ErbB3 is required for ductal morphogenesis in the mouse mammary gland

The epidermal growth factor receptor (EGFR) family of receptor tyrosine kinases, which consists of EGFR, ErbB2, ErbB3, and ErbB4, is important in many normal developmental processes and is often over-expressed or mutated in human cancer (for review, see [1]). Ligand binding stimulates homodimerization and heterodimerization of EGFR family members. The dimerization facilitates cross-phosphorylation of tyrosines within the cytoplasmic domains that become binding sites for downstream effector proteins. The proto-oncogene product ErbB2 is unique in that it does not bind soluble ligands and requires a heterodimer partner for ligand-dependent activation. ErbB3 is also unique because the kinase domain is catalytically inactive, so that ErbB3 also requires a heterodimerization partner for activity. Once activated, ErbB3 is strongly linked to prosurvival signaling through the phosphatidylinositol 3'-kinase (PI3K)/Akt pathway. ErbB3 has six binding sites for the p85 adaptor subunit of PI3K. PI3K catalyzes formation of 3' phosphoinositides that recruit the protein kinase Akt/protein kinase B to the membrane for activation by phosphorylation. The activated Akt is involved in cellular processes including survival, cell growth, and proliferation (for review, see [2]).

The ErbB2/ErbB3 heterodimer, one of many possible ErbB heterodimer combinations, is important in tumorigenesis [3‐6] (for review, see [7]). ErbB3 is a preferred heterodimerization partner for the proto-oncogene ErbB2, and the ErbB2/ErbB3 heterodimer is highly biologically active and pro-tumorigenic in vitro [8]. ErbB2 and ErbB3 are often co-over-expressed in breast, ovarian, colorectal, and bladder cancers [9‐13].

Recent advances in the treatment of cancers driven by ErbB alterations include the use of small molecule tyrosine kinase inhibitors and antagonist antibodies [14]. Surprisingly, ErbB3 signaling is often crucial to the success or failure of ErbB tyrosine kinase inhibitors, despite the lack of intrinsic ErbB3 kinase activity. In breast cancers with amplified ERBB2, ErbB3 augments ErbB2 signaling through strong coupling to survival pathways that complement the signals emanating from ErbB2, thus allowing the cancer cells to escape from inhibition therapy [15]. In all likelihood, ErbB3 signaling will also contribute to resistance to single and dual specificity EGFR and ErbB2 inhibitors (for example, lapatinib) [16], as they come into clinical use for trastuzumab-resistant ERBB2-amplified breast cancer, and to pertuzumab, an antibody inhibitor of ErbB2/ErbB3 heterodimerization [17]. In non-small-cell lung carcinoma (NSCLC) cells with mutated EGFR, amplification of the MET oncogene activates prosurvival ErbB3 signaling, which allows escape from tyrosine kinase inhibition [18] (for review, see [19]). Given these findings, downregulation of ErbB3 signaling in human cancers is critical to successful ErbB-targeted tyrosine kinase inhibitor therapy.

The importance of ErbBs in human breast cancer is linked to their importance in normal mammary gland biology. Understanding these normal functions and their regulators will lead to better prognostication and treatment decisions, and to identification of novel therapeutic targets. However, the individual role played by each receptor in mammary gland development has been difficult to determine. This is because the individual receptors and their agonists display complex and overlapping expression patterns, and because the receptors are required for multiple developmental activities, some of which are essential for prenatal viability [20, 21]. Both EGFR and ErbB2 are highly expressed and co-localized in all major cell types in the pubescent mouse mammary gland, but at maturity they are differentially localized, with EGFR in the stroma and ErbB2 in the epithelium [20]. ErbB3 and ErbB4 are only expressed at low levels in postpubescent mammary glands from virgin mice, but they are expressed at higher levels during pregnancy and lactation [20, 22]. All four of the ErbBs are expressed in the late pregnancy and early lactation glands. This differential expression pattern implies important roles for EGFR and ErbB2 in the developing mammary gland, and for ErbB3 and ErbB4 in the later stages of mammary gland development and differentiation.

