|Home | About | Journals | Submit | Contact Us | Français|
Epithelial monolayers are major determinants of three-dimensional tissue organization and provide the structural foundation for the body plan and all of its component organs. Epithelial cells are connected by junctional complexes containing the cell adhesion molecule E-cadherin. These adherens junctions mediate stable cohesion between cells but must be actively reorganized to allow tissue remodeling during development. Recent studies demonstrate that junctional proteins are dynamically turned over at the cell surface, even in cells that do not appear to be moving. The redistribution of E-cadherin through spatially regulated endocytosis and exocytosis contributes to cell adhesion, cell polarity, and is also involved in cell rearrangement. Here we describe recent progress in understanding the roles of the vesicle transport machinery in regulating cell adhesion and junctional dynamics during epithelial morphogenesis in vivo.
Epithelial monolayers are major determinants of tissue structure that provide the basis for many aspects of tissue morphogenesis. The remodeling of epithelial tissues drives fundamental morphogenetic changes including elongation of the body axis [1,2], branching of the lung [3,4], and closure of the neural tube [5,6]. Epithelial tissues are characterized by strong intercellular adhesion that would seem to present a barrier to cell rearrangement. The primary mechanism of adhesion in developing epithelia is the trans-dimerization of E-cadherin molecules on adjacent cell surfaces, which assemble into junctional complexes through the association of the Ecadherin intracellular domain with β-catenin and α-catenin. Recent studies demonstrate that adherens junctions are continually turned over through the cellular vesicle transport machinery [7-9], a mechanism that could facilitate the dynamic reorganization of cell interactions during processes of cell division, cell death, and cell rearrangement. Here we focus on recent studies in Drosophila, where this unexpectedly dynamic process is required for stable adhesion as well as for cell polarity and rearrangement during epithelial morphogenesis.
The cellular vesicle transport machinery delivers E-cadherin and other transmembrane proteins from sites of protein synthesis in the endoplasmic reticulum to specific destinations at the cell surface . Vesicle trafficking can also remove E-cadherin from the plasma membrane by endocytosis [7-9]. The internalization of surface E-cadherin into early endosomes requires the vesicle scission factor dynamin and the Rab5 GTPase. E-cadherin in the early endosomes is then either routed to the lysosome for degradation or to recycling endosomes in a process that requires the Rab11 GTPase. E-cadherin in recycling endosomes is returned to the cell surface by fusion of exocytic vesicles with the plasma membrane (Figure 1A).
Several lines of evidence demonstrate that E-cadherin is dynamically trafficked to and from the plasma membrane, even in the absence of obvious changes in cell shape or cell interactions. When E-cadherin is selectively labeled at the surface of epithelial cells in culture, labeled protein becomes detectable inside the cell, indicating that E-cadherin protein is internalized from the cell surface . After removal of the surface label and further incubation, labeled E-cadherin is detected on the cell surface, demonstrating that internalized protein is recycled back to the plasma membrane. These results have been extended in vivo, where imaging experiments in epithelial cells of the Drosophila pupal notum demonstrate that labeled E-cadherin at the basolateral cell surface is internalized and recycled to the apical membrane (Figure 1B) . Disruption of E-cadherin trafficking results in an irregular distribution of adherens junctions and ultimately leads to a loss of epithelial adhesion [12-16]. Together, these results indicate that E-cadherin turnover is required to maintain a dynamic and functional population of E-cadherin at the plasma membrane.
