Summary
Small GTPases are known to regulate hundreds of cell functions. In particular, Rho family GTPases are master regulators of the cytoskeleton. By regulating actin nucleation complexes, Rho GTPases control changes in cell shape, including the extension and/or retraction of surface protrusions and invaginations. Protrusion and invagination of the plasma membrane also involves the interaction between the plasma membrane and the cortical cytoskeleton. This interplay between membranes and the cytoskeleton can lead to an increase or decrease in the plasma membrane surface area and its tension as a result of the fusion (exocytosis) or internalization (endocytosis) of membranous compartments, respectively. For a long time, the cytoskeleton and plasma membrane dynamics were investigated separately. However, studies from many laboratories have now revealed that Rho GTPases, their modulation of the cytoskeleton, and membrane traffic are closely connected during the dynamic remodeling of the cell surface. Arf- and Rab-dependent exocytosis of specific vesicles contributes to the targeting of Rho GTPases and their regulatory factors to discrete sites of the plasma membrane. Rho GTPases regulate the tethering of exocytic vesicles and modulate their subsequent fusion. They also have crucial roles in the different forms of endocytosis, where they participate in the sorting of membrane domains as well as the sculpting and sealing of membrane flasks and cups. Here, we discuss how cell surface dynamics depend on the orchestration of the cytoskeleton and the plasma membrane by Rho GTPases.
Introduction
The cytoskeleton not only provides structural architecture in the cytoplasm, but also forms a cortical shell around the cell periphery that is composed primarily of actin filaments that are strictly associated with the cytoplasmic face of the plasma membrane. This association, which depends primarily on many weak bonds between membrane proteins (e.g. integrins, adhesion proteins and ion transporters) and specific actin-associated proteins (e.g. ankyrin, filamin, myosins, spectrins, talin and ezrin), provides the plasma membrane with specific properties that are lacking in intracellular membranes (Fig. 1). The plasma membrane exhibits high mechanical resistance and high tension. Moreover, it undergoes frequent structural changes, with local extension and retraction of protrusions and the formation of invaginations. These changes, which take place in response to a variety of different stimuli, accompany and sustain changes of cell shape (for reviews, see Sheetz et al., 2006; McConnell and Tyska, 2010; Parsons et al., 2010). Further changes of the plasma membrane are induced by the bidirectional traffic of vesicles to and from the cytoplasm, that is, various types of endocytosis and of constitutive, as well as regulated, exocytosis. In this Commentary, we will refer to these structural changes taking place at the plasma membrane, with or without the insertion or removal of vesicles, as membrane dynamics. When talking about membrane dynamics, together with the associated cytoskeletal changes, we will use the term surface dynamics.
The idea that the strict contact between the actin cytoskeleton and the plasma membrane provides a mechanism that underlies the dynamic events of the cell surface was proposed a few decades ago (Bretscher, 1996). However, it was not fully taken into consideration for quite a long time. In fact, the cortical cytoskeleton and the plasma membrane were often investigated separately. Laboratories working on the machinery involved in actin remodeling have paid only limited attention to the concomitant changes of the plasma membrane. Conversely, many laboratories actively carrying out research in the field of endocytosis and exocytosis have focused exclusively on the processes of membrane fusion and fission (for reviews, see Hong, 2005 and Malsam et al., 2008, respectively).
In the last few years the situation has progressively changed. It has been shown that, in order to induce protrusions and invaginations of the cell surface, remodeling of the cytoskeleton requires the cooperation of membrane-associated components, such as amphipathic proteins, which often contain membrane-deforming Bin-amphiphysin-Rys (BAR) domains (reviewed by Graham and Kozlov, 2010; Roberts-Galbraith and Gould, 2010; Suetsugu and Gautreau, 2012). Furthermore, the cortical cytoskeleton, which previously was considered primarily as a barrier, was recognized as also having a positive role in the surface traffic of vesicles (reviewed by Momboisse et al., 2010; Villanueva et al., 2012; Nightingale et al., 2012).
In this Commentary, we focus on a key aspect of surface dynamics, namely the role of the small GTPases, in particular Rho family GTPases, which have been recognized as master regulators of cytoskeletal remodeling. In the first part, we discuss the activity and distribution of the Rho GTPases in the cortical layer of the cell and illustrate their cooperation with small GTPases of the Ras, Rab and Arf families. In the second part, we discuss the role that vesicle traffic (i.e. endocytosis and exocytosis), has in membrane dynamics and cell mechanics, and we emphasize the regulation of these events by Rho GTPases.
