2015 Volume 40 Issue 2 Pages 69-77
The actin-cytoskeleton plays a critical role in various biological processes, including cell migration, development, tissue remodeling, and memory formation. Both extracellular and intracellular signals regulate reorganization of the actin-cytoskeleton to modulate tissue architecture and cellular morphology in a spatiotemporal manner. Since the discovery that activation of Rho family GTPases induces actin-cytoskeleton reorganization, the mode of action of Rho family GTPases has been extensively studied and individual effectors have been characterized. The actin-binding protein IQGAP1 was identified as an effector of Rac and Cdc42 and is the founding member of the IQGAP family with two additional isoforms. The IQGAP family shows conserved domain organization, and each member displays a specific expression pattern in mammalian tissues. IQGAPs regulate the actin-cytoskeleton alone and with their binding partners, thereby controlling diverse cellular processes, such as cell migration and adhesion. Here, we introduce IQGAPs as an actin-cytoskeleton regulator.
The Rho family GTPases function as molecular switches cycling between an active (GTP-bound) and inactive (GDP-bound) form in a manner similar to other small GTPases, including the Ras family. Guanine nucleotide exchange factors (GEFs) and GTPase activating proteins (GAPs) catalyze the conversion from the GDP- to GTP-bound form and the conversion from the GTP- and GDP-bound form, respectively. Since the discovery that manipulation of the activity of the Rho family GTPases (Rac, Cdc42, and RhoA) results in the drastic reorganization of actin filaments and adhesions in cultured cells (Ridley and Hall, 1992; Ridley et al., 1992), many researchers, including us, have been intensively studying the mode of action of the Rho family GTPases. Biochemical and genetic screens have identified critical effectors of Rho family GTPases that mediate the activation of the Rho family GTPases and the reorganization of the actin-cytoskeleton. These effectors include IQGAPs, Specifically Rac1-associated protein 1 (Sra-1, also known cytoplasmic FMR1 interacting protein 1), Rho-kinase (also known ROK, ROCK), Dia (Diaphanous related formin), p21-activating kinase (PAK), Wiskott–Aldrich syndrome protein (WASP) verprolin homologous (WAVE) complex, and WASP/N-WASP. In this review, we focus on the role of IQGAPs in the regulation of the actin-cytoskeleton. The mode of action of Rho family GTPases and their other effectors in the regulation of actin-cytoskeleton is discussed in several excellent reviews (Kaibuchi et al., 1999; Jaffe and Hall, 2005; Ridley, 2006).
IQGAP is an evolutionarily conserved multidomain protein, and three IQGAP isoforms have been identified in mammals (Weissbach et al., 1994; Brill et al., 1996; Wang et al., 2007). Among the IQGAP family, IQGAP1 has been extensively studied, and our understanding relies mainly on the evidence from IQGAP1. The domain organization of IQGAP1 is highly conserved in the IQGAP family consisting of an amino-terminal calponin homology domain (CHD) region, an IQGAP-specific repeat motif, a domain with two conserved tryptophan (W) residues, a calmodulin-binding IQ motif, a RasGAP related domain (GRD), and a RasGAP C-terminus (Fig. 1). Each member of the IQGAP family displays a distinct expression pattern in mammalian tissues. IQGAP1 is ubiquitously expressed, while the expression of IQGAP2 is restricted to the liver and testis and IQGAP3 is found mainly in the brain and lung (Weissbach et al., 1994; Brill et al., 1996; Wang et al., 2007). As expected from the conserved domain organization in the IQGAP family, genetic depletion of either IQGAP1 or IQGAP2 alone showed no drastic phenotypes in mice, and in IQGAP2 knockout mice IQGAP1 is overexpressed to compensate for the loss of IQGAP2 (Li et al., 2000; Schmidt et al., 2008). Although a study regarding IQGAP3 knockout mice has not been reported, these data suggest a functional redundancy between IQGAPs in mammals.
