Cell migration, phagocytosis and cytokinesis are mechanically intensive cellular processes that are mediated by the dynamic assembly and contractility of the actin cytoskeleton. GAPs (GTPase-activating proteins) control activities of the Rho family proteins including Cdc42, Rac1 and RhoA, which are prominent upstream regulators of the actin cytoskeleton. The present review concerns a class of Rho GAPs, FilGAP (ARHGAP24 gene product) and its close relatives (ARHGAP22 and AHRGAP25 gene products). FilGAP is a GAP for Rac1 and a binding partner of FLNa (filamin A), a widely expressed F-actin (filamentous actin)-cross-linking protein that binds many different proteins that are important in cell regulation. Phosphorylation of FilGAP serine/threonine residues and binding to FLNa modulate FilGAP's GAP activity and, as a result, its ability to regulate cell protrusion and spreading. FLNa binds to FilGAP at F-actin-enriched sites, such as at the leading edge of the cell where Rac1 activity is controlled to inhibit actin assembly. FilGAP then dissociates from FLNa in actin networks by myosin-dependent mechanical deformation of FLNa's FilGAP-binding site to relocate at the plasma membrane by binding to polyphosphoinositides. Since actomyosin contraction is activated downstream of RhoA–ROCK (Rho-kinase), RhoA activity regulates Rac1 through FilGAP by signalling to the force-generating system. FilGAP and the ARHGAP22 gene product also act as mediators between RhoA and Rac1 pathways, which lead to amoeboid and mesenchymal modes of cell movements respectively. Therefore FilGAP and its close relatives are key regulators that promote the reciprocal inhibitory relationship between RhoA and Rac1 in cell shape changes and the mesenchymal–amoeboid transition in tumour cells.
Over the last half-century, we have learned that the self-associating muscle proteins actin and myosin mediate the mechanical behaviours of many non-muscle cells [1–5]. Examples of such mechanical activities include locomotion, phagocytosis and cytokinesis [6–8]. Unlike the relatively static architecture of these proteins in striated muscle, non-muscle cell actin assembles reversibly from monomers into linear filaments. Numerous actin-binding proteins regulate actin assembly and organize filaments into diverse three-dimensional cross-linked network arrays to form morphologically distinct cell protrusions such as lamellae and filopodia [9–11]. Contraction of actin networks by myosin creates characteristic cell deformations associated with the mechanical behaviour described above [12–14]. Biochemical research on purified actin cytoskeletal components revealed that cell signalling intermediates including calcium, phosphorylation and phosphoinositides regulate the activities of the actin-binding proteins including myosin to mediate actin remodelling and contractility in vitro [15–19]. The discovery by Ridley and Hall 20 years ago that activation of particular Rho family GTPases creates distinctive actin-based cell structures provided the opportunity to determine how Rho proteins integrate the actions of previously identified signalling molecules to assemble these structures [20,21]. The number of components involved and the complexity of their interactions has made progress in capitalizing on this activity slow, but it has been steady . The present review focuses on how one group of these components, members of a Rho GAP (GTPase-activating protein) (also known as ARHGAP) family, FilGAP (ARHGAP24 gene product) and its close relatives (ARHGAP22 and AHRGAP25 gene products), mediate dynamic and spatial control of GTPase activities upstream of the generation of signalling cascades regulating actin assembly and myosin contractility [23–25] (Figure 1). The naming of one of these representatives as FilGAP is based on its association with FLNa (filamin A), the most abundant and ubiquitous member of an actin-binding protein family involved in the determination of actin filament architecture and that through binding to multiple other cell components links cell structure to cell function. FLNa binding of FilGAP importantly determines its spatial activity in the cell. This regulation helps to explain the basis of cell polarity associated with locomotion (defined as ‘mesenchymal cell motility’) and transitions between mesenchymal motility and a movement mode associated with cell blebbing termed ‘amoeboid’ [25,26].
