|Home | About | Journals | Submit | Contact Us | Français|
In response to directional stimulation by a chemoattractant, cells rapidly activate a series of signaling pathways at the site closest to the chemoattractant source that leads to F-actin polymerization, pseudopod formation, and directional movement up the gradient. Ras proteins are major regulators of chemotaxis in Dictyostelium; they are activated at the leading edge, are required for chemoattractant-mediated activation of PI3K and TORC2, and are one of the most rapid responders, with activity peaking at ~3 s after stimulation. We demonstrate that in myosin II (MyoII) null cells, Ras activation is highly extended and is not restricted to the site closest to the chemoattractant source. This causes elevated, extended, and spatially misregulated activation of PI3K and TORC2 and their effectors Akt/PKB and PKBR1, as well as elevated F-actin polymerization. We further demonstrate that disruption of specific IQGAP/cortexillin complexes, which also regulate cortical mechanics, causes extended activation of PI3K and Akt/PKB but not Ras activation. Our findings suggest that MyoII and IQGAP/cortexillin play key roles in spatially and temporally regulating leading-edge activity and, through this, the ability of cells to restrict the site of pseudopod formation.
The ability of amoeboid cells such as Dictyostelium cells, neutrophils, or macrophages to sense and respond to directional chemical cues and move up a chemoattractant gradient is central to a wide range of cellular processes (Jin et al., 2008 ; Sallusto and Baggiolini, 2008 ). Cells must efficiently integrate multiple signaling responses to coordinate F-actin–mediated protrusion at the leading edge and actomyosin contraction at the cell's posterior (Kehrl, 2006 ; Van Haastert and Veltman, 2007 ; Janetopoulos and Firtel, 2008 ). In Dictyostelium, Ras is required for efficient directional sensing and chemotaxis, lies upstream of and regulates PI3K (phosphatidylinositol 3-kinase) and TOR Complex 2 (TORC2), and is activated at the leading edge (Lee et al., 1999 , 2005 ; Sasaki et al., 2004 , 2007 ; Bolourani et al., 2006 ; Kae et al., 2007 ; Van Haastert and Veltman, 2007 ). Other pathways (Rap1 and guanylyl cyclase), which have been linked to myosin II (MyoII) regulation, are also activated at the leading edge. Rap1 functions, in part, to disassemble MyoII filaments at the leading edge through the activation of the Ser/Thr kinase Phg2, whereas cGMP, the product of guanylyl cyclase, mediates MyoII assembly and function in the cell's posterior (Rebstein et al., 1997 ; Gebbie et al., 2004 ; Bosgraaf and Van Haastert, 2006 ; Kortholt et al., 2006 ; Jeon et al., 2007 ; Kortholt and Van Haastert, 2008 ). Of these leading-edge pathways, Ras activation is the earliest known response downstream from the receptor and heterotrimeric G proteins in Dictyostelium with activity peaking at ~3 s after stimulation (Sasaki et al., 2004 ).
When cells are given a uniform (global) stimulation of chemoattractant, pathways that regulate the formation of the leading edge are rapidly and uniformly activated along the cell's cortex. In both Dictyostelium and neutrophils, these pathways are amplified through positive feedback loops that involve F-actin, causing signal amplification and stabilization of the leading edge, which are critical for cells to move up shallow, weak chemoattractant gradients (Rubin and Ravid, 2002 ; Sasaki et al., 2004 ; Van Keymeulen et al., 2006 ; Brandman and Meyer, 2008 ). In concert with these responses, cortical MyoII is disassembled through phosphorylation contemporaneously with F-actin polymerization (Moores et al., 1996 ; de la Roche and Cote, 2001 ; Funamoto et al., 2002 ; Iijima and Devreotes, 2002 ). In chemotaxing cells, cortical MyoII is spatially distributed along the lateral sides and backs of cells, areas in which leading-edge signaling pathways are normally not activated, where it provides cortical tension (rigidity) and contraction of the posterior during cell movement (Wessels et al., 1988 ; Egelhoff et al., 1996 ; Stites et al., 1998 ; Xu et al., 2001 ; Laevsky and Knecht, 2003 ; Reichl et al., 2008 ).
Leading-edge (front) and trailing edge (back) responses are also spatially restricted and balanced. In neutrophils, pathways regulated by the heterotrimeric G protein Gi control leading-edge responses, including PI3K, Rac, and F-actin polymerization, whereas G11/12, regulates RhoA, MyoII assembly, and contraction in the posterior (Weiner et al., 2002 ; Li et al., 2003 , 2005 ; Van Keymeulen et al., 2006 ). Inhibition of one of these pathways causes up-regulation of the other. Dictyostelium cells lacking MyoII (mhcA− cells) are unable to inhibit lateral pseudopodia and produce them simultaneously along the lateral sides of cells, in addition to being defective in contracting their posteriors and often exhibiting decreased polarity (Wessels et al., 1988 ). Mammalian cells in which RhoA activity or the activity of the Rho kinase (ROCK) is inhibited exhibit similar phenotypes (Xu et al., 2003 ; Wu et al., 2009 ). PTEN, a negative regulator of the PI3K pathway, localizes to the sides and posterior of chemotaxing cells and is thought to play a role in restricting phosphatidylinositol (3,4,5)-triphosphate [PI(3,4,5)P3] accumulation to the leading edge (Funamoto et al., 2002 ; Iijima and Devreotes, 2002 ; Zhang et al., 2008 ). The RasGAP Dd-NF1 also plays a key part in temporally controlling RasG and downstream PI3K signaling, is required for establishing a defined leading edge, and is important for directional sensing.
