The formation of the first epithelium was an essential step for animal evolution, since it has allowed coordination of the behavior of a cell layer and creation of a selective barrier between the internal medium and the outside world. The possibility of coupling the cells in a single layer has allowed morphogenetic events, such as tube formation, or gastrulation, to form more complex animal morphologies. The invention of sealed junctions between cells has allowed, on the other hand, creation of an asymmetry of nutrients or salts between the apical and the basal side of the epithelial layer. Creation of an internal medium has led to homeostasis, allowing the evolution of more complex physiological functions and the emergence of sophisticated animal shapes. During evolution, the origins of the first animals coincided with the invention of several protein complexes, including true cadherins and the polarity protein complexes. How these complexes regulate formation of the apicolateral border and the adherens junctions is still not fully understood. This review focuses on the role of these apical polarity complexes and, in particular, the Crumbs complex, which is essential for proper organization of epithelial layers from Drosophila to humans.
- epithelial cells
- polarity complexes
for about a billion years, life on Earth was characterized by the primacy of unicellular organisms, first represented by bacteria and Archaea. The emergence of Eukaryota some 2 billion years ago was the first step toward animal evolution, with the possibility of further developing multicellularity. Multicellularity has been acquired by several different eukaryote phyla (e.g., algae, filamentous mushrooms, Dictyostelium, and choanoflagellates), allowing cells to cooperate more closely than in populations. It seems that the critical step for the building of more complex animal organisms came with the advent of epithelial layers. This review addresses the role of the Crumbs protein complex, which is essential for the organization of epithelial layers and animal development. How this complex and other protein complexes have emerged as essential regulators of epithelial cell morphogenesis remains unknown and is the next challenge in epithelial biology.
Epithelia are layers of cells exhibiting concerted polarity; they are connected by cell-cell junctions based on proteins of the cadherin family and rest on a basement membrane that is also unique to Metazoa (29). This polarity is evidenced by the existence of an apical surface usually facing the external medium and a basolateral surface in contact with other cells, the basal lamina, and the internal medium (for review see Ref. 64). All Eumetazoa, including some sponges, have epithelial layers (36, 41), indicating that the metazoan ancestor also possessed epithelial cells. No bona fide epithelial layer can be observed in eukaryotes before the metazoan lineage, and even plants and fungi, which develop as large multicellular organisms, rely on external cell walls and direct cell-cell connections to build their cell layers. However, precedents for preepithelial organization among Protozoa such as Dictyostelium (12) indicate that acquisition of epithelial features and multicellular transitions was a gradual process (67).
Thus the invention of cadherin-based epithelia can be traced back about 680 million years and has allowed the blooming of all metazoan shapes and species. This invention was also a major step toward complex organism homeostasis, since one of the major roles of epithelia, besides coordination of cell movement, is to provide a selective barrier between the external world and the rest of the body (for review see Ref. 64). Epithelial cells are able to regulate the flux of water, salts, and nutrients across their layers and have acquired specialized membrane-bound enzymes and transporters to digest extracellular nutrients and import them to the internal medium, allowing the specialization of different cell types and tissues. This specialization has also offered protection from a changing environment, allowing extrapolation of the concept of homeostasis to the organism level, and not just to single cells or cell populations in a liquid environment. Cell specialization has also led to controlled mobility with the appearance of muscle and nerve cells.
One key feature of epithelial cells is the existence of cell-cell junctions, which allowed a mechanical bond between cells and the regulation of barrier function between the outside world and the inside of the body (for review see Ref. 7). All these junctions are connected to the intracellular cytoskeleton, composed of actomyosin networks, microtubules, or intermediate filaments (63). While several types of junctions can be observed in the metazoan lineage, all these junctions provide signaling functions to inform and coordinate the cells within the same layer to allow collective movement during morphogenesis (23, 44), control of organ size (5, 86), and proliferation (47). In addition, in most Metazoa, some specialized junctions regulate the paracellular flux of water and larger molecules to allow a more rapid or more selective uptake of nutrients, as well as to protect from mixing of the external and internal media. Thus the different junctional complexes that have been developed by epithelial cells play a central role in tissue homeostasis, in terms of physiology and overall size and shape.
