Rho GTPases, phosphoinositides, and actin

Abstract

Rho GTPases are well known regulators of the actin cytoskeleton that act by binding and activating actin nucleators. They are therefore involved in many actin-based processes, including cell migration, cell polarity, and membrane trafficking. With the identification of phosphoinositide kinases and phosphatases as potential binding partners or effectors, Rho GTPases also appear to participate in the regulation of phosphoinositide metabolism. Since both actin dynamics and phosphoinositide turnover affect the efficiency and the fidelity of vesicle transport between cell compartments, Rho GTPases have emerged as critical players in membrane trafficking. Rho GTPase activity, actin remodeling, and phosphoinositide metabolism need to be coordinated in both space and time to ensure the progression of vesicles along membrane trafficking pathways. Although most molecular pathways are still unclear, in this review, we will highlight recent advances made in our understanding of how Rho-dependent signaling pathways organize actin dynamics and phosphoinositides and how phosphoinositides potentially provide negative feedback to Rho GTPases during endocytosis, exocytosis and membrane exchange between intracellular compartments.

Introduction

Intracellular membrane traffic governs most aspects of cell homeostasis and behavior by appropriately and accurately transporting vesicles between membranous organelles. The diversity of organelles and the vast array of transported components imply that vesicle delivery has to be stringently regulated to guarantee the fidelity and efficiency of vesicle transport and targeting. Since the early 1990s, two main classes of proteins have been identified as “master regulators” of membrane trafficking: the Rab and Arf subfamilies belonging to the small GTPases of the Ras superfamily, and proteins from the family of soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs). Work over the years has demonstrated that these proteins constitute spatial landmarks of vesicular pathways and regulate many aspects of membrane trafficking, including cargo selection during vesicle budding, vesicle transport along actin and microtubule filaments, vesicle tethering to target membranes, and eventually membrane fusion to deliver vesicle contents.^1,2 At the same time, another subfamily of the small GTP-binding proteins of the Ras superfamily, the Rho GTPases, has emerged as new regulator of the actin cytoskeleton, one of the major short range carriers of vesicles in trafficking pathways.^3,4 Therefore, Rho GTPases are also potential regulators of membrane trafficking.

Since the identification of RhoC in 1985,⁵ the family of Rho GTPases has expanded to 20 members, divided into 8 subfamilies (Rho, Rac, Cdc42, RhoD/F, Rnd, RhoU/V, RhoH, and RhoBTB),⁶ which now tend to be classified into two major groups, the canonical (Rho, Rac, Cdc42, RhoD/F) and the atypical ones (Rnd, RhoU/V, RhoH, and RhoBTB).⁷ This classification has evolved from their distinct regulatory modes. The canonical class follow the general scheme of GTP hydrolyzing enzymes, cycling between an inactive GDP-bound and an active GTP-bound form with the aid of guanine nucleotide exchange factors (GEF) and GTPase activating proteins (GAP).^8-10 The GDP-bound form is predominant and mostly found in complex with a guanine dissociation inhibitor (GDI). The GDI stabilizes Rho proteins in the cytosol by masking the post-translationally added lipid moiety that serves to anchor Rho GTPases in cellular membranes.^10,11 However, Rho GDI regulation is limited to RhoA, Rac1, Rac2, and Cdc42.¹² In contrast, no GEF, GAP, or GDI have been clearly identified for atypical Rho GTPases and biochemical studies indicate that they are mostly in an active state, bound to GTP and associated with membranes. Their mode of regulation (either positive or negative) is controlled at the transcriptional level and/or by targeted degradation. Compared with canonical Rho proteins, most atypical Rho GTPases possess additional domains that mediate protein-protein interactions and these are likely to be important for their regulation and function.^7,13

The first evidence of vesicular trafficking controlled by Rho GTPases came from seminal works of Alan Hall’s group who demonstrated that active Rac1 stimulated the uptake of extracellular fluid by macropinocytosis.³ Since then, Rho GTPases have been implicated in many different aspects of membrane trafficking, that rely (or may rely) on Rho-dependent reorganization of the actin cytoskeleton. This occurs through the interaction of Rho GTPases with actin nucleators from the formin and the WASP family, which regulate actin polymerization.^7,14-16 In addition, Rho GTPases interact with various kinases and phosphatases that play a role in regulating actin dynamics and phosphoinositide turnover and both these processes are crucial for membrane trafficking.^17,18 Comprehensive reviews are available on the function of Rho GTPases in the regulation of particular membrane trafficking process.^15,19-23 However, work over the last two decades has shown that Rho GTPases regulate virtually all kinds of exocytic and endocytic processes including constitutive-, polarized-, and regulated-exocytosis, clathrin-mediated endocytosis (CME),^24,25 detergent resistant membrane (DRM)-dependent endocytosis,²⁶ pinocytosis,²⁷ macropinocytosis,³ and phagocytosis.²⁸ In this review, we compile current evidence indicating how Rho GTPases may control vesicle progression through these trafficking pathways by regulating local actin dynamics, phosphoinositides turnover, or function of complexes involved in vesicle tethering or fusion. Figure 1 summarizes where Rho GTPases intervene in each type of exo- and endocytosis pathways.

