BIOLOGICAL OVERVIEW

Genghis khan (Gek) was isolated in a search for proteins that physically interact with Drosophila small GTPases Rac1 and Cdc42. The sought after protein was likely to be an effector of Drosophila Cdc42 because of its role in actin polymerization. The role of Cdc42 and other small GTPases (Rho and Rac) in cytoskeletal dynamics has remained something of a mystery. In mammalian fibroblasts these proteins regulate the actin polymerization that underlies the respective formation of stress fibers (Rho), lamellipodia (Rac), and filopodia (Cdc42). These proteins are also involved in the activation of kinase cascades and in the regulation of cell proliferation. Perturbations in the activity of these GTPases in yeast, flies, and mice result in the disruption of many biological processes, including yeast budding, axon and dendrite outgrowth, myoblast fusion, epithelial cell shapes, and tissue polarity establishment. Pak, a protein kinase effector for both Rac and Cdc42, functions in mitogen-activated protein kinase activation, but not in cytoskeleton regulation, suggesting that Pak is the effector for Rac and Cdc42 in the activation of the MAP kinase cascades. The effectors of Rac and Cdc42 for the regulation of actin polymerization have not yet been characterized (Luo, 1997 and references).

Mutants in gek are defective in oogenesis. Expression of constitutively active and dominant negative mutants of Drosophila Cdc42, as described in Murphy (1996), results in a defective cortical F-actin around nurse cells and free floating ring canal components, similar to what is observed in gek mutants. Egg chambers homozygous for gek mutations exhibit abnormal accumulation of F-actin and are defective in producing fertilized eggs. The F-actin rich ring canals between nurse cells and between the oocyte and the nurse cells are present in egg chambers mutant for gek. However, the cortical F-actin that surrounds the nurse cells appear abnormal, and ectopic F-actin blobs are frequently observed in the nurse cells and oocytes. The phenotypes caused by overexpressing dominant mutants of Cdc42, however, are more severe than those of gek mutants and are associated with a block in oogenesis at earlier stages of development, as compared with loss of function gek mutants. This raises the possibility that Gek is not the only effector for Cdc42 in regulating actin polymerization. Potential candidates for additional Cdc42 effectors include the recently identified WASP family of proteins (Luo, 1997).

RhoGAP19D inhibits Cdc42 laterally to control epithelial cell shape and prevent invasion

Cdc42-GTP is required for apical domain formation in epithelial cells, where it recruits and activates the Par-6-aPKC polarity complex, but how the activity of Cdc42 itself is restricted apically is unclear. This study used sequence analysis and 3D structural modeling to determine which Drosophila GTPase-activating proteins (GAPs) are likely to interact with Cdc42 and identified RhoGAP19D as the only high-probability Cdc42GAP required for polarity in the follicular epithelium. RhoGAP19D is recruited by α-catenin to lateral E-cadherin adhesion complexes, resulting in exclusion of active Cdc42 from the lateral domain. rhogap19d mutants therefore lead to lateral Cdc42 activity, which expands the apical domain through increased Par-6/aPKC activity and stimulates lateral contractility through the myosin light chain kinase, Genghis khan (MRCK). This causes buckling of the epithelium and invasion into the adjacent tissue, a phenotype resembling that of precancerous breast lesions. Thus, RhoGAP19D couples lateral cadherin adhesion to the apical localization of active Cdc42, thereby suppressing epithelial invasion (Fic, 2021).

