Mechanism for Sequential Responses to Hormonal Therapy in Breast Cancer
Clinical observations demonstrate that women with breast cancer often respond to subsequent endocrine manipulation after resistance to initial hormonal therapy develops. As a mechanistic explanation for these findings, we hypothesized that human breast tumors can adapt in response to the pressure exerted by endocrine therapy with development of hypersensitivity to estradiol. To understand the signaling pathways responsible, we examined estrogenic stimulation of cell proliferation in a model system and provided in vitro and in vivo evidence that long-term deprivation of estradiol (LTED) causes adaptive hypersensitivity. Even though the estrogen receptor α (ERα) is markedly up-regulated in LTED cells, the enhanced responses to estradiol do not appear to involve mechanisms acting at the level of transcription of estrogen-regulated genes. We found that ERα co-opts a classical growth factor pathway and induces rapid nongenomic effects that are enhanced in LTED cells. Estradiol binds to cell membrane-associated ERs, physically associates with the adaptor protein Shc, and induces its phosphorylation. In turn, Shc binds Grb2 and Sos, which result in the rapid activation of mitogen-activated protein kinase. These nongenomic effects of estradiol produced biological effects, as evidenced by Elk-1 activation and by morphological changes in cell membranes. The mechanistic pathways involved in adaptive hypersensitivity suggest that inhibitors of the mitogen-activated protein kinase and phosphatidylinositol-3-OH kinase pathways might prevent the development of adaptive hypersensitivity and allow more prolonged efficacy of endocrine therapies.
Women with breast cancer who initially respond to first-line endocrine therapy and then relapse may experience tumor regression on sequential use of second-, third-, and even fourth-line therapies (1, 2) . As an example, premenopausal women initially responding to surgical oophorectomy often respond subsequently to aromatase inhibitors, tamoxifen, and then progestins before ultimately developing complete resistance to hormonal therapy. We postulate that the pressure exerted by endocrine manipulations causes tumors to undergo adaptive changes that initially allow them to respond to additional hormonal therapies and ultimately results in complete hormone independence. We have conducted systematic studies to examine the adaptive mechanisms responsible for these sequential responses.
The primary strategies for treatment of hormone-dependent breast cancer include blockade of estrogen synthesis with aromatase inhibitors or of estrogen receptor (ER)-mediated action with anti-estrogens. As a model system for our studies, we have used MCF-7 breast cancer cells exposed to estradiol (E2) deprivation over a prolonged period to mimic the effects of blockade of estradiol synthesis (3, 4, 5) . These studies demonstrated that breast cancer cells exhibit inherent plasticity and can adapt to the pressures of endocrine therapy by up-regulating ER-mediated and growth factor-related signaling pathways.
Clinical observations in premenopausal women with breast cancer suggested the hypothesis that breast cancer cells can adapt to conditions of treatment by developing enhanced sensitivity to estradiol. Hormone-dependent breast cancers often regress in response to surgical removal of the ovaries, a treatment that lowers circulating plasma estradiol from ∼200 pg/ml to 10–15 pg/ml (1, 2) . In response to this acute deprivation of estradiol, tumors regress for 12–18 months, on average, before they begin to regrow. Second-line therapy with aromatase inhibitors often induces additional tumor regression by lowering estradiol concentrations further to 1–5 pg/ml (1) . These observations are best explained by the concept of enhanced sensitivity to circulating estradiol. Specifically, 200 pg/ml estradiol is required to stimulate tumor growth before oophorectomy, whereas levels of 10–15 pg/ml are sufficient to cause tumor proliferation after adaptation 12–18 months later.
