|Home | About | Journals | Submit | Contact Us | Français|
BAR proteins comprise a heterogeneous group of multi-domain proteins with diverse biological functions. The common denominator is the Bin-Amphiphysin-Rvs (BAR) domain that not only confers targeting to lipid bilayers, but also provides scaffolding to mold lipid membranes into concave or convex surfaces. This function of BAR proteins is an important determinant in the dynamic reconstruction of membrane vesicles, as well as of the plasma membrane. Several BAR proteins function as linkers between cytoskeletal regulation and membrane dynamics. These links are provided by direct interactions between BAR proteins and actin-nucleation-promoting factors of the Wiskott-Aldrich syndrome protein family and the Diaphanous-related formins. The Rho GTPases are key factors for orchestration of this intricate interplay. This review describes how BAR proteins regulate the activity of Rho GTPases, as well as how Rho GTPases regulate the function of BAR proteins. This mutual collaboration is a central factor in the regulation of vital cellular processes, such as cell migration, cytokinesis, intracellular transport, endocytosis, and exocytosis.
The Bin-Amphiphysin-Rvs (BAR) domain proteins, or BAR proteins, form a multi-member group of proteins with diverse cellular roles.1 The common denominator is the presence of the BAR domain, which confers binding to lipid membranes.2-4 BAR domains function both as membrane-targeting motifs and as sculptors of lipid bilayers, to provide the supporting framework for curved membranes. Membrane deformation is required for membrane fission and fusion events during endocytosis, exocytosis, and vesicularization of endomembranes, such as in the endoplasmic reticulum and the Golgi complex.1 This review focuses on a subgroup of BAR proteins; namely, those BAR proteins that have direct influence on Rho GTPase signaling, either by providing the activation cues to Rho GTPases, or by inhibiting Rho GTPase activities. Some BAR proteins also have binding domains for Rho GTPases, and can thereby function as Rho effectors in various signaling cascades (Fig. 1).
There has been a relatively intense focus on the taxonomical aspects of BAR domains, and the current paradigm includes 6 classes of BAR proteins: the classical BAR, N-BAR, BAR-PH, PX-BAR, F-BAR, and I-BAR proteins.1 BAR domains are, however, not particularly well conserved in terms of their amino-acid sequence, although they all share a similar 3-dimensional fold and the core structure of a homo-dimer. The BAR domain consists of 3 extended α-helices that fold into crescent-shaped antiparallel dimers. The degree of curvature varies between the different classes of BAR domains. The classical BAR, N-BAR, and BAR-PH domains have a higher degree of curvature compared with the F-BAR and PX-BAR domains, which form a more shallow-curved “banana” shape. In contrast, the I-BAR proteins form structures that resemble “rugby balls” more than bananas. The crescent-shaped BAR domains form a concave surface with a surplus of positively charged residues that can bind to negatively charged head groups on lipid membranes. The classical BAR and N-BAR, BAR-PH and PX-BAR domains fall into this category. In contrast, although a majority of the F-BAR domains bind lipid bilayers through concave surface some of them bind through their convex surface and thereby give rise to an inverse curvature.1 This latter is also true for I-BAR proteins; all of them, with the exception of a protein known as Pinkbar interact through convex surfaces and trigger the formation of lipid membranes with inverse curvature.