Although some progress has been made, the mechanisms that restrict the activation of leading-edge pathways to the front of the cell, and thereby inhibit pathways from being activated along the lateral sides and posterior, are still not well understood. The observations that cells lacking MyoII (mhcA− cells) are unable to properly inhibit the formation of lateral pseudopodia suggested the possibility that components of the cytoskeleton may be important negative regulators of leading-edge function. To investigate this possibility, we examined the spatial and temporal control of Ras and PI3K signaling in cells in which MyoII function is either lost or altered and in cells lacking the IQGAP/cortexillin complex, which plays an important role as an F-actin cross-linker and in regulating cortical mechanics. IQGAPs are found in all eukaryotes and are required for cytokinesis in Dictyostelium, Saccharomyces cerevisiae, and Schizosaccharomyces pombe. IQGAPs contain a conserved a RasGAP-related Rac1-GTP–binding domain near their C-terminus, a C-terminal effector-binding domain, and an N-terminus of varying length (Adachi et al., 1997 ; Lee et al., 1997 ; Eng et al., 1998 ; Faix et al., 1998 ; Osman and Cerione, 1998 ; Briggs and Sacks, 2003 ; Bensenor et al., 2007 ; Brandt and Grosse, 2007 ). In Dictyostelium, activated Rac1 (Rac1-GTP) binding to DdIQGAP1 (DGAP1) leads to the recruitment of a dimer of the F-actin bundling protein cortexillin I (ctxI) to a Rac1-GTP/IQGAP1/ctxI complex at the cleavage furrow during cytokinesis and is involved with the control of MyoII localization (Faix et al., 1998 , 2001 ; Ren et al., 2009 ). DdIQGAP1 has also been linked to cell motility and morphogenesis; the expression level of IQGAP1, but not DdIQGAP2 (GAPA), has been correlated to the rate of cell motility.
We have found that MyoII and IQGAP/ctx play distinct and independent roles in regulating leading-edge function. MyoII is important for controlling Ras activity, and cells lacking MyoII have greatly extended Ras activation and exhibit Ras and PI3K activation randomly along the cell cortex, which we suggest is responsible for the adventitious formation of lateral pseudopodia. In contrast, cells lacking specific combinations of IQGAPs and ctxs exhibit normal activation of Ras but high and extended activation of PI3K and elevated F-actin levels. These studies provide new evidence that MyoII and IQGAP/ctx are important in restricting at least some leading-edge pathways to the site on the cortex closest to the chemoattractant source.
We have investigated the spatial and temporal control of two key leading-edge pathways, Ras and PI3K. Our findings support a role for MyoII in regulating the spatiotemporal control of Ras signaling and for IQGAP/ctx in inhibiting and spatially restricting the extent of activation of PI3K but not Ras, suggesting that different F-actin regulators play distinct roles in controlling leading-edge function.
Cells were starved for 2 h, pulsed with 100 nM cAMP for 5 h, and treated with 1 mM caffeine for 30 min before stimulation with 100 mM cAMP. Cytoskeletal proteins were isolated as proteins insoluble in the detergent Triton X-100 as described previously (Meili et al., 1999 ; Steimle et al., 2001 ). The protein pellets were dissolved by being boiled in 2× SDS-PAGE sample buffer, run on 8% acrylamide gels, and stained with Coomassie blue. Protein bands were scanned, and changes in actin and MyoII content were quantified using Image Gauge software (Fuji, Stamford, CT). All experiments were repeated at least three times on separate days with internal wild-type cells assayed contemporaneously to provide an internal reference.
To produce aggregation-competent cells, log-phase vegetative cells were washed twice with 12 mM Na/K phosphate buffer and resuspended to a density of 7 × 106 cells/ml in Na/K phosphate buffer (Insall et al., 1994 ). Cells were pulsed with 30 nM cAMP at 6-min intervals for 5 h. The cells were spun down and resuspended to 2 × 107 cells/ml in Na/K phosphate buffer, pH 6.2. The cells were stimulated with 1 μM cAMP, and time points were taken as indicated. All experiments were repeated at least three times on separate days with internal wild-type cells assayed contemporaneously to provide an internal reference.
Assays for Akt/protein kinase B (PKB) and PKBR1 were performed as described previously (Meili et al., 1999 , 2000 ). We determined the relative activity of Akt/PKB after cAMP stimulation using H2B (Histone 2B) as a substrate. We measured 32P phosphate incorporation using a Typhoon 9400 phosphorimager (GE Healthcare, Piscataway, NJ). All experiments were repeated at least three times on separate days with internal wild-type cells assayed contemporaneously to provide an internal reference.
We performed the chemotaxis analysis as described previously using cAMP-stimulated cells (Funamoto et al., 2001 ). Eppendorf “femtotip” micropipettes containing 150 mM cAMP were used without applied pressure for the chemotaxis assays. We did the computer analysis by using DIAS software (Soll Technologies, Iowa City, IA; Wessels et al., 1988 ). At least five cells were analyzed for at least three separate movies taken on separate days, and the chemotaxis assays were performed at least three times on separate days.
We performed phalloidin staining as described previously (Chung et al., 2000 ). Pulsed cells were spotted on a coverslip and let stand for 10 min. Cells were fixed with 3.7% formaldehyde for 10 min and then permeablized with 0.2% Triton X-100 for 1 min. We stained the cells with TRITC-labeled phalloidin containing 0.1% bovine serum albumin (BSA) and 0.2% Tween 20 for 1 h. After washing, we observed the cells with a 63× oil immersion lens on a Leica microscope (Deerfield, IL).
Ras pulldown assays were done as described previously (Sasaki et al., 2004 ). The cell extract was incubated with 10 μg of glutathione S-transferase (GST)-RBD (Rac-binding domain) on glutathione-agarose beads at 4°C for 30 min in the presence of 5 mg/ml BSA. We washed the beads three times. Ras proteins were separated on a 12% SDS-PAGE gel and immunoblotted with anti-Pan-Ras antibody. All experiments were repeated at least three times on separate days with internal wild-type cells assayed contemporaneously to provide an internal reference.
The ctxI, ctxII, and ctxI/II null strains were obtained from the Dictyostelium stock center. The mhcA− strain was from the stock center and another was also created in our laboratory. The phenotypes of the two strains were indistinguishable. The constructs and Southern blots of the newly created strains are shown in Supplemental Figure 6. Double knockouts were made by sequential use of Bsr and Hygro cassettes at the same insertion point in the knockout construct.