Our group and other investigators have sought to directly demonstrate the phenomenon of adaptive hypersensitivity and to determine the mechanisms involved using model cell culture systems (3, 4, 5, 6) . In our studies, wild-type MCF-7 cells are cultured over a prolonged period in estrogen-free medium to mimic the effects of primary endocrine therapy. This process involves long-term estrogen deprivation (LTED), and the adapted cells are known as LTED cells. In response to estradiol deprivation, MCF-7 cells initially stop growing but then, months later, adapt and grow as rapidly as wild-type MCF-7 cells maximally stimulated with estradiol. We attribute this effect to the development of hypersensitivity to estradiol with regrowth in response to residual amounts of estrogen in the charcoal-stripped culture medium.
We have provided direct evidence of hypersensitivity by showing that a 4-log lower concentration of estradiol can stimulate proliferation of LTED cells compared with wild-type MCF-7 cells (Fig. 1<$REFLINK> ; Ref. 5 ). It is of interest that the dose-response curves are bell shaped. Increasing concentrations of estradiol initially stimulate, whereas higher concentrations begin to inhibit cell number. We have recently demonstrated that the decrease in cell number in response to higher doses of estradiol represents estradiol-stimulated apoptosis (7) . Thus, both proliferation and apoptosis become more sensitive to estradiol in the LTED cells. To confirm that these effects can occur in vivo as well, we also demonstrated that LTED cell xenografts grown in nude mice are hypersensitive to low doses of estradiol (4) .
Mechanisms for Adaptive Hypersensitivity to the Proliferative Effects of Estradiol
Hypersensitivity could involve modulation of the genomic effects of estradiol acting on transcription, nongenomic actions involving plasma membrane-related receptors, cross-talk between growth factor and steroid hormone-stimulated pathways, or interactions among these various effects. Enhanced receptor-mediated transcription of genes related to cell proliferation could be involved as the levels of ERα increased 4–10 fold during LTED (3 , 8) . Accordingly, to examine directly whether enhanced sensitivity to E2 (17β-estradiol) in LTED cells occurred at the level of ER-mediated transcription, we quantitated the effects of estradiol on transcription in LTED and in wild-type MCF-7 cells (3 , 8) . As transcriptional readouts, we measured the effect of E2 on c-myc message levels, progesterone receptor, and pS2 concentrations, and on ERE-CAT (estrogen response element-chloramphenicol acetyl transferase) reporter activity (Fig. 2)<$REFLINK> . Although basal levels of pS2 and progesterone receptor were increased, no shift to the left in estradiol dose-response curves for any of these responses occurred when comparing LTED with wild-type MCF-7 cells (Fig. 2)<$REFLINK> . These data suggest that the hypersensitivity of LTED cells to estradiol does not occur at the level of ER-mediated gene transcription.
We next considered the possibility that adaptation might involve dynamic interactions between pathways using steroid hormones and growth factors for signaling. Both estradiol and various peptide growth factors are mitogens for breast tissues (9) . These effects are believed to result from the genomic effects of estradiol to stimulate transcription of early response genes such as c-myc and growth factors such as tumor growth factor α (TGFα). Growth factors result in the activation of mitogen-activated protein kinase (MAPK), which directly and indirectly enhances the degree of phosphorylation of the ER. MAPK directly phosphorylates serine 118 and also stimulates Elk and Rsk activity, which phosphorylate serine 167 (10) .
Our initial approach used MAPK as a surrogate marker to determine whether growth factor pathways were up-regulated in LTED cells. We demonstrated MAPK up-regulation directly by measuring the level of activated MAPK in LTED cells in vitro and in LTED xenografts in nude mice (3, 4 , 11) . We further demonstrated that activated MAPK is implicated in the enhanced growth of LTED cells because inhibitors of MAPK such as PD98059 or U0126 block the incorporation of treated thymidine into DNA. These data suggest that an increase in activated MAPK participates in the adaptive hypersensitivity process. As proof of this hypothesis, we stimulated activation of MAPK in wild-type MCF-7 cells by administering TGFα (Ref. 8 ; Fig. 3<$REFLINK> ). Initial characterization data demonstrated increases in MAPK in MCF-7 cells with doses of TGF α ranging from 0.1 to 10 ng/ml and blockade of this effect with the MAPK inhibitor PD98059 (8) . Administration of TGFα at a dose of 10 ng/ml caused a 2-log shift to the left in the ability of estradiol to stimulate the growth of wild-type MCF-7 cells. To demonstrate that this effect related specifically to MAPK and not to a nonspecific effect of TGFα, we coadministered PD98059. Under these circumstances, the 2-log left shift in estradiol dose response returned to baseline (Fig. 3B)<$REFLINK> .
