Epithelial cells are polarized along their apical–basal axis. Much of the cellular machinery that goes into establishing and maintaining epithelial cell polarity is evolutionarily conserved. Model organisms, including the fruit fly, Drosophila melanogaster, are thus particularly useful for the study of cell polarity. Work in Drosophila has identified several important components of the polarity machinery and has also established the surprising existence of a secondary cell polarity pathway required only under conditions of energetic stress. This work has important implications for the understanding of human cancer. Most cancers are epithelial in origin, and the loss of cell polarity is a critical step towards malignancy. Thus a better understanding of how polarity is established and maintained in epithelial cells will help us to understand the process of malignant transformation and may lead to improved therapies. In the present chapter we discuss the current understanding of how epithelial cell polarity is regulated and the known associations between polarity factors and cancer.
Multicellularity requires the organization of cells into specialized tissues. The cells that make up epithelial tissue are polarized along their apicobasal axes, and this polarity is crucial to their function, which may be absorptive (as in the gut), secretory (as in the glands) or protective (as in the skin). In epithelial cells polarity is driven by mutually exclusive apical, lateral and basal cortical regions. These regions are defined in large part by two cell–cell junctions which act to separate them: the AJ (adherens junction), which joins cells together in an epithelial sheet, and the septate junction (in insects) or TJ (tight junction) (in vertebrates), which acts as a barrier to paracellular diffusion (Figure 1). Notably, the arrangement of these junctions typically differs between insect and vertebrate cells. In the former, the AJ usually localizes apically to (above) the septate junction. In vertebrates, the opposite is true (Figure 1).
Despite this key difference, the regulation of epithelial polarity is substantially conserved between flies and humans. In fact, much of our understanding of epithelial polarity derives from studies in Drosophila. This model system has a number of advantages over cell-culture-based work. Flies can be easily dissected for convincing imaging of cells within the context of a tissue. They are also genetically tractable, so that polarity factors may be identified and manipulated. Work in flies, Caenorhabditis elegans and other systems has defined a number of signalling modules acting at each cortical region to establish and maintain epithelial polarity. Recent work in Drosophila has also revealed the surprising finding that regulation of polarity requires an additional signalling pathway under conditions of cellular energy deprivation.
The factors that control epithelial polarity are under increasing scrutiny for their association with cancer. Approximately 80–90% of all human tumours are epithelial in origin . The progression of cancer to malignancy and metastasis is marked by the EMT (epithelial–mesenchymal transition), a change in cell architecture and behaviour characterized in large part by the loss of cell polarity. The loss or misregulation of cell polarity factors may then be a key event in tumour progression. Investigation of the low-energy polarity pathway is likely to prove particularly important for understanding malignancy, as tumour progression characteristically includes a period in which cancerous cells are deprived of energy. In the present chapter we review the factors that regulate polarity in both normal and low-energy conditions, and what has been learned so far about the relationship of these factors to cancer.
Apical signalling factors
The Crb (Crumbs) complex
The Crb complex is comprised of three key components, Crb, Sdt (Stardust) and PATJ [PALS1 (protein associated with lin seven 1)-associated TJ protein], as well as the less well-defined components Lin7, Moesin, Yurt and βH-Spectrin (Figure 2) . Crb is critical to establishment of the apical domain. The loss of Crb function prevents formation of the apical domain entirely, whereas Crb overexpression results in an expansion of the apical domain and corresponding loss of the lateral domain (reviewed in ). Recent work from the Laprise laboratory demonstrates that this effect relies on a balance of signalling between Crb and a signalling module consisting of PI3K (phosphoinositide 3-kinase) and the Rho-family GTPase Rac1 .
A mammalian orthologue of Crb, called Crb3, appears to play an analogous role to Drosophila Crb in the establishment and maintenance of epithelial cell polarity. Overexpression of Crb3 promotes expansion of the apical domain in vertebrate cells . An additional function of Crb3 in vertebrate cells is the organization of TJs [6,7].