What are the mechanisms that direct E-cadherin trafficking? Two classes of spatial inputs have been shown to affect this process. First, the subcellular distribution of vesicular compartments correlates with sites of active protein turnover in some tissues (Figure 2D,E) [13,17-20], suggesting that the localization of vesicular compartments can direct protein delivery to specific regions at the cell surface. Second, localized molecules can target endocytosis and exocytosis to specific surface domains (Figure 2A–C). For example, the conserved multi-protein exocyst complex is thought to provide a spatial landmark on the plasma membrane for the fusion of exocytic vesicles from the Golgi and the recycling endosome [21,22]. Exocyst proteins mark sites of active membrane and protein delivery, including the growing bud tip in S. cerevisiae [23,24], intercellular junctions in mammalian epithelial cells [25,26] and the apical rhabdomere in Drosophila photoreceptors . In Drosophila mutants defective for specific subunits of the exocyst complex, E-cadherin becomes trapped in recycling endosomes in the pupal notum  and embryo , suggesting that the exocyst is required for the delivery of Ecadherin to the cell surface (Figure 2C). E-cadherin's cytoplasmic binding partner β-catenin/Armadillo directly interacts with Sec10 , the exocyst subunit which, together with Sec15, mediates the tethering of exocytic vesicles to the exocyst complex . This interaction could promote the trafficking of E-cadherin by linking vesicles containing E-cadherin and β-catenin to the exocyst to promote their fusion with the plasma membrane. Alternatively, this interaction could recruit the exocyst to sites of β-catenin localization at the cell surface, targeting membrane trafficking more generally to the region of the adherens junctions.
In addition to the spatial regulation of E-cadherin exocytosis, endocytosis is also spatially regulated by the PDZ domain protein Par-6 and the atypical protein kinase C (aPKC), two evolutionarily conserved regulators of cell polarity. Par-6 and aPKC assemble into a complex and are recruited to the apical cell cortex through an interaction between Par-6 and the active GTP-bound form of the Cdc42 GTPase . Apical aPKC phosphorylates and inactivates basolateral proteins, which contributes to the segregation of the plasma membrane into complementary apical and basolateral domains [29,30]. In addition to this role, recent studies demonstrate that the Par-6–aPKC–Cdc42 complex regulates E-cadherin endocytosis in the Drosophila pupal epithelium (Figure 2B) [14,15]. Disruption of endocytosis in mutants for shibire, the Drosophila dynamin homolog , prevents the release of endocytic vesicles into the cytoplasm. Live tracking of the fate of surface E-cadherin in dynamin mutant cells reveals that E-cadherin is trapped in aborted endocytic intermediates [14,15]. Mutants for par-6, aPKC, or cdc42 phenocopy these defects in E-cadherin endocytosis, and direct trafficking assays demonstrate that endocytosis originates from the apical junctions [14,15]. Endocytosis can also occur at the basal surface in these cells (Figure 1B) , indicating that there are other spatial inputs into this process.
Cdc42 is known to bind and activate WASp, which promotes actin nucleation by the Arp2/3 complex , and the Arp2/3 complex has been implicated in endocytosis [33,34]. Consistent with these interactions, Wasp and Arp3 are required for E-cadherin endocytosis in the Drosophila pupal notum [14,15]. The Cdc42-interacting protein Cip4  associates with WASp and dynamin and is also required for E-cadherin endocytosis . Taken together, these data suggest a model in which Cdc42 recruits dynamin, Wasp and the Arp2/3 complex to adherens junctions to promote E-cadherin endocytosis.
Although these studies identify several key players in E-cadherin trafficking, gaps in our understanding remain. For example, Cdc42 is required for the apical localization of Par-6 and aPKC [14-16,28], raising the question of how Cdc42 activity is spatially regulated to target endocytosis to the apical cell surface. In the early C. elegans embryo, the Cdc42 GTPase activating protein PAC-1 negatively regulates Cdc42 at contacting cell surfaces, restricting the Par-6–aPKC–Cdc42 complex to the outer, non-contacting cell surfaces . Another issue is that expression of dominant-negative Cdc42 stimulates endocytosis in the Drosophila embryo, as judged from an elevated rate of uptake of a fluorescent dye . This result stands in contrast to the cdc42 mutant phenotype in pupae, and raises the possibility of tissue-specific differences in the role of the Par-6–aPKC–Cdc42 complex in endocytosis.