IQGAP domain organization and expression pattern in tissues. (A) Schematic presentation of the domain organization of the IQGAP family. The domain structures and amino acid homologies are indicated. The total amino acid length is indicated to the right of each structure. CHD, calponin homology domain; IQ repeats, IQGAP-specific repeats; WW, double tryptophan motif; GRD, RasGAP-related domain, RasGAP C, RasGAP C-terminus. (B) The tissue expression patterns of IQGAP family members. This table is based on immunoblot analysis using specific antibodies (Wang et al., 2007). The plus number reflects the expression level. Notably, the expression levels among the members are incommensurable because of the differences in the sensitivities of the antibodies. (C) The binding regions of selected IQGAP1 interacting proteins are represented. Erk, extracellular-signal-regulated kinase; MEK, MAPK (mitogen-activated protein kinase) kinase; ELC, essential light chain. Of note, the binding region of N-WASP is under the debate (see text).
IQGAP1 was originally identified as a molecule with four calmodulin-binding IQ motifs and a GAP domain for Ras (Weissbach et al., 1994). The name “IQGAP” originates as a result of this domain organization, but no report has convincingly demonstrated its GAP activity toward any small GTPases, which is in agreement with the finding that the GAP domain of IQGAPs lacks the arginine finger that is critical for GAP activity (Kurella et al., 2009). Subsequent studies revealed IQGAP1 to be an effector of Rac1 and Cdc42, and the GRD domain in IQGAP1 was later shown to stabilize Rac1 and Cdc42 in the GTP-bound active form (Noritake et al., 2005; White et al., 2012). The other domains in IQGAPs mediate interactions with numerous partner proteins (Brown and Sacks, 2006). Indeed, IQGAP1 appears to be a hub of multiple signaling pathways mediating diverse biological processes. For example, IQGAP1 has been reported to function as a scaffold for the mitogen-activated protein kinase pathway by associating with B-Raf, MAPK/ERK kinase (MEK1/2) and extracellular signal-regulated kinase 2 (Erk2) (White et al., 2012). Furthermore, IQGAP1 has been shown to interact with components of adherens junctions (E-cadherin and β-catenin) (Kuroda et al., 1998; Fukata et al., 1999, 2001; Noritake et al., 2004) and microtubule (MT)-associating proteins (cytoplasmic linker protein, CLIP-170 (Fukata et al., 2002); CLIP-170-associating protein, CLASP (Watanabe et al., 2009); adenomatous polyposis coli protein, APC (Watanabe et al., 2004)). We and other researchers have demonstrated that IQGAP1 regulates E-cadherin-mediated intercellular adhesions in epithelial cells and MT organization in polarized migrating cells downstream of Rac and Cdc42 (Noritake et al., 2005; Jausoro et al., 2012; White et al., 2012). Thus, the IQGAP family is one of the critical regulators involved in mediating Rho family GTPases and their reorganization of adhesions and the cytoskeleton, including actin filaments (see below).
The activity of IQGAPs is controlled by various intracellular signals. It is widely accepted that Rac and Cdc42 facilitate the function of IQGAP1 in actin-crosslinking and determine its subcellular localization, such as lamellipodia at the front of migrating cells or intercellular adhesion sites in epithelial cells (Kuroda et al., 1996; Watanabe et al., 2004). In contrast to Rac and Cdc42, the binding of Ca2+/calmodulin appears to inactivate IQGAP1. Calmodulin primarily associates with the four tandem IQ motifs of IQGAP1 and a low affinity site in the CHD (Ho et al., 1999; Li and Sacks, 2003). Although, in general, calmodulin-binding proteins that contain the IQ motif have a higher affinity for the Ca2+-free form of calmodulin (apo-calmodulin), IQGAP1 preferentially associates with Ca2+/calmodulin rather than Ca2+-free calmodulin (Joyal et al., 1997). Importantly, calmodulin abrogates the association of IQGAP1 with Cdc42 only in the presence of Ca2+ (Joyal et al., 1997). Increasing the intracellular Ca2+ concentration enhances the interaction between calmodulin and IQGAP1, with a concomitant reduction in the binding of IQGAP1 to Cdc42. Thus, IQGAP1 integrates Ca2+/calmodulin and Cdc42/Rac1 signals in the regulation of the actin-cytoskeleton.