Although an important component, MAPK did not seem to be solely responsible for hypersensitivity to estradiol. Blockade of this enzyme did not completely abrogate hypersensitivity. Accordingly, we examined the phosphatidylinositol-3-OH kinase (PI3K) pathway to determine whether it, too, was up-regulated in LTED cells (11) . In preliminary experiments, we determined that LTED cells exhibit an enhanced activation of Akt, P70 S6 kinase, and 4EBP-1 (all components of the PI3K pathway). Dual inhibition of PI3K with LY40029 and MAPK with U0126 shifted the level of sensitivity to estradiol more dramatically: more than 2 logs to the right (11) . On the basis of these observations, our working hypothesis is that adaptive hypersensitivity involves the joint activation of the PI3K and MAPK pathways.
Additional studies examined why activated MAPK levels might be elevated in LTED cells. Up-regulation of MAPK could reflect a constitutive activation of growth factor receptors, an increase in the endogenous secretion of growth factors, or other mechanisms. We reasoned that inhibition of MAPK with a pure anti-estrogen would rule out the possibility of constitutive activation of growth factor receptors or growth factor secretion. Accordingly, we administered the pure anti-estrogen fulvestrant and examined the level of activation of MAPK in LTED cells. Surprisingly, fulvestrant returned the level of activated MAPK back to the level seen in wild-type MCF-7 cells (8) . This observation ruled out constitutive growth factor effects in LTED cells and suggested an interaction between ER-mediated functions and the elevation of MAPK.
Nongenomic Membrane ER Actions
One possible mechanism to explain the activation of MAPK would be through nongenomic effects of the ER acting at the level of the cell membrane (12) . We postulated that membrane-associated ER might use a classical growth factor pathway to transduce its effects in LTED cells. The adaptor protein Shc represents a key modulator of tyrosine kinase-activated peptide hormone receptors (13, 14) . As an upstream regulator of MAPK, Shc transduces mitogenic and differentiation signals from a variety of tyrosine kinase receptors such as those for epidermal growth factor (EGF), nerve growth factor (NGF), platelet-derived growth factor (PDGF), and insulin-like growth factor (IGF), to downstream kinase cascades (15) . On receptor activation and autophosphorylation, Shc binds to the adaptor proteins Grb2 and Sos, which results in activation of Ras, Raf, Mek, and MAPK (15) .
We postulated that estrogen deprivation might trigger activation of a nongenomic, estrogen-regulated, MAPK pathway that uses Shc. We used MAPK activation as an end point with which to demonstrate rapid nongenomic effects of estradiol. The addition of E2 stimulated MAPK phosphorylation in LTED cells within minutes. The increased MAPK phosphorylation by E2 was time- and dose-dependent, being greatly stimulated at 15 min and remaining elevated for at least 30 min. Maximal stimulation of MAPK phosphorylation was at 10−10 m E2 (15) .
We then examined the role of peptides known to be involved in growth factor signaling pathways that activate MAPK. Shc proteins are known to couple tyrosine kinase receptors to the MAPK pathway, and activation of Shc involves the phosphorylation of Shc itself (15) . To investigate whether the Shc pathway was involved in the rapid action of estradiol in LTED cells, we immunoprecipitated tyrosine phosphorylated proteins and tested for the presence of Shc under E2 treatment. E2 rapidly stimulated Shc tyrosine phosphorylation in a dose- and time-dependent fashion with a peak at 3 min. The pure ER antagonist fulvestrant blocked E2-induced Shc and MAPK phosphorylation at 3 min and 15 min, respectively. The time frame suggests that Shc is an upstream component in E2-induced MAPK activation. To provide direct evidence of the necessity of Shc for MAPK activation, a dominant-negative Shc abrogated the ability of E2 to rapidly stimulate MAP kinase (15) .
