Antitumor Action of Physiological Estradiol on Tamoxifen-stimulated Breast Tumors Grown in Athymic Mice¹

Adjuvant TAM⁴ has revolutionized breast cancer therapy, but the duration of treatment remains controversial. The recent overview analysis in 1998 (1) demonstrated that increasing adjuvant TAM therapy from 1 to 5 years is more effective in increasing survival, but a recent NSABP trial has shown that 5 years of TAM as adjuvant therapy for node-negative breast cancer carried a better disease-free and distant disease-free survival than >5 years of treatment (2). Large clinical trials will be needed to resolve this question, but it appears that some breast tumors acquire resistance to TAM after 5 years of therapy. Drug resistance to TAM can be manifest in many ways, one of which is TAM-stimulated growth. There is some clinical evidence that TAM-stimulated growth can occur during the treatment of advanced breast cancer (3, 4). In the laboratory, long-term TAM treatment results in TAM-stimulated growth in MCF-7 breast tumors (5, 6). After 1 year of continuous TAM treatment, tumors grow in response to both TAM and E₂ (MCF-7TAM, a tamoxifen-stimulated MCF-7 tumor;Ref. 5). However, no laboratory model is available that replicates the clinical use of 5 years of adjuvant TAM. We have addressed the issue and developed two transplantable MCF-7 tumor lines that have been exposed to TAM for >5 years by serially transplanting TAM-stimulated tumors into athymic mice treated with TAM. The two lines MCF-7TAM (7) and MT2 (an MCF-7 tamoxifen-stimulated tumor with an Asp351Tyr mutant estrogen receptor; Refs. 8 and 9) both grow in response to TAM, but we have discovered that E₂ actually produces a tumoricidal action after 4–5 years of TAM that is much more effective at reducing tumor growth than stopping TAM treatment (7). However, a proportion of the tumors regrow in response to E₂, and retransplantation has allowed us to examine the effectiveness of TAM if it is again used as an antitumor agent. Additionally, we have noted previously that the MT2 tumor line had a single point mutation in the ER, Asp351Tyr (9) that enhances the estrogen-like properties of TAM (10). We,therefore, asked the question of whether the mutant ER was essential for the TAM-stimulated response of the tumors by an examination of Asp351Tyr in TAM-stimulated tumors and then when they had been converted to E₂-stimulated tumors.

This unique observation of a tumoricidal effect of E₂ on two independent TAM-stimulated MCF-7 tumor lines raises the possibility that the clinical effectiveness of long-term TAM may not result from indefinite treatment but from the cessation of therapy at the appropriate time. A woman’s own estrogen may then produce a tumoricidal action on the supersensitized cells after TAM is cleared from the body. Additionally, these data in the laboratory also suggest that estrogen treatment of recurrent breast cancer after the failure of long-term TAM (on the presumed development of ER-positive, TAM-stimulated tumors) may result in a tumor regression that returns the breast cancer cells to estrogen responsiveness.

The MCF-7 tumors used in these experiments were derived by inoculation of 1 × 10⁷ MCF-7 cells(originally obtained from Dr. Dean Edwards, University of Texas, San Antonio, TX) into estrogenized athymic mice as described previously (8, 11). MCF-7, MCF-7TAM, and MT2 breast tumors were maintained as serially passaged solid tumors in ovariectomized BALB/c athymic nude mice (Harlan Sprague Dawley, Madison, WI), 4–5 weeks of age, implanted with either E₂ or TAM capsules (8, 12). Tumors were routinely passaged by removing a>1.0-cm-diameter tumor from an animal and mincing the remaining viable tumor into ∼1-mm³ pieces in a bath of cold RPMI 1640. Tumor pieces were then implanted with a 13-gauge trocar into the axillary mammary fat pads of the mice. At the time of tumor transplantation, all animals were also implanted s.c. with a silastic capsule containing either TAM or E₂, depending on the specific experiment.

For tumor harvests, animals were killed by CO₂inhalation and cervical dislocation, and tumors were removed and snap-frozen in liquid nitrogen. Frozen tumor specimens were stored at−80°C until use.