Origins of Epithelia
Evolution of junctions.
The organization of eukaryotic cells into epithelial layers is a common feature of all Eumetazoa, and even sponges possess epithelial cells. Epithelial cells are characterized by the presence of an apical and a basolateral domain, separated by a set of apicolateral junctions, and these traits are only observed together in the animal phyla that are thought to have been derived from a common ancestor about 680 million years ago. This hypothesis is reinforced by the fact that bona fide classical cadherin proteins connected intracellularly to the actin cytoskeleton via a β-catenin binding domain are present only in this phylum (14, 52). However, it is remarkable that nonclassical cadherin genes predate the metazoan lineage, since they can be found in choanoflagellates (52). Classical cadherins bind to β-catenins, and the proteins encoded by cadherin genes in at least two sponges, Oscarella carmela and Amphimedon queenslandica, have predicted cytoplasmic domains capable of also binding β-catenin (14, 52). Thus it is very likely that cadherin-catenin complexes have played a crucial role in the development of animal multicellularity and epithelialization. It is noteworthy, however, that, in a clade that separated before the origins of Metazoa, at least one Amoebozoa, Dictyostelium discoideum, exhibits, upon starvation, a developmental stage with a fruiting body comprising a stalk and spores and that stalk formation involves generation of a pseudoepithelium (21). In the layer of cells surrounding the stalk tube, there is an asymmetric distribution of cellulose synthase, a transmembrane protein, and knockout of a β-catenin homolog (Aardvark) or knockdown of the homolog of α-catenin leads to mislocalization of the cellulose synthase (12). The lack of either catenin homolog does not, however, disrupt the junctions, which can be observed by electron microscopy, indicating that these junctions are not based on the metazoan cadherin-catenin-F-actin complex (12). Previous ultrastructural observations showed, however, that these junctions are similar to adherens junctions (AJs) (21). Thus, in D. discoideum, it is possible to observe an attempt at the emergence of epithelial organization that predates cadherin-catenin AJs, and the fact that this epithelium-like tissue is only a transient stage might have prevented it from evolving further. Future identification of the exact nature of the junctions in this organism will help us understand how epithelial junctions might have appeared independently of Metazoa.
Thus typical AJs are found only in Metazoa and are characterized by dense plaques at cell-cell contacts made of F-actin and catenins, with the cadherins forming the intercellular bonds (for review see Refs. 25 and 35). On the basis of these minimal requirements and ultrastructural description, it can be said that, from Porifera (sponges) to mammals, there is a highly conserved organization of AJs in the presence of classical cadherins, α- and β-catenins, and p120 catenins (14, 52). This suggests that a common ancestor (for review see Ref. 53) already had this molecular toolkit to build AJs after the likely separation from the choanoflagellates. The fact, however, that α- and β-catenin homologs can be identified in D. discoideum without having a role in the building of junctions (12) suggests that it was the acquisition of classical cadherins that formed the link between cell-cell contacts and the F-actin cytoskeleton by recruiting preexisting α- and β-catenin. The question that remains is how and when the classical cadherin cytoplasmic domain that can bind to β-catenin was added to the extracellular cadherin domain. Alternative splicing or domain shuffling could explain the emergence of a new type of cadherin, and further genomic analysis will likely provide the answer.