Small GTPases
Small GTPases are ideal regulators of a large number of local intracellular events. By hydrolyzing GTP they act as molecular switches that control the activity and function of a variety of specific targets, including enzymes, scaffolds and accessory proteins. The activity of these GTPases is strictly controlled by various regulatory proteins: guanine nucleotide exchange factors (GEFs), which convert the inactive GDP-bound enzymes into active GTP-bound forms; GTPase-activating proteins (GAPs), which stimulate GTP hydrolysis, thereby converting the active GTPases into their inactive forms; and guanine nucleotide dissociation inhibitors (GDIs), which exist for all GTPases apart from the ADP-ribosylation factor (Arf) family and segregate the GTPases into the cytosol, away from their membrane sites of action, by masking the prenyl membrane-binding groups of the molecules.
Over 150 small GTPases have been identified in human cells, and these can be divided into different families, including the Ras, Rab, Arf and Rho families, which will be mentioned in this Commentary. Each family member can propagate multiple signaling pathways, and its final effects depend on the expression of its effectors and other factors that are associated with specific cell types (Goldfinger, 2008).
The Ras, Rab, Arf and Rho families of small GTPases
Despite considerable differences among their members, each of the GTPase families can be broadly considered as being specialized in distinct cellular processes. The Ras GTPases, which include the three well-known oncoproteins H-Ras, K-Ras and R-Ras, the Rap proteins and a few other GTPases, have key roles in cell survival, proliferation and differentiation (reviewed by Goldfinger, 2008; Johnson and Chen, 2012). The functional differences among the members of this family depend, in part, on their different subcellular distribution (Omerovic et al., 2007).
So far, 60 mammalian Rab GTPases have been characterized. All of them participate in the regulation of the different steps involved in membrane traffic, from the assembly of transport vesicles at the donor compartment to membrane fusion at the acceptor compartment. In each membrane compartment, the Rabs interact with diverse effector proteins that are required to select the membrane cargoes, to promote the movement of carrier vesicles, and to recognize the correct target membrane for the fusion (reviewed by Hutagalung and Novick, 2011). In addition to their association with membranes, Rabs interact, both directly and through specific adaptor proteins, with microtubule-based kinesin and dynein motor complexes (Horgan and McCaffrey, 2011).
The small GTPases of the Arf family include 26 Arf and Arf-like proteins. Arf proteins differ from other GTPase families, because they possess an N-terminal amphipathic helix that is crucial for their membrane binding. Their activity, therefore, always occurs in strict proximity to their membrane targets. GTP-bound Arf GTPases recruit coat proteins, lipid-modifying enzymes, tethers and other effector molecules, which modify their function, and they have key regulatory roles in both membrane traffic and organelle structure (reviewed by D'Souza-Schorey and Chavrier, 2006; Donaldson and Jackson, 2011).
There are 20 Rho GTPases in mammals, including the three best characterized, RhoA, Rac1 and Cdc42. Rho GTPases are master regulators of actin cytoskeleton remodeling (Maekawa et al., 1999; Geissler et al., 2000) (reviewed by Heasman and Ridley, 2008). Because of their key role in cell surface dynamics, we will discuss this GTPase family in detail below.
Rho GTPases – mechanisms of action and localization
Remodeling of the cortical actin cytoskeleton, which is triggered in response to extracellular signals, is a key process that is directly regulated by the Rho small GTPase family. This process is mediated by numerous actin-binding proteins, which are arranged in large multimolecular complexes that also include membrane-binding and signaling domains. Cdc42 and Rac1 stimulate the nucleation of actin filaments by activating specific actin-nucleation-promoting factors (NPFs), such as the (neural) Wiskott-Aldrich syndrome proteins (N-WASP and WASP) and Wiskott-Aldrich syndrome protein family member (WAVE, also known as SCAR). Through this mechanism, the two GTPases affect the function of the actin-related protein 2/3 (Arp2/3) complex, which promotes nucleation and elongation of branched actin networks. In addition, formins – another group of NPFs – are activated downstream of both RhoA and Cdc42 and induce nucleation and elongation of actin bundles (reviewed by Lee and Dominguez, 2010; Campellone and Welch, 2010). The different Rho GTPases might operate in a coordinated fashion to affect the actin cytoskeleton. In particular, RhoA often acts as an antagonist of Rac1 and Cdc42 (Kozma et al., 1997; Tsuji et al., 2002; Ohta et al., 2006; reviewed by Brown et al., 2006; Guilluy et al., 2011). Moreover, the small GTPases regulate the activity of effector enzymes, such as the protein kinases Rho-associated coiled-coil forming kinase (ROCK) (activated by RhoA) and p21-activated kinase (PAK) (activated by Rac and Cdc42). These effectors, in turn, can then contribute to the coordinated control of the GTPases. For example, in various cell types, Rac1 operates under the stable inhibitory control of ROCK (Yamaguchi et al., 2001; Tsuji et al., 2002; Takefuji et al., 2007). Therefore, inhibition of either RhoA or its effector ROCK can result in Rac1 activation (Jeon et al., 2010; Jeon et al., 2012).
A crucial aspect of Rac1 and Cdc42 activity is their activation-dependent accumulation at discrete sites in the plasma membrane. At these sites, they colocalize with other proteins, including their GEFs, for example TRIO (triple functional domain protein), VAV, DOCK1 (dedicator of cytokinesis 1) and PREX1 (for phosphatidylinositol 3,4,5-trisphosphate-dependent Rac exchanger 1) (see, among others, Premkumar et al., 2010; Neubrand et al., 2010; Tolias et al., 2011; Citi et al., 2011). As expected, downregulation of these GEFs results in the inhibition of Rac1- and/or Cdc42-dependent processes (Côté and Vuori, 2007). The unique concentration of many proteins at the sites of Rac1 or Cdc42 activity depends on multiple processes, including the accumulation of enzymes such as phosphatidylinositol 3-kinase (PI3K). This results in the assembly of domains rich in phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P3] that act as specialized platforms for plasma membrane dynamics. Protein recruitment to these platforms is not limited to small GTPases and their GEFs, but also includes NPFs, other actin-binding proteins, proteins with polybasic clusters and amphiphylic proteins that are necessary for the deformation of membranes (Heo et al., 2006; Takabayashi et al., 2010; reviewed by Saarikangas et al., 2010; Suetsugu and Gautreau, 2012). The local cooperation between actin remodeling and membrane deformation underlies dynamic processes at the cell surface, causing either protrusion (e.g. the formation of lamellipodia or filopodia) or invagination (see the endocytosis section below) of the plasma membrane (for a review, see Suetsugu and Gautreau, 2012).
Cooperation between small GTPases in the regulation of surface dynamics
Rho GTPases, through their ability to remodel the cytoskeleton, are often involved in the regulation of membrane dynamics in discrete surface domains. This function first emerged when Rho GTPases were shown to cooperate with GTPases from other families, in particular with Rabs and Arfs, which are involved in various phases of membrane trafficking (reviewed by D'Sousa-Schorey and Chavrier, 2006; Stenmark, 2009).
Mechanistically, the contributions of the Rab and Arf GTPases are remarkably diverse, ranging from the integration of vesicular transport and cytoskeletal remodeling to the direct regulation of Rho family GTPases (Table 1). Numerous findings regarding the function of Arf6, which is abundant in the cell cortical area, have been summarized in an excellent review by Myers and Casanova (Myers and Casanova, 2008). More recently, the interaction of GTPases from different families has been shown to operate through the assembly of large complexes that induce multiple coordinated effects. Some of these effects depend on membrane traffic. Cytoplasmic vesicles, which are assembled through an Arf6-dependent process, have been shown to be required for both the exocytic transport of Cdc42 and its GEF β-PIX (also known as ARHGEF7) and the recruitment of Rac1 to the polarity complex between Par6 and atypical protein kinase C (aPKC) alpha (Osmani et al., 2010). Moreover, the complex containing RhoA and its GEF Syx translocates to the leading edge of migrating cells through a form of exocytosis that is dependent on another GTPase, Rab13. In turn, Rab13 has been shown to have a role in cell signaling through its interaction with growth factor receptors, such as the vascular-endothelial growth factor (VEGF) receptor 2 (Wu et al., 2011) (see Table 1).
GTPases | Mechanism regulated by the GTPases | Process regulated by the GTPases | References |
Arf1 and Rac1 | Co-activation of the WAVE regulatory complex | WAVE-dependent actin-driven bead motility (cell-free assay) | (Koronakis et al., 2011) |
Arf1 and Cdc42 | Arf1-dependent regulation of Cdc42 by ARHGAP10 | Actin-mediated clathrin-independent endocytosis | (Kumari and Mayor, 2008) |
Arf6 and Cdc42 | Arf6-dependent exocytosis of Cdc42 and its GEF β-PIX at the leading edge | Actin-driven polarized astrocyte migration (wound-healing assay) | (Osmani et al., 2010) |
Arf6 and Rac1 | ARNO-mediated Rac1 activation | Fibronectin-induced haptotactic epithelial cell migration | (White et al., 2010) |
Arf6 and Rac1 | Rac1-PIP5K-induced Arf6-dependent endocytosis | Clathrin-independent plasma membrane invagination and tubulation | (Vidal-Quadras et al., 2011) |
Rab5 and Rac1 | Transition from Rac1-PtdIns(3,4,5)P3 to Rab5a-PtdIns(3)P for macropinocytic cup closure | Macropinocytosis in macrophages | (Yoshida et al., 2009) |
Rab13 and RhoA | Rab13-dependent translocation of RhoA and its GEF Syx to the leading edge | VEGF-mediated endothelial cell migration | (Wu et al., 2011) |
GTPases | Mechanism regulated by the GTPases | Process regulated by the GTPases | References |
Arf1 and Rac1 | Co-activation of the WAVE regulatory complex | WAVE-dependent actin-driven bead motility (cell-free assay) | (Koronakis et al., 2011) |
Arf1 and Cdc42 | Arf1-dependent regulation of Cdc42 by ARHGAP10 | Actin-mediated clathrin-independent endocytosis | (Kumari and Mayor, 2008) |
Arf6 and Cdc42 | Arf6-dependent exocytosis of Cdc42 and its GEF β-PIX at the leading edge | Actin-driven polarized astrocyte migration (wound-healing assay) | (Osmani et al., 2010) |
