Wnt Signaling can Substitute for Estrogen to Induce Division of ERα-positive Cells in a Mouse Mammary Tumor Model

Abstract

The interaction of estrogen with the estrogen receptor (ER, principally ERα) induces growth of human breast tumor cells. In contrast, ERα-positive cells have been described as non-dividing cells in normal breast (though estrogen stimulation of ERα cells directs the division of neighboring cells). However, there is a small sub-population of cells in normal mammary tissue that are ERα-positive, that can divide, and therefore share this property with human breast tumor cells. In order to investigate their pattern of growth regulation, we measured the fraction of dividing ERα⁺ cells during normal growth and compared that to glands stimulated by oncogenic Wnt effectors. First, we found there was no difference between the rate of division of ERα⁺ cells and ERα⁻ cells, whether the population was responding to estrogen or Wnt mitogens. The proportion of dividing ERα⁺ mammary epithelial cells was increased (10×) in response to pregnancy, and similar increases were observed in response to ectopic Wnt signaling. We propose that Wnt signaling can substitute for estrogen to drive total population growth (that includes ERα⁺ cells). Although the E-ERα-derived mitogenic response is situated in a minority of the luminal cells, and the Wnt-LRP5/6 –derived mitogenic response is situated in a minority of basal cells, overall, the growth response of the mammary epithelial population is remarkably similar.

Introduction

One of the defining features of the majority of human breast tumors is that they express (and over-express) the nuclear hormone receptor estrogen receptor-α (ERα). Non-pregnant levels of estrogen, in combination with ERα, are sufficient to drive cell division in human breast tumor cells, and indeed aromatase inhibitors or tamoxifen (inhibitors of E- ERα trans-activation) arrest or kill ERα⁺ tumor cells. Interestingly, although growth of normal cells is also estrogen-dependent, the vast majority of dividing cells are ERα-negative. To explain this, one hypothesis suggests that the ERα⁺ subpopulation (5–10%) induces growth, in paracrine fashion, in the ERα–negative majority. This is supported by observation of xenografts of human cells in the presence of estrogen, and the growth of ERα-knockout cells in chimeric transplants in mouse [1; 2]. Clearly, understanding the local mediators of cell growth are important to understanding the etiology of breast cancer.

A minority of dividing cells (1.0 – 0.05 %; with and without estrogen, respectively) that are ERα⁺ / PR⁺ / Ki67⁺ in human [3] and ERα⁺ / Ki67⁺ in mouse (this study; [4]. Various investigators have proposed that these cells represent a stem/progenitor cell type and suggest that their growth regulation be established separately from the cell majority. For example, using engineered mice, Ewan et al (2005) showed that the division of this sub-population was suppressed by TGFβ signaling. Thus, in normal mice, these cells were associated with lower cell-associated TGFβ and loss of TGFβ signaling (absent nuclear R-SMAD) [4]. In mice with a gain of function in TGFβ signaling, the fraction of dividing ERα⁺ cells in mammary glands was decreased (6-fold) and, vice versa, increased (16-fold) in mammary glands with loss of function of TGFβ-signaling. Thus, the proliferative behavior of these estrogen-dependent cells can be modified by alterations in other signaling pathways.

Here, we have turned to transgenic mouse models with ectopic expression of Wnt signaling to test whether this pathway is important to the control of this ERα⁺ cell minority. Wnt signaling is necessary for the growth of mammary ductal trees in virgin mice (at low ambient estrogen), and is key to the maintenance of mammary stem cells [5]. The receptors for canonical Wnt signaling only exist in the basal population (that includes functional regenerative stem cell activity). Ectopic Wnt signaling promotes stem cell accumulation, and is subsequently highly oncogenic [6]. Whereas Ewan et al (2005) showed that TGFβ inhibited the division of this group of cells, we show here that ectopic Wnt signaling induces their division to levels usually associated with juvenile ductal development, or pregnancy-associated lobuloalveolar development, the two phases of active growth for mammary glands. We propose that ectopic Wnt signaling can functionally substitute for estrogen-dependent growth, producing estrogen-independent tumors.