The most noticeable cellular effect caused by the activity of BAR domains is their induction of membrane tubulation.4 This can be seen in vitro, by incubation of isolated BAR domains with liposomes, as well as in vivo, seen as the formation of extended invaginations and tubulations of the plasma membrane and of endomembranes. Membrane tubulation occurs, for instance, during endocytosis and the extension of tubules, which are normally regulated by proteins with membrane-fission activity, such as Dynamin5 (Fig. 2). However, the mechanisms by which BAR-domain-induced membrane deformation drives membrane tubulation are not clear. According to an interesting study, the membrane binding and membrane bending abilities of F-BAR domains are independent. According to this model, F-BAR domains do not necessarily bind to curved membranes; instead, F-BAR domains bind along their side to flat membranes. Subsequently, oligomerization of F-BAR domains results in the relocation of the curved surface to the face of the membrane, and hence drives membrane deformation.6 This ability not restricted to F-BAR proteins, N-BAR proteins like endophilin, can also assemble into regular helical scaffolds to promote membrane curvature, however, in the latter case the oligomerization was stabilized through interactions between the N-terminal helices of endophilin.7 BAR domains do not only induce membrane tubulation, they can also promote the formation of stable phosphoinostide-containing microdomains on lipid bilayers.8
The BAR proteins coordinate membrane dynamics and cytoskeletal reorganization and this coordination is central during cell migration, cytokinesis, intracellular transport, endocytosis, exocytosis, cell polarity, transcription, and tumor suppression. Rho GTPases are another group of proteins with pivotal roles in these vital cellular processes. The aim of this review is to summarize the current knowledge of the interplay between BAR proteins and Rho GTPases.
The Rho GTPases form a subfamily of the Ras superfamily of small GTPases, and these proteins cycle between their inactive GDP-bound conformations and active GTP-bound conformations.9 The GDP-bound Rho proteins are sequestered by GDP dissociation inhibitors (GDIs). Cell activation signals often impinge on this complex, and result in the release of the Rho GTPase, which can then be activated by guanine nucleotide exchange factors (GEFs). The GTPase activating proteins (GAPs) catalyze the hydrolysis of GTP, and thereby the return to the inactive GDP-bound conformation (Fig. 1).9 It is still not entirely clear how BAR proteins collaborate with Rho GTPases, or for that matter, how Rho GTPases communicate with BAR proteins. This picture is complicated because BAR proteins and Rho GTPases tend to interact in multi-protein complexes that involve additional components. These can be proteins that regulate actin polymerization, most notably the actin-nucleation-promoting factors (NPFs), such as the Wiskott-Aldrich syndrome protein (WASP) family, and the Diaphanous-related formins (DRFs).10-12 Both NPFs and DRFs contain binding sites for Rho GTPases, which means that it is difficult to determine if signaling occurs directly from the BAR proteins or from the actin regulators. There are several excellent reviews focusing on all known functions of BAR proteins, in particular they for a good entrance point for readers interested in the genetic aspects of BAR proteins.13-15 Instead, this review will discuss the interplay between Rho GTPases and BAR proteins, and will consider some novel and intriguing ideas on this complex but important matter.