Arf6 and Rac1 | ARNO-mediated Rac1 activation | Fibronectin-induced haptotactic epithelial cell migration | (White et al., 2010) |
Arf6 and Rac1 | Rac1-PIP5K-induced Arf6-dependent endocytosis | Clathrin-independent plasma membrane invagination and tubulation | (Vidal-Quadras et al., 2011) |
Rab5 and Rac1 | Transition from Rac1-PtdIns(3,4,5)P3 to Rab5a-PtdIns(3)P for macropinocytic cup closure | Macropinocytosis in macrophages | (Yoshida et al., 2009) |
Rab13 and RhoA | Rab13-dependent translocation of RhoA and its GEF Syx to the leading edge | VEGF-mediated endothelial cell migration | (Wu et al., 2011) |
The accumulation of active Rho GTPases in specific plasma membrane regions also occurs through mechanisms other than exocytosis. Crosstalk between Arf and Rac1 or Cdc42 takes place through the regulation of GAPs and GEFs that are specific for the Rho family (Myers and Casanova, 2008; White et al., 2010) during the cooperation between the two Rho GTPases in the polymerization of actin filaments at the leading edge of migrating cells (Koronakis et al., 2011). A complex between β-PIX and the Arf GAP GIT1 (G protein coupled receptor kinase 2 interacting protein 1) has been shown to induce cell shape changes in responses to a variety of stimuli by integrating the signals initiated by members of different GTPase families (Frank and Hansen, 2008). The β-PIX–GIT1 complex interacts with additional scaffolds to yield large dynamic molecular assemblies, which tune the protrusive activity at the cell edge (Asperti et al., 2011). Finally, Rho GTPases cooperate with Ras to regulate axon outgrowth (reviewed by Hall and Lalli, 2010). Furthermore, they cooperate with RalA, a small GTPase of the Ras family that is activated downstream of oncogenic Ras, in the assembly of exocyst components (reviewed by Wu et al., 2008, and see below).
In summary, the role of Rho GTPases can no longer be considered separately from those of other small GTPases, in particular those from the Rab and Arf families. Rather, a strict cooperation seems to take place in many cases, which assures the integration of actin remodeling in a wealth of dynamic processes occurring at and near the cell surface.
Plasma membrane dynamics
So far, small GTPases have been shown to have crucial roles in actin cytoskeleton remodeling events that result in changes at the cell surface, such as plasma membrane protrusions and invaginations. As long as the cell surface changes are relatively small, they can occur without a major change in the extent of the cell surface (i.e. its area). Under these conditions the levels of constitutive exocytosis and endocytosis of vesicles at the plasma membrane, match each other. However, robust changes of cell shape require the area of the plasma membrane to be rapidly expanded or reduced through increased exocytosis or endocytosis. In the following sections, we will discuss the links between cytoskeleton remodeling and these forms of plasma membrane dynamics, which are also regulated by Rho family GTPases. Additional processes that also depend on the interaction between the plasma membrane and the cytoskeleton, and the accompanying changes of membrane tension will be briefly considered in a later section.
Exocytosis
Exocytosis is conceptually simple, as it refers to processes that lead to the fusion of an intracellular membrane with the plasma membrane. A detailed view, however, is more complex. Traditionally, the process is divided into constitutive and regulated exocytic pathways. It is still unclear whether constitutive exocytosis occurs through heterogeneous pathways, but it has been well established that regulated exocytosis does occur through a variety of mechanisms. In addition to the well-known secretory exocytosis, which is employed by specific organelles that are typically found in neural and non-neural secretory cells, many other forms of exocytosis exist in various cell types. Many cells express secretory lysosomes as well as distinct types of non-secretory, exocytic vesicles that account for the recycling of proteins (e.g. receptors, channels and transporters) to the plasma membrane and for the expansion of the cell surface. Such exocytic processes are necessary for cytokinesis, motility, neurite outgrowth and wound healing (Chieregatti and Meldolesi, 2005). For many years, the cortical cytoskeleton has been envisaged as a physical barrier that exocytic vesicles have to pass in order to get into contact with the plasma membrane prior to exocytosis. Recent studies, however, have revealed that the initial steps of the exocytic process are strictly interconnected with cytoskeletal remodeling (reviewed by Wu et al., 2008; Ory and Gasman, 2011). In particular, exocytosis has been linked with actin depolymerization that is induced by the RhoA GAP Gem interacting protein (GMIP) in the proximity of the vesicles that are destined to be discharged (Johnson et al., 2012). Following the remodeling of the cortical cytoskeleton, the vesicle and plasma membranes are brought into direct contact through specific tether complexes. Finally, this step is followed by fusion of the two membranes (Fig. 2).