The adapter protein Shc may directly or indirectly associate with ERα in LTED cells and thereby mediate E2-induced activation of MAPK. We considered this likely in light of recent evidence of ERα membrane localization (12 , 15) . To test this hypothesis, we immunoprecipitated Shc from nonstimulated and E2-stimulated LTED cells and then probed immunoblots with anti-ERα antibodies. Our data showed that the ERα/Shc complex preexisted E2 treatment and that E2 time-dependently increased this association (15) . In parallel with Shc phosphorylation, we observed a maximally induced association between ERα and Shc at 3 min. MAPK pathway activation by Shc requires Shc association with adapter protein Grb2 and then further association with Sos. By immunoprecipitation of Grb2 and detection of both Shc and Sos, we demonstrated that the Shc-Grb2-Sos complex constitutively existed at relatively low levels in LTED cells but was greatly increased by treatment of cells with 10−10 m E2 for 3 min.
Because ERα, Shc and MAPK are all involved in E2 action in MCF-7 cells, we wished to determine upstream components responsible for Shc phosphorylation. We suspected that c-Src might phosphorylate Shc in response to E2, because Shc has been reported to be a substrate of Src tyrosine kinase in HEK-293 cells (16) . On the other hand, we wished to demonstrate a role for ERα in this process and to exclude MAPK as the cause of Shc phosphorylation. Accordingly, we examined the effects of PP2 (a specific Src kinase inhibitor), fulvestrant (also referred to as ICI) and PD98059 (a specific MAP kinase inhibitor) on E2-induced phosphorylation of Shc. In the presence of the inhibitors, MCF-7 cells were stimulated with vehicle or 10−10 m E2 for 3 min, and the status of Shc phosphorylation was examined (15) . Both PP2 and ICI effectively inhibited E2-induced Shc phosphorylation, implying that Src family kinases and ERα are required for Shc activation. As expected, PD98059 did not influence the phosphorylation status of Shc, suggesting that it functions downstream of Shc. No effects of these inhibitors were apparent in the absence of E2 stimulation. Taken together, these results indicate that both ERα and Src are upstream components of Shc functionality and their involvement is required for Shc phosphorylation. Each of these inhibitors was capable of reducing the rate of cell proliferation in LTED cells.
Biological Effects of Nongenomic Pathway Activation
To provide evidence that the ERα-Shc-MAPK pathway exerts biological effects, we evaluated the role of MAPK on the activation of Elk-1, which is a transcriptional factor that is phosphorylated and activated by MAPK. When activated, Elk-1 serves as a downstream mediator of cell proliferation. The phosphorylation of Elk1 by MAPK can up-regulate its transcriptional activity. By cotransfection of LTED cells with both GAL4-Elk and its reporter gene GAL4-luc), we were able to show that E2 dose-dependently increased Elk-1 activation at 6 h, as shown by luciferase assay (15) .
We also wished to demonstrate biological effects on cell morphology. It has been reported that cell mobility is controlled by a network of membrane-initiated signals, such as activation of the Shc-Ras-MAPK pathway (15) . Recently, Smilenov et al. (17) reported that cell focal adhesions are highly dynamic structures. Cells can rapidly respond to the stimulation by growth factors and show reorganization of their cytoskeleton and cell shape. To examine E2 effects on reorganization of the actin cytoskeleton, we visualized the distribution of F-actin by phalloidin staining and also redistribution of the ERα localization in LTED and MCF-7 cells (Fig. 4)<$REFLINK> .Untreated LTED cells expressed low actin polymerization and a few focal adhesion points. After E2 stimulation, in contrast, the cytoskeleton underwent remodeling associated with formation of cellular ruffles, lamellipodias and leading edges; alterations of cell shape; and loss of mature focal adhesion points. A subcellular redistribution of ERα to these dynamic membranes on E2 stimulation represented another important feature of LTED cells. The ER antagonist fulvestrant at 10−9 m blocked E2-induced ruffle formation as well as the redistribution of ERα to the membrane, with little effect by itself (i.e., in the absence of estradiol). Therefore, these studies further demonstrated the rapid action of E2 with respect to dynamic membrane alterations in LTED cells.