TAM was administered either p.o. or s.c. (via a 2-cm silastic capsule filled with TAM), depending on the type of experiment. TAM was administered p.o. at doses ranging from 500 to 1500 μg as a suspension of 10% PEG 400/Tween 80 (99.5% PEG/0.5% Tween 80)and 0.9% carboxymethyl cellulose (0.05 ml). The capsule produced blood levels of TAM of 30–40 ng/ml (12). E₂ was administered s.c. via either a 0.3- or 1-cm silastic capsule filled with E₂, implanted s.c. into the animal’s back. The capsules produced serum levels of 83.8 and 379 pg/ml, respectively (13), and were used to mimic post- or premenopausal levels of E₂. In one experiment, an E₂ pellet (1.5 mg) from Innovative Research of America (Toledo, OH) was used to replicate pharmacological doses of E₂. Serum levels of 1044 ng/ml were noted (13). The E₂ and TAM capsules were replaced every 6–8 weeks (13).

Tumor measurements were performed weekly using Vernier calipers. Tumor cross-sectional area was calculated using the formula: length/2 ×width/2 × π. Mean tumor area was plotted against time in weeks to monitor tumor growth.

Total RNA was prepared from tumors using the TRIzol reagent (Life Technologies, Inc.). SSCP was performed using the methods originally described by Orita et al. (14, 15) but with minor modifications (16). Tumor total RNA (1 μg) was reverse transcribed in a 20-μl reaction containing 50 mm Tris-HCl (pH 8.3), 75 mmKCl, 3 mm MgCl₂, 0.5 mm each dATP, dCTP, dGTP, and dTTP, 10 mm DTT, 3 μmoligo(dT)12–14, 100 units of placental RNase inhibitor, and 200 units of Moloney murine leukemia virus reverse transcriptase. The DNA fragment containing the 351 codon was generated by PCR from 5 μl of reverse transcription mixture using two primers,5′-GAGACATGAGAGCTGCCAAC-3′ and 5′-GGGTGCTGGACAGAAATGTG-3′. Control amplification was carried out on 50 pg of double-stranded DNA, coding for either a wild-type ER (HEGO) or a mutant ER (HETO), which contains a single G-to-T point mutation at nucleotide 1559. The PCR cycles and conditions have been described previously (9).

An aliquot of ³²P-labeled DNA was digested with XbaI restriction endonuclease. After restriction digestion,an aliquot of cut and uncut DNA was diluted 1:4 with 0.1% SDS, 10 mm EDTA. Half of each diluted sample was mixed 1:1 with nondenaturing loading buffer (50% glycerol, 0.05% bromphenol blue, and 0.05% xylene cyanol), and the remaining half was mixed 1:1 with denaturing loading buffer (95% formamide, 20 mm EDTA, 0.05% bromphenol blue, and 0.05%xylene cyanol). The samples were electrophoresed on a 6% neutral polyacrylamide gel with 14–16 W constant power for 4 h.

Differences in mean tumor area between groups were measured using ANOVA, followed by unpaired Student’s t test.

A MCF-7TAM tumor, which had been passaged in TAM-treated animals for 5 years and then frozen in PEG solution before storage in liquid nitrogen, was thawed and implanted into 4–5-week-old athymic mice and treated with TAM (2-cm capsule) until the mean tumor size reached∼0.4-cm². At this point, the TAM capsule was removed, and the animals were randomized into two groups of drug treatment. One group was implanted with an E₂ (1 cm) capsule that delivers E₂ levels of 379 pg/ml serum (13) and the other half a TAM (2 cm) capsule (30–40 ng/ml). The serum levels of E₂ and TAM were within the ranges documented in premenopausal women (17)and those taking TAM as adjuvant therapy (17, 18). After 2 weeks, the tumors began to regress in the E₂-treated mice, and after 5 weeks, the tumors were ∼0.1 cm² in size (Fig. 1 A). Interestingly, after 6 weeks of E₂ treatment, 8 of 18 tumors started to grow in the E₂ group. We designated this line MCF-7TAME (a newly estrogen-responsive MCF-7TAM tumor).

To ensure that E₂ treatment of MCF-7TAM tumors was not equivalent to stopping TAM alone, we repeated the first experiment but included a no-treatment group (Fig. 1,B). Six weeks after stopping TAM treatment or starting treatment with E₂, the treatment groups readily displayed a divergence of growth characteristics. The no-treatment group of tumors was significantly larger than the E₂-treated tumors at week 11 of the experiment (week 6 of treatment; Fig. 1 B).