Materials and Methods

Mouse mammary samples

Samples were collected at the ages and timepoints indicated from transgenic and control mice. Wnt effector transgenic mice were described in Liu et al (2004); these strains express either Wnt1 ligand or ΔNβcatenin (a non-degradable canonical Wnt effector) under the control of the mouse mammary tumor virus LTR (MMTV-LTR). MMTV-driven gene expression is induced at low levels in embryonic mammary placodes, increases during puberty, and rises dramatically during pregnancy and lactation. All the tumors used in this study were analyzed for their H-Ras status, and no mutations were identified. Other later-developing tumors from Wnt1-transgenic mice were mutant for H-Ras (11/18 of those developing after 100 days); unpublished data and [7]). To test the effect of ovariectomy on Wnt-induced hyperplastic glands, 8 week-old mice were ovariectomized and their glands isolated for analysis 4 weeks later. To test the effect of ovariectomy on Wnt-induced tumors, mice with palpable Wnt-induced tumors (0.5–0.8 cm) were anesthetized, their tumors were biopsied to generate pre-ovex samples and simultaneously ovariectomized. When the masses were at least 1 cm diameter, they were excised to compare pre-ovex and post-ovex cell dynamics. All mouse procedures were done under NIH ACUC guidelines.

Immunocytochemistry

Primary antibodies were: anti-ERα from Santa Cruz Biotechnology (rabbit polyclonal antibody made to the C-terminus of ERα, MC20, SC-542; working concentration was 1µg/ml in Tris buffered saline-0.05% Tween; TBS-Tw), Ki67 (mouse monoclonal antibody, proliferation marker; cloneB56, BD Pharmingen; 25 µg/ml), Keratin 5 (Rabbit anti-K5, Covance, Madison, WI) and Keratin 8 (rat anti-K8 (Troma-I), Developmental Studies Hybridoma Bank (University of Iowa). Secondary antibodies were Alexa 546 goat anti-mouse IgG (20 µg/ml) and Alexa Fluor 488 goat anti-rabbit IgG (10 µg/ml) from Molecular Probes (Eugene, OR).

Samples were fixed (overnight) in paraformaldehyde (3.7% in PBS) at 4°C, embedded in paraffin and sectioned (10 µm). Rehydrated sections were treated for antigen retrieval (boiling in sodium citrate buffer (0.01 M, pH 6), as described by Zymed (www.invitrogen.com). Tissues were permeabilized in a solution of 0.2% Triton-X in PBS for 10 mins and washed. Non-specific binding was blocked by incubating slides for at least an hour at room temperature with 10% sheep serum in TBS-Tw. A mixture of both primary antibodies (diluted in TBS-Tween) was incubated with tissues overnight at 4°C, slides were washed three times for 5 minutes in TBS-Tween and incubated with secondary antibodies for 1 hour at room temperature. Slides were washed and counterstained for nuclear DNA with TO-PRO-3 (0.5 µM; Molecular Probes), for 10 minutes at room temperature. Immunofluorescent stains were visualized on a confocal microscope (BioRad MRC1024), using Kr/Ar laser lines at 488, 568 and 647 nm. There was no overlap between fluorescent stains (determined from single stains). Staining was done manually by one operator, and results were scored by 2 blinded operators.

To score ERα⁺, Ki67-positive and ERα⁺ Ki67 double positive nuclei, at least 2000 –5000 (epithelial cell) nuclei (the n (number of nuclei) is shown for all data sets in Supplementary Table 1) were counted separately in up to 30 fields chosen at random from tissue sections spaced approximately 100 µM apart. The statistical analyses that apply to each comparison are shown in Figure Legends.

Results

For this study, we recorded the fractions of ERα⁺ cells, of dividing (Ki67-positive) cells, and the proportion of dividing ERα⁺ cells for each physiological condition. The gross appearance of our sample tissues is shown in Fig. 1 (paraffin-embedded samples stained with H&E).