The use of confocal microscopy and immunofluorescence provided a dynamic means of assessing ERα location and alterations in response to E2 (Fig. 4)<$REFLINK> . Accordingly, we focused on the regions contiguous to the cell membrane. Under basal conditions, a faint green staining (i.e., ER immunofluorescence) could be observed along the cell membrane of the cells (Fig. 4)<$REFLINK> . In marked contrast, E2 appeared to translocate ERα into the region along the membrane ruffles, as indicated by the strong appearance of green staining. As shown by merging the red (actin) and green (ERα) views, the ERα appeared as yellow (Fig. 4F<$REFLINK> , insert b), indicating colocalization with actin in the membrane ruffles. Striking also was the translocation of the ERα into the “fist-like” region of the pseudopodia as shown by both the green staining and yellow merged views. Fulvestrant blocked E2-induced ERα membrane translocation but exerted little effect under basal conditions. To provide further proof of nongenomic ER-mediated effects, we constructed a series of designer ERs that localized specifically to the nucleus, cytosol, or cell membrane (18) . Only the membrane ER responded to exogenous estradiol with MAPK activation.
Mechanism for Localization of ERα to the Plasma Membrane
Data presented at the recent Endocrine Society National Meeting provided preliminary evidence that the IGF type 1 receptor (IGF-IR) is involved in the ability of estradiol to translocate the ERα to or near the plasma membrane (19) . We postulated that ERα binds to Shc and Shc, in turn, binds to sites on the IGF-IR. In support of this concept, we demonstrated that both Shc and ERα coimmunoprecipitate when using an anti-IGF-IR antibody in estradiol-pretreated cells. The binding of ERα to IGF-IR occurs within 3 min of the addition of 10−10 m estradiol. Knock-down of Shc with siRNA methodology abrogates the ability of ERα to bind to the IGF-IR. Confocal microscopic techniques also demonstrated colocalization of ERα with Shc and actin in the cell membrane. Knock-down of Shc with the siRNA methodology also blocked this effect.
Enhanced Nongenomic Effects in LTED Cells
From the data reviewed, we conclude that membrane-related ERα plays a role in cell proliferation and in the activation of MAPK. It seemed likely then that LTED cells might exhibit enhanced functionality of the membrane ERα system. As evidence of this, we examined the ability of estradiol to cause the phosphorylation of Shc in wild-type and MCF-7 cells and also to cause association of Shc with the membrane ER (15) . As previously reported (15) , we demonstrated a marked enhancement of both of these processes in LTED as opposed to wild-type cells. At the present time it is not clear what is responsible for the enhancement of the nongenomic ER-mediated process. Up-regulation of the amount of ERα is likely to be one factor responsible. We have shown by a variety of methods that there is a 4–10-fold elevation of ERα in LTED cells. Whether other processes are involved in addition is not currently clear.
Adaptation Rather Than Cell Selection
The development of hypersensitivity could reflect the selection of clones of cells with properties of hypersensitivity or epigenetic changes in the overall population of cells as a result of the process of adaptation. To address this issue, we have attempted to revert hypersensitive cells back to their wild-type phenotype by reexposure to exogenous estradiol. When LTED cells are grown in the presence of estradiol either in vivo or in vitro, the cells revert back to the wild-type phenotype (5 , 11) . These observations suggest that MCF-7 cells can undergo epigenetic or adaptive changes in response to estrogen deprivation or reexposure and that clonal selection is unlikely.