F-BAR proteins were previously known as Pombe Cdc15 homology (PCH) proteins, which might be of interest to those following the early literature on the subject.16 The founding member of the F-BAR proteins, Cdc42-interacting protein 4 (CIP4), was originally identified in a yeast 2-hybrid screen for Cdc42 interaction partners.17 The subsequent characterization of CIP4 resulted in the identification of a domain in the N-terminus that can be found in a number of eukaryotic proteins, including the non-receptor tyrosine kinases FER and Fes; hence the name FER-CIP4 homology (FCH) domain. C-terminal to the FCH domain, there is a potentially α-helical region that was identified and named the Cdc15-like coiled-coil.17 These motifs are now collectively known as the FCH and BAR (F-BAR) domain.18,19 CIP4 was shown to function as a Cdc42 effector, and one specific role of Cdc42 is to regulate the subcellular localization of CIP4, something that Rac1 and RhoA do not do.17 Subsequently, 2 additional CIP4-like proteins were identified, known as FBP17 and Toca-1. The main function of the CIP4-like proteins is at the interface between cytoskeletal dynamics and membrane trafficking. For instance, CIP4 and FBP17 have important, but complex, roles in the regulation of clathrin-mediated endocytosis.20
A seminal study by Ho and colleagues placed Toca-1 as a key link between Cdc42 and the actin regulator N-WASP.21 Toca-1 was ascribed a key role in the regulation of Cdc42-induced N-WASP-Arp2/3-driven actin polymerization. In non-stimulated cells, the WASP-interacting protein (WIP) binds N-WASP in an autoinhibited conformation. Cell activation results in the dissociation of the N-WASP–WIP complex, with the release of the autoinhibited N-WASP. The role of Cdc42 appears to be to alter the affinity of Toca-1 for N-WASP and/or for WIP. In addition, translocation of Toca-1 to the plasma membrane or to internal membranes via the F-BAR domain might also alter the affinity of Toca-1 for the N-WASP–WIP complex.21 Indeed, Cdc42 might bind both Toca-1 and N-WASP, according to a recent study, which demonstrated that Cdc42, Toca-1, and N-WASP can form a trimer complex.22
Cdc42 is not the only Rho GTPase member that binds F-BAR proteins. Importantly, different Rho GTPases can target F-BAR proteins to distinct cellular locations, and this targeting might contribute to the regulation of the CIP4-like proteins. Cdc42 can relocalize CIP4 to the plasma membrane, as well as to cytoplasmic membrane tubules. In contrast, the Cdc42-related GTPase TCL targets CIP4 predominantly to the plasma membrane, where CIP4 localizes to membrane protrusions and lamellipodia.23 The specific targeting of CIP4 is important and its subcellular localization can differ between different cell types. In fibroblasts and epithelial cells, CIP4 is localized to cytoplasmic vesicles and tubules, whereas in cortical neurons, CIP4 localized to lammellipodia and peripheral protrusions.24 This difference in subcellular localization of CIP4 mediated by selective binding to different Rho GTPases might constitute a mechanism for the regulation of CIP4 function. Moreover, there are differences between the 3 CIP4-like proteins in their specificities of interaction. FBP17 does not bind Cdc42, despite a high degree of similarity in the proposed Cdc42-binding domain. Instead, FBP17 has been shown to bind TCL in a yeast 2-hybrid assay; however, it is not known if this interaction also occurs under physiological conditions.23 Furthermore, in contrast to CIP4 and Toca-1, FBP17 interacts and cooperates with the atypical Rho member Rnd2 in the regulation of dendritic spines in hippocampal neurons.25
Several recent reports have demonstrated a role for the CIP4-like proteins in endocytosis and trafficking of cell-surface receptors, such as the epidermal growth factor receptor, the platelet-derived growth factor receptor, and the transferrin receptor.19,23,26 In addition, the CIP4-like proteins have roles in trafficking of the glucose transporter GLUT4. GLUT4 is the major insulin-responsive glucose transporter for the removal of glucose from the bloodstream.27 Under resting conditions, GLUT4 has mainly a cytosolic distribution, but within minutes of insulin stimulation, the balance between exocytosis and endocytosis of GLUT4 has already shifted in favor of exocytosis, and as a result, about 50% of GLUT4 is relocated to the plasma membrane. Several studies have helped to decipher the pathway downstream of the activated insulin receptor that results in GLUT4 translocation. The critical step in this pathway is the activation of the Rho family member TC10.28,29 Interestingly, this activation occurs via the Rap1-specific GEF C3G, which has been shown to function as a GEF also for TC10.27 TC10, in turn, binds a splice variant of CIP4 (CIP4/2), which is needed for insulin-dependent translocation of GLUT4.29 Importantly, CIP4 does not bind GLUT4 directly; instead, this cellular effect is mediated via a protein, Gapex-5, that is a GEF for another category of small GTPase, Rab31. In the absence of insulin, Gapex-5 maintains Rab31 in its active, GTP-bound state, which in turn maintains a significant pool of GLUT4 storage vesicles.29 Upon insulin stimulation, TC10 is activated and relocates CIP4–Gapex-5 to the plasma membrane. As a result, the amount of active Rab31 is decreased, and GLUT4 is targeted to the plasma membrane.30