Fig. 1. Morphology of samples (illustrated with H&E-stained sections from paraffin-embedded samples).

A. Virgin adult glands, showing the ductal tree, embedded in a predominantly fatty matrix. Total colonization of the fat pad is approximately 5%. B. In pregnant mice (12 days p.c), the ductal tree sprouts lobuloalveolar side-branches. C. After 3 months, transgenic mouse glands expressing the oncogene, ΔNβcatenin, develop ductal hyperplasia. D. As early as the mammary rudiment stage, transgenic mice expressing the proto-oncogene, Wnt1, show hyperplastic ductal overgrowth. Shown here is an example from an adult 12 week-old gland. E. After 12 months, 96% of [MMTV- ΔNβcat] mice develop solid, differentiated adenocarcinomas. F. After 8 months, 77% of [MMTV-Wnt1] mice develop differentiated, microacinar adenocarcinomas with highly reactive stromal infiltration.

Patterns of ERα⁺, Ki67⁺ and dividing ERα⁺ cells in normal mammary tissue

During ductal extension and during pregnancy, we confirmed that the majority of ERα⁺ cells were luminal (defined by their position), and that the majority of dividing cells were not ERα⁺ (Fig. 2). Though it is clear that estrogen is absolutely required for cell division (see Fig. 6), and it is likely that cells with nuclear ERα⁺ are the estrogen-responsive population, the dividing luminal cells tend to occur in clusters at a distance from the ERα⁺ cells. These properties are also typical of human breast tissues. Also in common with human breast tissues [3], there were a number of parabasal ERα⁺ cells (Fig. 2) that were also (sometimes) Ki67⁺. Dividing ERα⁺ cells are rare: 10–20% of total cells are ERα⁺, 1 – 10% of total cells are dividing, and only 0.1 – 1.0% of total cells are ERα⁺ Ki67⁺ double positive (DP) (their appearance and distribution is shown in Figs. 2 and 3, and their quantitation in Fig. 4).

Fig. 2. Double-staining for ERα- and Ki67-positive cells in normal glands.

A. Virgin ductal trees. The ductal lumen is indicated, together with the usual bilayered epithelial lining, comprising of luminal cells bounded by myoepithelial cells, which are bounded in turn by the basement membrane. Nuclei are stained blue, and the relative colors of the specific stains are indicated (ERα, green; Ki67, red). In the top panel, * shows an (uncommon) parabasal dividing cell, together with clusters of dividing luminal cells (marked #). Examples of parabasal double positive (DP) cells are shown in the middle and bottom panels. B. Pregnant glands. Samples from mice 10 (top) and 15 (bottom panel) days post-coitus are shown. The arrangement of ERα-positive and –negative cells is approximately the same as it is in ducts from virgin mice, together with the distribution of DP cells.

Fig. 6. In the absence of estrogen, ectopic Wnt signaling drives growth.

A. Mice with palpable Wnt-induced tumors were biopsied (PRE), ovarectomized, and their growth observed. When the masses were 1 cm diameter, they were excised (POST), and the relative proportions of ERα⁺, Ki67⁺ and DP cells assessed. One tumor was divided into 4 separate samples to assess intra-tumoral variation. B–E. Immunocytochemical evaluation of dividing luminal and basal cell populations in ovariectomized mice. Mammary tissues were stained with lineage-specific markers (keratin 5, basal cells, stained blue; keratin 8, luminal cells, stained red) to provide a cellular context for the Ki67 stain (Ki67, stained green). Two examples are shown for each physiological condition (B–E). Normal ducts from adult virgin mice showed occasional Ki67+ cells (B; same frequencies as observed throughout this study), whereas ovariectomized mice showed no dividing cells (D; the diffuse luminal green stain is non-specific). Wnt1 transgenic glands (13 weeks) showed an increased fraction of Ki67+ cells compared to control (C), that was not reduced after ovariectomy (E). F. Scheme of growth regulation in mammary epithelial cell populations. Basal cells respond to Wnt signals to create the total heterogeneous population, whereas luminal cells respond to estrogenic signals to create a grossly similar population (especially with respect to their ERα expression).