One hypothesis to explain the increase in lateral pseudopodia and the elevated F-actin response is that mhcA− cells might be unable to restrict leading-edge signaling pathways to the site of the cell closest to the chemoattractant source, resulting in the simultaneous activation of these responses at many sites along the cell's cortex, which then lead to the formation of multiple F-actin protrusions. To test this hypothesis, we examined the kinetics and extent of activation of the Ras and PI3K pathways known to regulate F-actin polymerization in Dictyostelium (Chen et al., 2003 ; Sasaki et al., 2004 , 2007 ; Zhang et al., 2008 ). Figure 1, A and B, depicts the analysis of Ras activation using a pulldown assay, which is extended in mhcA− cells compared with that in wild-type cells (Table 1). Both strains exhibit a rapid increase in Ras-GTP levels upon chemoattractant stimulation, when wild-type cells rapidly decrease to basal levels by the 20-s time point, as described previously (Sasaki et al., 2004 ). In contrast, in mhcA− cells, Ras-GTP levels remain elevated for an extended time.
We then studied the extent and kinetics of Ras activation in vivo using the real-time Ras-GTP reporter green fluorescent protein (GFP)-RBD, which binds to Ras-GTP and translocates to the plasma membrane in response to Ras activation, where it binds to Ras-GTP. Figure 1C shows the quantitation of these assays. Both strains exhibit a rapid increase in Ras-GTP levels upon chemoattractant stimulation, which in wild-type cells peak at ~3–4 s after stimulation and then rapidly decrease to near-basal levels by ~10 s, as described previously (Sasaki et al., 2004 ). In contrast, in mhcA− cells, the activity peaks later and exhibits a broader peak and a gradual decrease, consistent with the kinetics observed using the pulldown assay.
To determine if pathways downstream from Ras are extended in a similar manner, we used plasma membrane translocation of a GFP-PH domain reporter of PI(3,4,5)P3 levels, GFP-PHPhdA, to study the activation of the Ras-dependent effector PI3K (Parent et al., 1998 ; Meili et al., 1999 ; Funamoto et al., 2002 ). Figure 1D illustrates that GFP-PHPhdA plasma membrane localization rapidly increases, peaking at ~5–6 s after stimulation in wild-type and mhcA− cells and then rapidly decreasing to basal levels in wild-type cells, whereas in mhcA− cells, GFP-PHPhdA remains at the plasma membrane for a considerably longer period of time. These findings suggest that the extended kinetics of Ras activation lead to extended PI3K activation. PI(3,4,5)P3 levels are negatively regulated by the phosphatase PTEN, which is associated with the plasma membrane in unstimulated cells and then transiently delocalizes from the cortex (Funamoto et al., 2002 ; Iijima and Devreotes, 2002 ). To distinguish between the extended PI3K activity being a function of extended Ras activity or changes in the kinetics of PTEN localization on the plasma membrane, we investigated the kinetics of cortical GFP-PTEN delocalization and rebinding in wild-type and mhcA− cells. We found the kinetics are indistinguishable, supporting the model in which extended PI3K activity is due to extended Ras activity (data not shown). However, we cannot distinguish between this model and one in which Ras-mediated PI3K activity is extended and PTEN enzyme activity is also decreased in mhcA− cells.
F-actin levels are regulated, in part, by the PI3K pathway (Iglesias and Devreotes, 2008 ; Janetopoulos and Firtel, 2008 ; Koelsch et al., 2008 ; Kortholt and Van Haastert, 2008 ; King and Insall, 2009 ). To determine if the elevated and extended PI3K pathway results in a change in the F-actin profile, we examined the kinetics of F-actin polymerization in wild-type and mhcA− cells. As shown in Figure 1D, wild-type cells exhibit reciprocal, bimodal kinetics of chemoattractant-stimulated F-actin polymerization and MyoII polymerization, consistent with previous studies (Hall et al., 1988 ; Moores et al., 1996 ; Steimle et al., 2001 ; Levi et al., 2002 ). F-actin accumulation exhibits a first peak at ~5 s, after which F-actin levels decrease to near-basal levels with a trough at ~20 s, followed by a lower, broad second peak at ~30–45 s (Figure 1D; Hall et al., 1988 ; Yumura and Fukui, 1998 ). In contrast, MyoII levels decrease slightly at 5 s, followed by a gradual rise, peaking at 20–40 s and then decreasing to basal levels as described previously (Bosgraaf et al., 2002 ; Park et al., 2004 ). We find that mhcA− cells exhibit a higher basal level of F-actin. On chemoattractant stimulation, the rapid increase in F-actin levels is similar in wild-type and mhcA− cells (first peak), but the second F-actin peak is much larger in mhcA− cells and starts to rise before the first peak has decreased much (Figure 1E). These findings suggest that the level of the second F-actin peak, which has been linked to pseudopod protrusion (Hall et al., 1988 ), is modulated by assembled MyoII, consistent with a model in which MyoII must be disassembled at the site of F-actin polymerization for pseudopod extension.
Akt/PKB kinase activity depends on two Ras-mediated effector pathways: PI3K, which promotes the PIP3-dependent recruitment of Akt/PKB to the plasma membrane and regulates PDK1 phosphorylation of the Akt/PKB activation loop, and TORC2, which mediates Akt/PKB phosphorylation on the C-terminal hydrophobic motif (Meili et al., 1999 ; Funamoto et al., 2002 ; Lee et al., 2005 ; Kamimura et al., 2008 ). As depicted in Figure 2, A and B, and Table 2, Akt/PKB exhibits an elevated level and extended kinetics of Akt/PKB activation in mhcA− cells compared with those in wild-type cells, consistent with extended activation of Ras and PI3K. In addition, Dictyostelium has a second PKB-related enzyme that localizes to the plasma membrane constitutively through an N-terminal myristoylation, which makes PKBR1 activation PI3K-independent (Firtel and Meili, 2000 ). Like PKB, PKBR1 requires TORC2 to phosphorylate the C-terminal hydrophobic motif and to activate the enzyme. Figure 2, C and D, shows that, like Akt/PKB, PKBR1 activity is also highly elevated and extended, consistent with both TORC2 and PI3K being affected in mhcA− cells.