Mutations in exocyst proteins or the Par-6–aPKC–Cdc42 complex cause a breakdown of adherens junctions, which is at least in part attributable to a disruption of E-cadherin trafficking [12-16]. However, it is important to consider that many transmembrane proteins in addition to Ecadherin are transported through the vesicle trafficking machinery, some of which could stabilize adherens junctions indirectly, such as the apical Crumbs protein [13,16,37]. In addition, aPKC phosphorylates and inhibits the endocytic protein Numb [38,39], which regulates cadherin trafficking and adhesion in neuroepithelial cells . Moreover, aPKC also regulates adherens junctions by phosphorylating Crumbs and the basolateral proteins Lethal giant larvae (Lgl) and Par-1 . Unlike Numb, these substrates have not been implicated in cadherin trafficking, suggesting that the Par-6-aPKC complex may regulate adherens junctions through trafficking-dependent and independent mechanisms.
While membrane trafficking has primarily been characterized for its role in promoting adhesion in nonmoving cells, there are hints that junctional trafficking may also play an active role in cell rearrangement. The prepupal Drosophila wing disc is a highly disordered epithelium, with each cell contacting a variable number of neighbors. During pupal development this epithelium reorganizes to produce a more regular hexagonal lattice through the local remodeling of adherens junctions . Disruption of endocytosis in dynamin mutants blocks junctional remodeling and leads to gaps in surface E-cadherin, indicating that endocytosis is required to maintain epithelial adhesion in the Drosophila wing . Localized exocytosis may also contribute to the redistribution of adherens junctions, as the exocyst subunit Sec5 is enriched at proximal and distal cell boundaries and its localization is regulated by Flamingo, a planar cell polarity protein . These results suggest a model in which Flamingo recruits the exocyst to target E-cadherin recycling to specific cell boundaries as a potential mechanism for epithelial polarization or remodeling (Figure 1E).
The regulated trafficking of E-cadherin can also promote morphological changes at the tissue level during epithelial branch formation in the Drosophila trachea [19,42]. Measurements of junctional turnover using fluorescence recovery after photobleaching (FRAP) reveal that the turnover of α-catenin is lower in dorsal branch cells that intercalate during branch extension and higher in the dorsal trunk cells that do not . This tissue-level pattern of protein trafficking is specified by the trunk-specific transcription factor Spalt and its target dRip11, a positive regulator of Rab11 [19,43]. Increased trafficking through recycling endosomes is associated with higher levels of surface E-cadherin in the tracheal trunk cells (Figure 2D,E), consistent with the idea that exocytosis promotes adhesion. However, in contrast to other epithelial tissues in Drosophila, endocytosis in the trachea appears to weaken rather than strengthen adhesion. Disrupting endocytosis by expressing dominant-negative Rab5 or in mutants for dynamin leads to an increase in surface E-cadherin and the stabilization of cell–cell contacts . These results are consistent with a simple model in which endocytosis downregulates adhesion by removing E-cadherin from the cell surface (Figure 1D).
Defects in endocytosis are associated with a failure of cell intercalation in the tracheal branches, while defects in exocytosis lead to ectopic intercalation in the tracheal dorsal trunk . These results suggest that increased Rab11-dependent recycling of E-cadherin strengthens adhesion in the tracheal trunk (Figure 1C), allowing these cells to oppose the external forces that drive intercalation. E-cadherin trafficking in the Drosophila trachea may therefore act as a switch to permit or prevent cell rearrangement. Junctional trafficking could represent a general mechanism regulating cell interactions during morphogenesis, as trafficking upregulates adhesion to allow migration of the zebrafish prechordal plate  and organization of the mammalian cerebral cortex .
Adherens junctions are stabilized by several mechanisms in addition to protein trafficking, including the clustering of E-cadherin into adhesive microdomains and interactions with the actin cytoskeleton [45,46]. When epithelial cells first make contact in culture, E-cadherin accumulates in punctate microdomains at the contact site [47,48] along with a burst of F-actin that has been observed both in culture  and in vivo . FRAP and GFP tracking experiments in culture demonstrate that E-cadherin is highly immobile within microdomains, but each microdomain displays a remarkable mobility along the nascent interface . Activation of myosin II at the edges of growing contacts translocates E-cadherin microdomains from the center of the interface to the periphery, a mechanism that may expand the contact area between cells .