Fig. 3. Double-staining for ERα- and Ki67-positive cells in preneoplastic glands and tumors.

A. Glands from pre-neoplastic 6 week-old, and 12 week-old ΔNβcat mice, showing patches of ERα+ve cells (bracket) and DP cells. B. Staining patterns of ERα and Ki67 in ΔNβcat-induced tumors, showing clustering of dividing cells separated from ER+ve cells. C. Shown here is an example of a reactive site from a hyperplastic gland from a 6 week-old Wnt1-transgenic mouse, where cell-cell contacts are less organized, and dividing, ERα⁺ and DP cells are less restricted to their appropriate positions, annotated here is an ERα⁺ basal cell (*) and a giant parabasal cell (DP arrow). D. Two fields from Wnt1-induced tumors, one showing a high DP fraction (top). A large proportion of the cell population expresses some residual ERα (bottom panel).

Fig. 4. Quantitation of ER, Ki67 and DP populations.

Each value represents 2000–6000 cells (see Supplemental Table 1) and the samples and methodology is described in Materials and Methods. Normal virgin glands are shown in grey columns, glands from pregnant mice in pink columns. For preneoplastic glands from ΔNβcat- and Wnt1-transgenic mice, mice of different ages are shown with different color codes; yellow, 3 week old; green, 6 wk old; blue, 12 wk old; purple, 21 wk old. Tumors are color-coded earliest to latest, white, pale grey, dark grey, black (see Table 1). When populations appeared to fall into two groups, this was indicated with arbitrary brackets.

The fraction of ERα⁺ cells measured during very different phases of growth and development is remarkably consistent. Thus early in juvenile ductal development, where growth is estrogen-dependent but estrogen concentrations are low, the proportion of ERα⁺ cells is 15– 20% (3 week-old glands). After puberty, when estrogen concentrations increase, this declines to approx. 10% (Fig. 4), as expected from the feedback degradation of ERα⁺ protein after exposure to ligand. The number of dividing cells declines as the outgrowth of the ductal tree slows to a halt in the adult fat pad (from 7 to 1% total). Similarly, during lobuloalveolar development in pregnant mice (when estrogen levels are high) approximately the same proportion of cells- one out of eight - are ERα⁺. This is accompanied by a 10× increase in cell division (to 4% of the total cell population).

During both ductal extension and pregnancy, the fraction of dividing ERα⁺ cells was increased 10 fold, from approximately 0.06% of the total ERα⁺ cell population in quiescent virgin ductal trees to 0.6% in juvenile ductal trees, or late pregnancy (p= 0.0039; Fig. 4). To address whether the ERα⁺ cells enter the cell cycle with the same frequency as the total population, the fraction of ERα⁺ cells was multiplied by the fraction of Ki67⁺ cells to obtain a “calculated fraction of DP cells” (Fig. 5), for comparison with the observed frequencies of DP cells. Since this frequency was not statistically different from one (DP actual/calculated for ductal populations from juvenile glands= 0.79±0.23, and for lobuloalveolar populations from pregnant glands=0.89±0.06), we can conclude that ERα⁻ cells and ERα⁺ cells are equally likely to divide. Therefore, the mitotic effect generated by the E-ERα⁺-dependent interaction creates an equal chance of cell division in all cells, irrespective of their ERα⁺ status.

Fig. 5. Evaluation of the rate of division of ER+ cells compared to the cell majority.

If division of cells occurs randomly (irrespective of their ERα⁺ status) then the product of the fraction of cells positive for ERα⁺ × fraction Ki67+ cells (the calculated DP fraction) will be similar to the observed DP fraction. If the actual/calculated DP ratio=1, we conclude that ERα⁺ and ERα⁻ cells divide equally in response to mitogenic signals.

