BIOLOGICAL OVERVIEW

Crumbs protein is essential for the biogenesis of the adherens junction and the establishment of apical polarity in ectodermally derived epithelial cells. The adherens junction is a multiprotein complex that attaches one cell to another in an epithelial layer. The junction is not spread randomly between the cells, but is found in a belt-like, zonular structure encircling the apical side of the cell.

The apical side of the cell faces the outside of the embryo, in opposition to the inward facing basal side. Placement of the adherens junction is critical because it signals to the cell which side is out and which side is in, preventing the mixing of apical cell membrane tissue with the biochemically distinct basal cell membrane, and thereby assuring cell polarity.

DE-cadherin (shotgun) and Armadillo comprise the two main constituents of the adherens junction. DE-cadherin is a homophilic adhesion transmembrane molecule that links the outside of the cell with the inside. Armadillo, the Drosophila homolog of beta-catenin, is a molecule that links the adherens junction with the cell's cytoskeleton.

crumbs mutants fail to establish adherens junctions and thus fail to establish epithelial cell polarity. Crumbs protein delimits the apical border, thus establishing the proper position for the border's placement . The defect in crumbs minus mutants results in a misdistribution of Armadillo and DE-cadherin, resulting in a disruption of tissue integrity. Curiously, despite the lack of adherens junction formation in such mutants, there is no accompanying loss of membrane polarity. This supports the view that membrane polarity exists prior to the formation of adherens junctions, and establishes the pattern of proper placement of the junction (Grawe, 1996).

Crumbs protein is distributed over the entire apical cell surface of epithelial cells and accumulates at the outer margin of the apical membrane where neighboring cells are in contact. However, no Crumbs protein is detected at the zonula adherens. This suggests that the polarizing activity of Crumbs arises from a direct or indirect binding of the Crumbs protein to adherens junction material at the outer rim of the marginal zone. The retention of adherens junction material in direct contact with the marginal zone would facilitate the formation of the zonula adherens from patches of adherens junction material that assemble through interaction with Crumbs protein (Tepass, 1996).

A conserved motif in Crumbs is required for E-cadherin localisation and zonula adherens formation in Drosophila. Expression of just the short membrane-bound cytoplasmic domain is sufficient to rescue major defects associated with the loss of crumbs function. The cytoplasmic domain of Crumbs is highly conserved in two putative crumbs homologs in C. elegans. To assess the significance of conserved residues, various point mutations and deletions were introduced into this region. Two functional domains were revealed: an amino-terminal region and the carboxy-terminal amino acids EERLI. Both are necessary for rescue of the crumbs phenotype. The EERLI motif interacts with Discs Lost (now redefined as Drosophila Patj), a cytoplasmic protein containing PDZ domains. Overexpression of the Crumbs cytoplasmic domain induces a transition from the single-layered epithelium to a multilayered tissue. This transition is associated with redistribution of the Drosophila homolog of the cell adhesion molecule E-cadherin, and depends on the presence of the EERLI motif (Klebes, 2000).

Two C. elegans genes that encode transmembrane proteins with multiple EGF-like repeats and short cytoplasmic domains were detected in the database. The cytoplasmic domains of both proteins, called CeCrb1 and CeCrb2, also consist of 37 amino acids, nine of which are conserved in all three proteins. A transgene (CD2-IntraCE), encoding the cytoplasmic domain of CeCrb1 fused to the rat transmembrane protein CD2 (which provides a transmembrane domain and a tag) was expressed in wild-type Drosophila embryos. The phenotypic consequences were compared with those induced by overexpression of a corresponding Drosophila fusion protein (CD2-IntraWT). Both CD2-Intra proteins induce the same phenotype, which is indistinguishable from that caused by the expression of Myc-IntraWT: the epidermis became multilayered and DE-cadherin and phosphotyrosine-containing epitopes are mislocalized. This shows that the functionally important regions responsible for inducing formation of a multilayered epidermis are conserved in the cytoplasmic domain of CeCrb1 (Klebes, 2000).

Data presented here suggest a model in which the Drosophila Crb protein organizes the assembly of an apically localized protein scaffold in epithelial cells that is required for the proper formation and localisation of the ZA. This scaffold includes the protein Dlt and probably other, as yet unidentified, proteins, its assembly depends on the carboxy-terminal segment of Crb. The model further suggests that the Crb-mediated control of DE-cadherin localization depends on interaction between the Crb cytoplasmic domain and the PDZ protein Dlt. Neither DE-cadherin nor Dlt are localized in crb mutant embryos, whereas both proteins are sequestered by mislocalized Crb. However Dlt remains apically localized after overexpression of DE-cadherin. The interaction of Crb with Dlt depends on Crb's carboxy-terminal motif, EERLI. This motif is also necessary for misdistribution of DE-cadherin upon Crb overexpression and for the rescue of crb mutant embryos. The presence of four PDZ domains in Dlt makes it an ideal partner for recruiting other proteins into a hypothetical Crb-dependent, membrane-associated protein network. PDZ domains have been shown to act as versatile organizers of multiprotein complexes. In many cases, the binding site of the interacting protein, often a transmembrane protein, is localized at its carboxyl terminus and ends with a hydrophobic amino-acid residue. Class I PDZ domains bind a conserved S/T-X-V motif (where X is any amino acid), whereas class II domains recognize ligands that carry a hydrophobic amino-acid residue at the -2 position. Since the Dlt-binding site in Crb differs from these motifs, the first PDZ domain of Dlt, which binds to Crb in vitro, may belong to a different class. The presence of the ERLI motif in both C. elegans homologs and the similarities between the phenotypes produced by overexpression of CD2-IntraWT and CD2-IntraCE in the Drosophila embryo suggest that this region might mediate comparable interactions in the nematode. Not surprisingly, a protein similar to Drosophila Dlt has also been detected in the C. elegans database, pointing to the possible conservation of additional components of the postulated protein network (Klebes, 2000).