Early studies suggested that only ErbBs EGFR and ErbB2 are important in early postnatal mammary development, because ErbB3 and ErbB4 are expressed at very low levels, and because neuregulins (NRGs), growth factors that bind exclusively to ErbB3 and ErbB4, were difficult to detect at early stages [20]. Nonetheless, there is evidence for ErbB3 or ErbB4 signaling in the pubescent mammary gland. Implantation of Elvax ethylene vinyl acetate (Elvax Ethylene, Dupont, Wilmington, Delaware, USA) slow release pellets containing NRG1, normally expressed during pregnancy, promoted ductal growth and alveologenesis in mammary glands of pubescent mice [23]. Because NRG1 binds ErbB3 and ErbB4 but not EGFR or ErbB2, this suggested that ErbB3 and/or ErbB4 are functional (but not necessarily active) at puberty. ErbB4 null mice, with the essential cardiac function supplied by transgene-driven expression of ErbB4 cDNA exclusively in the heart, survive past birth and are defective in lactation, but undergo normal development at puberty [24]. This indicates that ErbB4 is not essential during pubescent mammary gland development, and implies, by elimination, that ErbB3 is the active receptor in NRG1 implantation experiments.

Attempts to determine the role of ErbB3 in mammary gland development have been impeded by the embryonic lethality caused by ErbB3 gene disruption, which leads to cardiac and neurodevelopmental defects [25]. To determine the roles played by ErbB3 in mammary gland development, we chose to use a loss-of-function approach. Mammary buds isolated from ErbB3 null embryos at day 12.5 were transplanted into the cleared fat pad of prepubescent immunocompromised recipient mice. Transplanted ErbB3 null buds supported growth of an epithelial tree, but the ducts did not penetrate the fat pad to the same extent as ducts from a wild-type bud transplant. Also, the terminal end buds (TEBs) that pilot ductal extension through the mammary fat pad have an increase in apoptotic structures.

Analysis of ErbB3 function in postnatal development of the mammary gland has been hampered by the embryonic lethality of ErbB3 knockout mice. ErbB3^-/- null mice die in utero at day 13.5 or as late as days 16 to 18, depending on the strain background, because of cardiac cushion abnormalities [25]. However, postnatal development of ErbB3^-/- epithelium in ErbB3 wild-type stroma can be analyzed by transplantation. Mammary buds from day 12.5 ErbB3^-/- embryos were isolated and transplanted into prepubescent immunocompromised recipient mammary fat pads that had been surgically cleared of the endogenous epithelium. The contralateral gland of each recipient mouse was also cleared of epithelium and transplanted with an ErbB3 wild type (+/+) or heterozygous (+/-) mammary bud from a female littermate embryo as a positive control. At various times after transplantation, the transplanted glands were harvested from the recipient mice and the epithelial tree was visualized by carmine alum staining of whole mounted glands.

At 4 weeks after transplantation the epithelial trees resulting from ErbB3⁺ (+/+ or +/-) transplant had filled a majority of the fat pad (Figures 1a and 2a), whereas the ErbB^-/- transplant epithelial trees only filled a small fraction of the glands (Figures 1b and 2a). The total ductal length was measured for each of the paired contralateral gland outgrowths (four pairs). The length of the ErbB3⁺ outgrowths and the severity of the ErbB3^-/- penetration defect varied at 4 weeks (Figure 2a). The mean ErbB3⁺ outgrowth length was 412; the mean ErbB3^-/- outgrowth was 140. At seven weeks after transplantation, most of the ErbB3+ outgrowths (eight glands) had completely filled the fat pad (Figures 1c and 2b), but none of the ErbB3^-/- outgrowths (five glands) filled the fat pad (Figures 1d and 2b). The severity of the ErbB3^-/- ductal penetration defects varied. At 7 weeks, transplantation yielded two 'paired' glands from mice in which both ErbB3⁺ and ErbB3^-/- transplants were successful. For these pairs only, the mean ErbB3⁺ outgrowth length was 721; the mean ErbB3^-/- outgrowth was 528 (Figure 2b). For all of the glands assayed, including a number of 'unpaired' glands from mice in which only one of the two transplants survived, the mean ErbB3⁺ outgrowth length was 737; the mean ErbB3^-/- outgrowth was 406. This ductal penetration defect persisted in adult nonparous ErbB3^-/- glands (two glands; Figures 1f and 2e), and was not rescued by the influence of hormones during pregnancy or lactation (Figure 1h and 2e). ErbB3⁺ outgrowths (four glands) filled the fat pad and normal alveolar structures were evident at 1 day postpartum (Figures 1g and 2e). Even though the ErbB3^-/- outgrowth did not reach the edges of the fat pad, normal alveolar structures were present at 1 day postpartum (Figures 1h and 2e; four ErbB3⁺ glands and four ErbB3^-/- glands). On the histological level, the alveoli were distended, with milk in the lumens and lipid droplets in the cells, suggesting that both the differentiation and secretory pathways are normal in the +/+ and -/- glands (insets, Figure 1g,h). Analysis of lactation after 1 day postpartum is not possible because the epithelial outgrowth from the transplant is not attached to the nipple, so remodeling of the gland to the prepregnancy state occurs quite rapidly.