MyoII assembly/disassembly in Dictyostelium is regulated by the phosphorylation of three threonines (Thr1823, 1833, 2029) in the coil-coiled domain by a family of four MyoII heavy-chain kinases (MHCK-A–D). Phosphorylation of these three sites causes MyoII filament disassembly, whereas dephosphorylation leads to MyoII assembly (Egelhoff et al., 1993 ; Stites et al., 1998 ; de la Roche et al., 2002 ; Bosgraaf and Van Haastert, 2006 ). Substitutions of these threonine residues with alanines result in a MyoII that is constitutively assembled, whereas Asp substitutions produce a MyoII that does not assemble (Egelhoff et al., 1993 ; de la Roche et al., 2002 ). We found that Akt/PKB activation is suppressed in mhcA− cells expressing MyoII-3XAla, suggesting that constitutive MyoII assembly inhibits pathway activation (Figures 2, A and B; Table 2). In contrast, mhcA− cells expressing MyoII-3XAsp exhibit a pattern of extended activation similar to that observed in mhcA− cells and an activation peak that is between that of wild-type and mhcA−cells. These findings are consistent with a model in which MyoII assembly dynamics are required for the rapid Ras/PI3K pathway adaptation. We examined this further by studying Akt/PKB activation in cells lacking MHCKs, which have elevated levels of assembled MyoII (Yumura et al., 2005 ). Cells lacking MHCK-A exhibit activation kinetics that are slightly lower than those of wild-type cells (Table 2). However, disruption of MHCK-A, -B, and -C results in a considerable impairment of Akt/PKB activation, consistent with our model (Figure 2, A and B; Table 2).
We then investigated how MyoII activity might down-regulate Ras and PI3K activity by assaying Ras and Akt/PKB activation in mhcA− cells expressing MyoIIS456A, a MyoII mutant with impaired motor activity (Murphy et al., 2001b ; Reichl et al., 2008 ; Ren et al., 2009 ). We found that activation of Ras and PKB activity is extended in MyoIIS456A/mhcA− cells, although not to the extent observed in mhcA− cells, consistent with the reduced activity of this motor (Figures 1A and and2,2, A and B; Tables 1 and and22).
We demonstrated previously that F-actin polymerization is a component of a positive feedback loop that amplifies leading-edge responses and stabilizes a nascent leading edge, in part through the recruitment of PI3K (Sasaki et al., 2004 ). Pretreatment of cells with latrunculin B (LatB), an F-actin inhibitor, dramatically reduces the level of chemoattractant-mediated PKB activation in both mhcA− and wild-type cells (Figure 2A; Table 1; Sasaki et al., 2004 ), although PKB activity is still higher and more extended in mhcA− cells than in LatB-treated wild-type cells. These observations indicate that the extended PKB activation in mhcA− cells is not due to an enhanced positive feedback mediated by the elevated level of F-actin polymerization in these cells. Our findings are consistent with a model in which MyoII helps mediate the adaptation or restrict the activation of at least some early chemoattractant-mediated effector pathways.
To understand the effect of loss of MyoII on the spatial activation of Ras and PI3K pathways in vivo, we followed the spatiotemporal localization of GFP-RBD and GFP-PHPhdA in cells placed in a chemoattractant gradient. The localization of GFP-PHPhdA and the Ras-GTP reporter GFP-RBD are correlated with the position of new F-actin polymerization and pseudopodial protrusions (Parent et al., 1998 ; Meili et al., 1999 ; Sasaki et al., 2004 ). Whereas GFP-RBD and GFP-PH domain reporters localize almost exclusively to the site on the membrane closest to the chemoattractant source in wild-type cells, both reporters localize almost randomly in mhcA− cells (Figure 3). As mhcA−cells respond to the chemoattractant gradient, activated Ras and PI(3,4,5)P3 are often found in more than one domain on the cell cortex simultaneously. The domains of Ras and PI3K activity are generally much broader than those of wild-type cells, consistent with the inability of mhcA− cells to polarize efficiently.
To evaluate the role of elevated PI3K activity in mediating the increase in lateral pseudopod formation in mhcA−cells, we examined the chemoattractant behavior of these cells treated with LY294002. As shown previously, LY294002 treatment considerably reduces chemoattractant speed and the extent of pseudopod extension in wild-type cells compared with untreated cells (quantitation of chemotaxis using DIAS software in Table 3; Supplemental Figures 1 and 2; Takeda et al., 2007 , Bosgraaf et al., 2008 ). The LY294002-treated cells also undergo a moderate loss of directionality due to some pseudopodia localizing more randomly along the cell cortex. As described previously and reexamined here for comparison, mhcA− cells move more slowly and have a reduced directionality compared with wild-type cells because of a reduced ability to suppress lateral pseudopodia (Table 3; Supplemental Figure 2; Wessels et al., 1988 ). Treatment of mhcA− cells with LY294002, as in wild-type cells, inhibits PI3K activity in these cells as determined by the abrogation of PH domain localization to the cell cortex in response to chemoattractant stimulation (data not shown). Analysis of LY294002-treated mhcA− cells chemotaxing cells shows that these cells have a further reduction of speed (Table 3) and the amount of pseudopod extension (Supplemental Figure 2) compared with mhcA− cells. Although the LY294002-treated mhcA− cells exhibit a modest increase in directionality, we note that they have a reduced ability to suppress lateral pseudopod formation compared with wild-type cells. The observations suggest that inhibition of PI3K reduces the extent of lateral pseudopod formation in mhcA− cells; however, these cells still exhibit a reduced polarity and directionality compared with LY294002-treated wild-type cells. We expect this reduction in polarity and directionality is due, in part, to the reduced and aberrant localization of traction forces observed in mhcA− cells due to loss of cortical tension and the ability to properly organize the cell cortex along the lateral sides and posterior of the cells (del Alamo et al., 2007 ; Lombardi et al., 2007 ; Meili et al., 2010 ).