Besides AJs, epithelial cells possess other types of junctions, and one of the characteristics of most epithelia is the ability to form seals. Two types of related junctions, tight junctions (TJs) and septate junctions (SJs), are essential for this sealing. SJs, which are found in invertebrate epithelia, are characterized by a ladderlike structure by electron microscopy (for review see Ref. 7). Even if ultrastructural images of ladderlike junctions can be observed in various sponges (for review see Ref. 41), their molecular nature remains enigmatic. In particular, the existence of well-known markers of insect SJs, such as neurexins, has not been successfully demonstrated, indicating that these proteins might not be present in Porifera or their molecular organization is too different to be detected by homology (14). In Drosophila melanogaster, for example, the early embryos have AJs, whereas SJs develop later (35), suggesting that acquisition of SJs is linked to more complex physiological functions. TJs are also able to seal epithelial layers, providing for regulation of nutrient and water uptake essential for internal medium homeostasis; until now, these properties have been restricted to vertebrates (for review see Ref. 7). TJs are made of a family of transmembrane proteins, with four membrane-spanning domains called claudins, whereas no obvious homologs have been found before bilaterians, even though a protein from the claudin superfamily has been detected in a Porifera, A. queenslandica (14). Thus, TJ proteins might have predated the appearance of the bona fide structure observed in vertebrates, but some ultrastructural observations in different sponge species suggest that related junctions might already have occurred in Porifera (41). Despite the presence of molecular markers for AJs and, potentially, for TJs and clear epithelial organization, in homoscleromorph sponges, for example, there is no evidence that these tissues provide physiological functions such as the generation of an internal medium, as seen in bilaterians. The establishment of new sponge models amenable to in vivo and in vitro studies will help address this issue.
Epithelia also have desmosomes and gap junctions, which provide mechanical and electrochemical coupling, respectively, in the epithelial layer. However, these structures are not restricted to epithelia in vertebrates and appear to be events secondary to the advent of epithelia. Thus these cellular structures are not discussed further, and recent reviews on the evolution of eukaryote cell-cell communications have shown that plants, fungi, and animals have devised independent ways to connect cells directly (4).
Thus the formation of the first epithelial junctions is based on cadherin-catenin complexes, which are present in some Porifera but absent in choanoflagellates and in earlier-diverging eukaryotes, even if some attempts at transient epithelium-like organization can be observed in Amoebae. This means that a metazoan ancestor was dependent on cadherin-catenin complexes to build epithelial junctions and that further evolution added other junctional complexes, allowing new physiological functions in animals with more complex body plans.
Emergence of the polarity complexes.
In animal epithelia, concerted polarity is essential for cells in a monolayer to orient in the same direction (inside/outside) and, thus, perform their barrier function. The correct positioning of AJs along the apicobasal surface is also important to ensure that mechanical forces and stresses are balanced in the monolayer during morphogenetic events. In most Metazoa, coordination between the polarized organization of epithelial cells and the position of AJs is under the control of a set of proteins called polarity complexes (2).
APICAL POLARITY COMPLEXES.
There are two apical complexes, the atypical PKC (aPKC)-partition defective (Par) 3 (Par3)-Par6 and the Crumbs-Stardust (Sdt) or protein associated with lin7 1 (Pals1)-Pals1-associated TJ (PATJ) complexes, and one basolateral complex, the Scribble (Scrib)-Lethal giant larvae (Lgl)-Disc large (Dlg) complex (for review see Ref. 57). The aPKC-Par3-Par6, or Par, complex was first discovered in Caenorhabditis elegans and is essential for the first asymmetric division of the zygote, establishing cortical polarity opposite the Par1-Par2 complex (24, 31). It has been established that aPKC binds to Par6 and that this heterodimer can recruit Cdc42, leading to activation of aPKC kinase activity. Activated aPKC can phosphorylate Par3, regulating its association with the Par complex and other Par3-interacting proteins (for review see Ref. 9). It is noteworthy that Par3 interacts with nectins and junction-associated molecules, which are transmembrane proteins of the AJs, providing a direct link with this structure. Indeed, the Par complex is essential for the establishment and maintenance of AJs in vertebrates and invertebrates and is also involved in polarized migration (for review see Ref. 75). On the basis of in vitro and in vivo data, it is likely that this complex plays a role in organizing a link between proteins of the AJs and the actin and microtubule cortical network at the apicolateral domain and the migrating front (13, 27). The other apical complex, the Crumbs complex, is described below (see Crumbs Complex). It is, however, important to point out that the Crumbs and Par complexes act in coordination to establish and stabilize the apicolateral border and the AJs in the epithelial layer (for review see Ref. 3). The lateral polarity Scrib-Lgl-Dlg complex counteracts the function of the two apical complexes to establish the position of the apicolateral border and AJs in invertebrates and vertebrates (for review see Ref. 45). Thus epithelial cells in Eumetazoa have developed an intricate network of protein complexes to establish the border between the apical and lateral domains, allowing formation of AJs at this level (Fig. 1).