In addition to the ductal penetration defect, there was an increase in the number of branchpoints in the ErbB3^-/- outgrowths in comparison to the ErbB⁺ outgrowths (Figure 2c,d). Likewise, there was an increase in the number of TEBs in ErbB3^-/- glands (Figure 3a). TEBs are bulbous structures found at the invasive ends of an advancing ducts, and are major sites of proliferation and apoptosis in the pubescent mammary gland [29]. Normally, TEBs regress once the developing epithelial tree has reached the edge of the fat pad. At 4 weeks after transplantation, several large TEBs are evident in whole mounts of ErbB3⁺ outgrowths (Figure 3a, top left). however, by 7 weeks after transplantation (Figure 3a, middle left) the TEBs have mainly regressed and they are not present at 20 weeks after transplantation (Figure 3a, bottom left). TEBs are visible in ErbB3^-/- glands at 4 weeks (Figure 3a, top right) and at 7 weeks (Figure 3a, middle right). Interestingly, they have regressed by 20 weeks after transplantation (Figure 3a, bottom right), despite the incomplete extension through the fat pad. The number of TEBs visible in both carmine alum stained whole mounted glands (Figure 3b, top left) and across a series of sections from paraffin-embedded whole glands (Figure 3b, bottom right) was greater in ErbB3^-/- glands than in ErbB3⁺ glands. However the epithelial area of the TEBs was smaller for the ErbB3^-/- glands than for the ErbB3+ glands. This was observed in both the carmine alum stained whole mounts (Figure 3b, top right) and in the sections from the paraffin-embedded glands (Figure 3b, bottom right). TEBs from ErbB3⁺ glands were larger in size and epithelial area (Figure 3c, right) than TEBs from ErbB3^-/- glands (Figure 3c, left); the corresponding glands are marked with circles in Figure 3b (lower right). In the carmine alum analysis (three pairs), the mean area of the ErbB3⁺ TEBs was 631 whereas the mean area of the ErbB3^-/- TEBs was 221. In an analysis of the paraffin sections, the mean area of the ErbB3⁺ TEBs (six glands) was 668, whereas the mean area of the ErbB3^-/- TEBs (six glands) was 443. In analysis of only the paired paraffin sections (two pairs), the mean area of the ErbB3⁺ TEBs was 734 whereas the mean area of the ErbB3^-/- TEBs was 395.

A cross-section of a classical TEB reveals two distinct epithelial cell layers. The outer layer of cap cells at the advancing end of the buds merges into the more distal myoepithelial layer [30, 31]. Cap cells are active sites for proliferation in the TEB and the progenitors for the myoepithelial cells that line the duct and the interior body and luminal cells. The second layer encompasses the inner, multilayered body cells that line the lumen and are thought to be major sites of apoptosis in TEBs [32, 33]. Both of these cell types are apparent in ErbB3^+/+ TEBs at 4 weeks after transplantation (Figure 4a). The single layer of cap (and myoepithelial) cells is marked by expression of SMA and the adhesion molecule P-cadherin (Figure 4b,d), whereas multilayered body cells express the adhesion molecule E-cadherin (Figure 4c) [34, 35]. There is a large lumen surrounded by the body cells in each of the consecutive stained sections from the ErbB3^+/+ gland. The ErbB3^-/- TEBs have the same general organization, but they differ in the type and severity of the defect (Figure 4, bottom four rows). The first ErbB3^-/- TEB (Figure 4e–i) has a single layer of cap cells (Figure 4h) and a multilayered body cell compartment (Figure 4g); however, there are large vacuolated spaces between and the cap and body cell layers. The second ErbB3^-/- TEB (Figure 4j–n) also has some spaces between the cap and body cell layers, and the lumen is filled with cells. There is an influx of cells that express smooth muscle actin in the body cell compartment (Figure 4l) and do not express E-cadherin like the surrounding body cells (figure 4m). In a more severely defective ErbB3^-/- TEB (Figure 4o–s), there are fewer body cells. There is a single layer of cap cells surrounding a large space (Figure 4q), and few body cells (Figure 4p) that surround a small luminal space. Although the three 4-week TEBs in Figure 4 are from different mice, each of these three types of ErbB3^-/- TEB defects can be found in the same gland and in glands from different mice. The 7-week ErbB3^-/- TEB (Figure 4t–w) is disorganized in that there is a multilayered SMA-positive cap/myoepithelial cell compartment, with large spaces with the layers (Figure 4u). The normally multilayered body cells are a single layer surrounding a small lumen (Figure 4v).

