Organizing the cell cortex: the role of ERM proteins

Abstract

Specialized membrane domains are an important feature of almost all cells. In particular, they are essential to tissues that have a highly organized cell cortex, such as the intestinal brush border epithelium. The ERM proteins (ezrin, radixin and moesin) have a crucial role in organizing membrane domains through their ability to interact with transmembrane proteins and the cytoskeleton. In doing so, they can provide structural links to strengthen the cell cortex and regulate the activities of signal transduction pathways. Recent studies examining the structure and in vivo functions of ERMs have greatly advanced our understanding of the importance of membrane-cytoskeleton interactions.

Figure 1. Domain organization of ERM proteins and activation model

A| All ERMs have a similar domain structure. The domain organization of ezrin is represented here. The N-terminal FERM domain (blue) consists of three subdomains, designated F1, F2, F3 or A, B, C. The central ~150 residue region (yellow) is predicted to have a high α-helical content and propensity to assemble into a coiled-coil structure. This is followed by a linker region that is rich in proline residues in ezrin and radixin, but not moesin, and the protein terminates in the C-ERMAD (red) containing the F-actin-binding site (green). B | Ezrin is activated through PtdIns(4,5)P₂ binding and phosphorylation of threonine 567, which reduces the affinity of the N-terminal FERM domain (blue) for the C-ERMAD (C-Tail, red). This unmasks binding sites for F-actin (green), the cytoplasmic tails of specific membrane proteins, and the adaptor protein EBP50. As EBP50 binding to the FERM domain reduces the affinity of the FERM domain for the cytoplasmic tails of membrane proteins, these two ERM binding modes might be mutually exclusive as shown. The PDZ domains of EBP50 can bind additional proteins, including the scaffolding protein PDZK1. Moesin and radixin are thought to function in a similar manner and bind many of the same, or related, proteins, although in general they have not been as well studied. EBP50 and PDZK1 are believed to exist in a closed form. Upon unmasking of the EBP50 binding site on the ezrin FERM domain, the open conformation of EBP50 is stabilized and binds ezrin. A similar type of mechanism might also exist for PDZK1.

Figure 1. Domain organization of ERM proteins and activation model

A| All ERMs have a similar domain structure. The domain organization of ezrin is represented here. The N-terminal FERM domain (blue) consists of three subdomains, designated F1, F2, F3 or A, B, C. The central ~150 residue region (yellow) is predicted to have a high α-helical content and propensity to assemble into a coiled-coil structure. This is followed by a linker region that is rich in proline residues in ezrin and radixin, but not moesin, and the protein terminates in the C-ERMAD (red) containing the F-actin-binding site (green). B | Ezrin is activated through PtdIns(4,5)P₂ binding and phosphorylation of threonine 567, which reduces the affinity of the N-terminal FERM domain (blue) for the C-ERMAD (C-Tail, red). This unmasks binding sites for F-actin (green), the cytoplasmic tails of specific membrane proteins, and the adaptor protein EBP50. As EBP50 binding to the FERM domain reduces the affinity of the FERM domain for the cytoplasmic tails of membrane proteins, these two ERM binding modes might be mutually exclusive as shown. The PDZ domains of EBP50 can bind additional proteins, including the scaffolding protein PDZK1. Moesin and radixin are thought to function in a similar manner and bind many of the same, or related, proteins, although in general they have not been as well studied. EBP50 and PDZK1 are believed to exist in a closed form. Upon unmasking of the EBP50 binding site on the ezrin FERM domain, the open conformation of EBP50 is stabilized and binds ezrin. A similar type of mechanism might also exist for PDZK1.

Figure 2. Structures of FERM domains with bound ligands

A| The radixin FERM domain complexed with a peptide from the cytoplasmic tail of ICAM2. The β-strand binds to a groove on the F3 subdomain.. B | The moesin FERM domain with the C-terminal peptide of EBP50 bound. The tail of EBP50 forms an α-helix that binds to the surface of the F3 subdomain, . C–D | Two views of the C-ERMAD of moesin bound to the moesin FERM domain. The C-ERMAD binds the F2 and F3 subdomains through a β-strand followed by four helices (helix A, B, C and D). In the orientation corresponding to panels A and B, it can be seen that the β-strand occupies the same groove that binds ICAM2, and helix D binds to the same surface as EBP50. E | Full length insect (Spodoptera frugiperda) moesin dormant structure revealing the structure of the central α-helical region. F | A conceptual model of the activation of ERMs involving the complete dissociation of the C-ERMAD from the FERM domain and allowing the central the α-helical region to unravel and potentially span up to 25nm. The figures were assembled from Protein Data Bank files 1J19 (panel A), 1SGH (panel B), 1EF1 (panels C and D) and 2I1J (panel E).

Figure 2. Structures of FERM domains with bound ligands

A| The radixin FERM domain complexed with a peptide from the cytoplasmic tail of ICAM2. The β-strand binds to a groove on the F3 subdomain.. B | The moesin FERM domain with the C-terminal peptide of EBP50 bound. The tail of EBP50 forms an α-helix that binds to the surface of the F3 subdomain, . C–D | Two views of the C-ERMAD of moesin bound to the moesin FERM domain. The C-ERMAD binds the F2 and F3 subdomains through a β-strand followed by four helices (helix A, B, C and D). In the orientation corresponding to panels A and B, it can be seen that the β-strand occupies the same groove that binds ICAM2, and helix D binds to the same surface as EBP50. E | Full length insect (Spodoptera frugiperda) moesin dormant structure revealing the structure of the central α-helical region. F | A conceptual model of the activation of ERMs involving the complete dissociation of the C-ERMAD from the FERM domain and allowing the central the α-helical region to unravel and potentially span up to 25nm. The figures were assembled from Protein Data Bank files 1J19 (panel A), 1SGH (panel B), 1EF1 (panels C and D) and 2I1J (panel E).

Figure 3. In vivo functions of ERM proteins

A | Two examples, mitotic cells in culture and differentiating epithelial cells, where ERMs function by binding to and organizing the cortical actin cytoskeleton. In interphase cells, inactive, self-associated ERMs do not stabilize the interface between membrane proteins and the cortical cytoskeleton, yielding randomly oriented actin filaments and low cortical tension. As cell enter mitosis, ERM activation results in linkage of actin filaments to the cell cortex so that they lie parallel to the plasma membrane, increased in cortical tension, an associated rounding of the cell membrane, and proper positioning of the mitotic spindle. In differentiating D. melanogaster epithelial cells, ERM recruitment to the apical membrane by Bitesize, a synaptotagmin-like protein, results in recruitment and stabilization of actin filaments in the apical domain. These filaments in turn stabilize adherens junctions that assemble in the apical junctional region. Although not depicted it is likely that ERMs interact with as-yet-unidentified transmembrane proteins in mitotic cells and in the fly embryonic epithelium. B | Examples where ERMs organize both the actin cytoskeleton and other cortical molecules. In the D. melanogaster oocyte Moesin is necessary for cortical cytoskeleton integrity and cytoskeleton-dependent localization of posterior polarity determinants such as oskar mRNA. As in [A], Moesin likely associates with transmembrane proteins during the establishment of oocyte polarity. In human T cell activation, Moesin, which normally binds and localizes CD43, is transiently inactivated in the region where the T-cell binds to the antigen-presenting cell, leading to microvillar collapse and delocalization of CD43. Simultaneously, Ezrin is activated and recruits ZAP-70, the signalling tyrosine kinase ζ-chain associated protein, to the immunological synapse, where it phosphorylates other components of the T-cell activation pathway.