Membrane expression of the estrogen receptor ERα is required for intercellular communications in the mammary epithelium

Abstract

17β-Estradiol induces the postnatal development of mammary gland and influences breast carcinogenesis by binding to the estrogen receptor ERα. ERα acts as a transcription factor but also elicits rapid signaling through a fraction of ERα expressed at the membrane. Here, we have used the C451A-ERα mouse model mutated for the palmitoylation site to understand how ERα membrane signaling affects mammary gland development. Although the overall structure of physiological mammary gland development is slightly affected, both epithelial fragments and basal cells isolated from C451A-ERα mammary glands failed to grow when engrafted into cleared wild-type fat pads, even in pregnant hosts. Similarly, basal cells purified from hormone-stimulated ovariectomized C451A-ERα mice did not produce normal outgrowths. Ex vivo, C451A-ERα basal cells displayed reduced matrix degradation capacities, suggesting altered migration properties. More importantly, C451A-ERα basal cells recovered in vivo repopulating ability when co-transplanted with wild-type luminal cells and specifically with ERα-positive luminal cells. Transcriptional profiling identified crucial paracrine luminal-to-basal signals. Altogether, our findings uncover an important role for membrane ERα expression in promoting intercellular communications that are essential for mammary gland development.

Keywords: ECM; Estrogen receptor alpha; Mammary gland; Membrane initiated signaling; Paracrine signals; Stem cells.

Fig. 1.

The invasion of mammary fat pad is delayed at puberty in C451A-ERα mice. (A,B) Representative images of whole-mount mammary glands from (A) 5-week-old (wild type, n=8; C451A-ERα, n=8) and (B) 3- to 6-month-old virgin C451A-ERα and wild-type mice (wild type, n=9; C451A-ERα, n=8). Scale bars: 10 mm (right); 1 mm (left). (C) Quantification of fat pad filling in 5-week-old (two-way ANOVA, P<0.01; interaction, **P=0.0082) and 3-month-old (ns, not significant) C451A-ERα and wild-type mice. (D) Representative images of Hematoxylin and Eosin stained transverse sections of mammary glands from 3- to 6-month-old virgin wild-type and C451A-ERα mice. Scale bars: 250 µm and 50 µm (insets). (E,F) Quantification of the number of branching points (E) and of ductal diameters (F) on the ductal tree in 3-month-old wild-type and C451A-ERα mice (wild type, n=7; C451A-ERα- n=7, t-test; **P<0.01; ****P<0.0001). (G) Circulating levels of E2 and progesterone in 5-week- and 3-month-old mice. Expression levels above the dotted line were considered detectable (*P<0.05; **P<0.01; ns, not significant).

Fig. 2.

Absence of development of the C451A-ERα/GFP mammary epithelium after transplantation of ducts in wild-type mice. (A) Fluorescence stereomicrographs of contralateral inguinal wild-type mammary fat pads engrafted with mammary epithelium from C451A-ERα or wild-type littermates. Images of virgin or day 16-18 pregnant recipients obtained 8 weeks after transplantation. Scale bars: 1 mm. (B) Dot plots showing the extent of fat pad filling by the engrafted epithelia in virgin mice (n=22, non-parametric Mann–Whitney t-test, ***P<0.001) or pregnant mice (n=15, non-parametric Mann–Whitney t-test, **P<0.01). Experiments were repeated with four independent donors. Data are mean±s.e.m. (C) Representative immunostaining using Hematoxylin coloration (upper panel), and anti-ERα (middle panel) or anti-PR (lower panel) antibodies in mammary glands from 16.5 day pregnant mice engrafted with epithelium from C451A-ERα or wild-type mice. Scale bars: 100 µm. (D) The proportion of epithelial cells expressing ERα and PR following immunohistochemistry with anti-ERα and PR antibodies was quantified as a percentage of total epithelial cells (**P<0.01; ***P<0.001).

Fig. 3.

