Background
The orchestrated development of the mammary gland requires several molecularly controlled, evolutionarily conserved, and temporally distinct processes [
1], occurring to a large extent in adolescents and adults of mammalian species. The mammary gland is comprised of several distinct stem cell, progenitor cell, and mature mammary epithelial cell (MEC) subpopulations that each perform a specialized function at distinct developmental stages [
2‐
7]. During pregnancy, as the mammary gland prepares for synthesis and delivery of milk to newborns, an apically located (i.e., luminal) MEC subpopulation proliferates rapidly in response to systemic hormonal cues such as progesterone and prolactin (PRL), as well as locally derived cues such as neuregulin (NRG, also known as heregulin). At parturition, these milk-producing MECs, referred to as alveolar MECs (aMECs), undergo terminal differentiation for milk synthesis and delivery [
8]. This entire process is referred to as lactogenesis, a fundamental requirement for the survival of all mammalian species [
9‐
11].
Several studies demonstrated that PRL-mediated milk production requires signaling through a molecular axis involving PRL receptor (PRLR)-mediated activation of the intracellular tyrosine kinase (TK) Janus kinase 2 (Jak2), which phosphorylates the transcription factor signal transducer and activator of transcription 5A (STAT5A), an obligate transactivator of several milk protein-encoding genes, including
Csn2 (the gene encoding β-casein) [
8,
12‐
16]. Genetically engineered mouse models (GEMMs) confirmed that the PRL/PRLR/Jak2/STAT5A signaling axis is crucial for lactogenesis [
17‐
19].
Although STAT5A is potently activated by PRLR/JAK2 signaling, STAT5A also forms complexes with other receptors in MECs [
20,
21]. Among these is the receptor tyrosine kinase (RTK) ErbB4 [
22‐
24], a member of the epidermal growth factor receptor (EGFR) RTK family, comprised of EGFR, HER2/ErbB2, HER3/ErbB3, and HER4/ErbB4. Interestingly, ErbB4 activity in the mammary gland, specifically in luminal aMECs, peaks during pregnancy and lactation [
25], similar to what is seen for STAT5A. Multiple GEMMs of ErbB4 ablation in the mammary epithelium each display reduced expansion of alveolar structures during pregnancy, decreased terminal aMEC differentiation, impaired activation of STAT5A, lactation defects [
20,
26,
27], and phenocopying of the effects of PRL, PRLR, JAK2, or STAT5A loss. Conversely, increased ErbB4 kinase signaling activates STAT5A, even in the absence of pregnancy-related hormones [
28], confirming the role of ErbB4 in STAT5A-mediated lactogenesis.
ErbB4 is ligand activated by NRG family members (NRG1–4), and by certain EGF-like family members (heparin-binding EGF (HB-EGF), betacellulin (BTC), and epiregulin).
NRG1 expression by alveolar basal/myoepithelial MECs is induced at mid-gestation in response to p63, a master regulator of transcriptional programs directing cell fate. NRG initiates proliferation of adjacent aMECs through ErbB4/STAT5A activation [
23]. Similar to what was seen with ErbB4 loss, mouse models lacking NRG1α or NRG1β expression [
29], NRG bioavailability [
23], or HB-EGF bioavailability [
30] suffered decreased aMEC expansion during pregnancy. Conversely, NRG1-loaded slow release pellets implanted into mouse mammary glands induced precocious lactogenesis in nonpregnant mice [
31]. The NRG ligands also bind to ErbB3, inducing ErbB3 heterodimerization with other EGFR family receptors, and indeed with a growing list of heterologous RTKs [
32,
33]. Ligand-activated ErbB3 has been described as lacking intrinsic kinase activity [
34], but once phosphorylated by a heterodimeric partner it potently stimulates cell survival signaling [
35]. The role of ErbB3 in MEC cell survival is illustrated by distinct models of impaired ErbB3 signaling in the mammary epithelium during puberty, each causing impaired cell survival of ductal MECs and decreased lengthening of mammary duct.