The observation that E₂ caused tumor regression rather than tumor growth in the MCF-7TAM tumor raised the question that the animals rather than the tumor had altered over the past 10 years (5). To exclude this possibility, we used a bitransplantation technique we used previously to demonstrate the target site specificity of TAM in breast and endometrial tumors (19). Two groups of 10 mice were bitransplanted with MCF-7TAM and MCF-7TAM2 (a newly TAM-stimulated MCF-7 tumor) and treated with either E₂ (1-cm capsule; premenopausal levels) or TAM (1.5 mg/day). Both MCF-7TAM and MCF-7TAM2 tumors grew in response to TAM, but only MCF-7TAM2 grew in response to E₂ (Fig. 2). Thus,early TAM-stimulated tumors can use either E₂ or TAM to grow, but long-term TAM-stimulated tumors can only use TAM. However, do MCF-7TAME tumors that regrow during E₂ treatment respond to TAM treatment?

To determine the growth characteristics of these newly E₂-responsive tumors (MCF-7TAME), experiments were performed to study the effect of E₂ and TAM. In the first experiment, athymic mice were implanted with MCF-7TAME tumors and divided into the following four groups:E₂ capsule (1 cm; premenopausal levels);E₂ capsule plus TAM, 0.5 mg/day p.o.;E₂ capsule plus TAM, 1.5 mg/day p.o.; and vehicle treated (Fig. 3,A). The TAM doses were based on studies published previously (6, 13). Tumor growth was observed in the E₂-treated group, but both doses of TAM (0.5 and 1.5 mg/day) blunted E₂-stimulated growth. The higher dose of TAM was more effective in blunting growth. In the second experiment, the mice were divided into the same four groups as above, but an E₂ (0.3 cm; premenopausal levels) capsule was used (Fig. 3 B). Our goal was to establish that a lower dose of E₂ could be blocked more effectively by TAM. Tumor growth in the E₂ group was similar to that seen with the E₂ (1-cm) capsule group, and tumor size reached a mean tumor size of 1.3 cm² at 9 weeks. Similar growth rates as the first experiment were seen in the E₂ plus TAM 0.5 mg/day and E₂ plus TAM 1.5 mg/day groups, and once again the higher dose of TAM (1.5 mg/day) was more effective in suppressing E₂-stimulated growth. It appears that the MCF-7TAME tumors have evolved from exclusive TAM dependence to E₂ responsiveness with TAM now acting as an antiestrogen again.

The MT2 tumor line was developed in 1993 and had been serially passaged into TAM-treated animals since that time. We noted previously that the tumors are both E₂ and TAM responsive for growth (8). However, after 4 years of serial passage in TAM-treated animals, the growth response to E₂changed during the fifth year. Three consecutive experiments illustrate the development of supersensitivity of the MT2 tumors to the tumoricidal effect of E₂. In the first experiment, TAM and E₂ both stimulated tumor growth, and no significant difference in tumor size was seen between the E₂- and TAM-treated groups(P = 0.97) after 13 weeks (Fig. 4,A). TAM-treated MT2 tumors were retransplanted into new athymic mice and once again treated with TAM, E₂, or vehicle for 11 weeks (Fig. 4,B). Both E₂ and TAM stimulated growth; however, there was a difference in mean tumor size at 11 weeks between the TAM- and E₂-treated groups (1.02 versus 0.58 cm², respectively; P = 0.08). However, this did not reach statistical significance. In the third experiment, TAM-treated tumors were retransplanted into new athymic mice and treated with the same doses of TAM and E₂ for 10 weeks (Fig. 4 C). Tumor growth was observed in the TAM-treated group, and at 10 weeks the mean tumor size was 0.35 cm², whereas no tumor growth was observed in the E₂-treated group, even at 10 weeks (0.08 cm²; P =0.0005). The difference between these two groups was statistically significant. It was clear that E₂ had gradually lost its ability to stimulate tumor growth, although TAM still maintained its stimulatory effect. By the end of the third experiment,E₂ suppressed tumor growth, a phenomenon observed previously only in the MCF-7TAM tumors.