The frequency and regenerative potential of CD29hiCD24⁺ basal cells are impacted by the C451A-ERα mutation. (A) Representative flow cytometry dot plots of CD24⁺CD29^lo (luminal) and CD24⁺CD29^hi (basal) populations gated on CD45 and CD31 from mouse mammary glands obtained from 3-month-old wild-type and C451A-ERα mice. The percentages of luminal or basal cells were calculated within the CD31- and CD45-negative populations (wild type, n=25; C451A-ERα, n=25, t-test, **P<0.01). (B) Representative images of GFP-positive outgrowths arising from transplantation of 2000 double-sorted CD29^hi CD24⁺ cells from glands of virgin adult wild-type or C451A-ERα mice. Scale bars: 2 mm. The virgin recipient tissue was collected 8 weeks after transplantation into wild-type and C451A-ERα mice (right panel). (C) Dot plots show the percentages of fat pad filling by outgrowths 8 weeks after transplantation of different numbers of double-sorted CD29^hiCD24⁺ cells. Cells were injected into the cleared mammary fat pads of 3-week-old syngeneic recipients and collected 8 weeks after transplantation. Data were pooled from two independent experiments (two-way ANOVA, ***P<0.001). Data are mean±s.e.m. (D) Repopulating frequency of the transplantation of limited numbers of double-sorted CD29^hiCD24⁺ cells from the mammary glands of 12-week-old wild-type or C451A-ERα female mice (ELDA statistical test).

Fig. 4.

Hormone supplementation in C451A-ERα mice restores the frequency, but not the regenerative potential, of CD29^hiCD24⁺ basal cells. (A) Representative images of whole-mount mammary glands from ovariectomized wild-type and C451A-ERα female mice captured after 3 weeks of treatment with a combination of 17β-estradiol and progesterone; higher magnification images of the ductal tree are also shown. Scale bars: 2 mm (right); 200 µm (left). (B) Bar plots show the percentages of fat pad filling and thickness of ducts (µm) in ductal trees of ovariectomized mice treated with E2 and progesterone for 3 weeks (wild type, n=8; C451A-ERα, n=11, t-test; ns, not significant; ***P<0.001). (C) Circulating levels of E2 and progesterone in ovariectomized mice treated with E2 and progesterone for 3 weeks. Levels above the dotted red line were considered detectable (E2 wild type, n=7; E2 C451A-ERα, n=5; E2 and Pg wild type, n=6; E2 and Pg C451A-ERα, n=7; t-test; ns, not significant). (D) Representative images of Ki67, ERα and PR immunostaining in mammary glands from ovariectomized wild-type and C451A-ERα mice treated with E2 and progesterone for 3 weeks. The percentages of epithelial cells positive for Ki-67, ERα or PR are expressed relative to the number of total epithelial cells (wild type, n=6; C451A-ERα, n=6, t-test; ns, not significant). Scale bars: 200 μm. (E) Flow cytometry dot plots of CD24⁺CD29^lo (luminal) and CD24⁺CD29^hi (basal) populations gated on CD45⁻ and CD31⁻ cells from mouse mammary glands removed from ovariectomized wild-type and C451A-ERα mice following 3 weeks of treatment with a combination of 17β-estradiol and progesterone (wild type, n=19; C451A-ERα, n=21, t-test; data not significant). (F) Representative images of GFP-positive outgrowths arising from transplantation of 2000 double-sorted CD29^hiCD24⁺ cells from the mammary glands of ovariectomized (left panels) wild-type and C451A-ERα (right panel) mice treated with E2+progesterone. Scale bars: 2 mm (left); 200 µm (right). The virgin recipient tissue was collected 8 weeks after transplantation. (G) Percentages of fat pad filling by outgrowths 8 weeks after the transplantation of different numbers of double-sorted CD29^hiCD24⁺ cells. Cells were injected into the cleared mammary fat pads of 3-week-old syngeneic recipients and collected 8 weeks after transplantation. Data were pooled from two independent experiments (two-way ANOVA, ***P<0.001). (H) Repopulation frequency of the transplantation of a limited number of double-sorted CD29^hiCD24⁺ cells from the mammary glands of ovariectomized WT and C451A-ERα female mice treated with E2 and progesterone (ELDA statistical test).

Fig. 5.