The ability of ErbB3 to enhance cell survival is explained largely in part by its six binding sites for the p85 regulatory subunit of phosphatidyl inositol-3 kinase (PI3K) [
24], more than any other RTK. When phosphorylated, ErbB3 interacts with p85 to promote PI3K activity, which generates the second messenger phosphoinositol-(3,4,5)-trisphosphate (PIP3), causing Akt recruitment, Akt phosphorylation (via PDK1 [
36] and mTORC2 [
37], which also are recruited by PIP3), and Akt kinase activation. Akt sits at the apex of a signaling cascade that supports cell survival, growth, metabolism, and cell cycle progression [
38]. Several lines of evidence demonstrate that PI3K/Akt signaling supports lactogenesis of aMECs through cell survival. For example, models in which Akt signaling was impaired caused aMEC apoptosis, decreasing the capacity for lactogenic alveolar expansion during pregnancy, and causing premature mammary gland involution at parturition [
39‐
42], while models of increased PI3K/Akt signaling, for example, through expression of myristoylated Akt1 (Akt
myr) in the mammary epithelium, delayed the onset of postlactational cell death upon weaning [
43‐
45]. Given that ErbB3 potently activates PI3K/Akt signaling, it is possible that ErbB3 may be required for lactogenic aMEC expansion during pregnancy. However, its role in aMECs is not yet clear, as two mouse models of ErbB3 ablation have produced disparate results. Although lactogenic aMEC expansion during pregnancy occurred normally in embryonic mammary bud transplants from a classical model of ErbB3 ablation to wild-type recipients [
46], a knock-in model eliminating each of the ErbB3 p85-PI3K binding motifs within an otherwise intact ErbB3 reduced aMEC expansion during pregnancy, causing increased aMEC cell death and accelerated involution [
47].
We used a mouse model of ErbB3 ablation specifically from aMECs, finding delayed aMEC expansion due to decreased Akt-dependent survival of cytokeratin 8 (CK8)-positive aMECs, and decreased STAT5A-mediated expression of milk-encoding genes. Restoration of Akt signaling rescued survival of ErbB3-depleted aMECs, but did not rescue STAT5A induction or milk protein expression. However, signaling through both Akt and STAT5A were rescued by ErbB4 overexpression. In vivo, compensatory ErbB4 upregulation in ErbB3-deficient MECs dampened the impact of ErbB3 ablation, allowing lactation to proceed, albeit in a delayed fashion, through the expansion of CK8+/CK5+ double-positive aMECs. These studies highlight the key roles played by ErbB3 and ErbB4 in establishing and maintaining milk production.
Methods
Mice
All animal husbandry and experiments were performed in accordance with protocols approved by the Vanderbilt University Institutional Animal Care Committee using humane procedures.
ErbB3
FL
[
48] and
WAPi-Cre [
49] mice have been described previously. All mice were inbred to, or generated on, the Friend Virus B-type (FVB) background. Mouse experiments were approved by the Vanderbilt Institutional Animal Care and Use Committee. Female virgin mice were bred at 10–12 weeks of age to WT FVB male mice. Mating pairs were separated upon identification of a semen plug, indicating 0.5 days post coitus (d.p.c.). Parturition indicated lactation day 0 (L0), and litters were normalized to eight pups per litter.
Histological analysis
Right #4 inguinal mammary glands were formalin fixed and paraffin embedded, and sections (5 μm) were stained with hematoxylin and eosin. In-situ terminal dUTP nick end-labeling (TUNEL) analysis was performed on paraffin-embedded sections using the ApopTag kit (Millipore). Immunohistochemistry (IHC) on paraffin-embedded sections was performed using the following antibodies as described previously [
50]: ErbB3 (C-17; Santa Cruz Biotechnology), Ki67 (Santa Cruz Biotechnology), P-Akt S473 (Cell Signaling Technologies), P-STAT5A/B (Neomarkers), and P-ErbB4 Tyr1056 (Cell Signaling Technologies). Immunodetection was performed using the Vectastain kit (Vector Laboratories) according to the manufacturer’s instructions. Immunofluorescence staining was performed with primary and secondary antibodies diluted in 12% Fraction-V BSA (Pierce) and slides were mounted in SlowFade mounting medium containing DAPI (Invitrogen). All fluorescent secondary antibodies were highly cross-adsorbed, produced in goat, and used at a dilution of 1:200 for 20 min (Molecular Probes). Primary antibodies used were CK5 (10956, 1:500; Covance/Biolegend), CK8/18 (1:500; Fitzgerald Industries International), and ErbB3 (c-17, 1:200; Santa Cruz Biotechnology).