The TEB is the major site of proliferation and apoptosis in the developing mammary gland. The proximal cap cells are actively proliferating cells, whereas the body cells that line the duct undergo apoptosis to clear the lumen during growth of the duct [32, 33]. A decrease in proliferation or an increase in apoptosis could result in the ductal penetration defect associated with ErbB3^-/- transplants. Histological analysis of the H&E stained ErbB3^-/- TEBs revealed fragmented condensed nuclei suggestive of apoptosis (Figure 5, top right) in comparison with ErbB3^+/+ TEBs (Figure 5, top left). Hence, we enumerated apoptotic bodies in fields of H&E sections from ErbB3⁺ and ErbB3^-/- mammary glands. Of the seven null glands analyzed, the mean number of apoptotic bodies was 2.3 (standard deviation [SD] 2.0). Six wild-type or heterozygous glands had a mean of 0.8 apoptotic bodies (SD 0.5). The counts were also compared after normalization to the total number of epithelial nuclei in each sample. Among all glands analyzed, ErbB3^-/- TEBs (seven glands) had a mean ratio of apoptotic bodies to nuclei of 0.03 (SD 0.03), whereas ErbB3⁺ glands (six glands) had a mean ratio of 0.01 (SD 0.01). In the four pairs only, ErbB3^-/- TEBs had a mean ratio of apoptotic cells to nuclei of 0.02 (SD 0.01), whereas ErbB3⁺ glands had a mean ratio of 0.01 (SD 0.01).

The area of the ErbB3^-/- TEB with the most apoptotic nuclei also displayed an increase in cleaved caspase 3 positive cells (Figure 5). On average, ErbB3^+/+ TEBs had 2.15% cleaved caspase 3 positive cells, with a range of 0.4% to 5.4% positive. ErbB3^-/- TEBs had an average of 6.3% cleaved caspase 3 positive cells, with a range of 1.6% to 18.5% positive. Overall, then, there was a trend for null TEBs to show more signs of apoptosis, but the differences may not rise to the level of statistical significance.

Proliferation was not affected by the loss of ErbB3. Immunohistological analysis of BrdU labeling yielded similar results for ErbB3⁺ TEBs (29% BrdU positive) and ErbB3^-/- TEBs (32% BrdU positive; Figure 5, lower panels).

Deletion of ErbB3 in the epithelium of the mouse mammary gland resulted in reduced ductal outgrowth that was evident at puberty, and endured through adolescence, pregnancy, and early lactation, establishing that ErbB3 is absolutely required in the epithelium during pubescent mammary gland development. This phenotype is more severe than that induced by knockout of any of the three other ErbBs in mammary epithelium, underscoring the importance of ErbB3 in mammary development. The postnatal requirement of ErbB3 in the stroma was not addressed in this study. This ductal outgrowth/penetration phenotype varied in severity. Lobuloalveolar development was not affected in ErbB3-null glands. Lack of epithelial ErbB3 affected the phenotype of TEBs. TEBs were more numerous but smaller in size in the ErbB3^-/- outgrowths. Apoptotic cells were more common within the TEBs in the null glands, but there was no apparent change in the proliferative cells.

In related experiments, Qu and coworkers [36] inhibited ErbB3 expression in the mammary gland by using Cre recombinase to invert a transcriptional suppressor integrated in an intron. Mammary glands, analyzed at 4 weeks and 8 weeks, had lower ductal density, fewer branches, and fewer TEBs. TEB morphology was not discussed, and neither was development at later time points. The limited analysis, the fact that the experimental approach involved reduction, not elimination, of ErbB3 expression, and other technical differences make it difficult to compare that report with the present study directly.

The pubescent epithelial outgrowth/penetration defect we obtained was accompanied by an increase in the branch density in ErbB3-null glands. Mammary branching is modulated by diverse factors, including members of the EGFR family (for review, see [37]). The interaction between the ErbB family and matrix metalloproteinases (MMPs), in particular, is known to be important in branching. The EGFR promotes activation of MMP-2 and MMP-14, which are important for lung branching and morphogenesis [38]. In the mammary gland, MMP-2 is important for ductal elongation whereas MMP-3 promotes side branching [39, 40]. Disruption of the MMP network by the lack of ErbB3 could result in decreased ability of the ducts to penetrate the surrounding stroma.

Despite the importance of branching, other abnormalities in the ErbB3-null glands are likely also to contribute to the outgrowth/penetration defect. TEBs are epithelial structures found at the end of a growing duct in the pubescent gland and are responsible for growth and penetration of the normal epithelial tree into the surrounding fat pad. TEBs are evident at the tips of the advancing ducts until the edge of the fat pad is reached, at which point they regress. In the ErbB3-null glands there was an increase in the number of TEBs, probably due to the increase in branching and consequently in the number of ducts, but a decrease in the overall size of the TEBs. On a histological level, the ErbB3^-/- TEBs had a range of structural abnormalities that could account for their small size (Figure 4). Some of the TEBs were relatively normal in organization, with an outer single layer of cap and myoepithelial cells marked by P-cadherin expression, an inner multilayered body cell compartment marked by E-cadherin expression, and a central cleared luminal space. A majority of the TEBs had some structural defect including large spaces between the cap/myoepithelial cell layer and the body cell layer, and/or filled in luminal spaces. The most severe TEB abnormalities were detected at 7 weeks after transplantation. The cap/myoepithelial cell compartment appeared to become multilayered and invaded or co-inhabited the body cell compartment, which itself was numerically diminished. This could occur through loss of normal compartmentalization boundaries, and even invasion of the inner layers by the outer layer, perhaps accompanying loss of body cells. Dysregulation of the P-cadherin or E-cadherin adhesion systems could account for some of these structural abnormalities, because they are important for maintaining the boundaries as is netrin/neogenin signaling. It is possible that changes in differentiation of progenitor cells alter the balance of these populations.

With the information available, the most likely source of the TEB aberrations is an increased rate of apoptosis. A change in the normal balance between proliferation and apoptosis in the TEBs of a pubertal mammary gland would affect the ability of the epithelial tree to fill the fat pad. The proliferative cap cells are typically located at the distal end of the TEB, whereas the apoptotic cells are located in the inner body cell layer surrounding the hollow lumen [32]. The pattern of apoptotic cells in the TEBs suggests that apoptosis is responsible for clearing of the lumen within the growing duct. In the ErbB3-null glands, proliferation was not affected, but apoptosis as detected by cleaved caspase 3 immunohistochemistry and the presence of apoptotic figures in H&E-stained sections was increased in comparison to ErbB3⁺ glands. ErbB3 has six binding sites for the p85 adaptor subunit of PI3K and is strongly linked to prosurvival signaling through the PI3K/Akt pathway. The uncoupling of ErbB3 from the PI3K/Akt pathway in ErbB3-null outgrowths could result in the increase in apoptosis detected in the TEBs. An increase in apoptosis would explain the large number of cells in the normally cleared luminal space at 4 weeks after transplantation and then the apparent decrease in body cells numbers in TEBs at 7 weeks in ErbB3-null glands.