Wnt-induced preneoplastic glands and tumors

In the preneoplastic, virgin glands from transgenic mice expressing Wnt signaling effectors, the growth index was increased compared to non-transgenic controls (Fig. 4), to approximately that of actively growing normal tissues. After 5 months, the mammary epithelial cell populations in both transgenic strains showed the lowest growth indices (data points are color-coded for the age of the mouse at sacrifice; purple spots are 5 month old). Note that for these tumor models, clonal transformation arises in this background of general hyperplasia, resulting in tumor development (median latency of tumor development is 21 weeks (of age) in Wnt1-induced glands and 28 weeks in ΔNβcat-induced glands; [7]).

Both Wnt1- and ΔNβcat-induced prenoplastic glands, and even tumors, contained approximately normal ERα⁺ cell fractions (ie. approx 1/8, though some showed 1/3 and others 1/20). These tumors have a highly differentiated phenotype that includes cells resembling normal basal and luminal cells. Despite the massive over-production of cells during rapid tumor development characteristic of this model, there is a relatively normal distribution of basal and luminal cells, which even extends to the proportion of ERα⁺ luminal cells. The distribution of ERα⁺ cells included (infrequent) patches of cells that were all positive (for example, see bracket in Fig 3A). In contrast, in normal tissues, ERα⁺ cells tend to be distributed in a salt and pepper pattern. This reflects the disturbance of normal intercellular interaction and polarization that occurs when ectopic Wnt signals are derived from the luminal compartment [8].

Interestingly, the samples from Wnt-induced preneoplastic glands showed higher variance than normal populations from pregnant glands. For example, the ERα⁺ cell fraction was 14± 4.8% for lobuloalveolar populations from pregnant glands, whereas it was 10 ± 102% for Wnt1-induced preneoplasias (f test, 0.001). This variability was also characteristic of the ductal populations in juvenile glands (ER cell population 13.6±25.5%), though the trend was not statistically significant. The cell division pattern showed a similar trend (pregnant, 4±2.4%; ductal, 4.36±12.6%; Wnt-induced, 4.8±6.3; f test=0.06). This variance was extended to a bimodal high/low pattern in the tumors (see brackets Fig.4). The significance of this is unknown, but might imply the presence of an unidentified pulsatile growth signal that cooperates with Wnt signals in non-pregnant mice.

A comparison of the calculated DP fractions with those observed (Fig. 5) in Wnt-induced tumors showed that the division rate of ERα⁺ cells was in striking equilibrium with ERα⁻ cells (except in ΔNβcat-induced tumors, where ERα⁺ ve cells show a 2.2× higher mitogenic index that the ERα⁻ population).

Can Wnt signaling sustain cell division in pre-neoplastic and tumor tissues in the absence of estrogen?

To test whether cell division was estrogen-dependent in Wnt-stimulated transgenic glands, the Ki67+ cell fraction was assessed in Wnt1-transgenic mice after ovariectomy. In contrast to normal tissues, the rate of cell division for these populations did not change (compare Fig. 6C and E with 6B and D), showing that cell division was estrogen-independent.

In order to confirm that the tumors were not estrogen-dependent, mice with palpable lesions were ovariectomized. Unlike human ERα⁺ tumors, these tumors do not regress; our data was consistent with data published by other groups [9; 10]. The variance of these samples was high, and resembled that of the other samples from normal and Wnt-induced tissues (DP fraction either fraction of dividing ERα⁺ cells in Wnt-induced tumors was either high (≥ 20%) or low (≤ 5%) (Fig. 6A).

Discussion

Wnt signaling can induce the division of ERα⁺ cells

All mouse mammary growth and expansion has now been shown to be ERα⁺–dependent [1]. The ERα null mouse strain (ERKO) that was previously studied is now assumed to have hypomorphic ERα function. When that hypomorphic strain was crossed to the Wnt over-expressing strain used here, the partial ductal outgrowth characteristic of this strain was augmented by ectopic Wnt signaling [10; 11].