Cell culture
HC11 cells [
51] were cultured in DMEM: F12 (1:1) medium supplemented with 10% fetal bovine serum (Life Science), insulin (5 μg/ml) and dexamethasone (10 μg/ml) (Sigma-Aldrich) and human EGF (5 ng/ml; R&D Systems). To induce differentiation, cells were serum and EGF-starved for 24 h, and then treated with 5 μg/ml mouse PRL (Preprotech). In some cases, cells were treated with Nrg1b (EGF-like domain) (R&D Systems), neratinib, AZD6244, or BKM120 (all from SelleckChem). Where indicated, single-cell suspensions (5 × 10
5 cells) were infected with 10
6 pfu/ml of the adenoviral particles adenoviral myristoylated Akt1 (Ad.Akt
myr), adenoviral ErbB4 (Ad.ErbB4), and adenoviral green fluorescent protein (Ad.GFP), from Vector Biolabs. Knockdown of ErbB3 in HC11 cells using siRNA sequences was performed using 5 μM siRNA against ErbB3 (SASI_Mm01_0031804, SASI_Mm01_0031805, and SASI_Mm01_0031806; Sigma-Aldrich) or a scrambled sequence (siScr) transfected with Lipofectamine 2000 (Invitrogen). Stable knockdown of ErbB3 was achieved by lentiviral transduction of LVP158 (Gentarget, Inc.) HC11 cells were transduced with lentiviral particles for 48 h prior to puromycin selection. Pooled clones of puromycin-resistant cells were expanded for analysis of ErbB3 knockdown prior to experimentation.
Western blot analysis
Mammary glands and cells were homogenized in ice-cold 50 mM Tris (pH 7.4), 100 mM NaF, 120 mM NaCl, 0.5% Nonidet P-40, 100 μM Na3VO4, and 1× protease inhibitor mixture (Roche), sonicated for 10 s on ice, and cleared by centrifugation at 4 °C, 13,000 × g for 20 min. Protein concentration was determined using the bicinchoninic acid assay (Pierce). Proteins were separated by SDS/PAGE and transferred to nitrocellulose membranes. Membranes were blocked in 3% gelatin in TBS-T (Tris-buffered saline, 0.1% Tween-20) for 1 h, incubated in primary antibody in 3% gelatin overnight at 4 °C, washed with TBS-T, incubated in HRP-conjugated anti-rabbit (sc-2375, 1:10,000; Santa Cruz Biotechnology) or anti-mouse IgG (ab97240, 1:5000; Abcam), washed with TBS-T, and then developed using ECL substrate (Pierce). The following primary antibodies were used: P-ErbB3 (sc-135654, 1:500; Santa Cruz Biotechnology), P-MAPK/Akt/S6/Rab11 cocktail (1:1000; Cell Signaling Technologies), total Akt (C67E7, 1:1000; Cell Signaling Technologies), actin (1:2000; Cell Signaling), STAT5A and P-STAT5A/B (93585 at 1:500 and C11C5 at 1:500, respectively; Cell Signaling Technologies), and cleaved caspase 3 (D175, 1:500; Cell Signaling Technologies).
Gene expression analysis
Total RNA was harvested from cells and mammary glands using RNeasy (Qiagen). RNA quality was assessed using the Agilent Analyzer. RNA (1 μg) was reverse transcribed (RT2 First Strand Kit; Qiagen). Mouse primers for Stat5a, Erbb4, Erbb3, Csn2, and 36B4 were purchased from SABiosciences (Qiagen). Primer sequences are as follows: mErbB3, forward 5′-CGAGAACTGCACCCAAGG and reverse 5′-TCTGCTTGGCCTAAACAGTCT; mStaa5A, forward 5′-CGCCAGATGCAAGTGTTGTAT and reverse 5′-TCCTGGGGATTATCCAAGTCAAT; mCsn2, forward 5′-GGCACAGGTTGTTCAGGCTT and reverse 5′-AAGGAAGGGTGCTACTTGCTG; and mErbb4, forward 5′-CCATGGACCGGACCTGC and reverse 3′-GCTCCCTGTAGGCCATCTGG.