Co-injection of wild-type CD24⁺CD29^lo LCs with CD29^hiCD24⁺ basal cells from C451A-ERα mice restores their regenerative ability in transplantation assays. (A) Representative images of GFP-positive outgrowths arising from the transplantation of 2500 double-sorted CD24⁺CD29^lo LCs from wild-type (left panel) or C451A-ERα (right panel) mice co-injected with 2500 GFP-positive CD29^hiCD24⁺ basal cells from C451A-ERα mice. Cells from ovariectomized wild-type or C451A-ERα mice treated with E2+progesterone (Pg) for 3 weeks were sorted by flow cytometry. Scale bar: 2 mm. The wild-type virgin recipient tissue was collected 8 weeks after transplantation. (B) Percentage of fat pad filling by outgrowths 8 weeks after the transplantation of different numbers of double-sorted CD24⁺CD29^lo LCs from wild-type or C451A-ERα mice mixed with GFP-positive CD29^hiCD24⁺ basal cells from C45A-ERα mice. Cells were injected into the cleared mammary fat pads of 3-week-old syngeneic recipients and collected 8 weeks after transplantation. Data were pooled from two independent experiments (two-way ANOVA, **P<0.01, ***P<0.001). (C) Repopulating frequency of the transplantation of limited numbers of double-sorted CD24⁺CD29^lo LCs from wild-type or C451A-ERα mice mixed with GFP-positive CD29^hiCD24⁺ basal cells from C45A-ERα mice (ELDA statistical test). (D) Confocal images of mammary gland epithelium after immunostaining using anti-GFP (green), -K5 (red) and -K8 (magenta) primary antibodies. (E,F) Confocal images of mammary gland epithelium after immunostaining using anti-GFP (green), -ERα (MC20, red) and DAPI (cyan) in epithelium 8 weeks after co-injection of GFP-positive C451A-ERα MaSCs mixed with GFP-negative LCs from wild-type mice. Representative sections of ducts when only C451A MaSCs gave rise to basal cells (E) or when MaSCs C451A gave rise to both basal and luminal GFP positive, but no ERα-positive LCs were observed (F). Scale bars: 2 mm in A; 20 μm in D; 10 µm in E,F (left); 25 µm in F (right).

Fig. 6.

Large-scale analysis of the effects of C451A-ERα on gene expression in the CD24⁺CD29^lo LCs. (A) Heatmap of the global gene expression analysis in CD24⁺CD29^lo LCs from wild-type and C451A-ERα ovariectomized mice treated with E2 or E2+progesterone (Pg) for 3 weeks (wild type, n=5; C451A-ERα, n=5 in each condition). (B) Venn diagram of differentially expressed (up- and downregulated) genes. (C-E) Gene ontology analysis of the dysregulated genes using the GO database. (F) Gene expression analysis using qRT-PCR. Relative mRNA levels were normalized and presented as relative levels compared with expression in wild-type mice treated with E2. Samples used in the large-scale analysis were included and are complemented with two additional samples prepared using the same conditions (wild type, n=7; C451A-ERα, n=7; two-way ANOVA, *P<0.05, **P<0.01, ***P<0.001).

Fig. 7.

The wild-type CD24⁺CD29^lo Sca1⁺ CD133⁺ LCs are required for paracrine signaling to restore the regenerative potential of basal cells. (A) Representative gating strategy illustrating LCs (CD24⁺ CD29^lo) being subgated for Sca1⁺ CD133⁺ and Sca1⁻ CD133⁻ subpopulations. (B) Relative quantities of Esr1, Pgr and Areg RNA normalized to B2m, Hprt, Gusb and Tbp in wild-type Sca1⁺CD133⁺ LCs when compared with wild-type Sca1⁻CD133⁻ LCs (**P<0.01, ***P<0.001; Mann–Whitney test). (C) Representative images of GFP-positive outgrowths arising from the transplantation of C451A-ERα MACS with sorted Sca1⁺ CD133⁺ or Sca1⁻ CD133⁻ wild-type LCs. Scale bars: 2 mm. (D) Confocal images of mammary gland epithelium after immunostaining with anti-GFP (green), anti-K5 (red) and anti-K8 (magenta) primary antibodies. Scale bars: 20 µm. (E) Confocal images of mammary gland epithelium after immunostaining using anti-GFP (green) and anti ERα (red) antibodies. Scale bar: 20 µm.