Phosphorylation is another mechanism by which IQGAP activity is regulated. Protein kinase C (PKC) ε can phosphorylate IQGAP1 in vitro and in vivo. This phosphorylation induces a conformational change in the C-terminal region of IQGAP1, thereby facilitating binding to Cdc42 (Grohmanova et al., 2004). Interestingly, Grohmanova et al. proposed that IQGAP1 is regulated by autoinhibition and that phosphorylation of Ser 1443 relieves the autoinhibitory conformation of IQGAP1 (Grohmanova et al., 2004). Consistently, a phosphomimetic mutant of IQGAP1 at the same site stimulates neurite outgrowth in NIE-115 cells (Li et al., 2005), indicating that phosphorylation in IQGAP1 controls its regulation of the cytoskeleton. Evidence regarding IQGAP1 phosphorylation has been increasing, and over forty phosphorylation sites in IQGAP1 have been reported in the PhosphoSitePlus database (http://www.phosphosite.org/). Further studies are required to uncover additional information about this interesting regulatory mechanism for the IQGAP family.
The IQGAP family has been implicated in signaling downstream of receptor tyrosine kinases, such as epidermal growth factor (EGF), vascular endothelial growth factor, hepatocyte growth factor, and nerve growth factor receptors, as well as G-protein-coupled receptors such as the bradykinin and thrombin receptors (Brown and Sacks, 2006; Bensenor et al., 2007; Wang et al., 2007). In addition, the signals from those receptors increase the phosphorylation of threonine and serine residues in IQGAP1 (Yamaoka-Tojo et al., 2004; McNulty et al., 2011; Kohno et al., 2013). Furthermore, complement receptor 3 (CR3) recruits IQGAP1 to the sites of phagocytosis (Brandt et al., 2007). Recent evidence clearly shows that, several integrin signals modulate the activity of Rho family GTPases through IQGAP1 (Wickstrom et al., 2010; Jacquemet et al., 2013a, 2013b), indicating IQGAP1 as a critical scaffold for signaling from transmembrane receptors.
Consistently, IQGAP1 associates with EGF receptor through the IQ motif and the stimulation with EGF induces the phosphorylation of IQGAP1 at Ser 1443 depending on PKC. This phosphorylation ensures EGF receptor activation, whilst the elevated intracellular Ca2+ level abrogates the binding between IQGAP1 and EGF receptor presumably through calmodulin (McNulty et al., 2011). As discussed above, the phosphorylation at Ser 1443 in IQGAP1 has positive effects on IQGAP1 activity for the reorganization of actin cytoskeleton and the binding of Ca2+/calmodulin to IQGAP1 are supposed to have negative effects. Calcium signal from the receptor may regulate IQGAP1 activity in a spatiotemporal manner.
Though many upstream signals modulate the activity of IQGAP1 and its regulation of the actin-cytoskeleton, the spatiotemporal control of IQGAP has not been directly addressed because of the lack of biosensors, which can be used in living cells, including FRET probes for IQGAPs. Such tools would greatly enhance our understanding of the temporal and spatial control of the IQGAP family in vivo.
The first evidence for the role of IQGAP1 in the actin-cytoskeleton came from a study showing co-purification of IQGAP1 with cytosolic F-actin from bovine adrenal tissue (Bashour et al., 1997). In that study, the authors clearly showed that IQGAP1 binds to actin filaments and crosslinks actin filaments. At almost the same time, it was shown that this ability to crosslink involved IQGAP1 dimerization and/or oligomerization (Fukata et al., 1997). Through the binding of active Cdc42 or Rac1, these Rho family GTPases enhance IQGAP1 oligomerization (Fukata et al., 1997). Subsequently, a monomeric CHD of IQGAP1 was shown to be sufficient to bind actin filaments at high affinity. Consistently, the IQGAP3 CHD also binds to actin filaments (Wang et al., 2007). Although the link between IQGAP2 and actin filaments has not been directly addressed, IQGAP2 appears to have a similar activity for actin crosslinking. Thus, dimerized and/or oligomerized IQGAPs utilize multiple CHDs to crosslink/bundle actin filaments without any actin-related proteins (Fig. 2).