EVOLUTION OF THE POLARITY COMPLEXES.
How the polarity complexes have appeared is a major question in the evolution of animal phyla. Dlg and Par1 have been detected in the genome of Monosiga brevicollis, a choanoflagellate, while a precursor of Lgl has been identified in fungi and choanoflagellates, indicating that at least some of the lateral polarity complexes existed before the emergence of Metazoa (14). However, Scrib homologs are not anteriorly expressed in animals, suggesting that their addition to the set of lateral polarity complexes was necessary for the correct assembly of AJs. Apical polarity complexes, however, are, specific to Metazoa, since they are not identified in choanoflagellates, fungi, or Amoebae (14). Par6 and aPKC are present in Porifera such as A. queenslandica, indicating that both proteins belong to a common metazoan ancestor (74). Curiously, homologs of Sdt were not identified with enough confidence before the cnidarian lineage, leaving open the possibility that this adaptor protein for Crumbs appeared later during evolution (14). Future functional and structural studies will help clarify this point. Crumbs homologs are found in sponges, even though it is not clear if they are functional at this state of our knowledge (74). In conclusion, most genes of apical polarity complexes have appeared in parallel with classical cadherins, allowing for the control of AJ formation and positioning. It must be noted that these genes are called polarity genes, but, in fact, they are not required for cell polarity organization in single-cell organisms. Choanoflagellates are believed to be ancestors of animals through aggregation and formation of a primary epithelium, since their cell morphology is similar to that of the choanocytes found in sponges (33). Although they lack most of the so-called polarity genes, choanoflagellates show strong cell polarity, with an apical flagellum and an apical collar of actin-based microvilli (18), confirming that the Par and Crumbs complexes are not necessary for establishing this type of polarity. Thus one hypothesis is that the apical polarity complexes were acquired to allow a more complex organization of polarized cells, such as coordinated polarization and correct positioning of junctions, rather than polarity itself.
The Crumbs Complex
The Crumbs complex is the only polarity complex containing a transmembrane protein. Crumbs was first discovered as a mutant in D. melanogaster during a screen for genes essential for early embryogenesis and cuticle formation (78).
Crumbs and its partners (Pals1, PATJ, Par6, and aPKC).
Crumbs is a large transmembrane protein made of multiple EGF and laminin-like repeats related to Notch in the extracellular domain and of a highly conserved intracellular domain (for review see Ref. 3). The short (37- to 40-amino acid) intracellular domain contains two motifs that are essential for Crumbs function in vivo (34). The first motif is based on the ERLI sequence, a PDZ binding domain, and interacts with Sdt/Pals1 or Par6 (40, 66), two cytoplasmic adapters with PDZ domains (for review see Ref. 57). Sdt/Pals1 binds to PATJ, a multiple PDZ domain scaffold protein, through a conserved L27 domain and also to Par6, reinforcing the interaction between Crumbs and the Par complex (28). Par6 interacts with aPKC, and this complex can phosphorylate another conserved motif in the cytoplasmic domain of Crumbs (73) by direct binding to Crumbs or by indirect binding via Sdt/Pals1. This motif is based on a GTY sequence, connecting to the apical actin-based cytoskeleton through the actin-binding proteins of the Ezrin-Radixin-Moesin (ERM) family (48). Crumbs and its partners are depicted in Fig. 2.
Evolution of Crumbs and its complex.