CIP4-like proteins also participate in DRF-dependent actin polymerization. The most thorough study comes from the fission yeast Schizosaccharomyces pombe, where the CIP4 ortholog Cdc15 was shown to bind the formin Cdc12, and to have a role in the organization of the contractile actomyosin ring during cytokinesis.31 It should also be mentioned that FBP17 was identified in a screen for formin-binding proteins.32 Moreover, the CIP4-like proteins bind the DRF members Dia1, Dia2, and Dishevelled associated activator of morphogenesis (DAAM1), and CIP4 participates with Src, Rho GTPases, and DAAM1 in the formation of filopodia.33 The non-receptor tyrosine kinase FER interacts and phosphorylates RhoGDIα. The tyrosine-phosphorylated RhoGDIα can then no longer bind Rac1, which suggests that the FER-induced phosphorylation results in a defect in the Rac1-sequestering of RhoGDIα.34 This might, in turn, be expected to result in increased Rac1 activity; however, this hypothesis remains to be tested. Importantly, FER-induced phosphorylation cannot affect the RhoGDIα–Rac1 interaction once this complex has been formed.34 Nervous wreck, a protein first identified in Drosophila, has also been found to work in concert with cdc42 to orchestrate actin polymerization and endocytosis during synaptic growth in the fly larvae.35
CIP4, and presumably Toca-1, interacts with Cdc42 through normal effector-loop interactions. Another F-BAR protein, Pacsin2, was recently shown to bind Rac1 through another type of interaction. Interestingly, the hypervariable C-terminal region of Rac1 was shown to mediate the Rac1 interaction with the SH3 domain of Pacsin2 at cytoplasmic membrane tubules and early endosomes.36 There is a mutual regulatory function in this interaction, as Rac1 inhibition results in the accumulation of Pacsin2 at tubular structures. Moreover, knockdown of Pacsin2 results in increased GTP-bound Rac1, and in increased Rac1-dependent cellular responses, such as cell spreading and cell migration. This demonstrates that a possible additional function of BAR proteins is to sequester Rho GTPases, and thereby to modulate their functions (Fig. 1).
The I-BAR domain was initially called IMD (IRSp53-MIM homology domain) and was identified as an N-terminal region of similarity between insulin receptor tyrosine kinase substrate p53 (IRSp53) and missing-in-metastasis (MIM).37 Overexpression of the IMD domains of IRSp53 and MIM induced the formation of filopodia, which was considered to be the result of the bundling activity of the IMD domain.37 Structural analysis of the IRSp53 IMD domain revealed a coiled-coil domain that can form 180-Å-long “zeppelin-shaped,” or “rugby-ball-shaped,” dimers.38 Subsequent studies questioned the physiological significance of this actin binding activity, as it was shown to be negligible under physiological salt conditions; instead, it was demonstrated that the IMD domain promotes interactions with phosphatidylinositide-rich lipid bilayers.39 Similar to BAR domains, the IMD domain was shown to induce membrane tubulation; however, in contrast to the classical BAR domains, this membrane deformation occurred in the opposite direction, i.e., it resulted in membrane protrusions rather than invaginations, hence the IMD was renamed inverse BAR (I-BAR).39,40 The I-BAR proteins include IRSp53, MIM, insulin receptor tyrosine kinase substrate (IRTKS), and actin-bundling protein with BAIAP2 homology (ABBA).41 The BAR domain of the fifth member here, called Pinkbar, has strong structural similarity to the I-BAR domain, but has been shown to induce flat sheets of lipids rather than membrane protrusions, and can therefore be considered as a unique type of BAR domain.42