IQGAP1 crosslinks actin filaments. IQGAP1 directly binds to actin filaments through its calponin homology domain (CHD) in the N-terminus. The IQGAP1 dimer or Cdc42/Rac-mediated IQGAP1 oligomer utilizes multiple CHDs to crosslink and bundle actin filaments.
This direct regulation of actin filaments by IQGAPs seems to be evolutionarily conserved. IQGAP proteins in the budding yeast, Iqg1/Cyk1, localize to a ring structure at the mother-bud junction with myosin and are required for completing cytokinesis (Lippincott and Li, 1998). Similar roles of IQGAP proteins in contractile actin filaments have also been reported for Rng2 (fission yeast) and DGAP1 (Dictyostelium). Consistently, depletion of IQGAP proteins in various species results in a loss of locally organized actin filaments at the cytokinetic ring or at intercellular adhesion sites (Brandt and Grosse, 2007). Interestingly, recent study from mammalian cultured cells suggest the divergent roles of IQGAPs in cytokinesis (Adachi et al., 2014).
Several studies have implicated the IQGAP family in the modulation of local actin dynamics. Although the molecular mechanisms are not fully understood, three scenarios have been proposed for the regulation of actin dynamics by IQGAP1. IQGAP1 can enhance actin polymerization through interactions with two different actin-assembling proteins, the actin-related proteins (Arp) 2/3 and formins. Furthermore it has been suggested that IQGAP1 directly regulates actin assembly by itself. These three scenarios are intriguing because the former two actin polymerization machineries mediate independent functions for entirely different actin filament structures. Arp2/3 generates a branched actin meshwork from pre-existing filaments together with actin nucleation-promoting proteins, such as Wiskott-Aldrich syndrome protein (WASP), whereas formins processively catalyze actin nucleation from the barbed ends through their formin homology domain (FH) to produce linear filaments (Goode and Eck, 2007; Takenawa and Suetsugu, 2007; Rotty et al., 2013).
IQGAP1 accumulates at the front leading edges and is required for proper lamellipodia formation in motile cells, implicating IQGAP1 in the regulation of the actin-cytoskeleton in generating protrusive forces. Two studies published in 2007 successfully identified N-WASP as an IQGAP1-binding protein in migrating cells (Bensenor et al., 2007; Le Clainche et al., 2007). Although their binding regions are under debate, IQGAP1 stimulates N-WASP and Arp2/3-mediated actin branching and polymerization for lamellipodia formation (Fig. 3). How does IQGAP1 mediate these two functions? Similarly to IQGAP1, N-WASP is maintained in an autoinhibited state in the absence of stimulation and is also an effector of Cdc42. Activated Cdc42 directly binds to the GTPase binding domain (GBD) in N-WASP inducing a conformational change into an “open” state (Takenawa and Suetsugu, 2007). Because the binding region for IQGAP1 overlaps with the GBD, the interaction with IQGAP1 appears to relieve the autoinhibition thereby activating N-WASP and stimulating Arp2/3-mediated branched actin polymerization (Fig. 3). Intriguingly, activated GTP-bound Cdc42 enhances this IQGAP1 activity, presumably acting on IQGAP1 rather than N-WASP (Bensenor et al., 2007). Because activated Cdc42 alone can stimulate N-WASP and IQGAP1 stabilizes Cdc42 in an active form, it is possible that this machinery somehow effectively induces the actin meshwork composed of a branched actin-cytoskeleton under the control of Cdc42 (Fig. 3).