Recent genomic studies on Choanozoa, choanoflagellates, and sponges have shed light on the emergence of animal-specific genes. In particular, it seems that apical polarity proteins such as Crumbs and its partners first appeared in the metazoan clade. Crumbs is detected in Placozoa and Cnidaria, and a clear Crumbs homolog, AmqCrbC2, is present in A. queenslandica, a demosponge (14). In particular, AmqCrbC2, in addition to the multiple EGF and laminin-like domains, possesses the typical short cytoplasmic domain, with conserved GFY (instead of GTY) and ERLL (instead of ERLI) motifs that clearly distinguish Crumbs from proteins of the Notch/Serrate/Delta family (74). More surprising, it has been suggested that a Crumbs homolog is already present in Ministeria vibrans, a member of Filasterea (70), a clade from the Choanozoa that separated from the common branch containing choanoflagellates and Metazoa more than 600 million years ago. This claim for the presence of a Crumbs homolog was based, however, on the presence of EGF-like repeats from the putative extracellular domain of Crumbs, and no evidence for the presence of the typical cytoplasmic Crumbs domain was found. Thus it is likely that, in pre-Metazoa, extracellular proteins with domains related to the Notch family already existed, since the EGF and laminin-like repeats are more ancient than the emergence of Metazoa (14). One hypothesis for the emergence of Crumbs could be that a preexisting extracellular protein of the Notch family was fused by genomic rearrangements to a new transmembrane and cytoplasmic domain encoding the innovative functional domains of Crumbs. Alternatively, a change in splicing could have resulted in a new cytoplasmic domain in a Notch ancestor. Further genomic analysis of the premetazoan lineages is necessary to elucidate this critical point.
The evolution of the rest of the Crumbs complex shows that PATJ is also present in sponges as a multi-PDZ domain protein (74), while the identification of a Sdt/Pals1 homolog is still being debated (14). In the A. queenslandica genome, however, a closely related MAGUK, AmqMPP5/7, is predicted to substitute for most of the known Sdt/Pals1 functions (14). Since Crumbs interacts directly with the PDZ domain of Pals1 in mammals and of Sdt in flies, it is likely that a functional homolog interacting with Crumbs existed in early Metazoa such as Porifera. More genomic and functional studies from other sponge families, such as homoscleromorphs, that have epithelia-like structures will help clarify this point.
The animal Crumbs ancestor was likely already made of a large extracellular domain with multiple EGF and laminin-like repeats related to Notch proteins and of a small intracellular domain able to interact with the apical cortical actin cytoskeleton and with the polarity proteins Par6 and Sdt/Pals1 through the GTY and ERLI motifs, respectively. The primary function of this new protein is unknown, since the functions of Crumbs have been investigated only in invertebrates and vertebrates (for review see Ref. 3), which are bilaterians and, thus, appeared later. The various functions of the modern Crumbs (58) are described below (see Cellular functions of Crumbs), but it can be said that these functions are changing with animal evolution. In particular, several isoforms of Crumbs, with two major features that have had an impact on their functions, exist in vertebrates (Fig. 3A). One innovation is the duplication of the classical Crumbs prototype into Crumbs1 and Crumbs2, with a more restrictive tissue expression of Crumbs1 than Crumbs2 in most species (10). The other innovation is the loss of the typical extracellular domain for a very short domain, with no conserved structural domain, characterized by rich putative O-glycosylation sites (for review see Ref. 3). This event occurred at least 500 million years ago, since this short isoform is common to fish, amphibians, and mammals but is not found in arthropods (unpublished data). How this new Crumbs, called Crumbs3, has emerged is unknown, but it is reasonable to speculate that it must have lost the functions associated with the original large extracellular domain. Crumbs3 is widely expressed in animal tissues and, in particular, is detected in most, if not all, epithelial tissues, indicating that the cytoplasmic domain of Crumbs is the one isoform essential for epithelial layers (40, 43). This also indicates that the functions of the extracellular domain and the intracellular domain can be disconnected in vertebrates during evolution. Even more recently, about 200 million years ago, a new motif appeared in the cytoplasmic domain of Crumbs3 that is shared by all mammals whose genome has been sequenced so far (for review see Ref. 3; unpublished data). This motif is a poly-proline sequence, PPxP, located between the GTY and ERLI motifs, but its function is unknown (Fig. 3B). Whether acquisition of this new motif in the cytoplasmic domain of Crumbs3 is associated with a new function in mammalian epithelia remains to be determined. Alternatively, the introduction of a short poly-proline motif could be essential for better accessibility of the COOH-terminal ERLI motif, which might otherwise be impaired in mammals for unknown reasons. In addition to this Crumbs3 isoform called Crumbs3A, a totally new Crumbs3B evolutionary isoform has also been discovered in mammals (15). This Crumbs3B isoform, missing the PPxP motif, has a different COOH-terminal sequence that originates from an alternative splice to create a PDZ binding motif CLPI at the COOH-terminal end. This Crumbs3B has lost one of the main signatures of the Crumbs protein family, namely, the ERLI motif that binds to the Par6 and Sdt/Pals1 PDZ domains. This new CLPI binds to Importin, a protein involved in transport to the nucleus and in ciliogenesis (15). Thus it seems that this isoform has acquired new properties by a simple alternative splicing, changing the reading frame of the Crumbs3 cytoplasmic domain, providing for rapid evolution of this gene function.