We have shown that mammary epithelial populations can be induced to divide by ectopic Wnt effectors (either secreted Wnt1 ligand, or the intracellular non-degradable effector, ΔNβcatenin), to levels associated with normal processes of active growth (either ductal extension in juveniles, or expansion of the lobuloalveolar network during pregnancy). The pattern of division of ERα⁺ and ERα⁻ cells resembles that of normal growing epithelial cell populations. Thus, the fraction of dividing ERα⁺ cells increases from 0.06% in quiescent virgin ductal trees to 0.6% (a 10-fold increase) in juvenile expanding trees, or in glands from pregnant mice. In preneoplastic Wnt and ΔNβcat glands, this fraction can be as high as 1.4%.

Note that Wnt signaling is required for normal ductal stem cell activity [5]. Thus, in the absence of the Wnt signaling receptor LRP5, the ductal tree forms, however the adult ductal regenerative stem cell population is depleted (by ≥ 95%). Vice versa, ectopic Wnt signaling over-stimulates this population [6], leading to an accumulation of stem cell activity (10×) in the total cell population.

Although the overall growth rate and pattern of production of luminal and basal cells is very similar whether growth is induced by estrogen or Wnt ligands, it is unlikely that the cellular target for E-ERα-dependent growth is the same as the Wnt-Lrp5/6-dependent growth. Thus, our work has shown that the canonical (growth/tumor-inducing) receptors for Wnt ligands exist only on basal cells (Badders et al submitted). Vice versa, ERα is clearly only present in the nuclei of luminal cells. The most striking conclusion drawn from this study is that, irrespective of the cell type that receives the signal, the final growth and morphogenesis includes the same proportion of ERα⁺ and ERα⁻ cells, and the same overall structure/function (see scheme of Fig. 6F). Wnt signaling can clearly substitute for estrogen to create a mammary epithelial population indistinguishable in most regards from estrogen-dependent normal populations. The growth stimulation is remarkably symmetric whether derived from a basal or luminal origin.

Note that TGFβ has directly opposite effects on the division of ERα⁺ cells compared to Wnt signals. In fact, Wnt and TGFβ signaling are often in opposition developmentally [12]. The growth inhibitory effects of TGFβ are known to be very important for maintaining the quiescence of mammary ductal trees [13], and ectopic Wnt signaling may reverse this growth inhibition. Constitutively activated TGFβ induced a 6-fold reduction of dividing ERα⁺ cells, whereas a 90% ablation of TGFβ signaling (in TGFβ +/− mice) increased their number 16-fold.

Although it is formally possible that Wnt and estrogen substitute for each other at the molecular level, and indeed there is ample precedent for cross-talk of Wnt and steroid receptor signaling [14] [15] [16], we consider this less likely. Thus, if one or both of these pathways was operating in a non-canonical mode (via the Fzd or other alternate receptors for Wnt signaling, and/or a non-nuclear estrogen function, with or without ERα), they could be using similar/shared transactivation sites to induce growth-associated genes.

Accumulations of ERα⁺ cells are typical of post-menopausal women and preneoplastic conditions such as ductal carcinoma in situ, and many are dividing (the fraction of dividing ERα⁺ cells increases from 0.01% in premenopausal breast to 11% [17; 18]. This is also true for preneoplastic conditions in genetically modified mice [19]. We observed accumulations of ERα⁺ cells in [MMTV-ΔNβcat] preneoplastic glands, which deviate from the normal heterogeneous morphogenetic fields that characterize the distribution of ERα⁺ cells in mammary gland (ERα⁺ cells within each field appear to regulate the proliferation of surrounding ER- cells).

Are dividing ERα⁺ cells stem/progenitor cells?

When we compare the rate of division of ERα⁺ and ERα⁻ cells, they are not significantly different. In other words, all mammary epithelial cells (irrespective of their cell-autonomous ERα⁺ expression) divide at the same rate after exposure to estrogenic or Wnt-mediated growth stimuli.