Summary of Mechanisms for Adaptive Hypersensitivity to the Proliferative Effects of Estradiol
Long-term estradiol deprivation causes an up-regulation of the amount of ERα present and of processes involved in utilization of membrane-related ER. This results in an increased level of activation of the MAPK as well as the PI3K pathways. All of these signals converge on downstream pathways directly involved in cell cycle functionality and probably exert synergistic effects at that level. As a reflection of this, E2F1, an integrator of cell cycle-stimulatory and -inhibitory events, is hypersensitive to the effects of estradiol in LTED cells (11) . Our working hypothesis at present is that hypersensitivity reflects a downstream synergistic interaction of several pathways converging at the level of the cell cycle. An increase in the basal level of transcription of ER-regulated genes may also be involved in the process but does not represent the proximate cause of hypersensitivity, because transcriptional events respond to estradiol with similar dose-response curves in wild-type and LTED cells.
Adaptive Hypersensitivity to the Proapoptotic Effects of Estradiol
Our in vitro model of long-term estradiol deprivation demonstrates an inhibitory effect of higher but still physiological concentrations of estradiol on LTED cells (Fig. 1)<$REFLINK> . We postulated that these cells might also have become sensitized to the proapoptotic effects of estradiol. To test this hypothesis, we used an ELISA assay for the assessment of apoptosis and to compare the effects of estradiol in wild-type and LTED cells. The expected inhibition of apoptosis occurred in the wild-type cells but, in marked contrast, a dramatic enhancement of this parameter occurred in LTED cells. We confirmed the presence of apoptosis in LTED cells by the measurement of annexin V and the use of time-lapse microscopy (7) .
As a mechanistic explanation for the differential effects of estradiol in LTED and wild-type cells, we demonstrated that the death receptor Fas was up-regulated in the LTED cells. Because Fas ligand has an estrogen response element in its promoter region, estradiol stimulated the level of Fas ligand in both LTED and wild-type MCF-7 cells. Only the LTED cells contained a Fas receptor and could respond with apoptosis. As additional evidence regarding the role of the Fas receptor, an activating antibody, directed against Fas, stimulated apoptosis in LTED cells (7) .
Clinical Relevance of the Process of Adaptive Hypersensitivity
Primary endocrine therapies exert pressure on breast cancer cells to undergo adaptation. The objective responses to aromatase inhibitors after oophorectomy in women with breast cancer reflect enhanced sensitivity of breast cancer cells to the proliferative effects of estradiol (20) . Specifically, in response to oophorectomy, tumors cells become hypersensitive and develop the ability to regrow in the presence of postmenopausal levels of estradiol (i.e. 10–15 pg/ml). Aromatase inhibitors lower estradiol further to very low levels, below those supporting growth even of hypersensitive cells, and secondary tumor regression occurs. Although not proven definitively, responses to aromatase inhibitors after tamoxifen might also reflect estradiol hypersensitivity. Third-generation inhibitors are now available that are 100-1000-fold more potent than the first-generation inhibitor aminoglutethimide and block aromatase by 97–99% (2 , 20) . If the adaptive hypersensitivity hypothesis were correct, third-generation aromatase inhibitors would be more effective than first-generation agents for treatment of patients with breast cancer. One might also expect superiority to tamoxifen, because this agent exerts partial agonist activity that is enhanced by the adaptive hypersensitivity process. Both expectations have been substantiated in clinical trials. The use of a pure anti-estrogen would also serve to abrogate the effects of hypersensitivity to estradiol. Fulvestrant and other agents being developed would be expected to be active in women relapsing after oophorectomy or tamoxifen. Published studies demonstrate the efficacy of fulvestrant in patients relapsing after initial responses to tamoxifen (20) .