Discussion
We demonstrate here, using aMEC-specific ErbB3 ablation, that ErbB3 engages two key molecular signaling pathways at mid-pregnancy in the mammary gland, STAT5A and PI3K/Akt, which regulate two nonoverlapping aspects of lactogenic development during mid-pregnancy. We show that STAT5 signaling is a critical driver of lactogenic differentiation downstream of ErbB3, while PI3K/Akt signaling downstream of ErbB3 is needed for survival of the rapidly expanding aMEC population during pregnancy. These findings are consistent with a study in which ErbB3-to-PI3K signaling was eliminated through knock-in of an ErbB3 mutant lacking all PI3K binding motifs [
47]. Interestingly, previous studies confirm that ErbB3 is required in luminal progenitor populations [
48], which may give rise to early alveolar progenitor MECs. Because the previous study used a systemic ErbB3 mutant knock-in, then it is possible that impaired ErbB3–PI3K signaling would have altered the luminal progenitor population of MECs prior to commitment to the alveolar lineage, indirectly decreasing the aMEC population, interfering with lactogenesis, and potentially confounding the results. The studies performed herein addressed this possibility, using a Cre-expressing model confined to the committed alveolar lineage [
49], and thus sparing ErbB3 expression in any luminal progenitor populations that might not yet be committed to the alveolar lineage. However, the results shown here largely recapitulate what was seen with the systemic ErbB3 mutant knock-in model, supporting the notion that ErbB3 drives survival of luminal aMECs even after lineage specification.
We further distinguish our findings from those using the ErbB3 mutant knock-in model, as well as the ErbB3-null mammary bud transplantation model, by our observations that complete ErbB3 disruption leads to decreased expression and phosphorylation of STAT5A in aMECs. This phenotype may have been missed in the previous models, perhaps because the knock-in ErbB3 mutant remains capable of ligand-induced heterodimerization with other ErbB family members, which would be capable of NRG-mediated ErbB3–ErbB4 signaling, which was potentially sufficient to induce STAT5A upregulation. This hypothesis is supported by our observations here that ErbB4 can partially, but not fully, compensate for ErbB3 loss.
The PI3K/Akt pathway is a key signaling node for lactogenic expansion and differentiation of the luminal mammary epithelium. Numerous signaling pathways that regulate lactogenic development converge on PI3K/Akt, including the insulin-like growth factor 1 receptor (IGF1R), RANKL and RANK, integrins, and PRLR-to-JAK2-to-STAT5A pathways [
8,
11,
14,
33,
44]. In fact, transgenic mice overexpressing Akt or expressing constitutively active Akt
myr in the mammary epithelium displayed increased aMEC survival, delaying the onset of MEC apoptosis at weaning [
40,
41,
43,
45]. Although cell survival was not affected during pregnancy or lactation per se, increased activity of Akt aberrantly increased glucose uptake, glycolysis, and lipid production by aMECs, causing milk stasis and insufficient lactation [
43], suggesting that Akt supports functions beyond cell survival in aMECs.
We show here that ErbB3 is expressed in mammary glands at mid-pregnancy, a time point when NRG ligands [
23,
29,
31], ErbB4 [
20,
26,
27], and STAT5 [
15,
18,
19,
54] are induced. These findings suggest that ErbB3 may participate, perhaps as a heterodimeric partner of ErbB4, in the signaling cascade that activates STAT5A expression and phosphorylation, while also potently activating Akt phosphorylation at mid-pregnancy. This was confirmed in experiments in which ErbB4 upregulation rescued Akt and STAT5A dysregulation in response to ErbB3 ablation, whereas ErbB4 knock-down exacerbated the defects in Akt and STAT5A signaling caused by ErbB3 depletion (Fig.