Cellular functions of Crumbs.
Cellular functions of Crumbs are reviewed here in an attempt to link the cellular and tissue functions of the Crumbs proteins and provide new lines of investigation for an understanding of their roles in epithelial origins and evolution. This is not intended to be an exhaustive review of the Crumbs literature but, rather, an attempt to make sense of the many described functions of Crumbs and its complex in the light of the evolution of epithelia. The Crumbs complex has been studied for more than 20 years, and data about its roles in epithelial and other tissues have accumulated, allowing for a better understanding of the multiple functions of this complex.
ROLE OF THE CRUMBS COMPLEX IN FORMATION AND STABILIZATION OF AJS.
In D. melanogaster and mammalian epithelia, the Crumbs complex (Crumbs, Sdt/Pals1, and PATJ) is localized in the apical part of the apicolateral junctions above the localization of Baz/Par3 in flies (51) or of occludin in Caco-2 or Madin-Darby canine kidney (MDCK) cells (39, 40). This is also the case in the mouse retina, where the Crumbs1 complex is located just above the AJs between Müller glial cells and photoreceptors (81). This localization suggests a role in the definition of the apical border of AJs and TJs. Numerous studies have shown that at least Crumbs and Sdt/Pals1 are essential for the correct formation and stability of apical junctions. In D. melanogaster, Crumbs or Sdt loss of expression in the embryo results in the lack or the disruption of a belt of AJs, leading to disintegration of the epithelium (20, 77) and mislocalization of E-cadherin (34). In kidney epithelial (MDCK) cells, overexpression of Crumbs3A induced defects in the formation of TJs (40, 65), while Pals1 depletion induced a strong disorganization of E-cadherin trafficking and junction formation (83). Thus there is a clear connection between the expression of Crumbs and Sdt/Pals1 and the correct organization of E-cadherin at the apicolateral border to build a complete belt of AJs in epithelial cells. It is likely that this function was one of the primary functions of the Crumbs complex in the metazoan ancestor. PATJ, on the other hand, has a more subtle role in the biogenesis of AJs, and it seems to play a scaffolding role in flies and mammals (49, 56, 72). It must be noted that the E-cadherin-β-catenin complex and the Crumbs complex seem to have appeared concomitantly during animal evolution, suggesting that the E-cadherin complex needed another apical complex to control its localization and stability.
ROLE OF THE CRUMBS COMPLEX IN ORGANIZATION OF THE APICAL ACTOMYOSIN NETWORK.