Endocrine therapies are well tolerated but responses last only 12–18 months on average. Prevention of the adaptive process could potentially result in enhanced duration of endocrine therapy. A blockade of MAPK and PI3K during the use of aromatase inhibitors might then prevent the adaptive process. A key question is whether the blockade of upstream or of downstream events would be more efficacious in blocking the adaptive processes. One strategic approach would be to block upstream growth factors in pathways involving HER1, -2, -3, or -4; the IGF-I and II receptors; or the fibroblast growth factor family for each of these receptors. An alternative approach is to block the adaptive process at a downstream step involving MAPK and PI3K. We and others have presented data using a compound that inactivates Ras as a potential means of blocking both the MAPK and the PI3K pathways (21, 22) . This agent, farnesylthiosalicylic acid (FTS), inhibits the binding of GTP-Ras to galectin-1, a membrane anchor protein that allows GTP-Ras to become lodged in the plasma membrane (21) . When GTP-Ras is dislodged by FTS, it reenters the cytoplasm, is degraded, and becomes inactive (21) . Because both MAPK and PI3K are downstream of Ras, this agent blocks both steps. Preliminary data indicate that FTS is a potent inhibitor of growth of both LTED cells and an analogous hypersensitive prostate cancer cell line (11 , 22) . We envision the use of agents such as FTS to block the process of adaptive hypersensitivity. FTS has been found to have relatively little toxicity in rodents carrying a variety of tumors (21) . We believe that this agent may be ideal for providing the proof of principle that the blockade of downstream events may ultimately be more effective than the inhibition of individual upstream receptor-mediated steps.
Potential Use of the Proapoptotic Effects of Estradiol in Hypersensitive Cells
Women with breast cancer receive third-generation aromatase inhibitors over a period of 1 to 5 or more years. On the basis of our model system, we would expect the breast cancer cells in these patients to become sensitized to the proapoptotic effects of estradiol (7) . Under these circumstances, estradiol might stimulate apoptosis and cause tumor regression. From the 1940s until the early 1980s, high-dose estrogen in the form of diethylstilbestrol (DES) was the treatment of choice for postmenopausal women with breast cancer (1 , 20) . Clinical studies demonstrated that pre- and perimenopausal women rarely respond to this therapy, whereas responses increased as a function of the number of years after the onset of menopause. We postulate that the long period of time after menopause, in fact, represents long-term estradiol deprivation. One might then explain the responses to high-dose estrogen by its ability to induce apoptosis. If correct, one would expect that women receiving aromatase inhibitors long-term would also respond to high-dose estrogen with apoptosis (23, 24, 25, 26, 27) . This possibility is subject to experimental examination in a clinical trial, and we are planning to conduct such a study.
Vision of the Future
Our concepts regarding adaptive hypersensitivity could be applied in the future to develop an innovative approach to the treatment of hormone-dependent breast cancer. This would involve cyclic therapy with the intermittent use of estrogen-deprivation therapy with aromatase inhibitors, followed by high-dose estrogen (23) . The rationale for this approach is that third-generation aromatase inhibitors would cause cells to up-regulate pathways involving MAPK and PI3K. Present experimental data suggest that such cells have a more aggressive phenotype than do untreated cells. One would then specifically destroy the subset of cells that had adapted and developed hypersensitivity by inducing apoptosis with high-dose estrogen. An ideal regimen might then use an aromatase inhibitor in combination with MAPK and PI3K inhibitors initially. At appropriate times, one could then administer a pulse of high-dose estrogen to induce apoptosis and kill adapted cells. Because various high-dose estradiol administration regimens have been used over a 30-year period, the relative toxicities of these agents are well known, and the means of overcoming specific problems have developed (27) . Agents are currently available to block growth factor pathways and to administer estrogen.
Dr. James Ingle: Would you say that tumor progression on third-generation aromatase inhibitors is analogous to your LTED cells?
Dr. Santen: We would think so. Third-generation aromatase inhibitors block the levels of estrogen in patients as effectively as is possible. In the in vitro system we have removed estrogen as effectively as is possible. Because the cell culture system is only a model, it might not directly parallel the clinical situation. However, the model was developed to see what happens with profound estrogen deprivation and should parallel events in women.