Figure 1. Summary of membrane trafficking pathways regulated by Rho GTPases. The figure represents cell compartments and the vesicular pathways in which Rho GTPases have a regulatory role. They act by altering the dynamics of the actin cytoskeleton, or the composition in phosphoinositides of the membranes or both. To simplify the scheme, some vesicles are represented with several transmembrane proteins, which are not necessarily present in the same vesicle or even in the same cell type. RhoA, Rac1/2, and Cdc42 are mostly localized at the plasma membrane when activated (green circle). Cdc42 and TC10 are also found at the Golgi apparatus and Rac1 binds AP1A at the TGN. RhoB, RhoD, and TCL(RhoJ) are enriched in subsets of endosomes or endosome domains encompassing early, late and recycling endosomes. RhoB and RhoD selectively target Src kinases members to the plasma membrane and TCL ensure efficient TfR recycling. RhoC, RhoF, TCL, RhoU (Wrch-1), and RhoV (Chp) have similar membrane locations as other Rho members, mostly endosomes and plasma membrane, but their knock-down impairs constitutive secretion by an unknown mechanism. Despite common subcellular location, Rho GTPases differentially regulate endocytosis and exocytosis. Clathrin-mediated endocytosis is blocked by active RhoA and Rac1 (cross), but is unaltered by Cdc42, irrespective of its activation state. Macropinocytosis depends on Rac1 and RhoG, which cycling are dependent on PI(3,4,5)P₃ levels: RhoG for its activation and Rac1 for its deactivation (bicolor circle). Both FcR- and CR3-dependent phagocytosis require RhoG. Particle engulfment relies on Rac1, Rac2 (noted as Rac) and Cdc42 if bound to FcR or RhoA if bound to CR3. Importantly, completion of FcR-dependent phagocytosis depends on Cdc42 deactivation. Clathrin-independent endocytosis (CLIC/GEEC, DRM, caveolae) depends on different subsets of Rho GTPases. Although GPI-AP and IL2-R accumulate in cholesterol-enriched membrane domains, their endocytosis is differentially regulated by Rho GTPases. Cdc42 is dispensable for IL2R endocytosis, but its activation and/or deactivation cycle is needed for GPI-AP endocytosis. IL2R endocytosis needs PI(3,4,5)P₃-dependent Rac1 activation by Tiam-1, as well as RhoA activity. During cell migration, some integrin endocytosis is regulated by caveolin, which in turn regulates RhoA, Rac1, and Cdc42 activities or expression levels. In addition, RhoG is critical for caveolin-dependent integrin turnover at the plasma membrane promoting efficient cell migration in response to Syndecan-4. Finally, during exocytosis, different subsets of Rho GTPases control polarized and regulated exocytosis. TC10 (RhoQ) and Cdc42 are both required for polarized exocytosis and Cdc42 may directly control vesicle fusion by acting on SNARE proteins. During regulated exocytosis, differences exist between specialized secretory cells and the type of vesicle released. In response to insulin, GLUT4 exposure to the cell surface of adipocytes relies on Rac1, TC10, and Cdc42 with different activation kinetics: TC10 and Cdc42 act early to mobilize vesicles that are docked to the plasma membrane, whereas Rac1 recruits vesicle from the storage pool to sustain GLUT4 exocytosis. TC10 deactivation is necessary to complete vesicle fusion. In chromaffin cells and neutrophils, RhoA deactivation is needed to bring secretory granules to the plasma membrane and to allow vesicle fusion. Cdc42 and Rac1 are activated by secretagogues and are needed for efficient exocytosis: Cdc42 acts on actin polymerization and Rac1 increases the production of fusogenic lipids (phosphatidic acid) in chromaffin cells.

Rho GTPases and Endocytosis

Phospholipid metabolism and actin dynamics in endocytosis

Cell surface proteins, lipids, or extracellular fluids enter endocytic pathways by different mechanisms, which depend on various key molecules, such as specific receptors, clathrin, dynamin, caveolin, or lipid rafts, but also on actin dynamics and phosphoinositide metabolism.^29,30 Although the initial cue and the extent of actin assembly at an endocytic site differ between different modes of endocytosis,³¹ the general scheme for progression through endocytic pathways is conserved and requires dynamic actin remodeling and a sequential conversion of phosphoinositides. For example, actin polymerizes extensively to form a large cup around receptor-bound pathogens during phagocytosis or around extracellular fluid during macropinocytosis. In contrast, less polymerized actin is observed during clathrin-mediated endocytosis (CME) or clathrin-independent endocytosis.^31-33 In CME, actin is even dispensable at the initial step of cargo recruitment into coated pits. However, clathrin-coated vesicle maturation, budding, and progression along the endocytic route requires actin reorganization,^31,34-37 indicating that despite differences in amounts of actin polymerization, actin dynamics ensure efficient endocytosis.^34,36-38

The initial step of most endocytic processes depends on the synthesis of PI(4,5)P₂ and its subsequent conversion into different phosphoinositide species (PI(3,4,5)P₃, PI(3)P, PI(4)P, and PI(3,4)P₂) by phosphoinositide kinases and phosphatases.^18,35,39-44 For example, CME starts with the recognition of cargo and PI(4,5)P₂ by adaptor proteins like AP-2. Clathrin subsequently stabilizes the complex and, with the help of accessory proteins (epsins, endophilin, and amphiphysin, for example), the coated membrane bends to form a coated vesicle, which buds until concomitant dynamin-dependent fission and conversion of PI(4,5)P₂ into PI(4)P by 5-phosphatases (such as synaptojanin) occur.^45-48 During phagocytosis, the generation of PI(4,5)P₂ by type I phosphatidylinositol 4-kinase 5-phosphate (PIP5K) is also required for Fc receptor (FcR) clustering and initiation of particle engulfment.⁴⁹ Subsequent reduction of PI(4,5)P₂ by PLC, PLD, or PI 3-kinase is necessary for phagocytosis to proceed.⁵⁰ Therefore, phosphoinositide switches and actin dynamics constitute major regulatory elements for endocytotic processes.

By binding actin nucleators belonging to the WASp and the Diaphanous formins families, Rho GTPases have been shown to control the formation of large actin-based structures involved in the maintenance of cell shape and the control of cell migration.^19,51,52 In addition to being a signaling intermediate between Rho GTPases and actin, actin nucleators control actin dynamics at discrete steps of endocytic processes.^53-56 Noteworthy, actin binding proteins are also regulated by phosphoinositides, which can form anchoring sites at membranes and/or unfold proteins to locally promote or inhibit actin polymerization.^57,58 Finally, phosphoinositide conversion during endocytosis involves PIP5K,^59,60 class I PI 3-kinase,^61,62 isoforms of the phospholipase C family,⁶³ and phosphoinositide phosphatases such as synaptojanins^64,65 or OCRL,^66,67 all of which have been shown to bind and/or be activated by RhoA, Rac1, or Cdc42.^68-74 Therefore, by controlling phosphoinositide metabolism and actin polymerization, Rho GTPases are likely to play a critical role in regulating endocytosis. How Rho GTPases interfere with phosphoinositide metabolism and actin dynamics to regulate different endocytic events will be discussed next.