Crb3 is implicated in tumour development and metastasis. A study by Karp et al.  demonstrated that immortalized murine epithelial cells selected for tumorigenicity when transplanted into mice lost expression of Crb. This loss was associated with phenotypic changes characteristic of EMT: disrupted polarity, failure to form TJs, and the loss of contact inhibition in vitro. Retrovirus-mediated re-introduction of Crb3 rescued these features and prevented metastasis . Although these findings suggest that Crb3 protects cells from transformation by ensuring that cell polarity is maintained, it is important to note that the tumour suppressive function of Crb3 might be attributed only in part to its role in regulating polarity. Work in Drosophila has demonstrated that Crb is a regulator of the Hippo-Yorkie pathway, which controls tissue size and is also implicated in cancer development [9,10].
aPKC (atypical protein kinase C), Cdc42 (cell division cycle 42) and PAR (PARtitioning defective)-6
Three additional factors, aPKC, Cdc42 and the scaffolding protein PAR-6, also act at the apical cortex to regulate epithelial polarity (Figure 2). Although earlier work suggested that these proteins participate in a complex with the polarity factor Bazooka (PAR-3 in mammals), this view is being refined. Increasing evidence derived from flies and mammalian cells indicates that they work together with Crb to regulate Bazooka and other factors both through phosphorylation and by physical association (reviewed in ). Crb itself is a substrate for aPKC and this phosphorylation is required for apical localization of the Crb complex .
Accumulated evidence links these proteins to cancer. In flies, the expression of constitutively active aPKC causes epithelial disorganization and overgrowth, suggestive of tumorigenesis [11,12]. aPKCι, one of two human orthologues of Drosophila aPKC, is frequently amplified and overexpressed in colon carcinomas and non-small-cell lung cancers, suggesting that overactive aPKC contributes to carcinogenesis in humans, as it does in flies [13⇓–15]. More classical oncogenes can also function, at least in part, by disrupting the regulation of the PAR-6–aPKC complex. Activated ErbB2 associates with PAR-6–aPKC to promote the loss of epithelial architecture and reduced apoptosis in an in vitro model for breast cancer . Furthermore, PAR-6 acts downstream of TGFβ (transforming growth factor β) receptors and in partnership with the ubiquitin ligase Smurf to induce degradation of the cytoskeleton regulator RhoA, thus in turn promoting EMT .
Bazooka is required to position the AJ, which separates the lateral from the apical domain (Figure 2) . The positioning of Bazooka requires an intricate system of regulation. Bazooka appears to cycle between a state in which it is bound to PAR-6 and a state in which it is not . Recent work has demonstrated that disassociation is promoted by two events . The first is competition with Crb, which can bind PAR-6 and prevent interaction with Bazooka. The second is phosphorylation of Bazooka by aPKC, which weakens the interaction between Bazooka and PAR-6. . Similarly, during cell polarization in the developing Drosophila embryo, phosphorylation of Bazooka by aPKC weakens a transient association of Bazooka with Sdt, allowing for the Crb complex to form .
The mammalian orthologue of Bazooka is PAR-3. It appears to have a similar role in organizing polarity, although with an important difference. Whereas Bazooka is required for generating the AJ in flies, PAR-3 is required for the formation of TJs in mammals . As discussed earlier, these two types of junction have different functions, but their positions within the cell are similar (Figure 1). Thus it seems that Bazooka and PAR-3 are required to position the most apical junctional complex, regardless of its function.
A potential role for PAR-3 in human cancer has been addressed only recently. Two studies have shown that PAR-3 is down-regulated or mutated in a number of human carcinomas and cell lines [21,22]. The rescue of PAR-3 function in two such cell lines both re-established TJs and slowed cell growth . Thus PAR-3 appears to be acting as a tumour suppressive factor.
Basolateral signalling factors
Basal to the AJ, along the basolateral cortex, a number of pathways act in opposition to the apical signalling complexes in defining polarity. The kinase PAR-1 regulates Bazooka by targeting it directly, consequently preventing the AJ from expanding basally along the lateral cortex (Figure 2) . As with the Crb complex, regulation of PAR-1 by aPKC is also important. In mammalian cells, phosphorylation of PAR-1 by aPKC is required for PAR-1 localization and activity . The regulation of PAR-1 during tumour development has not yet been studied.
The Scribble complex
Lgl (lethal giant larvae), Dlg (Discs large) and Scrib (Scribble) make up the Scribble complex of proteins, another important basolateral polarity pathway (Figure 2) . These proteins are required for formation of the septate junction (reviewed in ). As with PAR-1, Scribble complex proteins act to oppose apical polarity signalling. Lgl binds to the PAR-6–aPKC complex to inhibit its activity and cortical association, whereas aPKC phosphorylation of Lgl excludes the latter from the apical cortex [26,27].