Studies on IRSp53 identified it as a potential missing link between WAVE2 and Rac1 in a yeast 2-hybrid screen for WAVE-interacting proteins.43 IRSp53 has been shown to have different sites through which it binds Cdc42 and Rac1; the Cdc42 binding occurs through a CRIB-like motif, whereas the Rac1 binding is via the I-BAR domain itself.43-45 Also, the interaction between Cdc42 and IRSp53 is dependent on Cdc42 being in its GTP-bound conformation, whereas IRSp53 binds Rac1 in a GTP-independent manner.46 None of the other I-BAR proteins bind Cdc42, but at least MIM and IRSTK bind both GTP-bound and GDP-bound Rac1 via their I-BAR domains.46,47 ABBA has also been shown to be involved in signaling via Rac1, and although it is not clear if the cellular effects are mediated by Rac1, ABBA binds Rac1 directly.48
There have been several observations that demonstrate increased Rac1 activation in response to overexpression of I-BAR proteins (i.e., MIM and ABBA); however, the mechanism involved here is not clear, as I-BAR proteins are unlikely to have GEF activity.46,48,49 One plausible mechanisms would be via RhoGEF proteins, such as Tiam1, which has been shown to direct IRSp53 to Rac1 signaling by enhancing IRSp53 binding to both active Rac1 and the WAVE2 complex. Tiam1 also promotes the relocalization of IRSp53 to Rac1-induced lamellipodia, rather than to Cdc42-induced filopodia. A link between Tiam1 and IRSp53 is further demonstrated by the observation that IRSp53 depletion in cells prevents the lamellipodia formation induced by Tiam1 overexpression or PDGF stimulation.50 Furthermore, IRSp53-dependent Rac1 activation can be mediated by Eps8 via a complex that includes WAVE2, Abi1, and the GEF Sos-1.51,52 There are indications that this pathway is involved in filopodia formation, as Cdc42-induced filopodia are abrogated by depletion of IRSp53 or Eps8.53 It is clear that IRSp53 can interact with mDia1 and WAVE2 in the production of filopodia.54 These findings implicate a model in which activated GTP-bound Cdc42 binds IRSp53 and promotes the localization and activation of Rac via RhoGEFs. Intriguingly, the Eps8-Abi-1-Sos-1 tri-complex has an important role in metastasis of ovarian tumor cells, presumably in association with IRSp53.51,55
The simplest way for BAR proteins to activate Rho GTPases would theoretically be by combining Rho GEF and BAR domains in a single protein. However, there is only 1 protein known to have this type of domain organization, Tuba, which is a Cdc42-specific GEF.56 In addition, Tuba functions as an adaptor protein, through its binding to numerous partners by virtue of its 6 SH3 domains. Tuba is ubiquitously expressed, and several splice-variants have been described.56 Its functions have been deduced from studies performed predominantly in neuronal and epithelial cells. Like many BAR proteins, Tuba binds Dynamin and members of the WASP NPFs, and in rat brain, Tuba has been shown to be concentrated at synapses, which suggests a role in synaptic vesicle dynamics.57,58 In epithelial cells, Tuba was shown to act in a functional context linked to junctional integrity and cell polarity.59,60 Epithelial organs consist of a monolayer of polarized epithelial cells that surround a central lumen.60,61 To produce an apical surface and a lumen, this polarization requires interactions between the membrane sorting machinery and signaling pathways that define cortical domains. Cdc42 is a master regulator in this process, predominantly through its interaction with the Par6-Par3-atypical PKC polarity complex. The role of Tuba is to provide activated Cdc42 at specific sites at the apical side of epithelial cells, thereby influencing the cell junction integrity and spindle orientation.60,61 One interesting feature of Tuba is that it lacks a PH domain, which is normally present together with the catalytic Dbl homology (DH) domain in most RhoGEFs.9 Instead, Tuba has a BAR domain that is adjacent to a DH domain. As, the BAR domain is required for the membrane targeting of Tuba, it has been suggested that the BAR domain functionally replaces the PH domain. This is an attractive hypothesis, although it remains to be determined whether the BAR domain has any specific role in the activation of the catalytic activity of the Tuba Cdc42 GEF domain.