Clathrin mediated endocytosis

The first evidence for a function of Rho GTPases in CME came from studies of transferrin uptake by cells overexpressing constitutively active Rho GTPases. Endocytosis of transferrin (Tf) was blocked by active RhoA and Rac1, but not by active Cdc42, and interestingly, this inhibition occurred independently of actin rearrangement.²⁴ With the identification of PIP5K and synaptojanin-2 (Synj2) as potential binding partners for RhoA and Rac1,^68,73,75 it was proposed that Rho GTPases control vesicle progression through the endocytic pathway by imbalancing PI(4,5)P₂ production. Rac1-dependent recruitment of Synj2 at the plasma membrane was indeed sufficient to diminish CME of Tf by increasing PI(4,5)P₂ hydrolysis⁷³ and knocking down Rac1 by siRNA increased Tf uptake,⁷⁶ indicating that Rac1 may negatively regulate CME of Tf by promoting PI(4,5)P₂ hydrolysis. However, despite the fact that RhoA and Rac1 stimulate PI(4,5)P₂ production through PIP5K activation^68,75 and that sustained production of PI(4,5)P₂ by overexpression of PIP5K is sufficient to increase Tf uptake,^68,75,77 RhoA and Rac1 activation blocked, whereas RhoA and Rac1 inhibition increase CME of Tf. These results are thus inconsistent with the sole function of Rho GTPase to control changes in PI(4,5)P₂ levels during CME. Rho GTPases may serve another function, such as regulating the recruitment of accessory proteins at endocytotic sites. For example, the endophilin-A1, a Bin-amphiphysin-Rvs (BAR) domain protein, which is necessary for the completion of clathrin-dependent endocytosis,^78,79 is a substrate for the Rho-associated kinase (ROCK1). When phosphorylated by ROCK1, endophilin-A1 cannot recruit Synj in clathrin-coated pits and results in defective endocytosis of the EGF receptor.⁸⁰

It should be mentioned that most experiments before early 2000s were performed using expression of Rho GTPases locked in either a constitutive active state or a dominant negative form. Although very useful and informative, particularly in the absence of readily available tools, caution has to nonetheless be taken when epitope-tagged GTPases mutants are overexpressed. They may be mislocalized and the extent of overexpression may disturb Rho GTPases pathways.^12,81-84 With the discovery of gene silencing in plants and animals and their use as a tool to knock-down the expression of endogenous proteins in mammalian cells,⁸⁵ single gene knock-downs, and also unbiased screen assays using large scale or even genome-wide RNAi libraries have been developed to systematically address the consequence of endogenous protein knock-down in biological processes. For example, the vesicular stomatitis virus (VSV) uses CME to enter the cell, and in an assay designed to identify kinases involved in VSV entry, silencing of PAK1—a well-known effector of Rac1 and Cdc42—was found to increase VSV infection, indicating that by analogy to Rac1 silencing for Tf uptake, Cdc42/Rac1 pathway may have an inhibitory effect on VSV entry.⁸⁶

A genome wide RNAi screening assay was also developed to identify molecular components that regulate endocytosis of both EGF and Tf. This study again highlighted the need for Rac1, but also for RhoD in both of these endocytic pathways. In contrast to the previous cited study,⁷⁶ knocking down Rac1 was reported to not increase Tf uptake but, like RhoD, to increase clustering of Tf-positive endosomes close to the nucleus suggesting that these GTPases interfere with vesicle displacement and intervene in endosome maturation process.⁸⁷ Interestingly, this assay confirmed that different subsets of adaptor proteins are required for CME of EGF and Tf.^87-89 For example, whereas knocking-down the clathrin heavy chain blocks both Tf and EGF uptake, silencing of AP-2 only inhibits Tf endocytosis. Likewise, knocking-down signaling intermediates or regulators of Rho GTPases usually interferes with both EGF and Tf uptake to similar extents, but some Rho regulators have a selective effect on either EGF or Tf uptake (Table 1).⁸⁷ This suggests that CME of a given receptor is selectively controlled by specific downstream signaling pathways dependent on the cycling of Rho GTPases that is determined by activation of diverse GAPs and GEFs. It remains to be determined whether defects observed in CME are indeed a consequence of an alteration in Rho GTPase cycling. In addition to their conserved Rho GEF or Rho GAP, most regulators of Rho GTPases possess other binding domains that may alter CME independently of Rho GTPase activity. For example, the Cdc42 GEF Intersectin1 and Interectin2 localize to clathrin coated pits through AP-2 binding and regulate both Tf and EGF receptor endocytosis, but the involvement of Cdc42 into CME is still rather elusive.^90-92

Table 1. Effect of knocking-down proteins involved in Rho GTPases pathways on transferrin and EGF endocytosis or on endosome distribution (data were extracted from searchable database at http://endosomics.mpi-cbg.de/; Collinet et al., 2010).

Proteins are color coded: Rho GEF in green, Rho GAP in red, Rho GTPase in blue, Rho effectors in black and Rho GDI in black and italicized. *, selective modulation of endocytosis but with accumulation of endosomes in the cell center. **, decrease in EGF endocytosis but increase in transferrin endocytosis.

Macropinocytosis

Unlike CME, macropinocytosis and phagocytosis require extensive actin rearrangements that form a cup at initial steps of the endocytic processes. After internalization, actin depolymerizes. Macropinocytosis is characterized by actin-dependent formation of dorsal membrane ruffles, which occur spontaneously or in response to many growth factors, including PDGF, EGF, M-CSF, or HGF.⁹³ Dorsal ruffles are different from peripheral membrane ruffles induced by the same growth factors. They depend on different signaling intermediates and their formation is delayed compared with peripheral ruffles. The Rho GTPase Rac1 is involved in both types of membrane ruffles by promoting actin polymerization, but dorsal ruffles relies on WAVE-1-dependent actin remodeling, whereas peripheral membrane ruffles depend on WAVE-2.⁵³ In addition, unlike peripheral ruffles, dorsal ruffles need functional Rab5 and PI 3-kinase indicating that phosphoinositide production and Rac1 activity may be correlated.⁹⁴