Dr. Ingle: The preclinical data would suggest fulvestrant is incredibly effective in LTED cells, but the clinical experience with fulvestrant does not quite fulfill that expectation.
Dr. Steven Come:I hope you can clear up the difference also in the ERα expression in the clinic versus in these models. At the St. Gallen conference of the International Breast Cancer Study Group, Mitch Dowsett made a point that they don’t see this massive up-regulation of the ER in the clinic even though the models of early tumor have this.
Dr. Santen:Maybe Dr. Ellis can answer. This would be with the neoadjuvant setting and I don’t know those data.
Dr. Matthew Ellis:The bottom line is that’s right with the 4-month exposure in our neoadjuvant studies. Both with tamoxifen and with aromatase inhibitor, we see down-regulation of ER. Curiously, it is much more profound with tamoxifen than the aromatase inhibitor. It could be that 4 months is not long enough to see this up-regulation with estrogen deprivation in clinical samples. What we should really be looking at is the ER expression in the patients who suffer relapses. We don’t have those paired samples; the paired samples that Mitch Dowsett is talking about are tamoxifen-treated, not estrogen-deprived tumors.
Dr. Robert Nicholson:I am interested in your concept that we should try and inhibit the downstream signal molecules recruited by growth factor receptors rather than the receptors themselves, because there is quite a bit of redundancy with respect to the growth factors that cells can use. However, when you administer growth factors to cells, multiple signal pathways are recruited, so how do you know that inhibiting any one of them might equally allow the other ones to become dominant and promote growth?
Dr. Santen:If we consider a single pathway, the MAP kinase pathway, we know that upwards of 20 ligands and receptors can activate that pathway. Under those circumstances, the downstream blocker might be more effective than the upstream blocker. FTS blocks both the PI3 kinase and the MAP kinase pathways and should block events using these pathways. However, we could certainly have a number of other signaling pathways that are involved in cell proliferation, and any one of those could ultimately take over. Based upon this, your point is a good one. At the present time, we can only focus on the pathways that have been shown to be up-regulated. Our working strategy will be to block these pathways downstream rather than upstream.
Dr. Jeffrey Green:Is there any evidence that there are mutations occurring in the ER during this process of generating the sensitive cells? Do you go through crisis and see a change in PTEN expression?
Dr. Santen:We have not systematically looked for mutations in the estrogen receptor. We have looked at the estrogen receptor with about five different monoclonal antibodies and not found major structural changes. We’ve looked at pS2, c-myc, and four different reporters, and we just don’t find that the transcriptional effects of estradiol exhibit altered dose-response curves. The TGF-α and the stable transfection MAP kinase up-regulation experiments would suggest that it doesn’t take a mutation of the estrogen receptor to cause this major shift. So we have really focused on factors that don’t directly involve the structure or the coactivators of the estrogen receptor.
Dr. Daniel Medina:You used the MCF-7 model. Have you, or has anyone else, looked at the LTED phenomena with another breast cell line that is highly metastatic?
Dr. Santen:We have not extensively examined a number of different cell types. Over the last 10 or 15 years, a number of models have been demonstrated to develop hypersensitivity to estradiol. Whether or not these cells develop a metastatic potential once they turn on the growth factor pathways is unclear. We suspect that they would, but no one has looked at that to my knowledge. The majority of the data regarding hypersensitivity have been in T47D, ZR-75-1, wild-type MCF-7, and cloned MCF-7 cells, but not in more aggressive tumor types.
Presented at the Third International Conference on Recent Advances and Future Directions in Endocrine Manipulation of Breast Cancer, July 21–22, 2003, Cambridge, MA.
Requests for reprints: Richard J. Santen, University of Virginia Health System, Department of Internal Medicine, Division of Endocrinology, P. O. Box 801416, Charlottesville, VA 22908; Phone: (434) 924-2961; Fax: (434) 924-1284; E-mail: