Moesin (membrane-organizing extension spike protein) was initially isolated from bovine uterus in 1988. It was first identified as a heparin-binding protein and thought to be extracellular (Lankes et al. 1988), but later shown to be intracellular (Lankes and Furthmayr 1991). Moesin belongs to the ERM family of proteins (ezrin, radixin, and moesin), a subclass of the large protein 4.1 superfamily that also includes protein 4.1, talin and merlin. The ERM family of proteins is credited with coupling filamentous actin to the cell membrane and has roles in numerous cellular processes involving modification of the cortical actin network in conjunction with the plasma membrane.
Distribution and Cell Biology
Moesin is found exclusively in vertebrates along with the related ERM proteins ezrin and radixin. These three paralogous proteins probably arose from the ancestral invertebrate ERM protein during the two rounds of whole genome duplication preceding vertebrate evolution (see Fig. 1). Ezrin, radixin, and moesin are highly conserved (sequence identities are 76%, 74%, and 81% for human ezrin/radixin, ezrin/moesin, and moesin/radixin, respectively). The human moesin gene (MSN) is located on the X chromosome (Xq11.2-q12) and has 12 exons, extending over 30 kb.
ERM physiological expression patterns vary between cell types, suggesting that ezrin, radixin, and moesin carry out unique roles; however, most biochemical studies and studies within cell lines and transfected cells suggest they have very similar interactions. Moesin, expressed to various degrees, is found in many cell types such as macrophages, lymphocytes, and fibroblastic, endothelial, epithelial, and neuronal cell types. In lymphocytes, monocytes, and neutrophils, moesin is the dominant ERM protein (ezrin is also expressed), while moesin is the only ERM protein found in platelets.
In the cell, ERM proteins concentrate near the plasma membrane at the places where actin is enriched. The ERM proteins provide a regulated cross-link between the actin cortical cytoskeleton and the plasma membrane. This actin-membrane coupling role implicates the ERM proteins in a wide range of cellular processes (such as establishing cell polarity, cell motility, apoptosis, cell signaling, morphogenesis, cancer metastasis, and viral/bacterial infection) and makes them especially important in the regulation and formation of cell surface projections, structures, and modifications (such as microvilli, lamellipodia, cleavage furrows, ruffling membranes, filopodia, and tubulogenesis).
Moesin is not randomly distributed within the cell, nor within the cortical cytoskeleton, but localized to very specific structures, including the core of filopodia, microvilli, microspikes, and retraction fibers. For example, in thymoma cells, moesin is found concentrated at microvilli tips, but is not associated with stress fibers (Amieva and Furthmayr 1995). Antisense knockdowns of moesin in mouse experiments indicate that, in contrast to ezrin and radixin, moesin is critically involved in microvilli structures, while ezrin and radixin knockdowns in the same cell lines demonstrate involvement in cell-cell and cell-substrate adhesion processes (Takeuchi et al. 1994).
Mutant mice deficient in moesin are reported to show no obvious structural or functional abnormalities. However, moesin homozygote knockout mice show threefold lower lipopolysaccharide-induced inflammation when injected with lipopolysaccharide (Amar et al. 2001). Moesin mouse knockout neutrophils display aberrant orientations in chemotactic gradients (Liu et al. 2015), and inhibition of small GTPases (Rac, Rho, and Cdc42) by activated moesin in neutrophils has been shown to be required for maintenance of cell symmetry and orientation (Liu et al. 2015). These results demonstrate that moesin is involved in immune response, and indeed moesin has been implicated in the processes of both bacterial and viral entry, as well as cancer progression (see “Pathogenesis” section below).