Membrane tethering
In order to undergo fusion, membranes need first to overcome their mutual repulsion. For this they require tethers. Tethers function as flexible molecular bridges linking the two membranes, thereby contributing to cargo selection and cytoskeletal interactions, as well as the localization and assembly of the soluble NSF attachment protein (SNAP) receptor (SNARE) complexes that are necessary for membrane fusion (Sztul and Lupashin, 2006; Wu et al., 2008; Bröcker et al., 2010). The best-known tethering complex at the plasma membrane is the exocyst. It operates in many constitutive and several regulated exocytic events, such as the discharge of vesicles that are rich in glucose transporter 4 (GLUT4) in response to insulin stimulation in adipocytes (Chen et al., 2011; Stöckli et al., 2011).
RalA, a small GTPase of the Ras family, was the first GTPase to be identified that controls the tethering of vesicles by the exocyst (Moskalenko et al., 2002). More recently, Rho GTPases have also been shown to have crucial roles in this process, acting both directly and indirectly through regulatory proteins [such as SH3BP1, a RhoGAP that downregulates Rac1 (Parrini et al., 2011)] and targets [such as IQGAP1, a scaffold protein that mediates part of the action of Cdc42 (Rittmeyer et al., 2008)].
The interactions between Rho GTPases and the exocyst induce specific effects in different cells. In epithelial cells, they operate on the lateral and not on the luminal surface. Therefore, Rho and the exocysts are essential for establishing polarity (Mellman and Nelson, 2008; He and Guo, 2009). In developing neurons, Rho GTPases and the exocyst induce axon specification and outgrowth but have no role in the regulated discharge of synaptic vesicles (Dupraz et al., 2009; Schwenger and Kuner, 2010). In pancreatic β-cells, they modulate insulin secretion (Rittmeyer et al., 2008; Chen et al., 2011). In addition to functioning in exocytosis, the exocyst complex also regulates actin polymerization (Zuo et al., 2006). In many cell types, therefore, it stimulates migration (Spiczka and Yeaman, 2008; Letinic et al., 2009), whereas in tumor cells it reinforces invasion (Sakurai-Yageta et al., 2008). Owing to its regulation by small GTPases, in particular by members of the Rho family, the exocyst therefore operates in a vast array of processes that affect cell surface dynamics.
Exocytic membrane fusion
The conserved mechanism through which vesicle membranes fuse with the plasma membrane is based on the complexes that form between the vesicle SNAREs (vSNAREs) and the target membrane SNAREs (tSNAREs). These SNARE complexes are then converted into channels that establish the continuity between the two participating membrane-bound compartments. The most common SNARE complex, which comprises the vesicle-associated membrane protein 2 (VAMP2) and the two tSNAREs syntaxin1 and SNAP25 in the plasma membrane, operates in the release of transmitters from neurons and neurosecretory cells (Jahn and Scheller, 2006; Malsam et al., 2008). In the acinar cells of salivary and other glands the vSNARE is VAMP8 (Wang et al., 2007). Tetanus-toxin-insensitive VAMP (Ti-VAMP, also known as VAMP7), which is involved in the exocytosis of secretory lysosomes, also participates in neurite outgrowth and in various forms of endocytosis (Martinez-Arca et al., 2001; Danglot et al., 2010).
The various forms of exocytosis are characterized not only by their different SNAREs, but also by the involvement of different Rho GTPases. In chromaffin and PC12 cells, Rac1 and Cdc42 reinforce the neurosecretory responses induced by Ca2+-dependent stimuli, whereas RhoA inhibits these responses (Ory and Gasman, 2011). Similar events occur in mast cells and neutrophils. However, in these cells the reinforcement of the Ca2+-dependent exocytosis depends on Rac2 (Abdel-Latif et al., 2005; Stratmann et al., 2010). In pancreatic β-cells, Cdc42 potentiates only the second, prolonged phase of insulin release in response to high glucose levels, and not the first phase (Wang and Thurmond, 2009). In HeLa cells, GTP hydrolysis by the Rho GTPase TC10, which is accelerated by the p190 Rho GAP, triggers regulated exocytosis (Kawase et al., 2006). In cultured hippocampal neurons, Cdc42 stimulates the exocytosis of Ti-VAMP-positive vesicles at growth cones, which results in neurite outgrowth (Alberts et al., 2006). The inhibition of the RhoA effector ROCK in PC12 cells, in SH-SY5Y neuroblastoma cells and in primary cultures of brain astrocytes induces robust exocytic responses, which result in fast Rac1-dependent and Ca2+-independent outgrowth of neurites (Racchetti et al., 2010; Racchetti et al., 2012).