Rho GTPases and their reciprocal relationship in cell morphology and migration
Rho GTPases, GAPs AND ACTIN REMODELLING
Rho, Rac and Cdc42 are ubiquitously expressed and are the most extensively researched Rho GTPases that regulate remodelling of the actin cytoskeleton [22,27–33]. RhoA activity promotes actin stress fibre bundle formation, in part through activation of ROCK (Rho-kinase) that phosphorylates myosin to activate its contractile activity (that also generates force for retraction of the cell posterior) and by inducing mDia to increase actin filament elongation . In certain cells, the hydrostatic pressure caused by myosin's contractile activity induces circumferential membrane blebbing and a rounded morphology characterized as amoeboid  (Figure 1). Rac1 activation, on the other hand, generates lamellar actin polymerization at cells’ leading edges, leading, in concert with other signalling reactions, to a familiar fan-shaped appearance of relatively flat cells in tissue culture conditions, more recently defined as a mesenchymal morphology [28,29,31–33,35,36]. Cdc42 also initiates actin assembly reactions leading to spike-like protrusions (filopodia) and actin-rich adhesion structures called podosomes [22,37–40]. The balance of these diverse activities orchestrates cell shape and migration. Although there are exceptions [41,42], there is a general antagonism between the Rac- and Rho-based activities and a coincidental activation of Rac and Cdc42 [43–48]. Cross-talk between GTPases can therefore alter the relative activity of a given Rho GTPase at a specific times and/or sites in cells. One intersection regulating the antagonism between Rho by Rac1 involves myosin. Rho–ROCK-mediated phosphorylation activates myosin contraction, whereas Rac–Pak (p21-activated kinase) 1 has an opposing effect, phosphorylating MLCK (myosin light chain kinase) and decreasing its activity [49–51]. A second control mechanism is at the level of GAPs, which can be activated by one GTPase to inactivate an opposing GTPase . FilGAP and ARHGAP22 gene products mediate reciprocal processes or specifically inactivate Rac downstream of Rho [24,25,52] (Figure 1).
STRUCTURE OF THE ARHGAP FAMILY
The ARHGAP family genes encode GAPs for ARH (Aplysia Ras-related homologue) (also called Rho) proteins, of which some, including the products of ARHGAP24 and ARHGAP25, were first identified in silico using the human genome database, but lacked functional definition until recently [23,53]. These human ARHGAP genes are located on three different chromosomes (ARHGAP24 on chromosome 4; ARHGAP22 on chromosome 10; ARHGAP25 on chromosome 2) and, through differential splicing, express a number of protein isoforms. In general, the long forms of these proteins contain an N-terminal PH (pleckstrin homology), a GAP activity region, a spacer, and a C-terminal CC (coiled-coil) domain that mediates protein dimerization. Shorter isoforms derived from alternative splicing lack functional PH domains [23,54,55] (Figure 2). In FilGAP, the long product of ARHGAP24, an FLNa-binding site resides C-terminal to the CC dimerization domain . The GAP and CC domains of FilGAP are separated by a region of 319 amino acids (residues 330–648) containing multiple serine and threonine phosphorylation sites. This connecting region is not well conserved in the protein products of ARHGAP22 and ARHGAP25 (Figure 2). Whether the CC domains of the proteins expressed by ARHGAP22 or ARHGAP25 generate dimers is unknown . Although not firmly established, the binding of FLNa to ARHGAP22 (but not to ARHGAP25) probably depends on the sequence immediately following the CC domain that has an alternating hydrophobic residue pattern complementary to a β-sheet structure in FLNa's FilGAP-binding domain  (Figure 3). The structure of the PH and GAP domains can be computationally predicted using known structures of other proteins as a template. Although the structural and biochemical characterization of the phospholipid specificity of the N-terminal PH domains have not been reported, strong similarity to the PH domains of Akt, cytohesin and GRP1 (general receptor for phosphoinositides-1), suggest that FilGAP's PH domain binds most strongly to PtdIns(3,4,5)P3 .