In epithelial columnar cells, there is a specific organization of the subapical F-actin cytoskeleton that builds a network below the apical membrane to provide more rigidity to the apical domain and anchor the apical microvilli (for review see Refs. 17 and 50). In D. melanogaster, Crumbs interacts with the F-actin cytoskeleton, and its depletion in the early embryo induces a mislocalization of β(heavy)-spectrin (48), an actin-binding protein associated with AJs (79). Overexpression of Crumbs in the primary epithelium of fly embryos redistributes the F-actin (48) and the β(heavy)-spectrin (84). While this functional link between Crumbs and the apical actin-based cytoskeleton has not been shown in mammals, there is a conserved molecular interaction between proteins of the ERM family, such as Moesin, Yurt, EPB-41L5, and Expanded, and the Crumbs complex (19, 37, 42, 48). This common property between vertebrates and invertebrates indicates that, even in mammals, Crumbs is able to bind and likely regulate the organization of the apical F-actin cytoskeleton. Recently, it has been shown that, in D. melanogaster, the asymmetric distribution of Crumbs and aPKC in the apical plane of epithelial cells controls the formation of actomyosin cables at places where Crumbs is lowest (68). PATJ has also been implicated in the regulation of myosin activity at the AJs (69), further strengthening the hypothesis that the Crumbs complex is an important regulator of the subapical cytoskeleton. There is, however, no evidence that this mechanism is conserved between invertebrates and vertebrates. The link between the Crumbs complex and the actomyosin apical cytoskeleton might provide a simple mechanism to couple the organization of the apical domain and the contractility of the cytoskeleton that is necessary to allow for morphogenetic events such as cell intercalation, gastrulation, and tube formation. Since this interaction involves the conserved GTY motif of Crumbs, the link between Crumbs and the apical cytoskeleton might be very ancient and could be a property of the Crumbs ancestor.
ROLE OF THE CRUMBS COMPLEX IN CILIOGENESIS AND MEMBRANE TRANSPORT.
One of the least expected functions of the Crumbs complex was first uncovered in epithelial cells in culture. Crumbs3A and Crumbs3B are involved in the formation of the primary cilium in renal epithelial (MDCK) cells and the Crumbs complex, including aPKC, interacting with kinesins that are essential for transport along the axonema (15, 16). Confirmation of this role of Crumbs proteins came from studies in zebrafish, where Crumbs2b and Crumbs3a are involved in cilia elongation in vivo (54). This role in ciliogenesis can be related to the observation that, among mouse, zebrafish, and D. melanogaster, Crumbs or Crumbs2 overexpression or loss of expression regulates the extension of the stalk membrane (in flies) and the inner segment (in mouse and zebrafish), respectively (1, 26, 60). Since overexpression of Crumbs in the epidermis or Crumbs3A in MDCK cells also leads to an increase in the size of the apical membrane (40, 84), it could be concluded that Crumbs proteins regulate apical membrane trafficking. This hypothesis is reinforced by the fact that Crumbs controls the level of myosin V, which transports rhodopsin to the rhabdomeres of the photoreceptors in flies (59). Whether this role in membrane transport and ciliogenesis is a property of the Crumbs ancestor or was acquired later remains an enigma. Cilia and flagella are very ancient organelles that developed millions of years before the appearance of Crumbs and the Par complex, suggesting that the role of these proteins in regulating ciliogenesis was a secondary function. This function could have been developed to coordinate the loss of the apical cilium and cell division in the integrated layer of the epithelium, since it is acknowledged that most cells have to lose their primary cilium to enter the cell cycle (for review see Ref. 32).
Tissue functions of Crumbs.
We have seen that the Crumbs complex has many different cellular functions, but how these functions are integrated in vivo remains a burgeoning field of investigation using new models such as the zebrafish and mouse. Indeed, for many years, the only in vivo data regarding the role of the Crumbs complex were derived from D. melanogaster and C. elegans.
ROLE OF CRUMBS IN GASTRULATION.
Crumbs was first discovered as a mutant (crb) in D. melanogaster, and its phenotype is very strong in the early embryo, inducing lethality during gastrulation (78). Crumbs is clearly essential for holding the epithelial cells together during the dramatic cell movements during gastrulation and germ band extension in the fly embryo (for review see Ref. 76). Interestingly, this Crumbs function is also conserved in mammals, since it was recently shown that knockout of Crumbs2 in the mouse leads to embryo death after gastrulation (85), suggesting that Crumbs functions in controlling rapid junction remodeling during intense epithelial cell layer movements. This could be one of the primary functions of the Crumbs ancestor, since gastrulation is a very common step in animal development. In C. elegans, however, mutation in one or both Crumbs does not lead to dramatic epithelial defects during early development (6, 71). This lack of severe epithelial defects might be due to the intrinsic nature of apicolateral junctions in C. elegans that have evolved differently from the arthropods and mammals. Future studies in different models more proximate to the animal origins will help address this issue.