A detailed kinetic analysis of macropinosome formation has provided spatiotemporal insight into the kinetics of phosphoinositide metabolism and Rac1 activation. Upon EGF treatment, PI(4,5)P₂ progressively increases in membrane ruffles and PI-3 kinase-dependent production of PI(3,4,5)P₃ peaks before cup closure.⁹⁵ Monitoring Rac1 activity by FRET microscopy during macropinocytosis showed that Rac1 activity is correlated with PI(3,4,5)P₃ production and both reach their maximum in dorsal ruffles prior to macropinosome closure.⁹⁶ Interestingly, light-dependent activation of Rac1 is sufficient to trigger macropinosome formation and increase PI(4,5)P₂ production, but the macropinosome could not close unless Rac1 is deactivated.⁹⁷ Since the cup cannot close in the absence of PI 3-kinase activity, a PI(3,4,5)P₃-sensitive Rac GAP may be required for completion of macropinocytosis. Due to the organization of their molecular domain, regulators of Rho GTPases activity are intricately linked to phosphoinositide metabolism. Indeed, the molecular signature of a Rho GEF is a tandem of DH-PH domain (except for GEF of the DOCK family), which confers nucleotide exchange activity toward Rho proteins via the DH domain, and binds to membrane with differential affinity for phosphoinositide species through the PH domain.^98,99 Although Rho GAPs have more diverse domains, they often possess lipid binding domains (PH, C2, PX, or BAR domains) in addition to their RhoGAP domain.⁹⁸ Up to now, only GEFs for RhoG (SGEF and PLEKHG6) have been found to be sensitive to PI(3,4,5)P₃ and involved in macropinocytosis.^100-102 Intriguingly, although stimulation of EGF and PDGF receptors induces Rac1-dependent dorsal ruffles and macropinocytosis, only RhoG is rapidly activated in response to EGF and necessary in this case for dorsal ruffle formation.¹⁰¹ This raises the question of whether dorsal ruffle formation and macropinocytosis induced by different growth factor are comparable. For example, dorsal ruffles induced by HGF have all the hallmarks of macropinosome initiation: they are dependent on PI 3-kinase, Rab5, and Rac1 activities. However, although macropinocytosis has been shown to be independent of clathrin,¹⁰³ HGF stimulation of cells silenced for clathrin did not form dorsal ruffles due to the absence of Rac1 activation on endosomes by one of its exchange factor Tiam-1.^104,105 Since fluid phase uptake was not systematically evaluated in those studies, correlating dorsal ruffle formation and macropinocytosis is difficult. One example of decoupling between membrane ruffles and macropinocytosis has been described for immature dendritic cells, which shows intense membrane ruffling coupled to macropinocytosis. When Rac is inhibited, macropinocytosis is blocked without disturbing membrane ruffling.¹⁰⁶

Although the correlation between actin rearrangements, phosphoinositides, and macropinocytosis needs to be clarified, Rac1 and RhoG have nonetheless emerged as important Rho GTPases for macropinocytosis. They may also participate in late phases of macropinocytosis, since their common effector PAK1 activates CtBP1/BARS, a potential dynamin counterpart needed for scission of the macropinosome.^107-109

Phagocytosis

Like macropinocytosis, phosphoinositide turnover and Rho GTPase cycling are critical for phagocytosis, but their involvement depends on the membrane receptor engaged to engulf particles or dead cells. The two best-characterized types, Fc receptor (FcR)- and CR3-dependent phagocytosis, activate different F-actin polymerization pathways triggered by Cdc42/Rac and RhoA signaling cascades, respectively.¹¹⁰ During FcR-dependent phagocytosis, PIP5Kα-dependent accumulation of PI(4,5)P₂ controls both FcR clustering required for particle attachment and actin polymerization, which initiates phagocytic cup formation.⁴⁹ In PIP5Kγ knockout cells, RhoA and Rac1 activities are up- and downregulated respectively, and either inhibiting RhoA or activating Rac1 restores particle attachment and phagocytosis.^41,49 This indicates that the defect in PI(4,5)P₂ production alters basal levels of Rho GTPase activities and that, although mostly involved in CR3-dependent phagocytosis,^111,112 RhoA may be needed at a discrete step of FcR-dependent phagocytosis for polymerizing actin and promoting initial FcR clustering.⁴⁹ FRET experiments designed to examine Cdc42, Rac1, and Rac2 activation showed that their activities are temporally segregated during phagosome cup formation and do not necessarily correlate with actin enrichment. At the time of particle binding and the initiation of particle engulfment, active Cdc42 and Rac1 are localized in the extending pseudopods where actin is enriched. During phagosome cup formation and before closure, active Rac1 and Rac2 are found around the phagosome in regions devoid of actin.¹¹³ PLC, PLD, and PI 3-kinase contribute to the reduction in PI(4,5)P₂ levels necessary for phagocytosis to progress.⁵⁰ PI 3-kinase is dispensable for Cdc42 and Rac1 activation during phagosome formation, but required for Cdc42 deactivation and phagocytic cup closure,¹¹⁴ indicating that, a PI(3,4,5)P₃-dependent Cdc42 GAP may be required, as observed for Rac1 during macropinocytosis. Interestingly, Rac1 has been recently shown to increase PI(3,4,5)P₃ levels by directly binding to the p110β subunit of PI 3-kinase.⁷⁰ As the p110β plays a major function in FcR-dependent phagocytosis in macrophages,¹¹⁵ an intriguing possibility would be that Rac might deactivate Cdc42 by stimulating the production of PI(3,4,5)P₃ and thereby ensures phagocytosis progression.

The involvement of Rac1 in FcR-dependent phagocytosis has been recently questioned in RNAi screening assay testing for all Rho GTPases members during FcR and CR3-dependent phagocytosis. In addition to Cdc42, FcR-dependent phagocytosis requires Rac2, but not Rac1. Instead, RhoG has been found to be necessary for phagocytic cup formation and is activated during both FcR and CR3-dependent phagocytosis.¹¹² Until recently, RhoG was only implicated in apoptotic cell clearance mechanism, triggering Rac-dependent actin remodeling by forming a multimolecular complex with the adaptor ELMO1 and the Rac GEF DOCK180.¹¹⁶ As RhoG acts upstream of both Rho and Rac,¹¹⁷ RhoG may have a more general function in phagocytosis by coordinating phosphoinositide signaling and Rho GEF activities in response to the engagement of specific receptors.¹¹⁸ It remains to be established whether a comparable signaling cascade exists between RhoG and Rac1 during apoptotic cell engulfment and FcR-mediated phagocytosis.

Other clathrin-independent endocytosis: CLIC/GEEC, DRM, and caveolae pathways

Endocytic pathways described in the previous section rely on extensive actin polymerization or clathrin-dependent endocytosis. RhoA, Rac1, Cdc42, and RhoG have been also shown to control clathrin-independent endocytosis,^{27,32,119,120} which mainly depends on actin dynamics, caveolae and cholesterol-enriched lipid clusters in the plasma membrane.²⁹ For example, Cdc42 controls GPI-anchored protein (GPI-AP) endocytosis, which is independent of clathrin, dynamin, or caveolin.²⁷ GPI-AP and Cdc42 need to be concentrated into cholesterol-rich nanodomains to promote local actin polymerization and direct GPI-AP into a specific endosomal compartment (GEEC) resulting from the fusion of uncoated tubulovesicular clathrin-independent carriers (CLIC).^27,32 Intriguingly, dominant-negative Cdc42 redirects GPI-AP uptake toward a clathrin-dependent endocytic route.²⁷ For normal GPI-AP endocytosis, Cdc42 needs to be deactivated by ARHGAP10, which is recruited by Arf1 to nascent endocytic vesicles. GRAF-1, another Cdc42 GAP¹¹⁹ is necessary for efficient GPI-AP uptake, but unlike ARHGAP10, GRAF1 is mostly located in tubulovesicular structures devoid of Cdc42, and not in Cdc42-positive pinocytic vesicles. The relationship between GRAF1 and Cdc42 is thus unclear, but progression of GPI-AP through the CLIC/GEEC endocytic pathway requires an intact Cdc42 activation cycle.^121,122 These studies also indicate that inhibition of a specific Rho GTPase may divert cargo from their normal route and highlight the versatility of endocytic processes.

