Index Introduction Materials and Methods Cell culture and antibodies Deep-etch immunoreplica EM Biochemical subcellular fractionation Results and Discussion Acknowledgments References

Introduction

The cell receives extracellular signals, including chemokines, growth factors, and the extracellular matrix, at the plasma membrane through receptors. On receipt of the signals, the cytoskeleton near the plasma membrane (membrane cytoskeleton) is reorganized in a signal-specific manner. The signal-dependent reorganizations of the cytoskeleton as well as adhesions play critical roles in embryogenesis, tissue development, and tumorigenesis. Extensive biochemical studies have revealed that the Rho family GTPases, especially Rac1, are key regulators in the reorganization of the membrane cytoskeleton (Jaffe and Hall, 2005; Watanabe et al., 2005). The Rho family GTPases function as a molecular switch by cycling between inactive GDP- and active GTP-bound forms. The activated GTPases at the plasma membrane transduce the signals to their effectors via direct interactions.

Genetic and biochemical approaches have identified a number of Rac1 effectors, including p21-activated kinase (PAK), Wiskott-Aldrich syndrome protein (WASP)-family verprolin-homologous protein (WAVE), IQGAP, and specifically Rac1-associated protein (Sra-1) (Jaffe and Hall, 2005; Kaibuchi et al., 1999). Each effector contributes to the cytoskeletal reorganization downstream of Rac1 by distinct mechanisms. IQGAP1 and Sra-1 interact directly with actin filaments in vitro and accumulate at the periphery in motile cells (Briggs and Sacks, 2003; Kobayashi et al., 1998; Mateer et al., 2003; Noritake et al., 2005). Sra-1 leads a rapid actin polymerization together with WAVEs which activate actin-related proteins 2/3 (Arp2/3) complex (Eden et al., 2002; Steffen et al., 2004; Takenawa and Suetsugu, 2007), whereas IQGAP1 directly cross-links actin filaments. Their regulation of actin filaments downstream of Rac1 is necessary for cellular morphogenesis (Kawano et al., 2005; Mataraza et al., 2003; Takenawa and Suetsugu, 2007; Wang et al., 2007; Watanabe et al., 2004). In addition to linkages of actin filaments, there is accumulating evidence of linkages of some Rac1 effectors with microtubules (MTs). IQGAP1 connects MTs to cortical actin filaments through plus-end-tracking proteins (+TIPs), such as cytoplasmic linker protein-170 (CLIP-170) and adenomatous polyposis coli, under the control of Rac1 (Fukata et al., 2002; Watanabe et al., 2004). In hippocampal neurons, a Sra-1/WAVE1 complex is transported along MTs through collapsin response mediator protein-2 (CRMP-2) and kinesin-1 (Kawano et al., 2005; Kimura et al., 2005). PAKs, serine/threonine protein kinases, whose activity is stimulated by the binding of active Rac1, modulate the activity of some MT-associated proteins through phosphorylation (Daub et al., 2001). Thus, Rac1 appears to regulate organization of the cytoskeleton, especially actin filaments and MTs. However, it remains unknown where activated Rac1 affects cytoskeletal architecture through its individual effector at molecular levels.

In this study, we dissected the localization of Rac1 and its effectors, IQGAP1 and Sra-1, beneath the substratum-facing surface by deep-etch immunoreplica electron microscopy (EM). These effectors showed similar, but distinct, localizations with Rac1, suggesting that Rac1 can regulate cytoskeletal rearrangement through IQGAP1 and Sra-1 in a specific manner.

Materials and Methods

Cell culture and antibodies

Vero cells were maintained in Dulbecco’s modified Eagle medium supplemented with 10% fetal bovine serum at 37°C in an air–5% CO₂ atmosphere at constant humidity. Rabbit polyclonal anti-IQGAP1 and anti-Sra-1 antibodies were described previously (Kawano et al., 2005; Wang et al., 2007). Mouse monoclonal anti-Rac1 antibody was obtained from Upstate (Billerica, MA, USA). We purchased 5- and 10-nm gold-conjugated secondary antibodies from GE Healthcare UK (Little Chalfont, Buckinghamshire, England).