The ErbB3-null pubescent mammary gland phenotype might not be predicted based on earlier studies showing only limited expression of ErbB3 in the pubescent gland. However, implantation of the NRG1 in the pubescent mammary gland stimulated epithelial growth [23]. Hence, one of the NRG receptors ErbB3 and/or ErbB4 can function at this stage of mammary development. Because loss of ErbB3 function, but not loss of ErbB4 function, compromised pubescent mammary development, it is likely that ErbB3 was activated by the implanted ligand. However, ErbB3 is devoid of kinase activity and requires a heterodimerization partner for activation. Among the potential heterodimerization partners EGFR, ErbB2, and ErbB4, only EGFR and ErbB2 are required for ductal outgrowth in the pubescent mammary gland. EGFR is expressed in both the epithelium and stroma, but it is only required in the stroma during pubescent mouse mammary gland development [41]. Hence, ErbB2 may be the unique co-receptor, but EGFR and ErbB4 may work redundantly with ErbB3.

The epithelial null phenotype for ErbB2 [26] most closely resembles the phenotype for ErbB3 null, consistent with the importance of ErbB2/ErbB3 signaling at pubescence. ErbB2-null mammary buds transplanted into wild-type fat pads develop into ductal tree that grossly resembles the ErbB3-null phenotype. However, in contrast to ErbB3 nulls, the defect was present in the pubescent gland, but was not evident in adult glands or pregnancy. ErbB2-null TEBs also had structural defects in the TEBs, characterized by a decrease in body cell number, an increase in the presence of cap/myoepithelial-like cells in the prelumenal compartment, and the presence of large luminal spaces. The similar requirement for ErbB2 and ErbB3 in the epithelium of the pubescent mammary gland strongly implies that the ErbB2/ErbB3 heterodimer is important in the epithelium at this stage in development.

The heterodimerization partner(s) dictates the range of ligands capable of activating ErbB3 in the pubescent mammary gland. Because ErbB2 does not bind soluble ligands, the ErbB2/ErbB3 heterodimer would need to be activated by ligand-bound ErbB3. However, because growth factors in this family are activated by proteolysis, further work will be required to determine the importance of these agonists in pubescent mammary development. Knockout studies have shown that NRG1 isoform α is required, but in adult maturation of the mammary gland [28].

If ErbB3 is heterodimerized with EGFR, then other ligands could be responsible for transactivation of ErbB3. Of the known EGFR ligands, amphiregulin has been shown to be important and upregulated during pubescent mammary gland development [42, 43]. With the mammary defects in amphiregulin-knockout mice, EGFR is probably signaling in the epithelium either as a homodimer or heterodimer with ErbB2 and/or ErbB3 in the pubescent mouse mammary gland.

ErbB family members are over-expressed because of amplification or have activating mutations in a variety of human cancers. Mutated EGFR has been found in a subset of NSCLCs, and ERBB2 is amplified and over-expressed in up to 30% of human breast cancers, which is associated with a poorer clinical outcome. In both contexts, the ErbBs are validated therapeutic targets. ErbB3 may be quite important in governing the signaling output of the activated ErbBs, initial response to therapies, and in the development of drug resistance. For example, some NSCLC patients with EGFR mutations with a good clinical response to EGFR inhibitors gefitinib or erlotinib initially, but later develop resistance. In one study, half of these patients developed a second activating mutation, whereas the other half had amplified MET, which encodes a receptor tyrosine kinase activated by hepatocyte growth factor [18] (for review, see [19]). The NSCLC tumors escaped tyrosine kinase inhibition because MET activates ErbB3 and the prosurvival PI3K kinase pathway. In human breast cancer cells that over-express ErbB2 and have also developed resistance to tyrosine kinase inhibitor therapy, there is active PI3K/Akt via ErbB3 signaling [15]. In summary, therapies that target ErbB tyrosine kinase activities, kinase inactive ErbB3 and the downstream PI3K/Akt signaling may evade inhibition. A better understanding of ErbB3 in normal mammary gland signaling will reveal the nature of endogenous regulators, and may be beneficial to development of future directed therapies.

The ErbB3^-/- mammary outgrowth defect was associated with a decrease in the size of the TEBs, and increases in branch density, in the number of terminal end buds, and in the number of luminal spaces. Proliferation rates were not affected by the lack of ErbB3, but there was a trend toward increased apoptosis in ErbB3^-/- TEBs. Hence, endogenous ErbB3 regulates morphogenesis of mammary epithelium, most likely through impact on cell survival and other ErbB3-regulated processes.