Although in theory an ERα⁺ estrogen-responsive progenitor cell could orchestrate the endocrine response of mammary tissue (and be subverted during tumorigenesis; [20]), in practice, the ductal stem cell-enriched fractions able to regenerate ductal trees in cleared fat pads do not express ERα⁺. Published data shows that there are at least two stem/progenitor cell types, a ductal stem cell measured using an in vivo isograft assay (mammary repopulating unit; MRU), and another clonogenic cell (colony-forming cell; CFC) that has been scored in vitro using sphere- or colony-forming assay [21; 22]. The MRU population sorted by current surface markers is only 5% pure, whereas the CFC fraction is almost pure (ie. most viable cells are clonogenic). None of the MRU fraction expresses measurable ERα protein (by immunohistochemistry) [23; 24]. Other sub-populations with a more indirect relationship to stem cells (either human or mouse) have been tested for their expression of ERα, and found to be positive, including SP cells, label-retaining cells (LRC), and parity-induced cells [25; 26; 27].

We considered whether there could be two types of ER cells, one terminally differentiated, and a distinct progenitor cell that divides in response to estrogenic stimuli. The frequency of DP cells in this study is 0.06–0.6%. If this is considered to be the growth fraction (5–10% of total cells) of a distinct subpopulation of cells that express ERα⁺, the size of that sub-population might be 0.3–6%, approx 1/50 total mammary epithelial cells. However, given the concordance of the rates of division of ERα⁺ and ERα⁻ cells, it is hard to reconcile the data with this complicated scenario.

Could ectopic Wnt signaling support the growth of tamoxifen-resistant ERα⁺ breast tumors?

Our results are consistent with those of others in confirming that the majority of biological features of ERα⁺ cells in mouse mammary glands are similar to those of human breast tissue. Thus, the arrangement and fraction of ERα⁺ cells is similar, the separation between the majority of ERα-expressing cells and dividing cells is identical, together with the dependence of the minority, ERα⁺ cells on endocrine support for their division [2; 25; 28].

When mouse tumors are found to express ERα, they are widely assumed to be estrogen-dependent, without proper functional evaluation. There may be only two cases where mouse models have been rigorously shown to develop ERα⁺, estrogen-dependent tumors; these are mice that over-express an ERα transgene, and mice with reduced p53 function [29; 30]. The fact that tumorigenesis is estrogen-independent in the [MMTV-Wnt1] model has been shown before: Bocchinfuso et al (1999) showed that post-pubertal ovarectomy did not affect Wnt1-induced tumor growth rates, and Zhang et al [9] also showed that the rate of growth of xeno-transplanted Wnt1-induced tumors was unaffected by ovariectomy (though the frequency of ER+ cells reduced to the normal 8% range if the primary tumor had a high (15%) ERα⁺ fraction; their data shown without comment, confirmed by this study). We propose that Wnt signaling can substitute for ovarian support to promote mammary growth (though the pattern of outgrowth is abnormal and hyperplastic). Wnt1-induced tumors are known to be “oncogene-addicted”, so that when Wnt is removed, the tumors involute [31].

Unlike other tumor types (like colorectal cancer), the tumor suppressors and oncogenes in the Wnt pathway (Apc, axin and βcatenin) appear to be unaffected in breast tumors (except perhaps as a late event). However, genes that regulate extracellular Wnt signaling do show consistent changes in breast tumors (including loss of the inhibitor, SFRP1, [32]. It is therefore possible that ectopic Wnt signaling could be supporting the growth of human breast cancers that are ERα⁺ but PR-negative, and show tamoxifen-resistance. We propose that the small molecule inhibitors of Wnt signaling that are under development by pharmaceutical companies may show efficacy for this patient group.

Abbreviations

ERα

estrogen receptor α

MECs

mammary epithelial cells

DP

double positive (ER+ve, Ki67+ve) cells