Introduction
Glandular epithelial cells, such as those in the mammary gland, are organized into secretory structures with an epithelial monolayer that surrounds a hollow lumen and has distinct morphological features, such as specialized cell–cell contacts and polarized distribution of organelles and membrane proteins. The apical–basal polarity of epithelial cells and the organization of the glandular architecture are often disrupted in glandular carcinomas. Although many studies have identified mechanisms by which oncogenes induce proliferation and/or cell survival in tumorigenesis, the mechanisms by which polarity and epithelial architecture are disrupted remain poorly understood. Recent studies, including those by Aranda et al.1 on page 1235 of this issue, have begun to address how at least one oncogene, the ErbB2 (also known as Neu and HER2) receptor tyrosine kinase, modulates epithelial polarity during oncogenesis1.
Studies in multiple model systems have identified a set of evolutionarily conserved protein complexes, including Dlg–Lgl–Scrib and Par6–Par3–atypical PKC (aPKC), that establish and maintain normal epithelial organization and function2. Interestingly, genetic studies in Drosophila have demonstrated that disruption of genes in the Dlg–Lgl–Scrib complex results in hyperplastic growth, suggesting that regulation of epithelial polarity may influence tumorigenesis3. In addition, mammalian Dlg and Scrib are targets of viral oncoproteins, such as HPV-E6, providing further support for this hypothesis4, 5, 6.
Mammary epithelial cells cultured in reconstituted basement membrane are able to form three-dimensional spheroids that resemble glandular structures7. Many of the oncogenes implicated in the development of glandular carcinomas can elicit phenotypes reminiscent of tumour morphology in this system. The ErbB2 oncogene induces hyperproliferation when activated by artificial homodimerization in the non-tumorigenic MCF-10A mammary epithelial cell line8. Oncogenic ErbB2 signalling in this system also leads to filling of the luminal space due to survival of the hyperproliferative cells within each acinus and to a disrupted, multi-acinar architecture (Fig. 1a, b). Interestingly, a multi-acinar phenotype is not induced by all oncogenes that promote hyperproliferation and cell survival; for example, coexpression of the proliferative oncogene cyclin D1 with the anti-apoptotic gene Bcl-2 leads to hyperproliferation and luminal filling but does not disrupt acinar architecture9. Thus, ErbB2 seems to use a specific signalling mechanism to disrupt acinar architecture. Although the function of the previously mentioned polarity complexes in establishing and maintaining polarity in developing epithelia has been well documented, a potential role for these complexes in oncogene-mediated transformation of epithelial cells has not been described. In this issue, Aranda et al.1 demonstrate that components of the Par6–Par3–aPKC complex are involved in ErbB2-mediated disruption of epithelial organization during oncogenesis. The authors show that forced dimerization of ErbB2, and its consequent activation, disrupts membrane polarity in kidney epithelial cells and leads to the formation of hyperproliferative, multi-acinar structures with filled lumens in mammary epithelial cells. In both cell lines, dimerization of ErbB2 causes dissociation of Par6 and aPKC from Par3, and their association with an ErbB2 receptor complex. Interestingly, in mammary epithelial cells with activated ErbB2, expression of a Par6 mutant that is unable to bind aPKC reverses the multi-acinar and cell-survival phenotypes without affecting hyperproliferation (Fig. 1c).
Figure 1: Effects of ErbB2 activation in MCF-10A acinar structures.
The interactions between ErbB2 and the Par6–Par3–aPKC complex and the consequences of these interactions on mammary epithelial acini are shown. (a) In normal growth-arrested structures, basement membrane proteins like laminin and collagen IV (red lines) are deposited on the basal surface of the outer acinar cells, nuclei (yellow ovals) are basally localized, Golgi (green ovals) are apically oriented and Par6, Par3 and aPKC are associated in a protein complex. Cells within the luminal space of the acinus do not produce matrix proteins and undergo apoptosis. (b) Dimerization of ErbB2 leads to the dissociation of Par6 and aPKC from Par3 and their association with the ErbB2 receptor complex. ErbB2 signalling leads to induction of cell proliferation, generation of multi-acinar structures, loss of Golgi orientation, deposition of basement membrane proteins around the surface of the inner acinar cells and filling of the luminal space due to survival of the inner acinar cells. (c) Expression of a Par6 mutant (Par6K19A) that is unable to bind to aPKC reverts both the multi-acinar and cell-survival phenotypes, but not hyperproliferation. (d) Loss of Scribble (Scrib) rescues the multi-acinar and inner cell survival phenotypes in the Par6K19A background.
Full size image (67 KB)A report published earlier this year by Guo et al. provided evidence that proliferation and polarity are regulated independently during ErbB2-induced oncogenesis10. The authors introduced a targeted deletion of the integrin 
4 signalling domain into a murine model of ErbB2-induced mammary carcinoma11. Loss of integrin signalling delayed the onset of mammary tumours, reduced tumour burden, and suppressed tumour progression and metastasis. Ex vivo analyses identified a 4–ErbB2 complex that enhanced activation of the transcription factors c-Jun and STAT3. Interestingly, c-Jun was required for the hyperproliferative effect of ErbB2, whereas STAT3 contributed to the disruption of epithelial adhesion and polarity, demonstrating that distinct signalling mechanisms are responsible for each phenotypic outcome of ErbB2 activation in these cells. Together with the work of Aranda et al.1, these data argue that disruption of polarity and hyperproliferation induced by ErbB2 involve distinct signalling pathways that cooperate to drive oncogenesis (Fig. 2). These studies raise questions as to whether integrin 4 functions in recruiting Par6–aPKC to ErbB2 and whether Par6 and STAT3 signalling are functionally coupled or are components of parallel pathways that regulate polarity.