The form and function of epithelial cells depends on their polarization into distinct apical, lateral, and basal domains by conserved polarity factors. This polarity is then maintained by mutual antagonism between apical polarity factors such as atypical PKC (aPKC) and lateral factors such as Lethal (2) giant larvae (Lgl) and Par-1. While many aspects of the polarity machinery are now well understood, it is still unclear how the apical domain is initiated and what role cell division control protein 42 (Cdc42) plays in this process. Cdc42 was identified for its role in establishing polarity in budding yeast, where it targets cell growth to the bud tip by polarizing the actin cytoskeleton and exocytosis toward a single site. It has subsequently been found to function in the establishment of cell polarity in multiple contexts. For example, Cdc42 recruits and activates the anterior PAR complex to polarize the anterior-posterior axis in the Caenorhabditis elegans zygote and the apical-basal axis during the asymmetric divisions of Drosophila neural stem cells. Cdc42 also plays an essential role in the apical-basal polarization of epithelial cells, where it is required for apical domain formation. Cdc42 is active when bound to GTP, which changes its conformation to allow it to bind downstream effector proteins that control the cytoskeleton and membrane trafficking. An important Cdc42 effector in epithelial cells is the Par-6-aPKC complex. Par-6 binds directly to the switch 1 region of Cdc42 GTP through its semi-CRIB domain (Cdc42 and Rac interactive binding). This induces a change in the conformation of Par-6 that allows it to bind to the C-terminus of another key apical polarity factor, the transmembrane protein Crumbs, which triggers the activation of aPKC's kinase activity. As a result, active aPKC is anchored to the apical membrane, where it phosphorylates and excludes lateral factors, such as Lgl, Par-1, and Bazooka (Baz). In addition to this direct role in apical-basal polarity, Cdc42 also regulates the organization and activity of the apical cytoskeleton through effectors such as neuronal Wiskott-Aldrich syndrome protein (N-WASP), which promotes actin polymerization, and myotonic dystrophy kinase-related Cdc42-binding kinase (MRCK; Genghis khan [Gek] in Drosophila), which phosphorylates the myosin regulatory light chain to activate contractility (Fic, 2021).

This crucial role of active Cdc42 in specifying the apical domain raises the question of how Cdc42-GTP itself is localized apically. In principle, this could involve activation by Cdc42 guanine nucleotide exchange factors (Cdc42GEFs) that are themselves apical or lateral inactivation by Cdc42GAPs. The Cdc42GEFs Tuba, intersectin 2, and Dbl3 have been implicated in activating Cdc42 in mammalian epithelia. Only Dbl3 localizes apical to tight junctions, however, as Tuba is cytoplasmic and enriched at tricellular junctions and intersectin 2 localizes to centrosomes. Thus, GEF activity may not be exclusively apical, suggesting that it is more important to inhibit Cdc42 laterally. Although nothing is known about the role of GAPs in restricting Cdc42 activity to the apical domain of epithelial cells, this mechanism plays an instructive role in establishing radial polarity in the blastomeres of the early C. elegans embryo. In this system, the Cdc42GAP PAC-1 is recruited by the cadherin adhesion complex to sites of cell-cell contact, thereby restricting active Cdc42 and its effector the Par-6-aPKC complex to the contact-free surface (Fic, 2021).

This study has analyzed the roles of Cdc42GAPs in epithelial polarity using the follicle cells that surround developing Drosophila egg chambers as a model system. By generating mutants in a number of candidate Cdc42GAPs, this study identified the Pac-1 orthologue, RhoGAP19D, as the GAP that restricts active Cdc42 to the apical domain. In the absence of RhoGAP19D, lateral Cdc42 activity leads to an expansion of the apical domain and a high frequency of epithelial invasion into the germline tissue, a phenotype that mimics the early steps of carcinoma formation (Fic, 2021).

In the absence of RhoGAP19D, both N-WASP and Gek are recruited to the lateral membrane, indicating that Cdc42 is ectopically activated there. This implies that RhoGAP19D is the major Cdc42GAP that represses Cdc42 laterally, because no other GAPs can compensate for its loss. This also suggests that the GEFs that activate Cdc42 are not restricted to the apical domain and can turn it on laterally once this repression is removed. This is consistent with the identification of multiple vertebrate GEFs with different localizations that contribute to apical Cdc42 activation. The current results therefore identify RhoGAP19D as a new lateral polarity factor. This leads to a revised network of polarity protein interactions in which RhoGAP19D functions as the third lateral factor that antagonizes the activity of apical factors, alongside Lgl, which inhibits aPKC, and Par-1, which excludes Baz/Par-3 (Fic, 2021).