We wanted to establish the level of E₂ needed to inhibit MT2 tumors. All previous studies with MT2 tumors had used 2-cm TAM sustained-release capsules that delivered serum levels of ∼40 ng/ml to maintain tumor growth (12). MT2 tumors were implanted into athymic mice and divided into the following groups: TAM(2-cm capsule); E₂ (0.3- and 1-cm capsule; post-and premenopausal levels, respectively); or an E2 pellet(pharmacological estrogen), and treated for 11 weeks (Fig. 5). We reported previously the E₂ levels produced by these different E₂ regimens: 83, 379, and 1044 pg/ml,respectively (13). As expected, tumors grew with the TAM capsule, reaching a mean tumor size of 0.6 cm² at 11 weeks. However, all E₂-treated groups exhibited little or no tumor growth. However, like the MCF-7TAM tumors,some MT2 tumors regrew in the E₂ 1-cm capsule(premenopausal levels) group after 13 weeks of E₂treatment. The fact that only tumors from the E₂1-cm capsule group grew indicated that the phenomenon of selection was random rather then related to the concentration of E₂.

Two additional experiments were conducted to evaluate the hormonal sensitivity of this newly MT2E (newly estrogen-responsive MT2) tumor using different doses of E₂ and one dose of TAM(0.5 mg/day). Our aim was to compare and contrast the actions of the compounds with the previously established MCF-7TAME tumors (Fig. 2). In the first experiment, athymic mice were implanted with MT2E tumors and treated with an E₂ 1-cm capsule (premenopausal levels), an E₂ 1-cm capsule plus with TAM 0.5 mg/day, and TAM 0.5 mg/day for 6 weeks (Fig. 6,A). Tumor growth was observed in the E₂-treated alone group, and TAM 0.5 mg/day blocked E₂-stimulated growth at 6 weeks, although this effect did not reach statistical significance (0.7 versus 0.45 cm²; P =0.11). TAM alone stimulated some tumor growth (0.2 cm²) at 6 weeks, demonstrating that TAM is a weak agonist in this system. To evaluate the effect of a lower dose of E₂, athymic mice were implanted with MT2E tumors and divided into the same treatment groups as the previous experiment,but a 0.3-cm E₂ capsule (postmenopausal levels)was used (Fig. 6 B). Tumor growth was observed in the E₂-treated group, and TAM 0.5 mg/day significantly suppressed this E₂-stimulated tumor growth but more effectively than seen with the E₂1-cm capsule (0.19 versus 0.4 cm²,respectively; P = 0.002). TAM blocks the stimulatory effect of E₂; however, higher doses of E₂ can reverse the antiestrogenic effect of TAM. The MT2 tumors had evolved from pure TAM dependence to E₂ responsiveness, as the MCF-7TAME tumors did. However, the MT2 tumors provided an opportunity to observe clonogenic changes in response to E₂.

MT2 tumors are known to contain a mutant ER (Asp351Tyr) in contrast to MCF-7TAM tumors, which contain wild-type ER (9). This mutant ER has been shown to enhance the estrogenic properties of TAM (9, 10). To determine what role this mutation played in E₂ responsiveness or whether it disappeared in E₂-responsive tumors, SSCP was carried out on the MCF-7TAM and MT2 tumors. Because MT2E tumors had regained sensitivity to TAM as an antiestrogen, our hypothesis was that a clone of cells grew into tumors that had lost their mutation and thus lost the TAM-stimulated phenotype. SSCP was carried out on the newly E₂-responsive MT2 tumors and compared with tumors that were still receiving TAM treatment as well as on MCF-7TAM and MCF-7TAME tumors. Tumor samples from MT2 tumors in the E₂ and TAM group produced bands that comigrated with those produced by amplification of the Asp351Tyr mutant ER cDNA(HETO; Fig. 7), demonstrating that the MT2 tumors had retained their mutated ER during clonal regrowth. On the other hand, tumor samples from MCF-7TAM tumors in the E₂ and TAM groups produced bands that comigrated exactly with those produced by amplification of the wild-type ER cDNA(HEGO). We conclude that the MT2 tumors retained their ER mutation,despite a phenotypic change in their responsiveness to TAM.