The mechanisms by which Rho GTPases and their effectors regulate exocytic events are numerous. For example, they can act directly on the SNAREs. In chromaffin cells and neurons, activation of the RhoA-ROCK pathway results in the phosphorylation of syntaxin1, which, in turn, leads to the association of this SNARE with negative regulators of exocytosis (Sakisaka et al., 2004). In PC12 cells, RhoA and Cdc42, respectively, induce inhibition and stimulation of actin remodeling and exocytosis of dense-core vesicles (Ory and Gasman, 2011). In pancreatic β-cells, Cdc42 binds directly to syntaxin 1 and VAMP2 (Wang and Thurmond, 2009). Phospholipids are also relevant in the modulation of exocytic events. Phosphatidic acid, which is generated following the Rac1-dependent activation of phospholipase D1, has been shown to bind the tSNARE syntaxin 1A and thus strengthens Ca2+-induced exocytic responses (Momboisse et al., 2009). The product of its de-phosphorylation, diacylglycerol, together with PtdIns(4,5)P2, activates MUNC13-1, a regulatory protein that promotes the docking and discharge of neurosecretory vesicles (reviewed by Ory and Gasman, 2011; Martin, 2012).
Finally, Rho GTPases and the cytoskeleton operate in the last step of secretory exocytosis, namely the discharge process. A thick actin layer assembles rapidly around docked and fused vesicles in a Cdc42-dependent way, and myosin-mediated contraction of this actin coat squeezes out the vesicular content. Without this process the vesicle content would be largely recycled to the cytoplasm by endocytosis (Nightingale et al., 2011; Nightingale et al., 2012).
Endocytosis
The interest in endocytosis has greatly increased in the last few years. The process occurs through heterogeneous mechanisms that include not only classical clathrin-mediated endocytosis, macropinocytosis and phagocytosis, but also a number of additional forms such as caveolar, flotillin-dependent and clathrin-independent (CLIC) endocytosis (Box 1). Moreover, at least some of these forms have been shown to operate not only during the internalization of extracellular materials and the recycling of membranes, but also in the regulation of cell signaling (Sigismund et al., 2012). Excellent reviews about the structure, dynamics and function of the various forms of endocytosis have recently been published, so we will not go into specific detail here (see Doherty and McMahon, 2009; Hansen and Nichols, 2009; Lajoie and Nabi, 2010; Otto and Nichols, 2011; Lim and Gleeson, 2011; McMahon and Boucrot, 2011; Flannagan et al., 2012). Instead, we will consider the various forms of endocytosis together, focusing on the role of the cytoskeleton and small GTPases on the sorting of their specific membrane domains, and on their sculpting of the membrane into flask- or cup-shaped invaginations. Following the formation of such invaginations, these plasma-membrane-attached structures are severed at the neck, and discrete vesicles are pinched off into the cytoplasm.
Generation of specific endocytic membrane domains
The generation of specific membrane regions is crucial for the specificity of various endocytic vesicles. In clathrin-mediated endocytosis, sorting depends primarily on the adaptor protein AP2, which binds and concentrates the cargoes and cargo adaptors at specific sites of the plasma membrane (McMahon and Boucrot, 2011). The caveolar and flotillin-dependent domains, which often form membrane invaginations, correspond to membrane rafts that are rich in cholesterol and caveolin or flotillin, respectively (Hill et al., 2008; Verma et al., 2010; Otto and Nichols, 2011). In addition, these domains are enriched in glycolipids and in proteins that are linked to the membrane by glycosyl-phosphatidylinositol (GPI) tails (Doherty and McMahon, 2009; Hansen and Nichols 2009; Nichols, 2009). Rafts are also important for the generation of the CLIC domains (Lajoie and Nabi, 2010). In caveolar and flotillin-dependent endocytosis, small GTPases and the cytoskeleton are involved at the sorting stage (see Ludwig et al., 2010; Stoeber et al., 2012; Echarri et al., 2012). For the internalization of CLICs, no information on the role of GTPases and/or the cytoskeleton is available yet.
Macropinocytosis and phagocytosis result in the internalization of much larger domains in areas of the plasma membrane that are delimited by membrane protrusions, such as membrane ruffles and lamellipodia (Lim and Gleeson, 2011; McMahon and Boucrot, 2011; Flannagan et al., 2012). The rapid remodeling of the cytoskeleton which sustains these protrusions depends on the recruitment of Rac1 and Cdc42 through Rab35-dependent transport along microtubule tracks (Shim et al., 2010).