The function of RhoGAP19D is very similar to that of its orthologue PAC-1, which inhibits Cdc42 at sites of cell contact in early C. elegans blastomeres to generate distinct apical and basolateral domains. Both RhoGAP19D and PAC-1 are recruited to the lateral domain by E-cadherin complexes, although the exact mechanism is slightly different. RhoGAP19D recruitment is strictly dependent on α-catenin, which links it through β-catenin to the E-cadherin cytoplasmic tail, whereas α-catenin (HMP-1) and p120-catenin (JAC-1) play partially redundant roles in recruiting PAC-1 to E-cadherin (HMR-1) in the worm. Nevertheless, in both cases, the recruitment of the Cdc42GAP translates the spatial cue provided by the localization of cadherin to sites of cell-cell contact into a polarity signal that distinguishes the lateral from the apical domain. Classic work on the establishment of polarity MDCK cells grown in suspension has revealed that the recruitment of cadherin (uvomorulin) to sites of cell-cell contact is the primary cue that drives the segregation of apical proteins from basolateral proteins. Furthermore, the expression of E-cadherin in unpolarized mesenchymal cells is sufficient to induce this segregation, although the mechanisms behind this process are only partially understood. The observation that RhoGAP19D directly links cadherin adhesion to the polarity system in epithelial cells extends the results of Klompstra (2015) in early blastomeres, strongly suggesting that PAC-1/RhoGAP19D plays an important role in the first steps in epithelial polarization (Fic, 2021).

Although PAC-1 and RhoGAP19D perform equivalent functions in early blastomeres and epithelial cells, there is one important difference between their mutant phenotypes. In pac-1 mutants, Par-6 and aPKC are mislocalized to the contacting surfaces of C. elegans blastomeres where Cdc42 is ectopically active. By contrast, Par-6 and aPKC are not mislocalized laterally in rhogap19d mutant Drosophila epithelial cells, even though lateral Cdc42-GTP does recruit two other Cdc42 effectors, N-WASP and Gek. Thus, lateral Cdc42 activity is sufficient to recruit Par-6/aPKC to the lateral domain in early blastomeres, but not in epithelial cells. Instead, it was observed that lateral Cdc42 activity in rhogap19d mutant follicle cells acts at a distance to expand the size of the apical domain. A likely explanation for this difference is the presence of Crumbs in epithelial cells. The interaction between Cdc42-GTP and Par-6 alters the conformation of Par-6 so that it can bind to Crumbs, which anchors the Par-6-aPKC complex to the apical membrane and activates aPKC's kinase activity. Although Par-6 presumably binds to Cdc42 laterally in rhogap19D mutants and undergoes the conformational change, it cannot be anchored laterally in the absence of Crumbs. This activated Par-6-aPKC complex can then diffuse until it is captured by Crumbs in the apical domain, thereby increasing apical aPKC activity, providing an explanation for why the apical domain expands in rhogap19d mutant cells. C. elegans has three Crumbs orthologues, but removal of all three simultaneously has no effect on viability or polarity. Thus, in contrast to Drosophila epithelial cells, C. elegans Crumbs proteins are not required for Par-6/aPKC localization and activation, suggesting that some other mechanism, such as Cdc42 binding, is sufficient to activate aPKC (Fic, 2021).

If the failure of active Cdc42 to recruit aPKC laterally in rhogap19d mutant cells is due to the absence of Crumbs in this region, there must be a mechanism to exclude Crumbs from the lateral domain. One proposed mechanism depends on Yurt (Moe and EPB41L5 in vertebrates), which is restricted to the lateral domain by aPKC and binds to Crumbs to antagonize its activity. However, no lateral recruitment of aPKC was observed in rhogap19d;yurt double-mutant cells. Thus, there must be some parallel mechanism that excludes Crumbs, Par-6, and/or aPKC from the lateral domain (Fic, 2021).

