression were controlled for by examining only BN cells. Furthermore, if the effect of PP2 and PD 98059 on micronucleus formation was unspecific, the kinase inhibitors should have influenced micronucleus frequencies also with genotoxic agents, which produce MN by releasing DNA-damaging metabolites. Our results with B[ a ]P, a well-known micronucleus-inducing chemical and genotoxic agent ( 28 , 41 ), show that this is not the case. The induction of MN by B[ a ]P was unaffected by either PP2 or PD 98059. Since B[ a ]P is unable to activate the ER or to influence the fast non-genomic signaling that occurs after ER activation ( 33 ), it appears that the suppressive effects of these inhibitors are specific to agents that activate the Src/Raf/Erk pathway.
The data presented here show that the classical ER pathway, where the steroid receptor functions as a ligand-dependent transcription factor, is not involved in micronucleus formation by estrogenic agents. However, this does not mean that ER activation is without significance for these processes. Estrogens can rapidly activate a number of signaling cascades and it is thought that this results from the interaction of a membrane-associated ER with Src and other kinases ( 18 , 29 , 42 ). However, little is known about the nature of this putative membrane ER. Some authors have suggested that they are a subpopulation of the nuclear ER α and β and others propose the existence of an unrelated receptor or even the involvement of growth factor receptor tyrosine kinases ( 29 , 43 , 44 ).
Interestingly, the ways in which this membrane-associated receptor interacts with known ER ligands and inhibitors are still not clear and seem to differ from the classic ER. For example, some compounds that are known to bind to and activate the nuclear ER fail to induce activation of signaling kinases in breast cancer cells ( 45 ). Also, a number of studies indicate that both TAM and ICI 182 780, which inhibit the genomic effects of the ER in the nucleus, are unable to prevent the triggering of cell signaling cascades by estrogens ( 44 , 46 ). Conversely, there is evidence that both TAM and ICI 182 780 are able to stimulate the Src/Raf/Erk pathway ( 44 , 46 , 47 ). It appears that disruption of nuclear events by ER antagonists is totally divorced from signaling through Src and other kinases, which can proceed even when receptor-mediated promotion of gene expression is blocked by TAM and ICI 182 780. Accordingly, it would be expected that the ER antagonists by themselves should stimulate micronucleus formation. Indeed, we observed that TAM induced MN, with a 2-fold increase over the solvent control. ICI 182 780 also led to slight increase, although this did not reach statistical significance. Thus, the activation of Src/Raf/Erk in MCF-7 cells by the ER antagonists may provide an explanation for the elevated micronucleus frequencies and for their inability to reduce micronucleus formation in the presence of steroidal estrogens. However, this requires further experimental study, not least because genotoxic mechanisms could also be at play in the case of TAM.
Our observations provoke questions as to the relevance and precise role of MAPK signaling in contributing to genomic instability. There is evidence that active Erk1/Erk2 is present in kinetochores during mitosis and regulates proteins such as Aurora B kinase, which has been implicated in chromosomal alignment, cytokinesis and spindle checkpoints ( 48 , 49 ). It has been shown that overexpression of Ras can promote micronucleus formation via Erk ( 21 ). Furthermore, in a process mediated via Erk, Raf-1 activation is able to cause faster progression through mitosis by overriding the spindle checkpoint in the M phase of the cell cycle ( 50 ). At this key stage of cell division, proper attachment of kinetochores to the microtubule of the spindle apparatus is regulated. Disruption of this process will lead to disorderly separation of sister chromatids with distribution of unequal chromosome numbers to the daughter cells. Normally, cell division is halted before all chromatids are properly aligned along the equator of the spindle apparatus, but premature cell cycle progression at this stage may promote MN formation. Key proteins involved in the regulation of mitosis are the Aurora kinases ( 51 ). Aurora B monitors the correct attachment of microtubules to kinetochores and is thought to be a key player in the spindle checkpoint ( 51 ). Interestingly, enhanced activation of the Src/Raf/Erk cascade may lead to inhibition of Aurora B kinase by decreasing the localization of Aurora B to kinetochores with a consequent decrease in mitotic index and a defective spindle checkpoint ( 50 ). A growing body of evidence suggests that such defects may promote aneuploidy and induce micronucleus formation ( 52 , 53 ).
The reports by Saavedra et al. ( 21 ) and Eves et al. ( 50 ) offer a possible interpretation of our results. If the proposed disruption of the spindle checkpoint by stimulation of Erk signaling via Raf is applicable to micronucleus formation by estrogens in MCF-7 cells, missegregation of whole chromosomes should occur resulting in a large fraction of MN containing kinetochores. Indeed, this is borne out by our experiments. We observed that E2, E3, EGF and BPA induced a significantly higher proportion of MN that stained positive for kinetochores (CREST-positive MN), when compared with controls or to B[ a ]P, a known clastogenic agent. On the other hand, a certain degree of DNA fragmentation yielding CREST-negative MN also occurred.
Our results echo those by Saavedra et al. ( 21 ), who reported that increased levels of activated Erk led to a disproportionate increase in the frequency of CREST-positive MN. They also observed that a small, but significant fraction of the MN was CREST negative, indicative of loss of acentric chromosome fragments from the cell nucleus. Saavedra et al. proposed that two mechanisms may be operating during the Erk-mediated induction of MN. First, the presence of centromeres and kinetochores in MN (CREST-positive MN) can be attributed to the improper segregation of whole chromosomes during mitosis, very likely through overriding of the spindle checkpoint. Second, the generation of acentric (CREST negative) fragments is indicative of clastogenic events. Constitutive activation of MAPK can stimulate the amplification of centrosomes, with formation of multipolar spindles and mitotic bridges that arise from dicentric chromosomes. Alternatively, clastogenicity could also occur through redox cycling of semiquinones derived from steroidal estrogens, with concomitant release of DNA-reactive intermediates leading to chromosome breaks. There is evidence that steroidal estrogens may induce DNA strand breaks by such a mechanism ( 4 ), and this may have contributed to the occurrence of CREST-negative MN. However, we believe that centromere amplification and not clastogenicity through DNA-reactive intermediates is the predominant mode of action operating here because EGF, an agent essentially unreactive to DNA, also induced acentric MN.
In summary, our findings show that classical nuclear ER activation does not play a significant role in micronucleus formation. Our data support the hypothesis that micronucleus induction by estrogenic chemicals results from activation of the Src/Raf/Erk pathway. This effect seems to be linked to overriding of the spindle checkpoint and consequent numerical changes in chromosomes, rather than chromosome fragmentation. More direct evidence of Aurora B inhibition and centrosome amplification by estrogenic agents is now required to substantiate these ideas. Of particular importance will be to assess whether the effect patterns observed here with steroidal estrogens and BPA apply to all agents that are able to interact with the ER. Our results may have some relevance to improving an understanding of the role of estrogenic agents in the later stages of breast cancer.
European Commission (contract number: QLK4-CT-2002-00603).
- B[ a ]P
benzo[ a ]pyrene
epidermal growth factor
estrogen receptor α
extracellular signal-regulated kinase
mitogen-activated protein kinase
- PD 98059
The authors would like to thank Martin Scholze for his help with the statistical analysis.
Conflict of Interest Statement: None declared.