IQGAP1 stimulates actin dynamics together with its binding proteins. A. IQGAP1 induces actin branching and polymerization with N-WASP. N-WASP activity is inhibited by the intramolecular association between the GTPase-binding domain (GBD) and verprolin-homology, cofilin-homology and acidic domain (VCA). Active Cdc42 relieves the autoinhibition in N-WASP and activates actin nucleation by the Arp2/3 complex. Binding of IQGAP1 to N-WASP accelerates branched actin polymerization. Notably, the binding regions in IQGAP1 and N-WASP are under debate, and this schematic representation is based on one study (Le Clainche et al., 2007). B. IQGAP1 stimulates actin elongation together with formin. Dia1, a member of the forming family, mediates actin polymerization from the barbed ends. In the resting state, Dia1 resides in an autoinhibited conformation. Active RhoA (GTP-bound) binds to the GTPase-binding domain (GBD), thereby relieving the autoinhibition between the GBD and Dia-autoregulatory domain (DAD). RhoA-bound Dia1 associates with IQGAP1 and mediates actin elongation at the formin homology 2 (FH2) domain from the barbed ends. Binding to IQGAP1 occurs most likely to recruit Dia1 to specific subcellular destinations.
Using interactome-based screening, IQGAP1 was found to associate with the Diaphanous-related formin Dia1 (Brandt et al., 2007). Surprisingly, IQGAP1 binds specifically to Dia1 but not to other formins such as Dia2 and Dia3. Dia1 is widely accepted as an effector of RhoA, and the binding of active RhoA releases the intramolecular autoinhibition of Dia1 between the N-terminal four armadillo repeat-containing Diaphanous inhibitory domain (DID) and the C-terminal diaphanous autoregulatory domain (DAD) (Goode and Eck, 2007). IQGAP1 interacts with the N-terminal region of Dia1, which partially overlaps with DID, and RhoA increases the association of Dia1 with IQGAP1, indicating that an active open conformation of Dia1 preferentially associates with IQGAP1. However, unlike the case of N-WASP, the binding of IQGAP1 to Dia1 is not sufficient to activate Dia1 to enhance actin polymerization. Rather, IQGAP1 recruits Dia1 to a specific subcellular location, such as the leading edges in migrating cells and the phagocytosis cup in macrophages (Brandt et al., 2007). There, the association of IQGAP1 may maintain Dia1 in an active open conformation to induce actin elongation (Fig. 3).
The two molecular mechanisms discussed above (with N-WASP and Dia1) facilitate actin nucleation; however, IQGAP1 was recently shown to inhibit actin polymerization from the barbed ends by itself (Fig. 4) although this particular function of IQGAP1 in cells has not been addressed. Carlier and colleagues found that IQGAP1 purified from bacteria, caps the barbed ends with a higher affinity for the ADP-bound terminal subunits in actin filaments (Pelikan-Conchaudron et al., 2011). In addition, this activity depends on the C-terminal half of IQGAP1 and the capping activity is inhibited by calmodulin irrespective of the Ca2+ concentration. These findings may corroborate those of another study showing that IQGAP1 can stimulate actin assembly through N-WASP and Arp2/3 (Bensenor et al., 2007). Taken together, given that Ca2+/calmodulin antagonizes signals from Rac1 and Cdc42 on IQGAP1, activated Rac1/Cdc42 may cancel out the capping activity of IQGAP1 to induce reorganization of the actin-cytoskeleton.
IQGAP1 prevents actin polymerization by capping the barbed ends. The C-terminal half of IQGAP1 binds to the barbed end of actin filaments and prevents actin polymerization. The binding of calmodulin (CaM) cancels this inhibitory effect on actin polymerization, irrespective of the Ca2+ concentration.
The interplay among IQGAP1, Dia1, and N-WASP raises an interesting possibility for their roles in promoting protrusion in migrating cells. A paper using FRET biosensors and computational analysis indicates that there is coordination between specific Rho family GTPases during cell protrusion (Machacek et al., 2009): RhoA is activated at the cell edges synchronous with edge advancement, whereas Cdc42 and Rac1 are activated several microns behind the edge with a delay of a few tens of seconds. During cellular protrusion, RhoA may activate Dia1 at the edges to induce actin elongation from the barbed ends. Thereafter, behind this region, IQGAP1 may facilitate Dia1-mediated actin elongation and simultaneously enhance N-WASP-Arp2/3-mediated formation of the branched actin-cytoskeleton. In addition, in the region where Cdc42 and Rac1 are activated, IQGAP1 may crosslink actin filaments to stabilize the actin-based meshwork, thereby generating protrusive forces at lamellipodia.