Deep-etch immunoreplica EM

The cytoplasmic surface of the substratum-facing plasma membrane was visualized by deep-etch immunoreplica EM essentially according to previously described methods (Enomoto et al., 2005; Morone et al., 2006). Vero cells were cultured on glass coverslips (2.5 mm²). Immediately after being unroofed by brief sonication (Sonifier S-150D; Branson, Danbury, CT, USA), the isolated substratum-facing membrane was fixed for 30 min in 0.1% glutaraldehyde/4% paraformaldehyde in buffer A (70 mM KCl, 5 mM MgCl₂, 3 mM EGTA, 30 mM HEPES buffer, pH 7.4). After being washed three times with buffer A, the samples were quenched and blocked with buffer B (100 mM NaCl, 30 mM HEPES, 2 mM CaCl₂) containing 1% bovine serum albumin (BSA; Sigma, St. Louis, MO, USA) and then labeled for 1 hr with primary antibody and with gold-conjugated secondary antibody in buffer B containing 1% BSA. After being washed three times with the buffer B (5 min each), specimens were further incubated with gold-conjugated secondary antibody in buffer B containing 1% BSA. In the case of double labeling, the mixture of different antibodies was used. The samples were then washed three times with buffer B (5 min each) and fixed again for 10 min with 1% glutaraldehyde in buffer B. After being washed twice with distilled water, the samples were quickly frozen with liquid helium, using a rapid-freezing device (Eiko, Tokyo, Japan) (Usukura and Yamada, 1987).

The frozen sample was placed in liquid nitrogen and then brought into the freeze-etch device (FR9000; Hitachi, Tokyo, Japan). After the ice covering was shaved off the surface of the sample with a prechilled glass knife in the chamber at –140°C under 5×10^–6 Pa, the exposed surfaces were etched (slightly dried) for 10 min at –90°C, and were then rotary shadowed with platinum and carbon. The platinum-carbon replica was removed from the glass coverslip in 1% hydrofluoric acid in distilled water. After the replicas were washed with distilled water containing detergent (Photo-Flo 200; Kodak, Tokyo, Japan), they were mounted on 200-mesh copper grids (Ted Pella, Redding, CA, USA) coated with polyvinyl formvar (Nisshin EM, Tokyo, Japan) and were observed with a transmission electron microscope (1200EX; JEOL, Tokyo, Japan; or H7100; Hitachi).

The labeled cytoplasmic surface images were used for quantification. The localization of Rac1 and its effectors was classified into four groups by the association of each label with visible structures: “membrane”, “actin filament”, “microtubule”, and “actin filament and microtubule”. Note that contents of visible actin filaments and microtubules differed in each cell. More than ten cells from three independent experiments were counted.

Biochemical subcellular fractionation

Subcellular fractionation of Vero cells was performed as described previously (Kawano et al., 1999). The precipitates resulted from centrifugation at 100,000 g were resuspended with the buffer containing 1% Triton, and additionally centrifuged at 100,000 g for 30 min. The resulting precipitates were used as the Triton insoluble fraction. The fractions were blotted with indicated antibodies.

Results and Discussion

We employed deep-etch immunoreplica EM to observe the cytoplasmic surface of the substratum-facing plasma membrane, the membrane cytoskeleton, and the locations of the cytoskeletal regulators. In the immunoreplica EM images of Vero cells, the membrane cytoskeleton primarily consisted of actin filaments and its associated proteins, as described previously (Morone et al., 2006). In addition, we could often observe MTs, a finding consistent with targeting of MTs to the plasma membrane on a nanoscale (Fig. 1) (Krylyshkina et al., 2003). Actin filaments and MTs were easily identified by their diameters and shapes (Fig. 1). Labeling of anti-actin or anti-α-tubulin antibody supported our observations (data not shown). Under these conditions, the isolated ventral membrane rarely had intermediate filaments. We could also distinguish, without labeling, characteristic membrane structures such as clathrin-coated pits (CCPs) and caveolae from their apparent morphologies. The gold conjugated antibodies are evident as white dots surrounded by fuzzy halos (magnified images in each figure), because we showed replica images with inverted contrast for easy understanding (Figs. 1–3, 5, 6).

View Details

Fig. 1. Labeling of Rac1 beneath the substratum-facing surface. A low magnification of the image is shown in A (left panel). Right panels are enlarged images from the left panel image (B–D) or from other images (E–G). Rac1 localized at microtubules (MTs), actin filaments, and the plasma membrane. Rac1 appeared to form a dimer at the plasma membrane. Green arrows indicate Rac1 labels. CCP, clathrin-coated pit. Bar in A, 200 nm.