The N-BAR protein Bin3 regulates myofiber size during early myogenesis.62 Bin3-knockout cells migrate over shorter distances than wild-type cells, and have reduced levels of activated Rac1 and Cdc42. This suggests a functional link between Bin3 and Rho GTPases, although the underlying mechanism is currently not known. Finally, RhoB has been linked to Bin1 in apoptosis induced by farnesyltransferase inhibitors.63
The Slit-Robo GAPs (srGAP1–3) were originally identified as binding partners for repulsion receptor roundabout 1 (Robo1). The Slit-Robo ligand-receptor system is expressed in the developing nervous system and has crucial roles, for instance, during axon guidance.64 The functions of the srGAPs appear to be mainly in the regulation of neuronal morphogenesis and synaptic plasticity. Slit binding to the Robo1 receptor is an important determinant for the regulation of neuronal migration, and srGAP1, which is a Cdc42-specific and RhoA-specific RhoGAP, is an important signaling molecule downstream of Robo1.65,66 SrGAP3 is a Rac1-specific and Cdc42-specific GAP and a binding partner of WAVE1.67-69 Interestingly, genetic analysis of patients with severe types of mental retardation has implicated SrGAP3, which is also known as MEGAP or WRP, as one of the defective genes, and as causative of the disease. Ablation of the WAVE1 gene results in sensomotor slowing in mice, which lends support to the existence of a srGAP-WAVE1-dependent pathway in mental retardation.68 SrGAP2 has been shown to regulate neuronal cell migration and to induce neurite outgrowth and branching.70 Interestingly, srGAP2 induces filopodia-like membrane protrusions that resemble those induced by I-BAR domains, despite having an F-BAR domain.71 This I-BAR–like activity appears unique for srGAP2 and srGAP3, while the srGAP1 F-BAR domain functions as a bona fide F-BAR domain.71 There is also a hierarchical relationship between the srGAPs, as srGAP2 acts through srGAP3 in neurite outgrowth in neuroblastoma cells.72 The human minor histocompatibility antigen1 (HMHA-1) is yet another example of an F-BAR protein with a RhoGAP domain. In this case, HMHA-1 was found to stimulate GTP hydrolysis of Rac1 and RhoA.73 In summary, these observations demonstrate that F-BAR domains can be functionally diverse and can promote the formation of both protrusions and invaginations.
RICH1 (also known as Nadrin) and RICH2 are 2 other RhoGAP-containing BAR proteins. They function as Cdc42-dependent and Rac1-dependent RhoGAPs.74 RICH1 was described as a binding partner for CIP4, as an example of the communication between different types of BAR proteins. RICH1 has a role in the regulation of calcium-dependent exocytosis in neuronal cells.75 In addition, RICH1 is required for the organization of apical polarity in cells. RICH1 binding to the scaffolding protein Angiomotin targets it to a protein complex at tight junctions that contains additional polarity proteins, such as Pals1, Patj, and PAR-3. Cdc42 is an important factor for apical cell polarity in epithelial cells, and regulation of Cdc42 by RICH1 is necessary for the maintenance of tight junctions.76 RICH2 also appears to have a role in cell polarity, through interactions with a lipid-raft-associated integral membrane protein known as CD317 (or tetherin). This protein is expressed at the apical surface of polarized epithelial cells, where it interacts indirectly with the underlying actin cytoskeleton through RICH2, EBP50, and ezrin. Knocking down CD317 results in increased Rac1 activity, which is accompanied by a loss of apical microvilli.77
The RICH1-related protein SH3BP1 participates in the regulation of cell migration by downregulation of Rac1 at the leading edge. Rac1 needs to be under a dynamic regime of cycling between its GTP-bound and GDP-bound conformations. Downregulation of SH3BP1 leads to decreased cell migration and disorganized peripheral protrusions.78 SH3BP1 also has a role in junctional assembly during epithelial morphogenesis. Knock-down of SH3BP1 expression results in aberrant membrane remodeling and cell:cell junction formation.79 Finally, Gmip, an effector for the Ras-like small GTPase Gem, has a role in the regulation of stress fiber dissolution and focal adhesion disassembly.80 This appears to be of particular importance during exocytosis, as downregulation of the RhoA pathway is a critical step during exocytosis of secretory granules in neutrophils.81