Domain structure of ARHGAP gene products
The FLNa–FilGAP complex and comparison of FilGAP's binding domain with other FLNa partners
The GAP activity of all the three gene products has been established. FilGAP, and its splice variant p73RhoGAP/RC-GAP72, share GAP domains, and stimulate the GTPase activities of Rac1 and Cdc42 in vitro. However, when expressed in cells containing FLNa, FilGAP activity preferentially targets Rac1 [24,59,60]. Recombinant GAP domains of proteins from ARHGAP25 also target Rac1 and show no activity for Cdc42 or RhoA in vitro , although, once again, Rac1 and Cdc42  are reported to be targets in cells. p68RacGAP, an ARHGAP22 splice variant product lacking a PH domain, specifically stimulates the GTPase activity of Rac1 in cells , and silencing it (full-length ARHGAP22) increases cellular GTP-Rac1 . In addition, a point mutation in the GAP domain of FilGAP (Q158R) has been linked to focal segmental glomerulosclerosis, which damages podocytes and causes proteinuria in affected kidneys . This mutant FilGAP has been reported to markedly increase the amount of GTP-Rac1 in cells .
FilGAP's C-terminal FLNa-binding site binds to the 23rd of 24 Ig-like repeats comprising FLNa's dimeric subunit structure (IgFLNa23) [23,56,63] (Figure 3). As with many other FLNa-binding partners, tight binding requires the participation of binding sites from both subunits of dimer molecules. Hence the interaction of a monomeric synthetic peptide (residues 723–736) mimic of the C-terminal FilGAP-binding region with FLNa is weak with a larger than millimolar dissociation constant  (Figure 3). Structural and biochemical analysis demonstrate that the FilGAP FLNa-binding peptide is a β-strand that sits in a groove formed by the C and D β-strands of IgFLNa23, in similar fashion to the β-strand-forming peptides of other FLNa-binding partners, platelet GPIbα (glycoprotein 1bα) and integrin β tails that bind in the corresponding CD grooves of IgFLNa17 and IgFLNa21 respectively [56,64,65]. Interestingly, FilGAP does not interact with FLNb or FLNc in which the grooves formed by the C and D strands in their Ig-23 repeats differ from that of FLNa. In the binding interface between FLNa and FilGAP, the amino acids indicated with asterisks in Figure 3(C) (Phe726, Thr728, Gly730, Leu732 and Val734) face the groove of the CD face (Figure 3A). These residues are conserved in the protein products from ARHGAP22, but not the ARHGAP25 gene (Figure 3C), suggesting that only the ARHGAP22 products are capable of interacting with FLNa, although their CC domains, which influences binding, also differ from FilGAPs (Figure 2).
LOCALIZATION AND TISSUE DISTRIBUTION OF ARHGAPs
FilGAP localizes in cells with FLNa in F-actin (filamentous actin)-enriched structures [24,26,60,66]. For example, stimulation of cultured cells with EGF (epidermal growth factor) or LPA (lysophosphatidic acid) induces the formation of multiple lamellae that concentrate both FLNa and FilGAP, but not a FilGAP mutant lacking 100 residues at the C-terminus . FilGAP also localizes in focal adhesions with vinculin and FLNa in podocytes [24,60]. The short FilGAP splice isoform p73RhoGAP2/RC-GAP72 also co-localizes with F-actin, in a process that again requires its C-terminal half [54,59], and concentrates primarily in focal adhesions in cells expressing low levels of the protein. Higher p73RhoGAP2/RC-GAP72 expression associates it with the entire stress fibre system of cells . In epithelial cells, RC-GAP72 accumulates at cell–cell adherens junctions . Since the short isoforms of these GAPs lack a PH domain, but possess an FLNa-binding sequence (Figure 2), they could also interact with FLNa. ARHGAP22 and ARHGAP25 gene products, on the other hand, do not appear to co-localize with F-actin [55,61,67], implying that they do not associate with FLNa. Translocation of the splice variant of the ARHGAP22 gene lacking a PH domain (p68RacGAP) from the endothelial cell cytoplasm into nucleus has been reported when co-expressed with Vezf1 (vascular endothelial zinc finger-1). p68RacGAP binds Vezf1 and this binding inhibits its transcriptional activity for the endothelin-1 promoter . Since p68RacGAP is a splice isoform of ARHGAP22, it is possible that other ARHGAP22 products interact with Vezf1.
