John A. Legg and Laura M. Machesky are at the University of Birmingham School of Biosciences, Division of Molecular Cell Biology, Birmingham, B15 2TT, UK. l.m.machesky@bham.ac.uk
The MRL protein family is emerging as a new connection between signalling, adhesion and cell motility. Two members, lamellipodin and RIAM, can regulate both actin assembly and cell adhesion, which raises many questions about how their interactions with Ena/VASP proteins, Ras-related GTPases and membranes orchestrate these key processes.
Signals from the environment are connected to cellular shape and movement through large protein assemblies at the plasma membrane that link to the actin cytoskeleton. This includes regulation of proteins such as the Ena/VASP protein family and the Arp2/3 complex that directly control actin polymerization. Both of these sets of proteins regulate cell protrusions termed lamellipodia, which are important for driving the leading edge of cells and enabling them to move and modulate their shape. In the October issue of Developmental Cell, Lafuente et al. and Krause et al. introduce a new family of proteins that contain membrane-association domains, signalling domains and Ena/VASP-binding sequences and which probably function as signal adaptors to modulate the actin cytoskeleton1,
2. The two proteins studied, Lamellipodin and RIAM (Rap1−GTP-interacting adaptor molecule, also known as RARP1, for retinoic acid response proline-rich protein 1; ref. 3) have a common domain structure, but seem to have distinct functions in signalling, motility and adhesion.
The Ena/VASP proteins (in mammals, Mena, VASP and EVL) are defined structurally by a proline-rich core domain, which is important for interaction with the small actin-binding protein profilin, flanked by amino-terminal EVH1 and carboxy-terminal EVH2 (Ena/VASP homology 1 and 2) domains. The EVH1 domain recognises the consensus (D/E)-FPPPP-X(D/E)(D/E) sequence and is important for the localization of Ena/VASP proteins to focal adhesion complexes. The EVH2 domain binds to both G- and F-actin and is thought to be important for elongation of actin filaments. To discover new potential EVH1-domain-binding proteins, Krause et al. screened databases for proteins that contain FPPPP motifs. Through this approach, they identified Lamellipodin, so named for its ability to colocalize with the Ena/VASP proteins in lamellipodia and filopodia. Closer examination of the sequence revealed that Lamellipodin is a homologue of the Caenorhabditis elegans Mig10 protein.
In a separate screen, Lafuente et al. used a yeast two-hybrid screen to identify new binding proteins for the small GTPase Rap1a. Using a constitutively active Rap1E63 mutant as bait, they identified a potential interacting protein and named it RIAM. RIAM also seemed to have the potential for cytoskeletal interactions: it contains six putative profiling-binding sites and six FPPPP motifs. They proceeded to show that RIAM could indeed bind Ena/VASP proteins, both in vitro and in vivo.
Lamellipodin, RIAM and Mig-10 share numerous structural properties and are related to the Grb7 adaptor proteins (Fig. 1). Owing to the similarity to Mig-10, Lafluente et al. have proposed naming the RIAM-related adaptor molecules the 'MRL family' (Mig10/RIAM/Lpd family). The MRL family contain pleckstrin-homology (PH) and Ras-association (RA) domains along with a more divergent proline-rich C terminus, which contains the potential profilin- and Ena/VASP-binding sites. All three proteins also contain a conserved patch of 27 residues that are predicted to form a coiled-coil region. Thus these proteins are likely to have both structural and functional similarities, although the binding specificities of each of these domains may be unique.
Figure 1. RIAM and Lamellipodin are part of the MRL protein family (Mig10/RIAM/lamellipodin).
(a) RIAM, Lamellipodin and Mig-10 share a conserved domain structure and are related to the Grb7, Grb10 and Grb14 adaptor proteins. PR1 and PR2, proline-rich domains; RA, Ras-association domain; PH, pleckstrin-homology domain; SH2, Src-homology-2 domain; yellow star, putative coiled-coil domain. (b) Dendrogram of the MRL and Grb7 full-length proteins.
In both studies, the roles of RIAM and Lamellipodin in vivo were further investigated using overexpression and knockdown experiments. When lamellipodia are deficient for Ena/VASP proteins, these structures protrude slower, but more persistently, leading to increased cell translocation rates4. Conversely, when Ena/VASP are overexpressed, lamellipodia contain longer, less branched filaments. Krause et al. found that when Lamellipodin is overexpressed the resulting phenotype is similar to Ena/VASP overexpression and that this phenotype can be suppressed by blocking Ena/VASP function. Knockdown of Lamellipodin with short-hairpin RNA showed severe lamellipodia and F-actin depletion, suggesting that Lamellipodin might signal to actin regulatory proteins.