Together with some other integral membrane proteins such as interleukin receptors, GPI-AP accumulates in detergent resistant membrane (DRM). However, endocytosis of GPI-AP and the interleukin 2 receptor (IL2R) are differentially regulated by Rho proteins. IL2R endocytosis required RhoA and Rac1, but not Cdc42.²⁶ Interestingly, whereas PI 3-kinase have been shown to be mostly required for large particle or fluid uptake occurring during phagocytosis and macropinocytosis, the p85 subunit of PI 3-kinase appears necessary for IL2R endocytosis by recruiting activated Rac1 to IL2R complex. Rac1 activation is mediated by the PI(3,4,5)P₃-dependent recruitment of Vav2 to IL2R endocytic vesicle,¹²³ and progression through the endocytic pathway is ensured by PAK1-dependent cortactin phosphorylation and formation of a cortactin N-WASP complex.^124,125 This indicates that PI(3,4,5)P₃-dependent Rac1 activation and local actin polymerization are essential for IL2R endocytosis.

Caveolins are integral membrane proteins that bind cholesterol and serve as building units for the formation of small (50–80 nm), rounded invaginations in the plasma membrane called caveolae. Caveolae are relatively stable structures, but can detach from the plasma membrane and form endocytic vesicles when triggered by specific signals.¹²⁶ Among the protein trafficking controlled by caveolins, the regulation of adhesion molecules such as integrins constitute a major anchorage-dependent growth checkpoint that is overridden in pathological conditions like cell transformation. Upon binding to the extracellular matrix, integrins are well-known activators of Rho GTPases and the absence of integrin engagement deactivates Rac1 and releases it from the plasma membrane.¹²⁷ In Caveolin-1 (Cav-1) knockout cells, Rac1 remains active at the plasma membrane even in the absence of integrin engagement indicating that Cav-1-dependent endocytosis regulates Rac1 activity and location.^128,129 Cav-1 may directly reduce Rac1 and Cdc42 activities by increasing Rac1 degradation or acting as a GDI for Cdc42.^130,131 RhoA activity is also indirectly altered by the increased Src activity observed in Cav-1 knockout cells. Src phosphorylates and activates p190RhoGAP, which diminishes RhoA activity.¹³² It is thus not surprising that in Cav-1 knockout cells, polarized cell migration is defective due to impaired turnover of adhesion structures.¹³² Interestingly, in response to syndecan-4, cell migration is also impaired when Cav-1 and RhoG-dependent endocytosis of inactive integrin complex is inhibited. This study places RhoG as a downstream effector of Cav-1 necessary to redeploy integrin complexes at the cell surface and subsequent activation of Rac at the leading edge of migrating cells to ensure directed cell migration.^133,134 Cav-1 appears to coordinate both integrin trafficking and Rho GTPase signaling. Thus blocking Cav-1 interaction with Rho GTPases may be an efficient way to reduce metastatic potential of tumor cells.¹³⁵

Finally, among pathogens, many viruses or bacteria produce virulence factors that alter Rho GTPases activities and/or exploit Rho GTPases pathways to infect and invade cells.^136,137 Systematic gene silencing experiments in host-pathogens interaction assays further point out to crosstalk that exist between Rho GTPases signaling and caveolins or cholesterol-rich membrane. For example, the SV40 virus uses caveolins to enter the cell. Systematic silencing of human kinases identified two Rho effectors implicated in SV40 entry. Knocking-down the Cdc42 effector ACK1 blocks whereas knocking-down PAK1 increases SV40 virus entry.⁸⁶ The bacteria Salmonella typhimurium invades gut tissues by injecting virulence factors into epithelial cells causing diarrhea. Some injected factors activate Rac1 and Cdc42 to trigger membrane ruffles and help bacteria to invade cells. In the absence of the coat protein complex I (COPI), cholesterol-rich membranes are redistributed from the plasma membrane to perinuclear region. As a consequence, Rac1 and Cdc42 are mislocalized and absent from the plasma membrane. Membrane ruffles cannot form and bacteria invasion is prevented.¹³⁸ Altogether these studies highlight the role of cholesterol enriched membrane in the control of Rho GTPase activities, which in turn may control the fate of plasma membrane components (proteins or pathogens) or caveolae-dependent endocytic routes.

Endocytic routes and vesicle recycling

Once the vesicle has formed at the plasma membrane, cargo are either recycled back to the plasma membrane or routed toward degradative pathways. It has long been thought that most cargo meet a common population of early endosomes, but recent evidence suggests that instead, early endosomes are a heterogeneous population of vesicles that may condition the fate of the cargo.^139,140 RhoB and RhoD have been localized to early endosomes based on their colocalization with Rab5. They control progression of vesicles through the endocytic pathway by promoting actin polymerization on endosomes that is initiated by Diaphanous-related formins in a Src-dependent manner.^141-143 In addition, RhoB controls the degradation of EGF and CXCR2 receptors, which is necessary to switch off their signaling.^144,145 RhoD has never been shown to direct vesicles toward degradative pathways, suggesting that it might at least in part, target them to a different set of endosomes. This is further emphasized by the subcellular localization of Src kinase family members. They possess different post-translational modifications that target them to different subsets of endosomes. Knocking-down RhoB or RhoD selectively limits the abundance of Src members in the plasma membrane indicating that RhoB and RhoD control and/or direct them to different subsets of endosomes.^145-147 Other Rho GTPases have been shown to control the fate of membrane receptors. For example, RhoJ (TCL) is required for Tf recycling to the plasma membrane.¹⁴⁸ The nature of the endosome subsets targeted by the Rho GTPases remains to be characterized.