Of the normal energy signalling complexes, the Scribble complex has potentially the clearest association with tumorigenic phenotypes in the fly. lgl, dlg and scrib are nTSGs (neoplastic tumour suppressor genes); loss-of-function in any of these proteins promotes both the loss of cell polarity and concomitant overproliferation, resulting in a phenotype reminiscent of human tumorigenesis. . When combined with overactive oncogenes, such as Ras, these tumours can become metastatic [28,29]. Likewise, the loss of human Scribble promotes an invasive phenotype in cells with overactive Ras .
The story is likely to be more complex in humans than in flies, as mammals have multiple orthologues of each Scribble complex member . However, the relevance of these factors to human cancer is underlined by a number of recent studies in mammalian systems. Decreased expression of the human orthologue of Lgl, Hugl-1, has been observed in multiple cancer types, with the degree of loss correlating with disease progression and metastasis [31⇓–33]. Likewise, human Scribble is commonly deregulated in mammary carcinomas, either at the level of expression or localization . Furthermore, both Scribble and Dlg-1 are known targets for viral oncoproteins, which may bind these proteins and disrupt their function (reviewed in ).
Yurt and the Coracle group
A second set of septate junction proteins, composed of Yurt and the Coracle group proteins Cora (Coracle), Neurexin IV and Na+/K+-ATPase is also involved in regulating epithelial cell polarity in Drosophila (Figure 2). Yurt appears to function in parallel with a module comprising the other three proteins, which, unlike Yurt, are required for septate junction formation [36,37]. Interestingly these proteins are required for polarity only during the organogenesis stage of embryo development, during which they act to oppose Crb signalling. In embryos doubly mutant for Yurt and a Coracle group protein, apical boundaries are stretched, as they are upon Crb overexpression . When Crb function is lost, mutation of Yurt and any one of the Coracle group genes rescues polarity .
The mammalian Lulu proteins (Lulu1/2) are candidate orthologues of Yurt. Recent work has demonstrated that these proteins localize basolaterally in epithelial cells, where they act to regulate cell shape . The expression and regulation of these proteins in cancer has not yet been examined.
The low-energy polarity pathway
LKB1 (liver kinase B1) and AMPK (AMP-activated kinase)
LKB1 and its target AMPK comprise a signalling module responsible for sensing low cellular energy, as indicated by a high concentration of cellular AMP, and act to inhibit processes that use up ATP and to activate processes that generate ATP. Under low-energy conditions, disruption of either AMPK or LKB1 function yields strong cell polarity defects in the follicle cell epithelium, a single layer of epithelial tissue that surrounds the developing oocyte . These defects are rescued by the expression of a constitutively active phospho-mimetic AMPKα in lkb1 mutants . Importantly, maintenance of cell polarity under normal-energy conditions is AMPK-independent; ampk mutant cells only exhibit epithelial polarity defects under starvation .
Cell polarity defects are also observed in ampk or lkb1 mutant Drosophila embryos, suggesting that at this stage of development the fly may be under low-energy conditions . In embryos, the myosin regulatory light chain [called Sqh (spaghetti-squash) in flies] is an important downstream mediator of LKB1-AMPK polarity signalling (Figure 3) . Conservation of this pathway in human cells has already been established; low-energy polarity signalling through LKB1-AMPK has been demonstrated in the human colon carcinoma cell line LS174T .
With regard to cancer, low-energy polarity is particularly worthy of attention. It has long been recognized that malignant cells derive their cellular energy from a much higher rate of glycolysis (a characteristic known as the Warburg effect) than do healthy cells. The transition to a high glycolytic state as tumour cells become malignant is probably initiated to compensate for a loss of available oxygen within the growing tumour and resultant depleted energy stores in the cell. At this transition point the low-energy polarity pathway may act as a crucial check against further transformation. The loss of this pathway would thus open the door to malignancy.
LKB1-AMPK signalling is specifically implicated in the suppression of epithelium-derived cancers. LKB1 is a well-known tumour suppressor in epithelial tissues. Mutation of LKB1 is associated with Peutz–Jeghers syndrome, a condition characterized by increased risk of epithelial cancers and gastrointestinal polyps, and LKB1 is also commonly mutated in non-small-cell lung cancers and cervical cancers [41,42]. The tumour suppressor function of LKB1 is thought to be mediated through AMPK .