Lafuente et al. showed that overexpressed RIAM colocalized with actin at the leading edge of the cell, consistent with the localization of Ena/VASP proteins. In Jurkat T cells, RIAM overexpression had two effects: it led to the formation of extensive lamellipodia and caused increased 1 and 2 integrin-mediated adhesion. This latter phenotype was particularly important given that the authors originally found RIAM based on its ability to bind to Rap1, which controls adhesion by 1 and 2 integrins. Rap1, the closest relative to Ras, is a small GTPase that is activated through numerous receptors, including receptor tyrosine kinases, cytokine receptors, cell adhesion molecules and heterotrimeric G-protein coupled receptors (reviewed in ref. 5). Furthermore, both Rap1 and Rap2 are implicated in actin-based processes, such as cell spreading, the extension of membrane projections and integrin-mediated cell adhesion6. Lafuente et al. proceeded to test whether the interaction of RIAM with Ena/VASP proteins might also be relevant for adhesion. It was not: although knockdown of RIAM using shRNA showed that RIAM was necessary for Rap1-induced adhesion, disrupting the interaction of the Ena/VASP proteins with RIAM had no effect on the ability of RIAM to mediate adhesion. Thus, the effect of RIAM on adhesion is likely to be through Rap1 and not Ena/VASP.
The study from Krause et al. suggests that Lamellipodin can also affect adhesion. In cell adhesion assays, Lamellipodin had the opposite effect to RIAM, negatively regulating adhesion. Unlike RIAM, the RA domain of Lamellipodin does not seem to interact with Rap1; although in another study, it interacted with K-Ras, N-Ras, H-Ras and R-Ras-3 through this region7. As RA domains have differing specificities for small GTPases (ref. 7), further studies will be required to assess which small GTPases physiologically associate with RIAM and Lamellipodin and how this affects their cellular activities.
In addition to the RA domains, RIAM and Lamellipodin contain PH domains — protein modules of 100−120 residues that bind to phosphoinositides — which could be used for interactions with lipid membrane compartments and/or with proteins (Fig. 2). Indeed, Krause et al. provide evidence that the PH domain in Lamellipodin binds to phosphatidylinositol(3,4)bisphosphate (PtdIns(3,4)P2). PtdIns(3,4)P2 levels can be regulated by extracellular signals and this lipid is preferentially distributed at the plasma membrane and at the nuclear membrane, although its role in cell motility is unclear8. Because the PH domain in lamellipodin localizes to the plasma membrane, Krause et al. propose that the binding of PtdIns(3,4)P2 to Lamellipodin could be a mechanism by which Ena/VASP proteins localize at the leading edge. No phosphoinositide has so far been found to bind to the PH domain of RIAM, although the combined interaction of a phosphoinositide at the PH domain and Rap1 to the RA domain could in theory be important for the localization of RIAM. PH domains have also been implicated in GTPase activation; for example, the PH domain of ELMO binds to the Rac GEF Dock180 and stabilises the nucleotide-free state of Rac, thereby aiding the GEF activity of Dock180 (ref. 9). This trimeric complex is also implicated in phagocytosis and cell migration. Thus, one possibility is that the PH domain of RIAM and Lamellipodin might also cooperate with a GEF to activate GTPase(s) associated with their RA domains.
Figure 2. Comparison of the domains and putative activities of Lamellipodin and RIAM at the plasma membrane.
Lamellipodin contains an RA (Ras-association) domain that may associate with Ras or R-Ras GTPases, a PH (pleckstrin homology) domain that associates preferentially with PtdIns(3,4)P2 and up to four binding domains for Ena/VASP proteins. Lpd seems to be crucial for Ena/VASP localization at the plasma membrane and for the dynamics of lamellipodial actin assembly. By contrast, the RA domain of RIAM associates with Rap1 and the PH domain has not been characterized. Association with Rap1 mediates -1 and -2 integrin adhesion. The C-terminal portion of RIAM contains several proline-rich motifs that could bind up to 6 Ena/VASP molecules and possibly link actin assembly to adhesion.
Previous studies have shown that the C-terminal EVH2 domain of Ena/VASP is sufficient to complement the loss of Ena/VASP function in random cell motility10 and that the EVH1 domain alone fails to mediate robust targeting of full-length Ena/VASP proteins to the leading edge. EVH1-domain binding to adhesion proteins, such as vinculin, has also been implicated in recruiting Ena/VASP proteins to focal adhesion complexes. Therefore, it will require further study to determine the importance of these interactions between the EVH1 domains of Lamellipodin or RIAM and the Ena/VASP proteins, as well as how these interactions affect the location and function of these proteins in actin-based protrusions.
The identification of RIAM/RARP1 as a protein upregulated by ATRA (all-trans retionic acid) differentiationof HL-60 cells3 adds to the mystery surrounding the function of MRL proteins. This study implicated RIAM in transcriptional regulation and possibly the differentiation of myeloid cells, including T-cells and megakaryocytes/platelets. Whether the other MRL proteins are also connected to differentiation and transcriptional regulation remains to be determined and understanding how this might tie in with their actin cytoskeletal functions will be equally important.
1 and 