Sculpting of membrane invaginations
The membrane invagination stage is regulated by the strict cooperation between the cytoskeleton, phospholipids and amphipathic membrane proteins, which often contain BAR domains that protrude from the cytoplasmic face of the membrane (reviewed by Roberts-Galbraith and Gould, 2010; Suetsugu and Gautreau, 2012). In clathrin-mediated invaginations, amphiphysin and endophilin are examples of such amphipathic proteins (Fütterer and Machesky, 2007). With regards to caveolae, examples are SDPR (serum deprivation-response protein) and PACSIN2 (Hansen et al., 2009; de Kreuk et al., 2011), whereas CLICs employ GRAF1 (GTPase regulator associated with focal adhesion kinase-1) (Doherty and Lundmark, 2009). Recently, BAR proteins have also been reported to be involved in the formation of macropinocytic (Wang et al., 2010) and phagocytic cups (Clarke et al., 2010).
The assembly of coated pits is accompanied by remodeling of the actin cytoskeleton. Activation of N-WASP by Cdc42, in complex with the actin nucleation and elongation factor Arp2/3, binds amphipathic proteins (Bu et al., 2009; Lee et al., 2010). Cdc42 and Rac1 also control actin remodeling, which is required for the formation of caveolae at the plasma membrane (Klein et al., 2009; de Kreuk et al., 2011). During CLIC-mediated endocytosis, actin remodeling is controlled by Cdc42 together with Arf1 (Doherty and Lundmark, 2009; Kumari and Mayor, 2008) (Table 1). By contrast Rac1, together with the Ca2+-binding protein calmodulin, has been shown to inhibit this process (Vidal-Quadras et al., 2011).
Rac1, which is activated locally following the initiation of macropinocytosis, is deactivated before closure of the macropinocytic cup (Yoshida et al., 2009) (Table 1). The assembly of the phagocytic cup appears to be controlled indirectly by regulators of small GTPases, such as the Rac GAP ARHGAP25 (Csépányi-Kömi et al., 2012) and the Rho GEF VAV2 (Arora et al., 2008). Interestingly, during their establishment, both the macropinocytic and phagocytic cups require an expansion of their limiting membrane, and this happens by the exocytosis of specific cytoplasmic vesicles (Falcone et al., 2006; Corrotte et al., 2006).
Fission of the neck and pinching off of endocytic vesicles
Once assembled, the flasks and/or cups detach from the plasma membrane through fission of their membranous neck. This process occurs through various mechanisms, only one of which, namely the process that is dependent on the GTPase dynamin, has been investigated in detail. Dynamin assembles around the neck of the flasks and squeezes it mechanically. This appears to induce a lipid-phase segregation in the membrane, which is necessary for the pinching off to take place (Liu et al., 2009; Ferguson and De Camilli, 2012). A large fraction of the clathrin-independent vesicles, however, pinch off independently of dynamin (Sandvig et al., 2011). In this case, the energy could be provided by the mechanochemical feedback predicted by a model initially developed for yeast (Liu et al., 2009). Such a feedback, triggered by the curvature of the membranes, would produce an interfacial force sufficient to pinch off vesicles. Once released, exocytic vesicles undergo traffic within the cytoplasm, which is regulated by members of the Rab and Arf GTPase families.
Membrane tension
A cellular property that is also connected to surface dynamics and that has begun to attract interest in the last few years, is membrane tension. Compared with intracellular membranes, the plasma membrane is maintained under high tension, which is necessary for a variety of cellular functions including migration, cytokinesis and cellular wound healing. Mechanistically, the tension depends only marginally on the in-plane elasticity of the phospholipid bilayer, and much more on the continuous remodeling of the cytoskeleton, which is transduced to the plasma membrane by actin-binding proteins, in particular by myosin-1 (Nambiar et al., 2009). As long as the cell remains at rest or under moderate stimulation, the changes in the cell surface area are absent or small, and the plasma membrane tension remains constant. When the cell surface area increases, for example upon stretching of the cell, the plasma membrane tension increases. This increase has recently been identified as the cause of a rapid burst of exocytosis followed by the ensuing de-tension of the plasma membrane to values of tension even lower than those of resting cells (Gauthier et al., 2011). Cycles of local cell surface expansion and retraction, sustained by fluctuations of tension, can be the trigger of mechanically regulated outside-in signals (Vogel and Sheetz, 2009), whose relevance in cell physiology and pathology is now starting to be investigated in more detail.