Evidence connecting AMPK with cancer is provided by epidemiological studies of patients with T2DM (Type 2 diabetes mellitus). These patients are commonly treated with the drug metformin, which decreases available energy in the cell by acting as a mitochondrial poison. While T2DM patients demonstrate a significantly increased cancer risk, long-term treatment with metformin is associated with a decreased risk of cancer in an approximately dose-dependent manner . Both the anti-cancer and anti-diabetic effects of metformin are mediated by AMPK .
Importantly, the cancer-limiting effect of AMPK stimulation is not restricted to T2DM patients. Metformin acts to inhibit the proliferation of breast cancer cells in vitro , and metformin and other AMPK-stimulating agents delay tumour onset in cancer prone Pten+/− mice (PTEN is phosphatase and tensin homologue deleted on chromosome 10) . In contrast, inhibition of AMPK accelerates tumorigenesis in these same mice . Several studies have also demonstrated decreased cancer incidence in mice fed a calorie-restricted diet. Cumulatively, these findings suggest the possibility that low-energy-induced AMPK signalling helps to protect cells from transformation by ensuring that polarity is maintained.
Located at the basal cortex, Dg is one of several receptors that interact with components of the ECM (extracellular matrix). Under normal energy conditions, neither Dg nor its ECM ligand perlecan are required to maintain epithelial polarity, but both factors are required for polarity under energetic stress (Figure 3) . The relationship between these factors and the rest of the low-energy polarity pathway is complex. Although Dg is required for the phosphorylation of Sqh, this activity is independent of AMPK. Furthermore, the expression of phospho-mimetic Sqh is not sufficient to rescue the polarity phenotype of Dg-mutant cells. Instead Dg may be necessary to regulate the intracellular localization of Sqh. In starved cells Sqh localizes apically in the presence of Dg, but basally in its absence.
Dg is also implicated in cancer. Its expression is frequently decreased in a variety of tumour types, indicating that its function is lost during cell transformation . Furthermore, exogenous overexpression of Dg inhibits the tumorigenicity of transformed human breast epithelial cells . This evidence suggests that Dg acts as a protective factor against cancer, perhaps through its role in regulating low-energy polarity.
The establishment and maintenance of epithelial cell polarity presents a rich and intricate problem for study, and the fruit fly has proved an outstanding tool. Polarity factors, of which many are known already, continue to be identified in Drosophila and other organisms. Current work is focused on deciphering the complex molecular relationships between these factors and between the pathways in which they participate. These pathways can appear to act in opposition, redundantly, in a tissue-specific manner, or most intriguingly, in a manner dependent on the energy status of the cell.
The connection between cancer and polarity regulation is evidently important, but only beginning to be researched in depth. Malignant transformation is marked by the loss of cell polarity, and as discussed above, several polarity factors are known to be lost or mutated in certain tumours. To date these findings are largely correlative and merit further exploration. The regulation of polarity under low-energy conditions deserves particular scrutiny, as links between energy status and cancer are well established at the level of both the organism and the cell. Several important questions surround the low-energy polarity pathway: (i) how is Sqh, which makes up part of an actin motor protein, involved in regulating polarity? (ii) How does the ECM receptor Dg, which is required independently of AMPK for the phosphorylation of Sqh, relate to other members of the pathway? (iii) What additional factors are involved?
As the answers to these questions emerge we are likely to have an improved understanding of malignant transformation and how to address it. Therapeutic strategies aimed towards protecting epithelial cell polarity may prove useful in the treatment and prevention of cancer, and we will continue to owe the fruit fly for its important role in illuminating human biology and disease.
• The machinery that controls the establishment and maintenance of epithelial cell polarity is largely conserved among animals.
• Drosophila is thus a useful model system for the study of polarity regulation, and much of our understanding of polarity is derived from work in flies and other model organisms.
• A number of polarity pathways act at different regions of the cell cortex, often in opposition with one another.
• The loss of epithelial cell polarity is a hallmark event of malignant transformation.
• Several polarity factors have been shown to be lost or misregulated in diverse tumour types.
• Under energetic stress, signalling mediated by LKB1, AMPK, Dg and Sqh (myosin regulatory light chain) is required for the maintenance of epithelial cell polarity.
• Multiple lines of evidence connect low-energy states with cancer, suggesting that low-energy polarity signalling may be of particular importance to tumorigenesis.
- © The Authors Journal compilation © 2012 Biochemical Society