Protein Features, Biochemistry, and Interaction Partners
The three ERM paralogs, ezrin, radixin, and moesin, share extremely similar structural characteristics, with very few differences in the N- and C-terminal domains (see Fig. 2a). The N-terminus forms a globular FERM (protein 4.1, ezrin, radixin, moesin) domain that contains a membrane-binding site, specifically phosphatidylinositol 4,5-bisphosphate (PIP2). The FERM domain is about 300 amino acids and comprised of three globular subdomains, called F1, F2, and F3, that together form a clover-leaf-shaped structure (see Fig. 2a–d). F1 is very similar in structure to ubiquitin, F2 is similar to the structure of acyl-CoA-binding protein, and F3 resembles a ligand-binding domain found in phosphotyrosine binding, pleckstrin homology, and enabled/VASP homology 1 (EVH1) domains. Characteristics of the F3 module are often found in cytoskeletal and cell signaling proteins that bind peptides or phospholipids. There is also a lipid interaction region on the F1 module (Ben-Aissa et al. 2012). The C-terminal domain is ∼100 amino acids in length and contains an F-actin binding domain (called the C-ERMAD, C-terminal ERM-association domain) of about 30 amino acids.
The N-terminal FERM and C-terminal domains are linked by an intermediate helical domain predicted to form a flexible α-helical coiled coil. While the sequence of the coiled-coil domain is not strictly conserved, its length is strictly conserved in all ERM proteins. Crystal structures of moesin, ezrin, or radixin have failed to capture a complete snapshot of the intermediate coiled-coil ERM domain (see Fig. 2). However, a single structure (PDB 2I1K, see Fig. 2e) from the related invertebrate ERM protein from the moth Spodoptera frugiperda confirms the predicted coiled-coil nature of this region, and this structure likely represents the conformation adopted in other monomeric ERM proteins. Some attempts at crystallizing full-length moesin, ezrin, and radixin have reported proteolysis within the coiled-coil domain, consistent with the prediction that this region contains flexible elements. The moesin protein is different in that it lacks a proline-rich region or polyproline track found near the C-terminus of the intermediate coiled-coil domain in both ezrin and radixin (Polesello et al. 2002). It is this polyproline region that is a substrate for the enzyme calpain, which cleaves both ezrin and radixin but not moesin.
All ERM proteins have sites that are known to be modified by phosphorylation. Moesin can be phosphorylated at threonine 558, a conserved residue in the C-terminal domain of the protein. However, the precise role of ERM protein phosphorylation is unknown (see also “Signaling” section below).
The high level of sequence identity shared by ERM proteins has made it difficult to distinguish different paralogs both biochemically and in vivo. A resounding complication in the ERM protein field is that many antibodies raised against one family member cross-react with other ERMs, although specific antibodies can be raised (Lankes and Furthmayr 1991). There also appears to be some functional redundancy within the ERM family, further complicating the interpretation of many studies investigating this protein family.
Moesin is reported to interact with numerous binding partners. However, many of these interactions are largely uncharacterized. These include other ERM family proteins (e.g., ezrin), molecular scaffold proteins (e.g., actin, Crb1), cell surface cluster of differentiation proteins (e.g., CD43, CD44, ICAM3 (CD50), VCAM-1 (CD106)), PDZ domain proteins (e.g., PDZD8), small GTPases (e.g., Rac, Rho, and CDC42), kinases (e.g., ROCK1, LRRK2), neutrophil cytosolic factors (e.g., NCF1 and NCF4), chemokine receptors (e.g., CXCR4, CCR5), lipid polysaccharide receptors (e.g., TCR4), and FERM-binding proteins (e.g., Crumbs).
The invertebrate ERM protein has been shown to antagonize the activity of the small GTPase Rho (Speck et al. 2003) and furthermore that there is a negative feedback loop between ERM activation and the Rho pathway. It is assumed that this pathway follows through the more specialized vertebrate ERM proteins: ezrin, radixin, and moesin.
Moesin is phosphorylated by a number of pathways (including RhoA/ROCK2 kinases, lymphocyte-oriented kinase (LOK), and the Erk pathway) at the conserved threonine 558 in the C-terminal domain. It is thought that activation of moesin occurs firstly by binding to PIP2. Activation is then thought to be stabilized by phosphorylation and that phosphorylation is required for cytoskeletal rearrangements. Phosphorylation of moesin occurs during a range of processes including platelet activation, regulation of epithelial polarity, and in mediating endothelial angiogenesis (Wang et al. 2016).
Identification of a key role for moesin in numerous and diverse disease states attests to its fundamental role in cell signaling. However, the underlying pathways involved and the role of moesin, as distinct from the other ERM proteins, are generally poorly understood.