Rho GTPases and Exocytosis

Vesicle formation at the Golgi apparatus

Membrane pinch-off from a donor compartment is necessary to generate vesicles that will incorporate cargo to be delivered to a target membrane. Coat protein complex I (COPI), complex II (COPII) and clathrin form a cage, which all contain an inner layer of adaptor proteins (AP). These coats drive the budding of vesicles at distinct locations.¹⁴⁹ Whereas clathrin coat function is restricted to post-Golgi membranes (plasma membrane, endosomes, and TGN), COPI and COPII coats act on endoplasmic reticulum (ER) and Golgi to form vesicular carriers that follow bidirectional transport. So far, among Rho GTPases, only Cdc42 has been localized in the Golgi.¹⁵⁰ By binding to COPI subunits, Cdc42 regulates ER to Golgi transport and, importantly, protein exit from the ER depends on Cdc42 cycling between an inactive and an active state.¹⁵¹ How Cdc42 controls ER to Golgi trafficking is still unclear, but one obvious possibility is the ability of Cdc42 to mediate local actin polymerization in early steps of vesicle formation by recruiting and activating the N-WASP-Arp2/3 complex.^152-154 Interestingly, actin polymerization at the Golgi, as well as the recruitment of Cdc42, requires Arf1, which in turn recruits ARHGAP10¹⁵⁵ and the machinery to promote vesicle scission.¹⁵⁶ As Cdc42-dependent actin polymerization at the Golgi inhibits dynein recruitment to COPI vesicles,¹⁵⁷ these data support a model in which local and transient Cdc42-dependent actin polymerization may help the coatomer bend membrane and form vesicles whereas Arf1-dependent inactivation of Cdc42 may favor vesicle formation and dynein-dependent transport on microtubules. In adipocytes, TC10 (RhoQ), a close relative of Cdc42, may also control secretory vesicle trafficking through N-WASP-dependent actin polymerization and COPI recruitment.¹⁵⁸ There is an intriguing parallel between GPI-AP endocytosis and the formation of secretory vesicles. Both involve Cdc42 deactivation, Arf1 and ARHGAP10 suggesting that this tripartite module may coordinate clathrin-independent endocytosis and secretory pathways. Notably, GPI-AP containing vesicles have been found to be the major membrane supplier for membrane expansion when cells spread during cell adhesion and during phagocytosis^159,160 and Cdc42 is needed for the recycling of major histocompatibility complex of class I.¹⁶¹

Clathrin and the AP-1 adaptor mediate the trafficking of specific cargoes from the TGN to the endosomal system.¹⁶² Recently, Rac1 was found to associate with AP-1A and promote actin polymerization at the TGN once activated by βPIX, a known GEF for Rac1.^163,164 These studies further identified a molecular network involving Arf1-dependent activation of Rac1, and subsequent N-WASP-dependent polymerization of actin which was necessary for the biogenesis of clathrin-AP-1 coated carrier formation at the TGN. Intriguingly, Cdc42 is not involved in this pathway indicating that Rac1 and Cdc42 control actin polymerization and protein transport along different routes of the secretory pathway. In addition to its effects on actin, Rac1 may also regulate the lipid composition of the TGN membrane. Rac1 colocalizes at the TGN with the bifunctional protein OCRL that possesses a RhoGAP domain and a 5-phosphatase activity for PI(4,5)P₂.⁷⁴ Whether Rac1 modulates OCRL activity or whether OCRL has a bona fide GAP activity toward Rac is unclear,¹⁶⁵ but OCRL and Rac may help to maintain PI(4)P levels at the TGN and enhance the binding of AP-1,¹⁶³ which, combined with local actin polymerization, may promote vesicle formation.

Constitutive and polarized exocytosis

The secretory pathway consists of transporting vesicles from the Golgi to the cell surface. The general mechanism to release vesicles containing secretory products is highly conserved, and relies on sequential steps at the plasma membrane, consisting of vesicle tethering, docking, priming, and finally vesicle fusing with the plasma membrane. Whereas constitutive exocytosis occurs constantly and maintains the plasma membrane composition, polarized exocytosis requires the abundant delivery of membrane and proteins to specific spatial landmarks. The first evidence of a role for Rho GTPases in vesicle fusion came from yeast studies in which temperature-sensitive Rho GTPase mutant fail to divide because of defects in bud growth in Cdc42 mutant and vesicle accumulation in the daughter cell in Rho3 mutant. In both cases, post-Golgi vesicles form normally, but do not fuse at budding site leading to a decreased supply of the membrane necessary for bud growth.^166,167 In vitro experiments have unraveled the potential mechanism of Rho function by showing that a blocking peptide against Rho3 and Cdc42 prevents vacuole fusion. Sequential analysis of the fusion reaction revealed that Rho3 and Cdc42 have no effects on vesicle tethering but alter the docking step and subsequent SNARE-dependent fusion events.^168-170 Interestingly, the actin cables necessary for delivering vesicles to budding sites are unaltered in these mutants, however, in vitro, actin polymerization on vacuoles is defective and prevents efficient fusion. Whereas actin cable formation depends on Cdc42 and formins,¹⁷¹ vacuole fusion rely on Cdc42 and the yeast WASp and WIP homolog, as well as Arp3.¹⁷² This suggests that Rho GTPase-dependent actin polymerization, as well as the type of actin filaments formed at specific sites of vesicle docking regulates vesicle fusion.

Cdc42 has been also shown to control polarized exocytosis in higher eukaryotes through the Cdc42-Par6-αPKC pathway.^173-176 In organs, epithelial cells are the building units of tubes such alveoli and cysts, and they are oriented with their apical surfaces facing the central lumen and their basolateral membranes joining neighboring cells. In a culture system recapitulating cystogenesis, Cdc42 provides the membrane necessary for lumen formation.¹⁷⁷ Interestingly, an unbiased RNAi screen directed against regulators of Rho GTPases has confirmed the importance of Cdc42 and Cdc42 GEFs (Intersectin2 and Tuba) during lumen formation and revealed the requirement for other Rho GEFs known to activate RhoA and RhoG (Lbc/AKAP13 and SGEF respectively).¹⁷⁸ While RhoA controls cell polarity by regulating cell contractility through the actomyosin system and by stabilizing tight junctions,^179,180 no function for RhoG in epithelial cell polarization has been reported yet. The fact that SGEF activity is dependent on PI(3,4,5)P₃ and that PTEN phosphatase, which converts PI(3,4,5)P₃ into PI(4,5)P₂, is required for apical localization of Cdc42 and normal cystogenesis¹⁷⁷ suggest that, like in endocytic processes, PI(3,4,5)P₃ levels need to be reduced to establish normal lumen formation. Whether, RhoG controls an early step of cystogenesis by initiating/promoting cell polarity is an interesting possibility, especially since RhoG participates in the establishment of front-rear cell polarity during cell migration.¹⁸¹