Data indicate that the EERLI motif is necessary, but not sufficient, to rescue the phenotype of crb mutant embryos. Rescue also requires an intact amino-terminal region of the cytoplasmic domain. It is tempting to speculate that the region containing the mutated amino acids may be involved in additional protein-protein interactions. A comparable situation is provided by a group of transmembrane proteins, including glycophorin C, beta-neurexin and syndecans, that have been identified, respectively, as ligands for the class II PDZ proteins p55, CASK and syntenin. The cytoplasmic tails of these proteins terminate in the tetrapeptides EYFI, EYYV and EFYA, respectively, and show additional conservation in their amino-terminal regions. For glycophorin C it has been demonstrated that the 12-residue sequence immediately adjacent to the membrane binds directly to protein 4.1, a member of the 4.1 superfamily, which includes, among others, the so-called ERM proteins (ezrin, radixin, moesin). The latter proteins provide a linkage between cell-surface receptors and the spectrin/actin cytoskeleton. The 12-residue sequence of glycophorin C that binds protein 4.1 includes a Gly8-Thr9-Tyr10 motif, which is Gly8-Ser9-Tyr10 in beta-neurexin and all syndecans. The corresponding region of Drosophila Crb also contains a Gly8-Thr9-Tyr10 motif at an equivalent position (Gly-His/Lys-Tyr in the C. elegans proteins); mutating Tyr10 to alanine completely abolishes the rescuing function. It is unlikely that the Drosophila protein 4.1 homolog, encoded by coracle, is a partner of Crb in wild-type embryos. Coracle is associated with septate junctions, which are localized basally to the ZA, and colocalizes with Discs Large, a PDZ-domain protein, and beta-neurexin IV. The amino-terminal region conserved between Drosophila and the two C. elegans homologs extends further, to Gly8-X9-Tyr10-X(11-15)-Glu16. The data clearly show that Glu16, which is also a charged amino acid in syndecans, glycophorin C and neurexin, is also of crucial importance for the rescuing function (Klebes, 2000).

There is a further indication of possible involvement of the amino-terminal region of the Crb cytoplasmic domain in interactions with other, as yet unknown, proteins closely associated with the plasma membrane. All CD2 fusion proteins used in this study fail to rescue crb mutant embryos, even those containing the full-length cytoplasmic domain. Whereas the Myc fusion proteins contain the Crb transmembrane domain, immediately followed by the cytoplasmic portion, CD2 fusion proteins contain the CD2 transmembrane domain and provide a spacer of 45 amino acids between the membrane and the cytoplasmic tail of Crb. This spacing could prevent interactions between the cytoplasmic segment of Crb and a hypothetical partner localized at the membrane. At present, however, it cannot be determined whether it is this spacer, the lack of the Crb transmembrane domain, or some other feature of the CD2 fusion protein that is responsible for the lack of rescuing function (Klebes, 2000).

Crb is the earliest zygotically expressed apical transmembrane protein, but nothing is known about the cis-regulatory sequences that target it to the apical face of the cell nor the mechanisms and proteins required for this process. Nothing is known about the function of the large extracellular domain; its overexpression in a secreted or membrane-anchored form (lacking the cytoplasmic domain) does not induce any mutant phenotype. Embryos devoid of maternal Dlt fail to localize Crb. Since the blastoderm epithelium of these embryos itself lacks cell polarity, however, all other defects, including improper Crb localisation, could be regarded as secondary effects. In embryos mutant for stardust, Crb is first expressed apically, but during germ band extension it is no longer detectable, making stardust a likely regulator for the maintenance of apical localization of Crb. In agreement with this, stardust mutant embryos develop a phenotype nearly identical to that of crb mutant embryos. Because the molecular nature of the stardust gene is not yet known, no information can be obtained about its relationship with Crb expression at present (Klebes, 2000).

GENE STRUCTURE

There are two RNAs produced by addition of different polyA tracts, but the source of this variation is not known (Tepass, 1990).

cDNA clone length - 7226 bases

Bases in 5' UTR -113 plus

Exons - five

Bases in 3' UTR - 505

PROTEIN STRUCTURE

Amino Acids - 2139

Structural Domains

crumbs encodes a large transmembrane protein with 30 EGF-like repeats and four laminin A G-domain-like repeats in its extracellular domain, suggesting its participation in protein-protein interactions. There is an N-terminal CAX pepeat. The cytoplasmic region consists of 28 amino acids (Tepass, 1990 and Knust, 1993).

Proteins encoded by the tumor suppressor fat gene, the neurogenic slit gene and crumbs gene of Drosophila contain domains homologous with modules identified previously in laminin A. These proteins of Drosophila have a number of features in common: they have large extracellular regions containing laminin A modules linked to epidermal growth factor-like domains, and they are all involved in cell-cell interactions that are crucial for correct morphogenesis of ectodermal tissues (development of midline neuroepithelial, organization of epithelial tissues etc.). Patthy has suggested that the laminin A-type modules of these proteins play important roles in interactions controlling ectodermal differentiation (Patthy, 1992).

date revised: December 10 2002

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

Please e-mail comments/corrections to brodyt@codon.nih.gov