Localized deactivation of moesin by myosin phosphatase has been shown to be essential for neutrophil polarization, enabling infecting bacteria to be engulfed by the cell (Liu et al. 2015). Moesin has also been implicated in the processes of viral entry including measles, rabies, and HIV. HIV infection is initially regulated by attachment of the viral particles to the target cell via association with CD4 and chemokine receptors on the plasma membrane (either CXCR4 or CCR5). Moesin knockdown has been shown to increase infection by R5-tropic HIV-1, a strain that uses only the CCR5 receptor, whereas ezrin knockdown has no effect and radixin knockdown decreases infection (Kubo et al. 2008). Upon HIV-1 viral entry, activated moesin also drives redistribution and clustering of CD4 and CXCR4 through promotion of F-actin redistribution (Barrero-Villar et al. 2009).
The ERM proteins have also been implicated in cancer progression. Moesin and ezrin are differentially distributed in the early steps of melanoma tumor cell invasion, suggesting a specific role for moesin in melanoma cell invasion (Estecha et al. 2009). Ezrin and moesin are also thought to work cooperatively to promote T-cell activation, where removal of CD43 and moesin is associated with immune synapse formation (Ilani et al. 2007; Shaffer et al. 2009).
Additionally, genome-wide association studies have identified a single nucleotide polymorphism in the noncoding moesin pseudogene 1 (MSNP1) as an autism risk factor. Noncoding RNA-encoded antisense to MSNP1, which can regulate expression of the moesin protein, is highly overexpressed in the postmortem cerebral cortex of individuals with autism (Kerin et al. 2012), and in human neural cell lines decreases neurite length and number, and alters gene expression of proteins involved in both protein synthesis and chromatin remodeling (DeWitt et al. 2016).
Moesin is still very much a mystery player in a number of cellular processes that link the membrane and the actin cytoskeleton. Future work defining the atomic details of moesin protein-protein and protein-lipid interactions may shed light on the mechanical processes involved in moesin biological function. However, it seems likely that the mechanism by which moesin functions is common to all ERMs and progression of the field may occur via studies involving radixin or ezrin as well. Major complications have arisen in understanding the ERM proteins due to their shared evolutionary origin, high sequence identity and consequent structural similarity, the possibility of functional redundancy, and the fact they are co-expressed in many cell types.
- Barrero-Villar M, Cabrero JR, Gordon-Alonso M, Barroso-Gonzalez J, Alvarez-Losada S, Munoz-Fernandez MA, Sanchez-Madrid F, Valenzuela-Fernandez A. Moesin is required for HIV-1-induced CD4-CXCR4 interaction, F-actin redistribution, membrane fusion and viral infection in lymphocytes. J Cell Sci. 2009;122:103–13.PubMedCrossRefGoogle Scholar
- Ben-Aissa K, Patino-Lopez G, Belkina NV, Maniti O, Rosales T, Hao J-J, Kruhlak MJ, Knutson JR, Picart C, Shaw S. Activation of moesin, a protein that links actin cytoskeleton to the plasma membrane, occurs by phosphatidylinositol 4,5-bisphosphate (PIP2) binding sequentially to two sites and releasing an autoinhibitory linker. J Biol Chem. 2012;287:16311–23. doi:10.1074/jbc.M111.304881.PubMedPubMedCentralCrossRefGoogle Scholar
- Kerin T, Ramanathan A, Rivas K, Coetzee GA, Campbell DB. A noncoding RNA antisense to moesin at 5p14.1 in autism. Sci Transl Med. 2012;128:128ra40. doi:10.1126/scitranslmed.3003479.Google Scholar
- Liu X, Yang T, Suzuki K, Tsukita S, Ishii M, Zhou S, Wang G, Cao L, Qian F, Taylor S, Oh MJ, Levitan I, Ye RD, Carnegie GK, Zhao Y, Malik AB, Xu J. Moesin and myosin phosphatase confine neutrophil orientation in a chemotactic gradient. J Exp Med. 2015;212:267–80. doi:10.1084/jem.20140508.PubMedPubMedCentralCrossRefGoogle Scholar