Figure 2: Schematic representation of a model for ErbB2-induced oncogenesis.
Activation of ErbB2 leads to disruption of epithelial polarity, protection from cell death and hyperproliferation. This model depicts signalling proteins that have been implicated in these events and their potential relationships. Studies described in the text suggest that ErbB2-induced hyperproliferation can be functionally separated from polarity disruption. These studies raise the questions as to whether 4 integrin functions in recruiting Par6–aPKC to ErbB2, and whether Par6 and STAT3 signalling are functionally coupled or are components of parallel pathways that regulate polarity. They also raise the possibility that the anti-apoptotic activity of ErbB2 is at least partially a consequence of polarity disruption.
The two model systems used in these studies exhibit different features of polarity and ErbB2 induces distinct alterations in each model. In three-dimensional culture, the MCF-10A mammary epithelial acini develop an axis of polarity in which the basement membrane is deposited and organized at the basal surface of the outer acinar cells and the Golgi orient to the apical side; however, the outer cells do not develop a discrete apical surface or tight junctions. ErbB2 activation in these cells disrupts this axis of polarity1. In contrast, the mammary carcinoma cells used by Guo et al. do develop tight junctions and in ErbB2-induced tumours, both the axis of polarity and the membrane polarity are disrupted10. Differences in ErbB2 effects on polarity in these models raises the question as to which aspects of polarity are most relevant to tumour progression in vivo.
It was previously shown that the development of an axis of polarity in MCF-10A acini correlates with the creation of a dichotomy between the outer and inner cells that eventually leads to death of the inner cells9. ErbB2 disrupts this difference between the outer and inner cells as suggested by the lack of polarization of the outer acinar cells1 and by the induction of matrix deposition by the inner cells8. The idea that disruption of polarity is involved in the ErbB2-induced survival of inner cells is supported by the high rate of apoptosis of inner cells in ErbB2 acini expressing a Par6 mutant unable to bind aPKC, and by the observation that cell death can be rescued by downregulating the polarity regulator Scribble (Fig. 1c, d). Interestingly, Scribble also rescues the multi-acinar phenotype of these acini, providing further evidence that disruption of polarity proteins can contribute to the multi-acinar phenotype (Fig. 1d). However, the consequences of the loss of Scribble on polarity in these cells have not been determined. Taken together, these studies raise the possibility that the anti-apoptotic activity of ErbB2 is at least partially a consequence of disrupting acinar organization — that is, ErbB2 may prevent apoptosis within acinar structures by disrupting the dichotomy between the outer and inner acinar cells. It is intriguing that expression of an anti-apoptotic protein in the Par6 mutant background partially restores the cell survival and multi-acinar phenotypes, suggesting that protection from cell death may, in turn, regulate cell polarity.
More detailed analyses of the molecular basis for the recruitment of Par6–aPKC to the ErbB2 receptor will provide further insights into the function of this complex in ErbB2-mediated polarity disruption. Recently, it was demonstrated that Par6 interacts with another cell-surface receptor, the TGF receptor, and that this interaction affects TGF-induced epithelial–mesenchymal transition (EMT)12. In this case, Par6 serine phosphorylation by the TGF receptor leads to interaction of Par6 with the ubiquitin ligase Smurf and the subsequent Par6–Smurf-dependent degradation of RhoA, a GTPase required for the maintenance of apical–basal polarity and cell–cell junctions in epithelial cells. This TGF-induced Par6-dependent loss of junctional polarity is independent of other aspects of TGF-induced EMT, namely Smad-mediated transcription. As described here, ErbB2 recruitment of Par6 leads to an outcome distinct from that of TGF receptor association with Par6. Whereas the TGF receptor–Par6 interaction is associated with an EMT phenotype, the ErbB2–Par6 interaction leads to a specific disruption of polarity without loss of epithelial character. These differences in outcome are likely to be due to many factors, including receptor complex localization, other receptor-associated proteins or collaboration with other receptor pathways.
It will be interesting to explore whether the interaction of Par6–aPKC with ErbB2 is direct or indirect and whether phosphorylation is required. It is also unclear whether changes in subcellular localization of ErbB2 and/or Par6–aPKC function in the loss of polarity. Aranda et al.1 suggest that ErbB2-mediated redistribution of Par6–aPKC away from Par3 may be sufficient to disrupt polarity in this context. Alternatively, altered signalling downstream of aPKC may be responsible for the observed effects. In the light of their data showing that the interaction of aPKC with Par6 is required for epithelial disruption, the latter mechanism is more likely.
All the reports discussed here, and an additional study on PI(3)K signalling in mammary tumour cell lines13, provide evidence that defects in epithelial polarity can be separable from other aspects of oncogenic signalling, such as hyperproliferation (Fig. 2). This not only suggests that aberrant epithelial architecture may be a distinct temporal event in tumour progression, but raises the possibility that signalling molecules in polarity pathways may provide novel targets for therapeutic intervention14. The study by Aranda et al.1 also raises interesting questions about the role of polarity and polarity proteins in regulating cell death. Does disrupting polarity itself protect cells from apoptosis or do polarity proteins signal directly to the cell-death machinery (Fig. 2)? Furthermore, do Par6 and other polarity proteins interact with different cell-surface receptors and, if so, what are the consequences for cell polarity and cell death? Future studies using a variety of model systems will shed light on these questions and on the importance of epithelial polarity in tumorigenesis.