The IQGAP/ctx complex is a key component of the cytoskeleton that also functions to cross-link F-actin and to mediate cortical tension (Ren et al., 2009 ; Kee and Robinson, unpublished observations). Before examining the potential role of IQGAPs and ctxs in spatially and temporally controlling leading-edge pathways, we felt it was important to define the composition of the Dictyostelium IQGAP complexes. Previous studies demonstrated that DdIQGAP1 (DGAP1) interacts with ctxI and Rac1 in vegetative cells (Faix et al., 1998 ; Weber et al., 1999 ). (For simplicity and to conform to Dictyostelium gene nomenclature, we have renamed DGAP1, GAPA, and a third IQGAP we have identified as DdIQGAP1, DdIQGAP2, and DdIQGAP3, respectively, and the respective genes as iqgA, iqgB, and iqgC.) To understand potential differences in function of DdIQGAP1 and DdIQGAP2 and to identify the components that interact with IQGAP1 and IQGAP2 during the aggregation stage of development when cells are competent to respond to cAMP as a chemoattractant, we used individually epitope-tagged DdIQGAP1 and DdIQGAP2 to purify complexes containing these proteins from vegetative cells and unstimulated and cAMP-stimulated developed (pulsed) cells. For both DdIQGAPs, we observed four major bands on silver-stained SDS-PAGE gels that were not found in the control samples (cells not expressing the epitope-tagged DdIQGAP; Figure 4), and the pattern of bands observed in samples from all of the cells were indistinguishable. Mass spectroscopy analysis revealed that the bands isolated using tagged DdIQGAP1 were DdIQGAP1, ctxI and ctxII, a previously undescribed cortexillin, ctxIII, and all three members of the Rac1 family: Rac1A, Rac1B, and Rac1C (Supplemental Figure 3). Using tagged DdIQGAP2 to isolate complexes, we identified DdIQGAP2, Rac1A, B, and C, and ctxI and II but not ctxIII (Figure 4). The low-mobility bands at ~200 kDa are most likely homodimers or a large multimeric complex of the IQGAP, because immunoblots of the gels revealed that the bands reacted with the epitope used to tag the IQGAP1 or IQGAP2, respectively. Complexes that contained both DdIQGAP1 and DdIQGAP2 were not observed. Two other proteins, HspE and rp17, were also found to be enriched in the complexes but were present at much lower stoichiometric levels than the other proteins and were not examined further (data not shown). Coimmunoprecipitation experiments using cells expressing the DdIQGAP1 or DdIQGAP2 tagged with one epitope and ctxIII with another epitope confirmed that DdIQGAP1, but not DdIQGAP2, coimmunoprecipated with ctxIII (Supplemental Figure 4).
To understand the potential roles each of these proteins play in chemotaxis, we undertook a more detailed examination of both iqgA− and iqgB− strains and created and studied iqgC−, iqgA−/B−, iqgA−/C−, and iqgB−/C− strains. Images of chemotaxing cells are presented in Figure 5(some strains not shown), and the quantitation of chemotaxis using DIAS software is found in Table 3. When placed in a chemoattractant gradient, iqgA− cells have a speed slightly, but reproducibly, higher than wild-type cells and exhibit more lateral pseudopodia than wild-type cells. In contrast, iqgB− cells exhibit no change in speed but have a reduced directionality due to a reduced ability to suppress lateral pseudopodia, similar to mhcA− cells (Table 3; Wessels et al., 1988 ). iqgC− cells exhibit chemotaxis parameters similar to those of wild-type cells. Conversely, the iqgA−/B− cells show very decreased speed, directionality, and cell polarity. These cells do not aggregate, consistent with their chemotaxis defects (data not shown). iqgB−/C− cells also have a 50% reduction in speed, although the reduction in directionality is less than that for iqgA−/B− cells. The iqgA−/C− cells exhibit good directionality and speeds similar to those of wild-type cells, although they are less polarized than wild-type cells and have an increased orientation change. Our findings suggest that DdIQGAP2 is important for polarity and is required with DdIQGAP1 or DdIQGAP3 for functional chemotaxis. DdIQGAP1 and DdIQGAP3 may have overlapping genetic functions during chemotaxis and in controlling cell polarity.
To better understand the basis for the chemotaxis defects of the IQGAP null strains, we examined chemoattractant-mediated F-actin polymerization and MyoII assembly (Figure 6, A and B, respectively). The analysis indicates that, compared with wild-type cells, iqgA− cells exhibit a slight reduction in the first F-actin peak and a small increase in the second peak. In iqgA− cells, the peak of assembled MyoII is reduced considerably compared with that of wild-type cells. In contrast, iqgB− cells show a small increase in basal F-actin levels, and a substantial increase in both F-actin peaks with a very small decline between the first and the second peak. Assembled MyoII is elevated in unstimulated iqgB− cells by ~30% and exhibits a much larger increase upon stimulation than that observed in wild-type cells. This chemoattractant-stimulated increase in assembled MyoII is elevated in iqgB− cells compared with that of wild-type cells even after normalization of the basal level of assembled MyoII.
iqgA−/B− cells show a pattern similar to that of iqgB− cells, suggesting that the loss of DdIQGAP2 is the major determinant of the iqgA−/B−cell phenotype. iqgC− has an F-actin profile similar to that of wild-type cells. In iqgA−/C− cells, the second F-actin peak is reduced compared with that in iqgA− cells, and iqgB−/C− cells have an F-actin response that is higher than that of wild-type cells but lower than that of iqgB− cells. iqgC− and iqgA−/C− have MyoII responses that are reduced compared with those of wild-type cells, whereas iqgB−/C− cells exhibit a response similar to that of iqgB− cells. The analyses of the single and double knockout strains demonstrate that the loss of IQGAP2 results in a considerable increase in the F-actin response and an enhanced MyoII response. Loss of IQGAP1 causes a moderate increase in the responses, whereas the loss of IQGAP3 suppresses the responses. These data suggest that IQGAP2 behaves as a negative regulator and IQGAP3 is a positive modulator of the F-actin and MyoII responses.
As we showed previously, loss of both ctxI and ctxII, but not either protein alone, leads to an increased number of turns but only a small effect on directionality, and no effect on cell polarity or the speed of chemotaxis (Figure 5; Table 3; Jeon et al., 2007 ). Loss of both proteins results in about a twofold increase in the second peak of F-actin polymerization and a modest reduction in the levels of chemoattractant-mediated MyoII assembly (Figure 5B). In contrast, loss of ctxIII causes a 50% reduction of speed and a decrease in directionality, but F-actin polymerization, MyoII assembly, and distribution of F-actin in phalloidin-stained cells are indistinguishable from those of wild-type cells. These findings suggest that ctxIII is an important regulator of chemotaxis but this regulation does not result from mediating the level or timing of the F-actin and MyoII responses (Figure 6, A and B; Table 3; data not shown).
To examine the spatial organization of F-actin in unstimulated, cAMP-responsive cells, we stained cells with FITC-phalloidin. Figure 6C illustrates that, like wild-type cells, iqgC− cells have a dominant F-actin–enriched pseudopod and a moderate level of filopodia (Figure 6C). iqgA− cells exhibit multiple thin, pseudopodia-like projects with F-actin localized at the ends of the projections. iqgB− cells have multiple, F-actin–enriched small projections, which may be filopodia. iqgA−/B−and iqgB−/C− cells are round and have multiple, small F-actin–containing projections, some of which appear similar to be micropinosomes, structures that are not normally observed in developed cells. iqgA−/C− cells have a combination of broad F-actin–containing, lamellipodia-like domains and enhanced, F-actin–containing filopodia-like spikes. These F-actin localization studies suggest that the three different IQGAPs play somewhat distinct roles in controlling the F-actin cytoskeleton.
Because of the abnormal patterns of F-actin polymerization and MyoII assembly, we studied the kinetics and level of Ras and PI3K activation in cells lacking different combinations of the three DdIQGAPs or ctx proteins. Figure 7, A and B, and Table 1 show that Ras activation is unaltered in two single iqgap null strains, a double mutant, and a double ctx null strain. When we examined PI3K activity indirectly by quantifying the level and kinetics of Akt/PKB activation, we found that the single iqgap and ctx null strains exhibited a normal level and kinetics of activation (Figure 7, C and D; Table 2; data not shown). However, cells lacking IQGAP2 in combination with either IQGAP1 or IQGAP3 (iqgA−/B− and igqA−/C− cells) exhibit highly elevated and extended PKB activation profiles. Cells lacking IQGAP2 and IQGAP3 (igqA−/C− cells) have a slightly extended activation profile but with a lower level of PKB activation than wild-type cells. Cells lacking IQGAP2 in combination with IQGAP1 or IQGAP3 also exhibit the highest basal (unstimulated cells) levels of PKB activity. As shown in Figure 7, C and D, and Table 2, ctxA−/B− cells display greatly extended kinetics of PKB activation but no increase in the peak level of activity.
To examine the possible basis for the extended activation of PKB in igqA−/B− cells, we studied the kinetics of PI(3,4,5)P3 levels by visualizing the subcellular localization of GFP-PHPhdA. We found, as with mhcA− cells, igqA−/B− cells exhibit a broad second peak that overlaps with the second peak of F-actin polymerization (Figure 8A). We then studied the kinetics of PI3K localization to the cell cortex in response to chemoattractant stimulation. As described previously (Funamoto et al., 2002 ), in wild-type cells, PI3K is predominantly cytosolic and transiently localizes to the plasma membrane in response to chemoattractant stimulation with a peak at ~6 s followed by a slow delocalization from the cortex as shown either by time-lapse examination of GFP-PI3K fluorescence at the cell cortex or by examining the change in GFP-PI3K found in the Triton X-100–insoluble fraction, a method that provides a more quantitative analysis of PI3K protein levels in the cortex (Figure 8, B and D). In contrast, we find that iqgA−/B− cells exhibit a highly elevated basal (unstimulated cells) level of PI3K at the cell cortex in both vegetative and aggregation-competent cells as indicated by level of GFP-PI3K fluorescence at the cell cortex and the amount of GFP-PI3K associated with the Triton X-100–insoluble fraction (Figure 8, C and D). On chemoattractant stimulation, iqgA−/B− cells still exhibit an increase in cortically associated GFP-PI3K with a peak at ~6 s. However, the cortical level of PI3K only decreases slightly before increasing again with a second peak at 15–20 s, after which cortical levels decrease to basal levels. The timing of the second peak coincides with the second broad peak of PI(3,4,5)P3 accumulation at the plasma membrane in these cells.
Rac1 and Cdc42 interact with mammalian IQGAPs. In Dictyostelium, which lacks Cdc42, only Rac1 has been found in the IQGAP-containing complexes that have been studied. An ~24-amino acid residue region (1054–1077) within the conserved RasGAP-related domain is required for Cdc42-GTP interaction with human IQGAP1 (Mataraza et al., 2003 ). Sequence comparison determined that this region has a high degree of conservation between DdIQGAP1, DdIQGAP2, and human IQGAP1 (data not shown). To examine the role of Rac1-GTP interaction with IQGAP function in vivo, we created two point mutations in two conserved, charged residues, in each IQGAP in which the Lys or Arg residues are changed to Glu, IQGAP1K273E, K282E and IQGAP2R308E, K317E, and found that the DdIQGAP1 carrying these mutations was no longer able to bind GST-Rac1A-GTP in a pulldown assay (Supplemental Figure 5). We then investigated the effect of Rac1 binding to the IQGAPs on PKB regulation. Expression of wild-type IQGAP2, but not the mutated IQGAP2R308E, K317E, in iqgA−/iqgB− cells does not complement the iqgA−/iqgB− cell chemotaxis defects or the extended PKB activation (Tables 2 and and3;3; Figure 7E; data not shown), suggesting that Rac1-GTP interaction is required for IQGAP function.