The exocyst complex, consisting of 8 subunits (Sec3, Sec5, Sec6, Sec8, Sec10, Sec15, Exo70, and Exo84), mediates the tethering of secretory vesicles at the plasma membrane before SNARE-mediated fusion occurs. This is particularly important for polarized exocytosis in which tethering of secretory vesicles and exocytosis is sustained at a given site to permit cell membrane expansion or polarized secretion. Interactions between Rho GTPases and the exocyst subunits were first reported in yeast. Exo70 and Sec3 form a landmark of bud formation and promote exocytotic-dependent cell surface expansion through their interaction with Cdc42, Rho3, and Rho1.^22,182 The relationship between Rho GTPases and the exocyst complex is conserved and controls different aspects of polarized exocytosis in mammals. It relies mostly on the interaction of Exo70 with two closely related GTPases, Cdc42, and TC10. During neurite outgrowth in response to NGF or IGF, the polarized exocytosis of vesicles at the growth cone requires the integral exocyst complex and TC10 activation.^183-185 Although the relationship between PI(3,4,5)P₃ and TC10 activation is unknown, it is noteworthy that PI 3-kinase activation is needed for growth cone expansion in response to IGF-1, and that PI(3,4,5)P₃ accumulates in the distal region to sustain vesicle fusion. This increases IGF-1 receptor exposure to the extracellular medium, which may contribute to self-reinforcement of neurite outgrowth in response to IGF.^184-187 Conversely, the phosphoinositide phosphatase PTEN induces growth cone retraction and neurite collapse.¹⁸⁸ TC10 is homogeneously localized at the plasma membrane, but is activated at sites of growth cone expansion where PI(3,4,5)P₃ is enriched and at discrete sites of spine formation where it recruits Exo70. In response to NGF, the TC10-Exo70 complex antagonizes Cdc42-N-WASP-dependent actin polymerization, which is, nonetheless, required for normal neurite growth. This indicates that a subtle balance between Cdc42 and Exo70-TC10 signaling has to be preserved.^184,189 Interestingly, TC10 deactivation by p190RhoGAP-A appears to be necessary for vesicle fusion.¹⁹⁰ Thus, once the vesicle has fused, Cdc42-N-WASP inhibition may be relieved to promote actin-dependent neurite elongation.

Focal exocytosis during phagocytosis or metalloprotease release in invadopodia also requires Cdc42 and a functional exocyst complex.^191,192 Interestingly, during phagocytosis, Rab11 overexpression is able to supplant both Cdc42- and N-WASP-deficiencies and rescue phagocytosis,¹⁹¹ suggesting that sustaining the membrane flow toward the plasma membrane may be sufficient to ensure phagocytosis. This mechanism supposes the presence of a functional tethering complex. Knocking-down the Exo70 subunit blocks Rab11-dependent membrane supply for phagocytosis and for transferrin receptor exocytosis, and overexpression of Exo70 is sufficient to override Rab11 knock-down.^191,193 The fact that trafficking and small GTPase pathways are interconnected for efficient delivery of vesicles is further exemplified during cystogenesis in which Rab11 controls the polarized localization of Cdc42 during lumen formation.¹⁹⁴ In addition, in cells knocked-down for PIP5Kγ, polarized exocytosis of integrin β1 at the leading edge of migrating cells is inhibited due to defects in Rab11-dependent supply of vesicles and local PI(4,5)P2 production and exo70 recruitment.¹⁹⁵ The role of Cdc42 has not been addressed in this context, but since knocking-down PIP5Kγ may alter Rho GTPases activity,⁴⁹ this provides an additional clue about the existence of a conserved framework in which Rab11, Cdc42 and maybe other Cdc42-like proteins such as TC10, cooperate to regulate membrane fusion at sites of membrane expansion. Finally, Rho GTPases may also control fusion steps by directly regulating the SNARE fusion machinery. For example, syntaxin is phosphorylated by the kinase ROCK, which inhibits neurite outgrowth by preventing vesicle fusion and SNARE complex assembly.¹⁹⁶ A recent siRNA screen has identified several Rho GTPases that may control constitutive secretion of transmembrane proteins. Knocking-down RhoC, RhoF, RhoJ (TCL), RhoU (Wrch-1), or RhoV (Chp) reduces exposure of secreted proteins at the cell surface.¹⁹⁷ Although the molecular mechanisms remain to be explored, this study clearly points out that other less well-characterized Rho GTPases are potential regulators of constitutive secretion.

Regulated exocytosis

In contrast to constitutive exocytosis, regulated exocytosis is triggered by a burst of intracellular calcium in response to an external stimulus. This mode of secretion occurs mostly in specialized secretory cells including neurons, neuroendocrine cells, and granulocytes, to name a few. Most secretory cells possess a dense cortical actin network, which acts as a barrier to prevent inappropriate fusion of secretory granules with the plasma membrane in resting conditions. Upon stimulation, depolymerization of cortical actin, together with local actin repolymerization, is needed for efficient secretion.^198-200 RhoA, Rac1, Cdc42, and TC10 have been shown to control different aspects of regulated exocytosis by modulating actin dynamics, but also phosphoinositide production in different cell systems. In mast cells, serotonin and histamine are stored in secretory granules that are released in response to antigen-mediated cross-linking of IgE. Activated Rac1 and Cdc42 stimulate exocytosis both by PLCγ-dependent production of Ins(1,4,5)P₃, which leads to a rise of intracellular calcium and PAK1-dependent actin remodeling at the plasma membrane.^201-203 In chromaffin cells, knocking-down Rac1, Cdc42 or their respective GEFs, βPIX, or Intersectin-1, inhibits secretagogue-induced exocytosis in PC12 cells.^204,205 Interestingly, whereas membrane depolarization induced by a high potassium concentration activated Rac1 and Cdc42, only Cdc42 was found to induce N-WASP-dependent actin polymerization at the plasma membrane.^204,206,207 Rac1 instead activates PLD1, which produces phosphatidic acid (PA) at the exocytic site facilitating secretory granule fusion.^204,208 In neurons, actin dynamics is not required for fast neurotransmitter release from synaptic vesicles,²⁰⁹ but Rac1 and PLD1 are both present on synaptosomes. Since both are required for neurotransmitter release,^210-212 this suggests that, like in chromaffin cells, Rac1 may regulate exocytosis through PA production at the plasma membrane rather than through actin dynamics and remodeling.