Although loss of RhoGAP19D only leads to a partial disruption of polarity, it causes the follicular epithelium to invade the adjacent germline tissue with 40% penetrance. This invasive behavior is not driven by an epithelial-to-mesenchymal transition, because the cells retain their apical adherens junctions and epithelial organization. Instead, the deformation of the epithelium seems to be driven by the combination of an increase in lateral contractility and an expansion of the apical domain, because reducing the dosage of Gek, which activates myosin II to drive the contractility, significantly reduces the frequency of this phenotype, as does halving the dosage of any of the apical polarity factors. The expansion of the apical domain makes the domain too long for the cells to adopt the lowest-energy conformation, giving them a tendency to become wedge shaped, which could drive the evagination. It is also possible that buckling of the epithelium contributes to invasion. Recent work has shown that epithelial monolayers under compressive stress and constrained by a rigid external scaffold have a tendency to buckle inward. The follicular cell layer is surrounded by an ECM that constrains the shape of the egg chamber and that should therefore resist expansion. In addition, the pulses of lateral contractility are likely to generate compressive stress because transiently reducing cell height while maintaining a constant volume will increase the cells' cross-sectional area, thereby exerting a pushing force on the neighboring cells. This compression coupled to the tendency to become wedge shaped due to apical expansion could therefore trigger the rare buckling events that initiate invasion. In support of this view, lateral contractility has been shown to drive the folding of the imaginal wing disc between the prospective hinge region and the pouch. This phenotype provides an example of how a partial disruption of polarity can induce cell shape changes that lead to major alterations in tissue morphogenesis (Fic, 2021).

The rhogap19d phenotype resembles the defects earliest observed in the development of ductal carcinoma in situ. In flat epithelial atypia (FEA), the ductal cells are still organized into an epithelial layer, but they display apical protrusions that are strongly labeled by the apical polarity factor Par-6. This suggests that the apical domain has expanded and bulges out of the cell, just as was observed in the rhogap19d mutant follicular cells. In the next stage, atypical ductal hyperplasia (ADH), the ductal cells start to invade the lumen of the duct while retaining aspects of normal apical-basal polarity. This again resembles the invasive phenotype of rhogap19d mutants, although overproliferation of the ductal cells probably also contributes to invasion in this case. Thus, these abnormalities, which can sometimes progress to ductal carcinoma in situ and breast cancer, mirror the effects of lateral Cdc42 activation. The RhoGAP19D human orthologues, ARHGAP21 and ARHGAP23, have been shown to bind directly to α-catenin and localize to cell-cell junctions. Furthermore, low expression of ARHGAP21 or ARHGAP23 correlates with reduced survival rates in several cancers of epithelial origin. It would therefore be interesting to determine whether these orthologues perform the same functions in epithelial polarity as RhoGAP19D and if their loss contributes to tumor development (Fic, 2021).

GENE STRUCTURE

PROTEIN STRUCTURE

Amino Acids - 1613

Structural Domains

Gek is a large protein. The N-terminus of Gek contains a predicted Ser/Thr kinase catalytic domain. This is followed by a large coiled-coil domain with sequence homology to myosin heavy chain, a Cys-rich domain similar to the phorbol ester/diacylglycerol binding domain of protein kinase C, and a plekstrin homology domain, which is found in many signaling molecules and used for protein-lipid interactions and the recruitment to the cell surface. Near the Gek C terminus resides the sequence that resembles a Cdc42/Rac interactive binding (CRIB) domain. Indeed, deletion of three residues in Gek, which correspond to three conserved residues of the CRIB domain, disrupts Gek's binding to Cdc42 (Luo, 1997)

date revised: 15 January 98

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.