Understanding the function of this family of proteins in a given cell type is complicated by the diverse tissue-specific expression patterns of the protein isoforms. Hence future experiments must dissect and account for all the GAPs in a given cell to define the total contribution they make to the differential regulation of GTPase activity. FilGAP is expressed ubiquitously in most cells and tissues, with the highest expression in kidney. Podocytes are particularly enriched in FilGAP, up-regulating its expression ~70-fold when they differentiate in situ [24,60]. In contrast, p73RhoGAP2/RC-GAP72 expression is restricted to vascular smooth muscle and endothelial cells , although its mRNA has been detected in a variety of tissues, with kidney again having the highest amount of mRNA . ARHGAP25 and ARHGAP22 genes are expressed in all tissues, but spleen and peripheral leucocytes have the highest level of ARHGAP25 products , whereas vascular endothelial cells have the highest levels of the ARHGAP22 gene products [55,68,69].
REMODELLING OF ACTIN CYTOSKELETON BY THE FilGAP PROTEINS
Rho, Rac and Cdc42 are the master regulators of actin cytoskeleton remodelling and actomyosin contractility in smooth muscle and non-muscle cells . Since GAPs switch off GTPase activation and terminate signalling, FilGAP and its family members should negatively regulate Rac and Cdc42 signalling cascades to suppress cellular protrusions and membrane ruffles (Figure 4). As expected, silencing of FilGAP in cells induces large lamellae and membrane ruffles, whereas forced expression of FilGAP has the opposite effect, suppressing lamellae formation and integrin-mediated cell spreading . Consistent with these findings, undifferentiated podocytes with low levels of FilGAP, have extensive membrane ruffling activity that is lost when they differentiate and FilGAP levels increase . Interestingly, expression or knockdown of a FilGAP splice variant lacking the N-terminal PH domain leads to the opposing results: transfected cells lose actin stress fibres and extend lamellae, whereas knockdown cells increase stress fibres, implying altered GTPase targeting [54,59]. Similar effects have been reported for other Rac1/Cdc42-specific GAPs such as n-chimerin and ARHGAP15. Besides altering membrane targeting, the loss of PH domain function could also influence cell motility at the level of phosphoinositide metabolism [16,70,71]. Since ARHGAP22 and ARHGAP25 gene products are RacGAPs that influence cell morphology and movement, direct effects on actin polymerization would be expected. Unfortunately, little is known concerning the cellular localization and/or dynamics of these two proteins during cell motility [61,67].
Remodelling of actin cytoskeletons by FilGAP and its close relatives
REGULATION OF GAP ACTIVITY BY PHOSPHORYLATION
ROCK phosphorylates five serine residues and a threonine residue in FilGAP downstream of GTP-RhoA  (Figures 2 and 4). Blunting phosphorylation of FilGAP by alanine substitution of all six serine/threonine residues suppressed GAP activity in vivo, although alanine substitution did not affect GAP catalytic activity or FLNa binding in vitro . Of the six phosphorylation sites in FilGAP, only Ser402 is conserved in the other family members (Ser395 in ARHGAP22 and Ser400 in ARHGAP25). Mutation of Ser402 to alanine reduced FilGAP activity in cells, but only modestly compared with the substitution of all six sites [24,26]. Since no differences in the RacGAP activity of the recombinant GAP domain alone have been observed in the presence or absence of the other FilGAP domains, it is unlikely that phosphorylation functions by relieving an autoinhibitory state . More likely is that phosphorylation of FilGAP leads to differential targeting of the RacGAP, either altering membrane binding or by exposing a new partner interaction. If this hypothesis is true, phosphorylation of FilGAP could regulate turnover of Rac through associations with other cellular components and/or by facilitating dissociation from FLNa under force [69,72] (see below).