5). This does not rule out the possibility that other EGFR family RTKs or ligand-activated PRLR might interact with ErbB3 and/or ErbB4 during lactogenesis, either as heterodimers or as multimeric signaling units. Because transgenic mice expressing a mammary-specific dominant-negative ErbB2 mutant have defective aMEC expansion and milk production [
58], while mice lacking the EGFR ligands amphiregulin or EGF display reduced aMEC growth during pregnancy [
59], it is possible that EGFR and/or ErbB2 might play a supporting role in aMEC expansion during pregnancy. This is an attractive hypothesis, given that ErbB4 upregulation occurred rapidly in
ErbB3
KO
mammary glands, as early as 16.5 d.p.c., but restoration of STAT5A and P-Akt was not evident until lactation day 1. However, we did not find compensatory upregulation of ErbB2 expression in HC11 cells expressing shErbB3 (Additional file
1: Figure S9A) or in
ErbB3
KO
mammary glands (Additional file
1: Figure S9B).
Because previous reports suggest that ErbB4 can exist within a physical complex with PRLR, leading to NRG-induced PRLR activation, and conversely PRL-dependent ErbB4 phosphorylation [
12], it is possible that ligand-activated PRLR could trans-activate ErbB4, resulting in low levels of ErbB4 activity that may partially compensate for ErbB3 loss. Consistent with this idea, treatment of HC11 cells with PRL induced low levels of ErbB4, ErbB3, and STAT5A/B phosphorylation (Additional file
1: Figure S10A, B), suggesting some level of cross-talk between PRL and ErbB RTKs. However, PRL treatment did not induce Akt phosphorylation. Further, ErbB3 knockdown interfered with PRL-mediated STAT5A/B phosphorylation, suggesting an important role for ErbB3 in PRL-mediated STAT5A/B phosphorylation, at least in the setting of a 30-month PRL treatment.
The observation that ErbB4 upregulation partially alleviated the impact of ErbB3 loss upon Akt phosphorylation and cell survival is consistent with previous studies demonstrating that full-length ErbB4 (the Cyt-1 isoform) harbors a PI3K binding motif capable of PI3K activation when phosphorylated at Tyr-1056. Notably, we identified potent upregulation of P-ErbB4 Tyr-1056 in
ErbB3
KO
mammary glands, Interestingly, inhibition of PI3K was unable to modulate STAT5A levels in HC11 cells, and restoration of Akt signaling in ErbB3-depleted HC11 was not sufficient to restore STAT5A expression, despite the previously described ability of Akt to enhance STAT5A stabilization and activity. Because restoration of Akt signaling rescued cell survival but did not rescue STAT5A expression, it is unlikely that selective death of STAT5A-expressing cells is the reason why STAT5A expression is low in the absence of ErbB3. The molecular mechanisms underlying decreased STAT5A expression remain unclear, and will require additional studies. However, we found that ErbB4 upregulation was capable of restoring STAT5A expression in the absence of ErbB3, highlighting the ability of ErbB3 and ErbB4 to compensate for one another. These observations are supported by our findings that the combined loss of both ErbB3 and ErbB4 further impaired Akt and STAT5A signaling in aMECs (Fig.
5f), thus supporting both cell survival (PI3K to Akt) and lineage commitment/differentiation (STAT5A).
Conclusions
In summary, we have found that the NRG receptor ErbB3 supports lactogenic expansion of the milk-producing alveolar mammary epithelium during pregnancy, similar to what was found for the only other known NRG receptor, ErbB4. Given the importance of lactation for the survival of mammalian species, it may not be surprising that compensatory signaling mechanisms exist, ones that will ensure adequate lactation even if a single pathway is compromised. The important roles played by the NRG ligands and their two receptors in driving lactogenesis of the mammary epithelium during pregnancy are becoming increasingly evident. Although PRL is a known activator of STAT5 in the mammary gland, these data showing ErbB3-dependent induction of STAT5 expression, taken with previous data demonstrating ErbB4-mediated phosphorylation and activation of STAT5, suggest that NRGs produced locally in the mammary microenvironment enable PRL-mediated STAT5A activation, while at the same time promoting PI3K/Akt signaling, which together amplify, sustain, and differentiate this unique cell population on which newborn mammals depend.