Conclusion
In this Commentary, we have summarized evidence that demonstrates that cell surface dynamics are not the result of either the remodeling of the cortical cytoskeleton or changes of the plasma membrane, as has been envisaged for quite some time. Instead, surface dynamics are the result of the integration of these two processes in time and space. Rho family GTPases are crucial in orchestrating the integration of these processes. They achieve this by operating coordinately with each other, as well as by collaborating, in many cases, with other GTPases, such as those belonging to the Rab and Arf families (Table 1). The activity of small GTPases, which is induced through their regulatory proteins, is exerted by the concomitant regulation of two strictly interconnected types of events: the nucleation and elongation of the actin filaments in the cytoplasmic layer immediately adjacent to the plasma membrane, and the fusion/fission of the plasma membrane with the membranes of specific cytoplasmic organelles.
The experimental evidence presented here is expected to contribute substantially to future developments in various areas of cell biology. The strict association of the plasma membrane with the cortical cytoskeleton contributes to a variety of properties (such as tension, resistance and plasticity) that distinguish the plasma membrane from intracellular membranes. However, even internal vesicles depend on the cytoskeleton for their function, and many aspects of their interaction with the cytoskeleton remain to be clarified. A fascinating development of future research could be the comparative (structural, molecular and functional) analysis of the interactions that occur in different regions of the cell. In terms of physiology, the orchestration of cell surface dynamics by Rho GTPases underlies a variety of functions, ranging from surface plasticity and resistance to cell adhesion, motility, neurite outgrowth and many others. Although these processes are being intensely investigated, their mechanisms are still partially unknown. Thus, the integrated perspective we have summarized might be useful to gain further insight into the regulation of these processes. Finally, we expect that the understanding of the interaction between the plasma membrane and the cytoskeleton will become increasingly important in cell pathology. So far, the interest has been largely restricted to cancer cells and their proliferation. Surface dynamics and its alterations appear, however, to be involved in many other, highly disruptive pathologies, from the early synaptic defects of neurodegenerative diseases, including Alzheimer’s disease, to multiple forms of autoimmunity.
Clathrin-mediated endocytosis (CME): CME occurs in most cells and is the best-characterized form of endocytosis. Triskelia that consist of three clathrin molecules self-assemble at the cytoplasmic surface of the plasma membrane to form coated pits, where cargoes are concentrated by the adaptor protein AP2. Following vesicle budding into the cytoplasm, the clathrin coat disassembles and the vesicles enter the intracellular endosomal pathways.
Endocytosis through caveolae: Caveolae are small flask-shaped invaginations that undergo endocytosis. Their markers (caveolins) are small hairpin proteins that are immersed in raft domains at the cytosolic membrane surface that expose amphipathic proteins that account for their curvature. Caveolae are abundant in adipocytes and endothelial cells. They have important roles in signal transduction, transcytosis and tumor growth.
Flotillin-dependent endocytosis: The oligomerization of flotillins, proteins that are associated with the cytoplasmic face of the plasma membrane, leads to the formation of membrane microdomains. These are mobile in the plane of the plasma membrane, exhibit raft properties and contain amphipatic proteins. These flotillin-containing microdomains can be endocytosed, but the mechanisms behind this are still largely unknown.
Clathrin-independent carriers: Clathrin-independent carriers (CLICs) are tubulovesicular systems that are rich in cholesterol (i.e. lipid rafts) and amphipatic proteins. They account for a considerable fraction of the endocytic events inside cells. They appear to function in the internalization of specific membrane cargos. CLIC-mediated endocytosis might include several distinct subtypes of endocytosis (e.g. dependent on Arf6 or specific to interleukine 2β receptor and possibly others).
Macropinocytosis: Macropinocytosis consists of the internalization of larger plasma membrane domains that are initially delimited by actin-rich linear protrusions from the plasma membrane (membrane ruffles) and which give rise to a cup that is destined to become sealed and form a cytoplasmic vacuole. Their function is the internalization of large volumes of extracellular fluid that can contain cell fragments, viruses, ectosome and exosome vesicles, and DNA plasmids.
Phagocytosis: Phagocytosis is a process that is only performed by phagocytic cells, such as macrophages and other types of leukocytes. It differs from micropinocytosis because of the larger size of its cup, which is initially limited by actin-rich pseudopodia that are generated following the direct interaction of the cell with the solid materials to be ingested. Expansion of the cup occurs by exocytosis of cytoplasmic vesicles. Following cup sealing, the phagosome fuses with lysosomes and initiates the digestion of its internalized material.
Funding
The I.d.C. and J.M. laboratories are supported by grants from the Italian Telethon Foundation ONLUS [grant numbers GGP09066 to J.M., GGP09078 to I.d.C.], the Italian Institute of Technology (IIT) to J.M.); and AIRC, the Italian Asssociation for Cancer Research [grant number 10321 to I.d.C.].