Maintenance of glucose homeostasis in the body relies on regulated exocytosis of insulin by β-pancreatic islet cells, which in turn stimulates the translocation of the glucose transporter GLUT4 from intravesicular stores to the plasma membrane in adipocytes and muscle cells. Rho GTPases are implicated in both processes. Insulin release in response to glucose increase requires Cdc42 and Rac1. Interestingly, the kinetics of Cdc42 and Rac1 activation is different with a rapid activation/deactivation of Cdc42 corresponding to the first phase of insulin release mobilizing docked secretory granules for fusion, and a slower Rac1 activation that is required for sustained release of insulin from the storage pool granules.^213,214 Differential activation may be regulated by differential binding and phosphorylation of RhoGDI or Cav-1 that interacts with Cdc42 bound GDP. Upon glucose stimulation, Cav-1 is phosphorylated and allows for β-PIX-dependent Cdc42 activation. Subsequently, Cdc42-activated PAK1 triggers the release of Rac1 from GDI to promote Tiam-1-dependent Rac activation and sustained insulin release.^131,215-217 In contrast to neuroendocrine cells in which Cdc42 and Rac1 have been localized to plasma membrane and Cdc42-dependent actin polymerization facilitates exocytosis, Cdc42 and Rac1 have been localized to secretory granules in β-pancreatic cells and Cdc42-dependent actin polymerization inhibits insulin release. Cdc42 may directly control vesicle fusion of docked vesicles by interacting in its active state with proteins from the SNARE complex,^218,219 whereas a Rac1-dependent actin rearrangement may be required to bring secretory granules close to the plasma membrane.²²⁰

In adipocytes and muscle cells, insulin triggers the translocation of the GLUT4 transporter from intravesicular store to the plasma membrane. This exocytosis process depends on actin rearrangement and two Rho GTPases, Rac1 and TC10.²²¹ In adipocytes, insulin stimulation triggers tethering of vesicles carrying GLUT4 to sites of exocytosis via the association of Exo70 with activated TC10.²²² Rac1 and TC10 are necessary for actin remodeling and vesicle translocation, but may act at two different steps dependent on two different classes of PI 3-kinases. Under physiological concentrations of insulin, active Rac1 stimulates GLUT4 translocation to the plasma membrane. Unlike TC10, Rac1 activation, requires ClassI PI 3-kinase and P-REX-1, a PI(3,4,5)P₃-regulated Rac GEF, providing a link between PI 3-kinase and Rac activity.²²³ On the other hand, translocation and docking of GLUT4-positive vesicles requires the formation of PI(3)P at the plasma membrane by actvating a TC10-dependent classII PI 3-kinase.^224,225 Since TC10 can control both the recruitment of vesicles at the plasma membrane by binding to the Exo70 exocyst subunit and actin polymerization through N-WASP activation,^{158,184,222,226} this implies a differential role for TC10 and Rac1 in the control of GLUT4 exocytosis. TC10 controls vesicle transport and docking during the initial phase, while Rac1 sustains exocytosis by maintaining high levels of GLUT4 at the plasma membrane. .

Finally, RhoA has been also involved in the control of regulated exocytosis in neuroendocrine cells and neutrophils. Overexpression of activated RhoA induces cortical actin polymerization, but unlike Cdc42, inhibits exocytosis in neuroendocrine cells.^227,228 Similarly, maintaining high levels of RhoA activity by silencing the Rho GAP GMIP in neutrophils prevents myeloperoxidase exocytosis from azurophilic granules,²⁰⁰ suggesting that RhoA needs to be deactivated for efficient exocytosis. Whether the observed actin depolymerization in response to stimulation is the result of Rho inactivation is still elusive. Alternatively, RhoA has been found to generate PI(4)P by activating PI-4 kinase on secretory granules.²²⁹ As in yeast, PI(4)P may be necessary for myosin-regulated transport of secretory granules to the cell periphery,²³⁰ and reduced RhoA activity may free secretory granules from preexisting actin filaments to favor their docking and fusion with the plasma membrane.

Conclusion

These advances in understanding Rho GTPases function in membrane trafficking emphasize their pleitropic role in endocytosis, exocytosis and vesicle segregation inside the cell. Signaling cascades are gradually being deciphered, and it does not come as a surprise that actin rearrangements constitute the cornerstone of Rho GTPase signaling. However, Rho GTPases also interfere with phosphoinositide signaling, which can then feedback on their own activities. How Rho GTPase signaling, actin rearrangements, and phosphoinositide signaling are spatiotemporally coordinated remains a challenging question. With their multidomain architecture, Rho GEFs and Rho GAPs are well suited for processing and integrating these multiple signaling entries and unraveling their regulatory roles will undoubtedly provide further insight into how this tripartite framework controls membrane trafficking.

Glossary

Abbreviations:

AP-2

adaptor proteins complex 2

BAR

Bin-amphiphysin-Rvs

CLIC

Clathrin-Independent Carrier

CR3

complement receptor 3

CtBP1/BARS

C-terminal-binding protein1/BFA–ADP-ribosylation substrate

CXCR2

chemokine (C-X-C motif) receptor 2

DH

Dbl homology domain

EGFR

epidermal growth factor receptor

FRET

Förster Resonance Energy Transfer

GPI-AP

Glycosylphosphatidylinositol anchored proteins

HGF

hepatocyte growth factor

IGF-1 receptor

insulin-like growth factor 1 receptor

M-CSF

macrophage colony stimulating factor 1

NGF

nerve growth factor

N-WASP

neural Wiskott-Aldrich syndrome protein

PAK1

p21 protein (Cdc42/Rac)-activated kinase 1

PDGF

platelet-derived growth factor

PH

pleckstrin homology domain

PI(3)P

Phosphatidylinositol-3-phosphate

PI(3,4)P2)

Phosphatidylinositol (3,4)-bisphosphate

PI(3,4,5)P3

Phosphatidylinositol-3,4,5-triphosphate

PI(4)P

phosphatidylinositol-4-phosphate

PI(4,5)P2

Phosphatidylinositol 4,5-bisphosphate

PLC

phospholipase C

PLD

phospholipase D

PX

Phox homology domain

ROCK1

Rho-associated kinase

TGN

Trans-Golgi network

WASP

Wiskott-Aldrich Syndrome Protein

WAVE

WASP-family verprolin-homologous proteins

10.4161/sgtp.29469