No ROCK-dependent phosphorylation of ARHGAP22 gene products has been detected, despite the finding that the GAP activity of the proteins depends on ROCK activity . However, phosphorylation of multiple serine and threonine residues in the connecting regions between the GAP and CC domain has been detected by MS analysis in certain cells (at http://www.phosphosite.org/) . Interestingly, phosphorylation of the conserved Ser395 of ARHGAP22-encoded proteins is downstream of Akt, and promotes binding to 14-3-3 protein. Other phosphorylation sites have also been implicated in 14-3-3 binding . Mutation of phosphorylation sites, including Ser395, disrupts binding to 14-3-3, alters the GAP activity of ARHGAP22 products, and reduces cell motility . Inhibition of ROCK phosphorylation, however, does not promote 14-3-3 binding (Figure 4). These results suggest that phosphorylation regulates the GAP activity of FilGAP and ARHGAP22 protein products in cells, but that the phosphorylation pathways and activation mechanisms differ (Figure 4).
FORCE-DEPENDENT FLNa–FilGAP INTERACTION
The GAP activity of FilGAP increases for Rac when co-expressed with FLNa in cells, whereas a mutant FLNa lacking the FilGAP-binding domain has equal GAP activity for Rac1 and Cdc42 . Hence the specificity of interaction is expected to change as mechanical force is applied on FLNa–actin networks in cells [72,74]. FilGAP binds to unstressed FLNa, and dissociates when FLNa–actin networks are subjected to mechanical shear strains in vitro that correspond to physiological applied forces, such as vessel stretch and fluid shear stress driven by blood flow or when myosin motors, embedded in networks, are activated to apply collapsing forces (Figure 5). The mechanism by which force regulates the FLNa–FilGAP interaction can be inferred from the nature of their interaction  (Figures 3 and 5). Tight complex formation between FilGAP and FLNa requires the correct spacing of the four binding surfaces, two of each that are donated by each partner, which is facilitated by the flexible hinge domain between FLNa repeats 23 and 24 (Figure 3), and the more compact structure of FLNa's C-terminus. FLNa is a flexible V-shaped molecule with high-affinity actin-binding sites on the free ends that are required to link filaments into orthogonal junctions. This interaction is promoted by a second low-affinity actin-binding site on repeat 10 in the rod 1 portion of FLNa that zips up the rod 1 portion of FLNa along the actin filaments at the joints [63,75]. When the filaments in these junctions are deformed by mechanical shear stress, they are pushed apart somewhat, spatially forming a gap between the two filaments. This deformation extends the two arms of the FLNa molecule, further opening the V-shape molecule. The resulting misalignment of the FilGAP–FLNa binding interfaces reduces the binding avidity, resulting in FilGAP dissociation (Figure 5). Hence the application of mechanical force dissociates FilGAP from FLNa, which attenuates GAP activity for Rac1. Testing this model now requires direct information as to where and when filamin molecules are reacting to force displacements in cells and the generation of reagents to following the dynamics of FilGAP. It also requires the development of a new generation of force-sensing probes that can be inserted into mechanically regulated proteins [76–78].
Mechanical force regulates FLNa–FilGAP interaction by altering the geometry of the binding interfaces
Another example of a mechanical transduction involving FilGAP is the suppression of Rac-dependent lamellae assembly that occurs when tensile forces are applied to integrins connected to collagen-coated magnetic beads. Force application recruits FilGAP to the bead–cell interface, where FLNa and actin concentrate, and FLNa binding has been shown to be required for this suppressive effect . The FLNa–integrin interaction is strengthened under mechanical forces and occurs specifically at mature adhesion sites [72,79], explaining the recruitment of FLNa. However, in this case, FilGAP is collected on FLNa, not dissociated . One possible explanation for this behaviour could be that the response to force is different in adhesion sites compared with an actin network. These experiments also look at cellular responses on a timescale that is much different from the binding interactions studied in vitro where force rapidly dissociates FilGAP from FLNa, e.g. a timescale of 1 s or less [72,80]. In contrast, accumulation of FilGAP to the sites around the beads was examined after 2 h of continuous force application , which may be sufficiently long to reorganize the surrounding actin cytoskeleton.