E-cadherin puncta are also observed in the Drosophila embryonic epithelium [52,53]. Recent work showed that, as in culture, these microdomains are relatively stable. Approximately 35–50% of E-cadherin protein is dynamically turned over within microdomains in the Drosophila epithelium, compared to 55–70% of the E-cadherin outside of microdomains , a measurement that reflects the contribution of both trafficking and diffusion. Both populations are more dynamic than has been reported for junctional proteins in the trachea, in which only 20–30% of α-catenin protein is mobile [19,42]. Although E-cadherin microdomains are largely immobile in wild-type embryos, disruption of actin or RNAi knockdown of α-catenin causes microdomains to move in lateral and basal directions , indicating that the actin cytoskeleton anchors E-cadherin microdomains at fixed locations on the cell cortex. An intriguing possibility is that microdomain mobility may be actively regulated during cell rearrangement, but further studies are required to determine whether these structures move in active cell populations in vivo.
A common theme emerging from in vivo studies of adherens junction regulation is that adherens junction proteins are dynamically turned over at the cell surface, even in epithelial tissues where adhesion is stable. Paradoxically, junctional turnover often appears to promote adhesion, as both endocytosis and exocytosis are necessary to maintain stable adhesion between cells in several Drosophila epithelia [12-16]. How does the trafficking of junctional proteins contribute to junctional stability? One possibility is that E-cadherin trafficking to the apical surface may offset its diffusion along the apical-basal axis , a mechanism that would maintain the apical localization of adherens junctions. Alternatively, junctional proteins may have a limited window of activity and require turnover to maintain an active population at the cell surface. In this model, passage of E-cadherin through the vesicular trafficking machinery could allow signals or post-translational modifications that restore its adhesive activity. A third possibility is that the mechanical forces generated by membrane dynamics could stabilize junctional complexes, reminiscent of the role of tension in stabilizing focal adhesions [56,57].
A dynamic junctional network also provides an opportunity to reorganize interactions between cells during development. Epithelia experience cell proliferation and turnover even in the absence of morphogenetic change, raising the possibility that junctional turnover permits the remodeling of adherens junctions as cells divide or are lost from the epithelial sheet. Patterned E-cadherin trafficking at the tissue level can also influence tissue morphogenesis by restricting cell rearrangement to specific regions, such as the dorsal branch cells in the Drosophila trachea. In this tissue, the directionality of cell rearrangement is determined by external forces from migrating cells at the branch tip [58,59]. By contrast, cell intercalation during axis elongation in the Drosophila embryo is driven by forces that are intrinsically generated by localized actomyosin contractility [60-62]. Adherens junctions in this intercalating population are planar polarized, with reduced E-cadherin and β-catenin at cell contacts that are disassembled during elongation . Further studies are required to determine whether polarized trafficking plays an instructive role in cell rearrangement.
Cell biological approaches have provided insight into the regulation of adherens junctions in cultured cells and inspired a number of hypotheses about how adherens junction regulation influences the establishment and reorganization of cell contacts. Recent work in Drosophila illustrates the power of genetic approaches in extending these findings to reveal unexpectedly dynamic junctional regulation in vivo. Testing the above models directly will require new tools to probe the composition, functional state, and subcellular trajectories of junctional complexes in real time using photoswitchable proteins and live sensors for protein interactions . A combination of live imaging with classical genetic and biochemical approaches will provide an opportunity to uncover how molecular interactions within junctional complexes are dynamically regulated in space and time to effect morphogenetic changes at the tissue level.
We are grateful to Buzz Baum and members of the Zallen lab for comments on the manuscript, and Richard Zallen for suggestions on Figure 1. F.W.P. is supported by a long-term fellowship of the Human Frontiers Science Program. J.A.Z. is supported by a Burroughs Wellcome Career Award in the Biomedical Sciences, a Searle Scholar award, a W. M. Keck Foundation Distinguished Young Scholar in Medical Research award, and NIH/NIGMS R01 GM079340.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.