ROLE OF CRUMBS IN REGULATING TISSUE GROWTH AND SIZE.
Recent studies using Crumbs mutants in D. melanogaster have shown that Crumbs regulates the size of several epithelial organs, such as the wing, eye, and head (38, 61). This action on organ growth and size is mediated through the Hippo pathway (8, 22, 42, 62). It has been proposed that the Crumbs complex acts as a sensor for cell density in epithelia to regulate the transforming growth factor-β-SMAD pathway through the Hippo cascade (82). While these important findings await functional confirmation in mammals, it has been shown that the Crumbs3A-PATJ complex binds to the tuberous sclerosis complex to regulate cell size in human intestinal cells (46). Thus several lines of evidence suggest that the control of epithelial cell and organ size by the Crumbs complex is a conserved function that could have played a role in the evolution of animal shape.
ROLE OF THE CRUMBS COMPLEX IN THE RETINA.
In humans, Crumbs play an important role in the retina, since CRB1 is the target of mutations inducing severe degenerations of the retina, called Leber congenital amaurosis and retinitis pigmentosa (RP12) (11). Starting from that observation, the role of Crumbs has been studied in mouse, D. melanogaster, and zebrafish; in these species, Crumbs genes are involved in the correct organization of the retina (1, 30, 54, 55, 81). The precise function of Crumbs that is first affected in the retina remains unclear, since the organization of the retina in all species studied relies on several key points: 1) correct assembly of AJs, 2) formation of photoreceptors possessing a connecting cilium in vertebrates, and 3) very active membrane transport toward the outer segment or the rhabdomere. Studies in the fish retina have, however, shown a new role for the extracellular domain of Crumbs in intercellular recognition and adhesion (87), which was reinforced recently in flies (80).
Conclusions and Perspective
With the genomic information that is available, we can establish that the Crumbs complex appeared at a crucial point in the emergence of epithelia in Metazoa in conjunction with the other apical polarity complex, aPKC, Par3, Par6, and the classical cadherins linked to β-catenins. Thus it seems that these protein cassettes were necessary from the outset in establishing a true epithelial layer, complete with junctions, basement membrane, and coordinated polarity, allowing for vectorial transport. The identity of the first organism composed of this epithelium remains unknown, and it is puzzling that all these different protein complexes were acquired in parallel, leading to all the existing animals. It is, however, possible that ancestors lacking some of these key elements may have existed, but their lineage did not survive, or that we have not yet identified them. Life of a multicellular organism is not possible without some plasticity, either during its development or during its growth, and precise mechanisms are necessary to modulate the adhesion between cells. The cadherin-β-catenin complexes provided such strong adhesion, and from the contemporary functions of the Crumbs complex, we may hypothesize that this complex was necessary to allow for this plasticity. In particular, it is highly likely that Crumbs provided an ancient signaling and integrator complex coupling external signals for growth with its extracellular EGF repeats, the subapical actin cytoskeleton with the GTY motif, and the AJs with the ERLI motif. Further functional genomics in Porifera (sponges), Placozoa, and Cnidaria are needed to elucidate which of the modern Crumbs functions are common to all Metazoa. It is very likely that the shared functions expressed by these early, diverging Metazoa, will be the most ancient ones.
This research was supported by the Centre National de la Recherche Scientifique (UMR7288) and Aix-Marseille University, a grant from the Association pour la Recherche sur le Cancer, a grant from A*MIDEX, and Coordination Theme 1 (Health) of the European Community FP7 Grant HEALTH-F2-2008-200234.
No conflicts of interest, financial or otherwise, are declared by the author.
A.L. prepared the figures; A.L. drafted the manuscript; A.L. edited and revised the manuscript; A.L. approved the final version of the manuscript.
I thank John Torday for critical reading of the manuscript.
- Copyright © 2013 the American Physiological Society