Rac1 appears to be activated at the plasma membrane and interacts with its effectors to control rearrangement of the cytoskeleton (Fukata and Kaibuchi, 2001; Jaffe and Hall, 2005). We first labeled the cytoplasmic surface of the plasma membrane with anti-Rac1 antibody. Rac1 localized at the plasma membrane, which was presumably the activated form (Fig. 1). Besides being detected at the membrane, Rac1 was also detected at the ends of actin filaments (Fig. 1C–E) and often associated with lattices of MTs (Fig. 1B, F, G, Table I). Our findings are consistent with findings of biochemical or light microscopic studies in which activated Rac1 at the plasma membrane induced actin polymerization and controlled the array of actin filaments and MTs (Jaffe and Hall, 2005; Watanabe et al., 2005).

To address where Rac1 regulates cytoskeletal arrangements, we next determined the location of IQGAP1 and Sra-1, effectors of Rac1. Although IQGAP1 cross-links actin filaments directly and connects them to MTs through +TIPs downstream of activated Rac1 (Briggs and Sacks, 2003; Mateer et al., 2003; Noritake et al., 2005), detailed ultrastructural analyses have never been performed in vivo before now. Beneath the substratum-facing surface, IQGAP1 predominantly localized beside actin filaments (Fig. 2). IQGAP1 was rarely detected at stress fibers that were composed of highly bundled actin filaments and lay mainly in the cytoplasm (data not shown). IQGAP1 sometimes localized at the intersections of MT ends and actin filaments, and beside MTs (Fig. 2E, G, Table I). Furthermore, IQGAP1 appeared to form a dimer and oligomer in vivo, judging from the number of labels at the same site. Because IQGAP1 has one calponin homology domain responsible for binding to actin filaments, the cross-linking activity of IQGAP1 may require its dimerization and oligomerization in vivo (Bashour et al., 1997; Fukata et al., 1997; Ren et al., 2005). However, it is possible that anti-IQGAP1 polyclonal antibodies and/or secondary antibodies recognized the different epitopes simultaneously against the single IQGAP1 molecule. IQGAP1 was recently shown to induce polymerization of branched actin filaments together with N-WASP and Arp2/3 (Bensenor et al., 2007; Le Clainche et al., 2007). In fact, IQGAP1 on actin filaments formed a complex with unidentified molecules (Fig. 2). These populations might represent actin polymerization machinery.

View Details

Fig. 2. Labeling of IQGAP1 beneath the substratum-facing surface. A low magnification of the image is shown in A (left panel). Right panels are enlarged images from the left panel image (B–D) or from other images (E–G). IQGAP1 appeared to form a dimer or oligomer along the membrane-associated actin filaments (B, F). IQGAP1 labels were often observed at the sides, ends, and intersections of microtubules (MTs) (D, E, G). Green arrows indicate IQGAP1 labels. CCP, clathrin-coated pit. Bar in A, 500 nm.

On the other hand, Sra-1 showed a characteristic distribution near the substratum-facing surface. Sra-1 was detected on the plasma membrane, actin filaments, and MTs, forming an oligomer (Fig. 3B–G, Table I). Sra-1 also localized on MTs beneath the plasma membrane. Sra-1 forms a complex with WAVEs to induce lamellipodia under the control of Rac1 (Takenawa and Suetsugu, 2007). Sra-1 on actin filaments might represent such a complex, but the nature of Sra-1 oligomers remains to be clarified. In our images of fibroblasts, Sra-1 was located on MTs, suggesting that Sra-1 is transported with the WAVE complex on MTs in nonneuronal cells.

View Details

Fig. 3. Labeling of specifically Rac1-associated protein (Sra-1) beneath the substratum-facing surface. A low magnification of the image is shown in A (left panel). Right panels are enlarged images from the left panel image (B–E) or from other images (F, G). Sra-1 was frequently detected along the membrane-associated actin filaments and at the plasma membrane. Sra-1 labelings were also observed at microtubules (MTs) (E). Green arrows indicate Sra-1 labels. CCP, clathrin-coated pit. Bar in A, 200 nm.

We carried out subcellular fractionation by the biochemical method. Rac1, IQGAP1, and Sra-1 were detected in the membrane fraction and the cytoplasmic fraction. Rac1 and Sra-1 were exclusively fractionated into the Triton insoluble fraction from the membrane fraction, but IQGAP1 was not (Fig. 4), suggesting the difference of spatial association between Rac1 and its effectors at the plasma membrane.