The mechanisms that spatially and temporally restrict the signaling pathways regulating pseudopod formation are not well understood. Localized activation of RasG and RasC causes the localized activation of PI3K, TORC2, and other effectors that play major roles in forming a functional pseudopod (see Introduction). Positive feedback loops involving F-actin amplify the responses, stabilizing the leading edge, whereas negative regulators, including RasGAPs and PTEN, help restrict these pathways in space and time. The spatial distribution of PTEN and MyoII in an increasing anterior-to-posterior gradient suggests other regulatory events are important in restricting leading-edge pathways.
We examined the role of two different cytoskeletal complexes that help mediate cortical integrity by cross-linking F-actin in controlling leading-edge pathways known to promote F-actin polymerization and cell polarity. We demonstrate that MyoII plays a role in turning off chemoattractant-stimulated Ras activity and the absence of MyoII leads to extended activation of Ras-GTP levels, to the downstream increased and extended activation of PI3K and Akt/PKB, and to F-actin polymerization. We suggest that the defects in the mhcA− cells' Ras/PI3K and probably also Ras/TORC2 signaling pathways constitute an underlying factor in the previously described inability of these cells to inhibit lateral pseudopod formation (Wessels et al., 1988 ). We attempted to examine the effect of deleting MyoII in RasC and RasG activation separately by expressing individually epitope-tagged versions of RasC and RasG in wild-type and mhcA− cells. Although we could overexpress tagged RasC in both strains, we could only overexpress RasG in wild-type cells. There was an insignificant extension of RasC activation, but as we were unable to overexpress RasG, we cannot determine if the effects observed are the result of extended activation of RasG and/or other Ras proteins.
Cells lacking MyoII exhibit an elevated basal level of Ras-GTP and a very extended activation profile. mhcA− cells expressing the MyoII uncoupler mutant MyoIIS456L, which has a 10-fold slower unloaded actin sliding velocity due to a one-quarter productive working stroke and threefold slower ADP release rate (under no-load conditions; Murphy et al., 2001a ), also exhibit extended Ras activity but not to the level of mhcA− cells. In our previous studies of S456L MyoII, we found that mechanical stress can partially rescue this motor. Therefore, the cell cortex of developed chemotaxing cells is likely to be weakly mechanically stressed, allowing this motor to provide partial function in restricting Ras and PI3K activity (Girard et al., 2006 ; Reichl et al., 2008 , Ren et al., 2009 ). As described previously, myoII null cells cannot restrict lateral pseudopod formation (Wessels et al., 1988 ). Our findings suggest that the lateral pseudopodia occur at sites of Ras/PI3K activation that would normally be inhibited from occurring in wild-type cells. In mhcA− cells, areas of the cortex lacking MyoII are permissive for Ras/PI3K activation, suggesting why in highly polarized cells a very localized MyoII cortex is less responsive to the formation of new pseudopodia along the lateral sides than less polarized cells, even though receptors/Gαβγ are present and become activated. As there is unregulated and extended activation of RasG and PI3K (e.g., nfA− and pten− cells, respectively; Funamoto et al., 2002 ; Iijima and Devreotes, 2002 ), we suggest that these lateral pseudopodia result from an inability to restrict Ras activity, leading to activation of a PI3K/F-actin pathway. Once activated, the previously described positive feedback loops involving F-actin in which PI3K and RacGEFs are recruited to these sites would amplify the response and cause the robust formation of a new pseudopod. We suggest that in wild-type cells, as a leading edge forms and the cell polarizes, MyoII is disassembled at this site through the activation of Rap1 and the recruitment of MHCKs (Jeon et al., 2007 ), whereas MyoII assembly and cortical localization are promoted along the lateral sides and posterior of the cell. Our findings suggest that one role for MyoII is to inhibit at least some leading-edge functions, which would help restrict leading-edge functions to the leading edge, whereas F-actin assembly at the new leading edge is involved in a positive feedback loop with PI3K (Sasaki et al., 2004 , 2007 ; Charest and Firtel, 2006 ). We suggest that this combination of positive and negative feedback loops between leading edge and lateral/posterior pathways is part of the mechanism that enhances cell polarity. As the cell starts to polarize in response to a directional chemoattractant signal, the localization of MyoII to the posterior helps restrict the response to the leading edge by blocking Ras activation. Whether this “inhibition” afforded by MyoII takes place by physically blocking access to the receptor/Gα/Gβγ (which are activated in proportion to the concentration of the chemoattractant) or through another mechanism is not known. Inhibition of PI3K with LY294002 suppresses cell speed and pseudopod extension in mhcA− cells as it does for wild-type cells, which is consistent with this model (this article; Takeda et al., 2007 , Bosgraaf et al., 2008 ). However, loss of MyoII also results in major changes in the level and organization of traction forces, presumably the result of loss of both F-actin cross-linking and motor function that plays a major role in the inability of mhcA− cells to suppress lateral pseudopodia (del Alamo et al., 2007 ; Lombardi et al., 2007 ; Meili et al., 2010 ).
We found that IQGAPs and ctxs, which bind and cross-link F-actin, also play an important role in controlling F-actin polymerization and MyoII assembly, and they are important in controlling the kinetics of the PI3K pathway. In agreement with previous studies, we observed that both IQGAP1 and IQGAP2 bind to one or more ctx proteins and Rac1. The association of ctx presumably provides the F-actin–binding activity found in the mammalian IQGAPs. We observe, however, that different ctx proteins interact with the different IQGAPs, which may indicate that individual IQGAP/ctx complexes have different functions. We determined that the individual mutant strains have different chemotaxis phenotypes, suggesting that they are not fully redundant. Mutating two conserved residues in the IQGAP1 RasGAP-related Rac1-GTP in the binding domain abrogates Rac1-GTP binding and the inability of the protein to complement the severe phenotypes of iqgA−/B− cells. This finding suggests that Rac-GTP binding is required for IQGAP function, consistent with models for the involvement of Rac1/Cdc42-GTP binding in mammalian IQGAP function.