In order to study the role of mechanical force in regulating protein function, it is essential to maintain or reconstitute force in situ or in vitro. This is not an easy task and has persisted as a major obstacle to progress in this field of research [72,76,77,81]. However, force-dependent redistribution of FilGAP could be measurable using fluorescently labelled FilGAP in cells.
SPATIOTEMPORAL DYNAMICS OF FilGAP REGULATORS
How do phosphorylation, PtdIns(3,4,5)P3 and force co-ordinately regulate FilGAP at a specific time and site in living cells? Although no study has addressed this question, the present review attempts to collate the current knowledge regarding the key molecules that regulate localization and activation of FilGAP in cell protrusions, the most extensively studied system.
In the traditional view of competing RhoA compared with Rac1 and Cdc42 activities in migrating cells, RhoA activation restricts to the retracting cell rear to promote tail contraction by activating myosin, whereas Rac1 and Cdc42 work at the front of the cell to promote actin polymerization and membrane protrusion [28,29] (Figure 6). Consistent with this hypothesis, GTP-RhoA has been detected along the sides and at the rear of polarized HL-60 cells differentiated into neutrophils, whereas the advancing cell front contained little detectable GTP-RhoA . However, other studies have shown that all three GTPases may be activated at the front of migrating cells such as fibroblasts, MDCK (Madin–Darby canine kidney) cells and HeLa cells [83–86]. Computational multiplexing analysis has revealed a dynamic pattern of Rho GTPase activation with high resolution and 1 s timescale within the cell edge where protrusion and retraction take place . RhoA activation increases and decreases in synchrony with protrusive and retractile movements in a confined 2 μm region at the cell edge. In contrast, the activities of Rac1 and Cdc42 persist for longer and cover a wider region of the cell edge, i.e. are spatially less coupled to protrusion. Activation is initiated at a distance ~1.8 μm from the edge and reach their peak of activation with a 40 s delay relative to protrusion . The spatial restriction and delay may be regulated by FilGAP concentrated by binding to a PtdIns(3,4,5)P3 that maintains Rac1 and Cdc42 in an inactive state in close proximity to plasma membrane [88,89] (Figure 6). Phosphorylation of FilGAP by ROCK and binding to FLNa would collaborate to promote further the suppressive effect of FilGAP on Rac1 in this region . Actomyosin contraction stimulated by Rho–ROCK would dissociate FilGAP from FLNa , thereby recycling FilGAP bound to FLNa to the cell edge, which attenuates its suppressive effect to facilitate actin polymerization to reinforce and stabilize newly expanded protrusions (Figure 6).
Spatiotemporal dynamics of molecules that regulate localization and activation of FilGAP during cell protrusion
MAT (MESENCHYMAL–AMOEBOID TRANSITION) IN CANCER CELL MIGRATION
Moving cells, particularly cancer cells, reciprocally switch between mesenchymal and amoeboid modes of cell movement (Figure 7). During mesenchymal migration, cells are polarized and elongated. Movement is driven by leading-edge protrusion by a Rac1-induced actin polymerization and the formation of focal interactions to ECM (extracellular matrix). Proteolysis of the ECM can weaken physical barriers to cell migration that are established near the cell front. The migration speed of the cells is slow (0.1–1 μm/min) [25,90]. In contrast, amoeboid movement is characterized by a more compact and blebbing morphology lacking consistent polarity, independence from a requirement for proteolysis of the ECM and high migration speeds (5–25 μm/min) [25,90]. Cells in amoeboid mode adapt their shapes to squeeze through pores in the ECM by Rho–ROCK-mediated actomyosin contractions at the cell rears  (Figure 7).
FilGAP- and ARHGAP22 protein-mediated transition between mesenchymal and amoeboid modes of tumour cell migration
siRNA (short interfering RNA) screens identified ARHGAP22 products as mediators that suppresses Rac1 downstream of RhoA activation in melanoma cells . Silencing of ARHGAP22 mRNAs increased active RacGTP levels 2-fold and promoted AMT (amoeboid–mesenchymal transition). In addition, pharmacological inhibition of ROCK and myosin independently increased the activity of Rac1 and converted the cells into an elongated morphology, suggesting that actomyosin contractility regulates ARHGAP22-mediated inactivation of Rac downstream of ROCK  (Figure 7B). Nevertheless, the precise biochemical mechanism leading to activation of ARHGAP22 products downstream of RhoA is not known. One possibility is that the Rho/ROCK/PTEN (phosphatase and tensin homologue deleted on chromosome 10) pathway regulates GAP activity of ARHGAP22 products [69,92–94] (Figure 4).