View Details

Fig. 4. Biochemical fractionation of Rac1 and its effectors. Vero cells were separated into cytoplasmic and membrane fractions, and Triton soluble/insoluble fraction of membrane fraction (see Materials and Methods). The fractions were blotted with indicated antibodies. N-cadherin, actin, and RhoGDI were used as a marker of membrane fraction, cytoskeleton, and cytoplasmic fraction, respectively. The loaded ratio is indicated at the bottom, except for actin and RhoGDI in cytoplasmic fraction, for which only a quarter amount was loaded. These results are representative of three independent experiments.

We further addressed the spatial linkage between Rac1 and its effectors on the cytoplasmic surface of the plasma membrane by double-labeling, using different sizes of gold-conjugated secondary antibodies. Most IQGAP1 labels did not colocalize with Rac1 labels, indicating they are free of Rac1 (Fig. 5). However, some IQGAP1 labels localized together with Rac1 at actin filaments and MTs (Fig. 5B–G). Because IQGAP1 is also an effector of Cdc42, another Rho family GTPase, and binds more to activated Cdc42 than to Rac1 (Kuroda et al., 1996), IQGAP1 lacking Rac1 labels might exist with Cdc42 or might localize at the membrane by different mechanisms. Our EM images showed that IQGAP1 stayed at the ends and sides of MTs and at the intersections of MTs with Rac1. IQGAP1 may comprehensively regulate polarized MT array by affecting plus ends as well as lattices of MTs at the plasma membrane. In the case of Sra-1 and Rac1, Sra-1 sometimes colocalized with Rac1 at actin filaments and on the plasma membrane, but rarely did so on MTs (Fig. 6B–G). It is possible that Sra-1 on MTs, which lacks Rac1, is transported along MTs, presumably through kinesin-1, to specific areas for regulating actin filaments. Sra-1 with activated Rac1 on the membrane or actin filaments would then cooperate with WAVEs to induce actin polymerization. Because membrane-associated Rac1 is thought to be the active form, Rac1 without IQGAP1 labels or Sra-1 labels might associate with other effectors such as PAKs (Fig. 1). Taken together with the biochemical fractionation (Fig. 4), Rac1 spatially regulates the cytoskeleton near the plasma membrane through a distinct effector for specific functions.

View Details

Fig. 5. Double-labeling of Rac1 and IQGAP1 beneath the substratum-facing surface. IQGAP1 and Rac1 were labeled with 10-nm and 5-nm gold-conjugated secondary antibodies, respectively. A low magnification of the image is shown in A (left panel). Right panels are enlarged images from the left panel image (B–D) and from other images (E–G). IQGAP1 together with Rac1 often associated with microtubules (MTs) (B, D) and actin filaments (C, E). Green and yellow arrows indicate IQGAP1 (10 nm) and Rac1 (5 nm) labels, respectively. Bar in A, 200 nm.

View Details

Fig. 6. Double-labeling of Rac1 and Sra-1 beneath the substratum-facing surface. Sra-1 and Rac1 were labeled with 10-nm and 5-nm gold-conjugated secondary antibodies, respectively. A low magnification of the image is shown in A (left panel). Right panels are enlarged images from the left panel image (B–D) and from other images (E–G). Sra-1 together with Rac1 was often associated with actin filaments (B, D) and located at the plasma membrane (C). Most Sra-1 labelings on microtubules (MTs) lacked Rac1 labels (E, F, G). Green and yellow arrows indicate Sra-1 (10 nm) and Rac1 (5 nm) labels, respectively. Bar in A, 200 nm.

In our study, we showed the localization of Rac1 and its effectors, IQGAP1 and Sra-1, beneath the substratum-facing surface. Our morphological approaches can yield spatial information on a nanoscale that is not possible with biochemical approaches. Combining our morphological approaches with the RNA interference technique and light microscopy will permit investigation of how the Rho family GTPases regulate the cytoskeleton and adhesions in an effector-specific manner.

Acknowledgments

We wish to thank Dr. J. Heuser (Washington University) and Dr. N. Morone (National Center of Neurology and Psychiatry) for their helpful discussions and technical advice. We also thank Dr. E. Mekada (Osaka University) for providing Vero cells, and K. Murase and T. Ishii for their technical and secretarial assistance. This research was supported in part by the Special Coordination Funds for Promoting Science and Technology (Japan Science and Technology Agency), a Grant-in-Aid for Scientific Research (Japan Society for the Promotion of Science [JSPS]), the Human Frontier Science Program, a Grant-in-Aid for Creative Scientific Research (Ministry of Education, Culture, Sports, Science and Technology [MEXT]), a Grant-in-Aid for JSPS Fellows (JSPS), and the 21st Century Centre of Excellence Programme (MEXT).