Mutations in the human OPHN1 gene, which encodes the BAR-containing and RhoGAP-domain-containing protein Oligophrenin-1, results in nonspecific X-linked mental retardation.82 OPHN1 is required for dendritic spine morphogenesis, and its Rho GAP activity appears to be important in the modulation of the length of the dendritic spines.83 Inactivation of OPHN1 results in increased RhoA activity that is associated with significantly increased length of the dendritic spines. This phenotype can be suppressed by inhibition of ROCK, which is downstream of RhoA. OPHN1 is a clathrin-mediated endocytosis regulator, and moreover, it binds the N-BAR protein Endophilin A1 and regulates the endocytosis of synaptic vesicles.84 Inhibition of OPHN1 results in reduced endocytosis, and more particularly, in reduced internalization of the AMPA receptor, which is a receptor for glutamate and is an established regulator of synaptic plasticity.85 Interestingly, RICH2 has also been shown to have a role in the regulation of AMPA receptor internalization, via an interaction with a postsynaptic scaffolding protein called Shank3. Knock-down of RICH2 using RNA interference, or disruption of the RICH2-Shank3 complex, interferes with the recycling of the AMPA receptor, and thereby the control of synaptic plasticity.86
GTPase regulator associated with focal adhesion kinase-1 (GRAF1) (also known as Oligophrenin-like 1) is an OPHN1-related BAR-domain-containing RhoGAP.87 In contrast to OPHN1, which is a GAP for RhoA, Rac1, and Cdc42, GRAF1 is a Cdc42-specific, and to some extent, a RhoA-specific, RhoGAP. GRAF1 binds FAK and was originally ascribed a role in the regulation of cell adhesion. More recently, GRAF1 has been shown to have a classical BAR domain, to localize to tubulo-vesicular structures in cells, and to participate in Clathrin-independent endocytosis.88 This is yet another example of a BAR protein that links Rho signaling to endocytosis.
The prevailing model during the 1970s stated that cell migration was driven through retrograde flow that arises from endocytosis at the trailing edge of a cell, followed by directed membrane insertion at the leading edge. Cells were believed to basically “float” on a sheet of lipids.89 This model has to a large extent been proven inadequate, and instead, cell migration has been shown to be driven by dynamic reorganization of the actin filament system.90 The cytoskeletal reorganization that controls cell migration is mainly orchestrated by dynamic interactions between the Rho GTPases, the WASP family of NPFs, and the DRFs. During recent years, the concept of membrane dynamics has reappeared in the field of cell migration. Cell migration clearly requires the very intricate coordination of cytoskeletal dynamics and the machineries that control endocytosis, exocytosis, and intracellular transport of endomembranes. The cooperation between the BAR proteins and the actin regulatory NPFs and DRFs that is brought about through the Rho GTPases provides the central players in this control.
Although BAR proteins are strongly associated with actin regulation and endocytosis, it is clear that they have other important cellular roles. Moreover, genetic ablation of some BAR proteins, most notably Bin1 and amphiphysin-I, has revealed that edocytosis is largely unperturbed in these mice.91,92 This indicates that the situation is rather complex and several BAR proteins are likely to work in concert. It is increasingly clear that BAR proteins are not only functioning to induce or sense membrane curvature, they can also induce the formation of lipid microdomains that inhibit the lateral diffusion of phosphoinositites.8 These stable microdomains are associated with the formation of eisosomes in yeast, i.e., large stable protein complexes that mark the site of endocytosis.93 BAR proteins could have a similar function in eisosome formation also in higher eukaryotes, something that will need further studies.
It is worth noting that BAR proteins are not only implicated in actin dynamics and endosome function, there is an emerging awareness that they regulate a great variety of vital cellular processes such as cell polarity, gene transcription, stress signaling, and tumor promotion. If they also communicate with Rho GTPases is not clear at the moment but we are likely to see an increase in the number of cellular contexts in which we see intensive collaborations between Rho GTPases and BAR proteins.
No potential conflicts of interest were disclosed.
P.A. has been supported by grants from the Swedish Cancer Society, the Swedish Research Council, and the Karolinska Institutet.