We have presented data to show that breast tumors in the laboratory progress through different stages of hormonal dependency over 5 years. During the first few years of the acquisition of a TAM-stimulated phenotype, the MCF-7 ER-positive tumors respond to E₂ and TAM equally (5). However, we discovered that serially transplanted tumors appear to acquire a paradoxical supersensitivity to physiological E₂administration after stopping TAM (7). E₂ causes a dramatic regression of tumors that is more effective than stopping TAM alone. However, tumor growth can be reactivated by E₂ in a proportion of tumors, and TAM again acts as an antitumor agent to prevent E₂-stimulated tumor growth. We have found that a second TAM-stimulated transplantable tumor, the MT2, which contains a mutant ER (9), is also susceptible to the tumoricidal effect of E₂ after 4 years of continuous TAM treatment. Similarly, some MT2 tumors regrew after E₂-induced regression, but TAM again blocked this E₂-stimulated growth. We therefore conclude that the hormone-responsive breast cancer cell line MCF-7 can progress cyclically during 5 years of TAM treatment. We propose the following stages of hormone sensitivity: (a) TAM acts as an antiestrogen by blocking tumor growth; (b) TAM stimulated tumors, which can grow with either TAM or E₂;(c) TAM exclusively stimulates tumor growth, but E₂ causes a dramatic regression of tumor size;and (d) E₂ stimulates the regrowth of some tumors, but again TAM blocks E₂-stimulated growth. We have summarized this working model in Fig. 8.

Interestingly, before the introduction of TAM, estrogens were considered the hormonal treatment of choice for advanced breast cancer in postmenopausal women (20, 21, 22, 23, 24, 25). Estrogens, such as DES and ethinyl estradiol, were found to induce tumor regression (20, 21, 22, 23, 24, 25), and this effect was dose dependent, with higher doses producing higher regression rates (23). Doses from 1.5 to 1500 mg/day of DES were used; therefore, the antitumor action of estrogen was pharmacological not physiological. However, with the introduction of TAM, several clinical studies compared TAM to DES for the treatment of advanced breast cancer in postmenopausal women (20, 23, 24), and all found the two agents equal in efficacy. However, patients suffered more side effects with DES, and thus it was felt that TAM was superior to DES for the treatment of advanced breast cancer.

Most laboratory research has focused on the mechanisms of estrogen-stimulated breast tumor growth, and surprisingly, there is little information about the mechanism of estrogen-induced tumoricidal actions. One exception is the T61 breast tumor model established by Brunner et al. (26, 27, 28, 29). Unlike the MCF-7 tumor cell line, which was derived from a pleural effusion (30),the T61 tumor was derived from a primary breast cancer. T61 tumor growth in athymic mice is ovarian independent, and both TAM and E₂ inhibit its growth (29). In these studies, tumor inhibition is dose dependent and varies with the specific doses of estrogen ranging from pharmacological to physiological (29). In our study, only physiological doses were needed to induce tumor regression or prevent tumor growth. We suggest that the repeated transplantation of our tumors in TAM-treated animals has resulted in the selection of an MCF-7 tumor that is now supersensitive to the cytotoxic effects of estrogen. It is, however,interesting to note that whatever the mechanisms, our MCF-7TAM tumors regress after ∼2 weeks of E₂ treatment, a time course noted by Brunner et al. (26) with the T61 tumors. However, MCF-7TAM and MT2 tumors progress through a cycle of hormonal sensitivity, and the studies reported here have not been done previously with the T61 tumor. In addition, there was no evidence that TAM could stimulate the growth of the T61 tumors.

It is important to point out that the mutant ER found in the MT2 tumor is retained in the E₂-selected tumor subline MT2E(Fig. 7). Mutated ERs have been associated with increased estrogenicity of TAM (9, 10). The MT2 tumors retain the Asp351Tyr mutation, despite their phenotypic change from TAM-stimulated growth(MT2) to E₂-stimulated growth and TAM-inhibited growth (MT2E). We must, therefore, conclude that other cellular factors are essential for the development of the MT2 TAM-stimulated phenotype. The fact that wild-type ER predominates in the MCF-7TAM-stimulated tumor suggests that a mutant receptor is only one of the many potential supporting mechanisms for drug resistance to TAM.