The IQGAP family also contributes to the regulation of the actin-cytoskeleton in neuronal cells. In dissociated hippocampal neurons, IQGAPs accumulate at actin-rich cellular structures such as axon growth cones and spines (Wang et al., 2007; Swiech et al., 2011). Consistently, IQGAPs are required for axon outgrowth and spine morphology (Gao et al., 2011). Although the above regulatory mechanisms that are performed by IQGAP1 in actin dynamics can be applicable to neuronal cells, it would be interesting to determine the underlying mechanisms in neurons.
The IQGAP family has emerged as a critical regulator in the dynamics of the actin-cytoskeleton by itself and/or through Dai1 and N-WASP. Those regulatory mechanisms appear to be essential for cell migration and epithelial polarity in cultured cells. The major regulators for IQGAP activity are the Rho family GTPases, Rac1 and Cdc42, and Ca2+/calmodulin. The development of highly sensitive biosensors has furthered our understanding of the regulation of the Rho family GTPases, but their interplay with Ca2+/calmodulin signaling is largely unknown. Addressing this question is required for a full understanding of the mode of action of IQGAPs in the regulation of the actin-cytoskeleton.
Consistent with the critical roles of IQGAPs in regulation of cytoskeleton including actin filaments and microtubules (Noritake et al., 2005), IQGAPs are involved in cell migration. In motile cells, IQGAP1 localizes at the leading edges and is well established to promote cell migration together with Rac1 and Cdc42 (Mataraza et al., 2003). Recent studies have identified additional molecular mechanisms where IQGAP1 controls cell migration by modulating the signaling from growth factor receptors (Yamaoka-Tojo et al., 2004, 2006; Kohno et al., 2013) and cell-substratum adhesion (Kohno et al., 2013; Schiefermeier et al., 2014), and by controlling protein trafficking (Osman, 2010). Although the details are referred to the recent review (Smith et al., 2015), the diverse actions of IQGAPs in cell migration reflect the scaffold function of IQGAPs through their multidomains that interact over 100 proteins (Hedman et al., 2015).
The founding member IQGAP1 interacts with the proteins, including Rac1, Cdc42, cadherin, and the components of MAPK pathway, that relates to cancer. It is well established that IQGAP1 implicates in cancer and tumorigegenesis (White et al., 2009). The accumulating evidence has suggested a variety of physiological roles of the IQGAP family. IQGAP1 is involved in glomerular filtration in kidney, neuronal morphogenesis and migration, cardiomyocyte hepertrophy and survival in heart, endothelial migration/proliferation and barrier integrity in vascular system, airway smooth muscle cell contraction in lung, and insulin secretion in pancreatic β-cells (Hedman et al., 2015). Despite the high sequence similarity in IQGAPs, IQGAP1 is an oncogene while IQGAP2 is a tumor suppressor (Johnson et al., 2009; White et al., 2009). Indeed, decreased IQGAP2 was observed in hepatocellular (Sun et al., 2008; White et al., 2010; Gnatenko et al., 2013), prostate (Xie et al., 2012), and gastric carcinoma (Jin et al., 2008). IQGAP2-null mice develop hepatocellular carcinoma, but IQGAP1 and IQGAP2 double knock out mice reverse their survival (Schmidt et al., 2008). Similar to IQGAP1, through Erk activation, IQGAP3 promotes proliferation of liver and mammary epithelial cells (Nojima et al., 2008; Kunimoto et al., 2009). Increased expression of IQGAP3 also correlates with invasion and proliferation of lung cancer cells (Yang et al., 2014). Generation of IQGAP3 and triple IQGAP knockout mice would facilitate our understanding for physiological functions of the IQGAP family.
We thank all members of the Kaibuchi lab for discussions and support, and Dr. Ellen O’Shaughnessy (The University of North Carolina at Chapel Hill, Department of Pharmacology) for proofreading and helpful discussions. This work was supported by KAKENHI (23123507, 20227006 to K.K., 20790279 to T.W.) and GCOE to K.K.
All authors have no conflict-of-interest and no financial disclosure with regards to this manuscript.