We characterized the chemotaxis phenotypes of IGAPA, B, and C and ctxI, II, and III and different double knockout combinations. iqgB− cells exhibit moderate chemotaxis defects as suggested previously: the cells are less polarized than wild-type cells and have a partial loss of lateral pseudopod inhibition. The phenotype of iqgA− cells is mild and iqgC− cells are even more polarized than wild-type cells and move slightly faster. iqgA−/B− cells exhibit severe polarity and motility defects, whereas iqgA−/C− cells have mild defects, although both strains exhibit multiple F-actin protrusions along the cortex of pulsed cells. Both iqgB− and iqgA−/B− display similar levels of increased basal and chemoattractant-stimulated F-actin and assembled MyoII, even though only the double knockout shows the severe chemotaxis phenotype.
In conclusion, we have identified key roles for components of the cortical system in controlling the levels and extent of leading-edge pathways. MyoII is required for restricting the activation of Ras along the lateral sides and back of the cell and, we suggest, suppresses the formation of lateral pseudopodia by inhibiting one of the earliest steps in the pathway. We cannot, however, exclude the possibility that loss of MyoII may also directly affect PI3K activation independent of an effect on Ras. The IQGAP/ctx/Rac1 complex plays a different role and modulates F-actin and MyoII.
The data presented here demonstrate that MyoII and ctx impact the signaling pathways of chemotaxis and provide evidence that they inhibit “front-end” signaling. Mechanistically, how might this cross-talk occur? First and foremost, the actin cytoskeletal network of a Dictyostelium cell is largely elastic with a mechanical phase angle of ~15° (where 0° is a solid and 90° is a liquid), implying that the cortex is largely solid-like (elastic; Girard et al., 2004 ; Reichl et al., 2008 ). Because signals can propagate through an elastic network over long distances and on much faster time scales than can occur by diffusion alone (Wang and Suo, 2005 ), an attractive idea is that this signal propagation occurs directly through the elastic network, which has properties defined in part by MyoII and ctx (Girard et al., 2004 ; Reichl et al., 2008 ). Both MyoII and ctx contribute ~20–30% to cortical viscoelasticity and to cortical tension. Particularly in dividing cells, mechanical stresses in the elastic network can also direct the accumulation of MyoII and ctxI, which work cooperatively to sense and accumulate locally in response to applied mechanical stress (Ren et al., 2009 ). Interestingly, IQGAP2, but not IQGAP1, is required for mechanosensitive localization of ctx and MyoII (Kee and Robinson, unpublished data). Thus, the IQGAP proteins clearly define different populations of ctx with different mechanosensitive properties. IQGAP1 may define a ctx complex that is more dependent on chemical signal transduction, whereas IQGAP2 defines a ctx pool that is more responsive to mechanical stress inputs. By then coupling distinct signaling elements to the different IQGAPs, these downstream signaling cascades would become part of a mechano-chemical signal transduction system. As IQGAP2 plays a greater role in chemotactic signaling, it is possible that at least part of this signal relay between the two ends of the polarized cell occurs from mechanical stresses propagated directly through the elastic cytoskeleton.
Internally generated mechanical stresses may help drive accumulation of MyoII and ctx at the rear of a chemotaxing cell. The cell may initiate chemotaxis signaling using chemical inputs that trigger new F-actin assembly at the front of the cell. As these newly nucleated filaments push on the front of the cell, mechanical stresses are then sensed by the MyoII-ctx-IQGAP2 system. Of course, this MyoII-ctx-IQGAP2 system would not accumulate at the front of the cell because of the local activation of myosin heavy-chain kinases. However, the mechanosensitive accumulation of these complexes along the lateral and rear cortex would serve as a positive feedback on the signal, and, because they also stiffen the network, these proteins would suppress pseudopodial protrusion along these cortical domains. Consistently, softening of the elastic cortex twofold by depletion of another cross-linker (dynacortin), which enriches at the front and in pseudopodia, leads to a loss of cell polarity and more pseudopodia extending over a broader distribution of angles (Kabacoff et al., 2007 ; Reichl et al., 2008 ). The MyoII-ctx-IQGAP2 system may then convey an inhibitory signal on PI3K, perhaps by providing a positive feedback on PTEN accumulation in the cortex (Pramanik et al., 2009 ).
Another aspect of how MyoII contributes to cortical dynamics is by modulating the active (superdiffusive) nature of the cortex (Girard et al., 2006 ). These superdiffusive properties were originally detected by tracking the fluctuations of surface-attached particles, which then report on the underlying cortical dynamics. mhcA− cell cortices are considerably depleted of these superdiffusive activities; however, depletion of dynacortin in a myoII null cell substantially restores them. Observations such as this suggest that MyoII antagonizes some F-actin–associated proteins, perhaps by “kicking” or “stirring” the actin filaments, which may loosen the F-actin cross-linkers and/or associated proteins (this feature is in contrast to the cooperative interactions between MyoII and ctx). Thus, MyoII potentiates, rather than drives, the superdiffusive activities much like traffic signals potentiate the flow of traffic through a city. This antagonistic relationship between MyoII and some actin-associated proteins such as dynacortin may help explain the enhanced actin assembly in the mhcA− cells. Through a process of elimination, although not directly proven, microtubules are proposed to be the source of the superdiffusive activities (Girard et al., 2006 ). Microtubules might make contact with the cortex and help modulate the chemotactic signaling. At least one protein, tsunami, is associated with the microtubule network and contributes to cell polarization during chemotaxis (Tang et al., 2008 ). Furthermore, if signaling complexes are coupled to F-actin–associated proteins that interact antagonistically with MyoII, then MyoII enrichment in part of the cell would help to shut off those signals. Clearly, Ras appears to be directly or indirectly inhibited through MyoII-mediated mechanochemical signal transduction, suggesting such a mechanism. In summary, the integration of mechanical with signaling observations suggests several novel hypotheses for how signal propagation through the cell may coordinate front and rear modules that reinforce cell polarity to drive efficient chemotaxis.
We thank Pascale Charest and Alessia Para for critically reading the manuscript and all members of the Firtel laboratory for helpful suggestions. This work was supported by the following grants: National Institutes of Health (NIH) R01 GM037830 to R.A.F. and NIH Grant GM066817 and American Cancer Society Grant RSG CCG-114122 to D.N.R.
This article was published online ahead of print in MoBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E10-01-0009) on April 7, 2010.