An examination of the interaction of TAM and E₂illustrates an important point about the therapeutic effectiveness of TAM as a competitive inhibitor of E₂ action. We have reported the competition between E₂ and TAM in vivo previously (12), and it is important clinically when considering TAM therapy in premenopausal women (17, 31). High doses of E₂ can potentially render the antitumor activity of TAM less effective. In the experiments evaluating TAM action in the second generation of MCF-7TAME tumors, higher doses of TAM (1.5 versus 0.5 mg/day) were more effective in suppressing E₂-stimulated growth (Fig. 4, A and B). In the experiments evaluating TAM action in MT2E tumors (Fig. 6, A and B), in which the TAM dose was held constant but the E₂ dose changed, a 1-cm E₂capsule (premenopausal levels) was more effective in reversing the antitumor effect of TAM than a 0.3-cm E₂ capsule(postmenopausal levels). This is powerful evidence that the effectiveness of the antiestrogenic effect of TAM is dependent on the relative concentration of E₂ and TAM. A 1-cm E₂ capsule produces E₂levels (379 pg/ml) equivalent to that of a premenopausal woman(150–350 pg/ml; Ref. 17), whereas a 0.3-cm E₂ capsule produces levels (83.8 pg/ml)equivalent to that of a postmenopausal woman (13). Perhaps TAM would be more effective in a premenopausal woman if estrogen levels were lowered. Some evidence to support this position has been obtained recently by comparing the action of TAM with TAM plus a luteinizing hormone-releasing hormone agonist to produce a medical oophorectomy. The theoretically combined therapy is superior (32), and a preliminary clinical report supports this view (33).

We believe it is appropriate to advance a cyclic model of hormone dependency in breast cancer that is based on our experiences with transplantable MCF-7 breast tumors over the past decade (Fig. 8). The transition from estrogen- to TAM-stimulated tumor growth occurs in some tumors within a year of TAM treatment, but the important observation we now report is the reproducible change from TAM-stimulated growth to E₂-inhibited growth. Some breast cancer cells then revert back to estrogen-stimulated growth, and TAM, once again,blocks estrogen-stimulated tumor growth.

Stoll (34) has described previously the regression of tumors during high-dose estrogen therapy, but the tumor eventually regrows, only to regress again when estrogen is removed. Patients can be palliated by intermittent estrogen and withdrawal over many years. Our laboratory results are a variation of this clinical observation,because the physiological estrogen is key to tumor regression after TAM failure. Estrogen treatment is superior to simply withdrawing TAM treatment (Figs. 1,B and 5). Another potentially important point is the return of the resistant tumor to TAM sensitivity. It is possible that patients could be maintained on TAM, with episodic periods of estrogen treatment. However, this hypothesis can only be validated through the clinical trials process.

The question of a mechanism to explain the actions of E₂ and TAM as antitumor agents is not a simple issue to answer. The mechanism of TAM-stimulated tumor growth must involve two components: a selection of cells that grow in response to the partial agonist actions of TAM; and a second requirement for estrogen-like angiogenesis to support tumor growth. TAM-stimulated tumors have an increased activation of the VEGF gene (35), which is an important component for angiogenesis. The mechanism of E₂-induced tumor cell death is also unclear. Indeed, a mechanism for the antitumor action of pharmacological doses of estrogen has never been solved.

The recent discovery of a second ER referred to as ER β (36) introduces a new dimension for the understanding of estrogen action. ER β has a structural organization similar to ERα, the classical ER (36), but there are important differences in the activating functions (37, 38) and the ligand-binding domains (39, 40, 41, 42). Most importantly, ER βand ER α are able to activate an AP-1 signal transduction pathway with TAM (43, 44, 45), which might be able to amplify the agonist actions of TAM to induce TAM-stimulated growth. However,E₂ inhibits ER β AP-1 pathways (44, 45); therefore, it is plausible that this could be a mechanism for the tumoricidal actions of E₂. Long-term TAM-stimulated tumors, such as MCF-7TAM and MT2, could contain cells selected for the ER β AP-1 pathways. It is difficult to test this hypothesis directly without reliable monoclonal antibodies to quantitate ER β, but it is possible to test the hypothesis experimentally. The pure antiestrogen ICI 182,780 stimulates the ER βAP-1 pathway (43, 44, 45); therefore, if the pathway is critical for TAM-stimulated tumors after 5 years of treatment, then ICI 182,780 should stimulate and not block tumor growth. Only about half of the ER α-positive tumors that fail TAM respond to ICI 182,780 with tumor regression (46, 47). Perhaps, tumor growth is supported by an activated AP-1 pathway in the tumors that progress. We are currently testing our hypothesis.