Given the similarity of the domain structure of ARHGAP22-encoded proteins to FilGAP and ARHGAP25 products, all ARHGAP family proteins may also mediate the antagonism between RhoA and Rac1 and regulate AMT. In melanoma cells, RacGTP levels increase modestly (1.2–1.3-fold) when mRNAs for FilGAP and ARHGAP25 expression is silenced, and cell morphology is not transitioned . In carcinoma cells, on the other hand, silencing of FilGAP promotes AMT, whereas forced expression of FilGAP induces the opposite following Rho/ROCK-mediated phosphorylation of FilGAP  (Figure 7A). The study cited also demonstrated that inhibition of myosin activity by blebbistatin significantly suppresses FilGAP-induced bleb formation and promotes AMT, presumably through FLNa-mediated mechanotransduction [26,72] (Figures 5–7).
Despite coincidental activation of Rac and Cdc42 in general, the role of Cdc42 in regulating MAT is distinct from that of Rac1 in the cells investigated to date . DOCK10 (dedicator of cytokinesis 10)/Cdc42/N-WASP (neuronal Wiskott–Aldrich syndrome protein) signalling induces actin polymerization, whereas DOCK10/Cdc42/Pak2 stimulates actomyosin contractility by activating myosin light chain and contributes to amoeboid motility. Cdc42 also activates Rac1 and can lead to mesenchymal movement . Further investigation is required to determine whether any of the ARHGAPs are involved in these pathways.
CONCLUSION AND PERSPECTIVES
The last 10 years have seen significant progress in our understanding of the functions of FilGAP and its close relatives in cell migration and MAT, yet many issues remain unresolved. (i) Although the atomic structure of each protein and subdomain (PH, GAP and CC domains) can be inferred on the basis of comparison with other related proteins, structural elucidation remains unreported except for the FLNa-binding site. In addition, the structures of the full-length proteins in complex with partners remain to be solved. (ii) The roles of the PH domain have not been addressed. Interaction with phosphoinositides through the PH domain may confine the GAP activity at a specific site in cells or may attenuate activity of other molecules downstream of phosphoinositide. Studies comparing the function of the long form with the splice variants lacking the PH domain should help to elucidate the function of the PH domain. (iii) Regulation of the GAP activity of ARHGAP22 products by actomyosin contraction has been hypothesized during MAT. However, the molecular mechanism of this regulation is unknown. FLNa could be a mediator for such regulation, but a direct interaction of ARHGAP22 proteins with FLNa has not been demonstrated. In line with this question, (iv) empirical support is required for the idea that mechanical force regulates GAP activity and translocation of FilGAP in living cells. (v) The kinases and phosphatases that regulate the phosphorylation of the ARHGAP protein family require clarification. Finally, (vi) since FLNa is a scaffold not only for GAPs but also for RhoGTPases, GEFs (guanine-nucleotide-exchange factors), integrins and many other effectors [65,72,74,96–101], understanding the co-ordination of all activities through FLNa is critical. These studies will help to define the mechanics of cell migration, as FLNa deficiency in cancer cells significantly reduces their migration and invasion . The ultimate goal of this work would be to help lead to a rationally designed drug to target specific FLNa–partner interfaces such as FilGAP to slow pathological cell migration processes.
dedicator of cytokinesis 10
phosphatase and tensin homologue deleted on chromosome 10
vascular endothelial zinc finger-1
I thank Dr Thomas P. Stossel and Dr John H. Hartwig for their critical reading of this paper before submission.
This work is supported by the HUSEC (Harvard University Science and Engineering Committee) Seed Fund for Interdisciplinary Science and the National Institutes of Health [grant number HL19749].