If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* This work was supported in part by a European Community Marie Curie Fellowship (to R. D.), by funds from the Verband der Chemischen Industrie and the Bundesministerium für Bildung und Forschung (to D. F.), and by funds from the Deutsche Forschungsgemeinschaft (to L.-C. H. and L. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ These authors contributed equally to this work.
Rac1b was recently identified in malignant colorectal tumors as an alternative splice variant of Rac1 containing a 19-amino acid insertion next to the switch II region. The structures of Rac1b in the GDP- and the GppNHp-bound forms, determined at a resolution of 1.75 Å, reveal that the insertion induces an open switch I conformation and a highly mobile switch II. As a consequence, Rac1b has an accelerated GEF-independent GDP/GTP exchange and an impaired GTP hydrolysis, which is restored partially by GTPase-activating proteins. Interestingly, Rac1b is able to bind the GTPase-binding domain of PAK but not full-length PAK in a GTP-dependent manner, suggesting that the insertion does not completely abolish effector interaction. The presented study provides insights into the structural and biochemical mechanism of a self-activating GTPase.
The small GTPase Rac acts as a binary molecular switch that cycles between an inactive GDP-bound state and an active GTP-bound state in response to a variety of extracellular stimuli. The interconversion between both states is controlled by nucleotide exchange and GTP hydrolysis. The structures of several GTPases in either state revealed that the switching mechanism depends on the conformational change of two regions, termed switch I and switch II (
). After GTP hydrolysis, release of the cleaved γ-phosphate allows the switch regions to relax into the GDP conformation. Guanine nucleotide exchange factors (GEFs), stimulating the GDP/GTP exchange, bind independently of the nucleotide-bound state (
). We tried to address what influence this insertion might have on the structure and the biochemical properties of Rac1b in comparison with Rac1. Therefore, we solved the crystal structures of Rac1b in the GDP- and GppNHp-bound conformations at 1.75 Å resolution. Furthermore we investigated nucleotide binding and hydrolysis of Rac1b and studied its regulation by the RacGEF Tiam1 and p50GAP and its interaction with the downstream effector PAK. The Rac1b structures explain the drastic changes of the biochemical properties of Rac1b, namely a dramatic decrease in nucleotide affinity and GTP hydrolysis. The presented data identify Rac1b as a predominantly GTP-bound form of Rac1.
Plasmids—The pcDNA3-FLAG constructs of human Rac1b, Rac1, and the respective constitutive active Rac1(G12V) mutant were generated by PCR and cloned via BamHI and EcoRI restriction sites. Rac1, Rac1ΔC (1-184), Rac1b, and Rac1bΔC (1-201) were cloned in pGEX4T1, using BamHI and EcoRI restriction sites. The DH-PH domain of Tiam1 (1033-1404) was cloned into pGEX4T1 using BamHI and XhoI restriction sites. The coding region of Tiam1 contains an internal BamHI site that was removed for the cloning procedure. pGEX-PAK1-GBD was kindly provided by J. Collard (
Preparation of Recombinant Proteins—Rac1, Rac1ΔC, Rac1b, and Rac1bΔC, the catalytic domain of p50GAP (amino acids 198-439), the Cdc42/Rac-interacting binding domain of PAK (amino acids 57-141), full-length PAK, and the DH-PH domain of Tiam1 were produced as glutathione S-transferase (GST) fusion proteins in Escherichia coli. All of the proteins were purified as described previously for Rnd3 (
Crystallization and Data Collection—Crystals of truncated Rac1bΔC (1-184) in complex with GDP and GppNHp (nonhydrolyzable GTP analog) were grown at 20 °C using the hanging drop method by mixing 2 μl of a 0.5 mm solution of the Rac1b G domain in 20 mm Tris/HCl, pH 7.5, 2 mm MgCl2, 2 mm dithioerythritol, 100 μm GDP or GppNHp with 2 μl of reservoir solution consisting of 100 mm Hepes buffer, pH 7.5, 18-30% polyethylene glycol 3350, and 2-6% isopropanol. The crystals of both complexes belonged to space group P212121 (a = 51.55 Å, b = 78.67 Å, c = 96.88 Å). For data collection at 100 K, the crystals were transferred to a solution containing reservoir solution and 10% glycerol. A cryo-protected crystal was then suspended in a rayon loop (Hampton Research) and flash frozen in liquid nitrogen. X-ray diffraction data were collected on an ADSC Q4 CCD detector at the beam line ID14-1 at the European Synchrotron Radiation Facility and were processed using XDS (
) and REFMAC5 were used. The residue ranges that were included in the final model as well as the corresponding R-factors are listed in Table I. For all four molecules the two additional N-terminal Gly-Ser residues caused by the thrombin cleavage site could be observed and were included in the model.
Fluorescence Measurements—Long time fluorescence measurements were monitored on a LS50B PerkinElmer Life Sciences spectrofluorometer, and rapid kinetics were measured with a stopped flow apparatus (Applied Photophysics SX16MV) as described (
). The dissociation of the fluorescent nucleotide from Rac proteins (0.1 μm) was measured by the addition of 200-fold excess of nonfluorescent nucleotide in the absence and the presence of 5 μm Tiam1 DH-PH at 20 °C. The equilibrium dissociation constant (Kd) for the PAK-GBD interaction with Rac1b was determined as previously described for the Ras-Raf kinase interaction (
). The measurements were carried out using 0.2 μm mantGppNHp-bound GTPase, 40 μm GppNHp, and increasing concentrations of PAK-GBD at 25 °C for Rac1 and at 10 °C in the case of Rac1b because of its fast nucleotide dissociation rate. All of the measurements were carried out in 30 mm Tris/HCl, pH 7.5, 5 mm MgCl2, 10 mm Na2HPO4/NaH2PO4 pH 7.5, 5 mm dithioerythritol. The observed rate constants were evaluated using Grafit (Erithacus software).
GTPase Assay—The intrinsic and GAP-stimulated GTP hydrolysis reactions were measured by HPLC on a C18 reversed phase column using a mixture of 80 μm nucleotide-free GTPase and 70 μm GTP in the presence and the absence of 8 μm GAP at 25 °C in 30 mm Tris/HCl, pH 7.5, 5 mm dithioerythritol, 10 mm Na2HPO4/NaH2PO4, 5 mm MgCl2 as described (
). The interaction of full-length GST-PAK with the Rac proteins was examined under the same conditions using purified proteins. The beads were washed four times and subjected to SDS-PAGE (15% polyacrylamide). Bound Rac proteins were detected by Western blot using a monoclonal antibody against Rac (Upstate Biotechnologies, Inc.).
RESULTS AND DISCUSSION
Rapid GEF-independent Nucleotide Dissociation Reaction of Rac1b—To investigate the influence of the 19-amino acid insertion on the nucleotide binding affinity, we first determined kinetic constants for the association of fluorescently (methylanthraniloyl- or mant-) labeled nucleotides to nucleotide-free Rac1b protein. This allowed us to monitor nucleotide association kinetics at increasing protein concentrations. As shown in Fig. 1A, the formation of the binary complex is not affected by the 19-amino acid insertion. The association rate constants (kon) for the binding of mantGDP and mantGTP to Rac1b were obtained by linear fitting of the observed rate constants at the given protein concentrations. They are only marginally slower than those of Rac1 (Table II).
To determine the intrinsic and GEF-accelerated nucleotide dissociation rates, Rac1b and Rac1 were loaded with the respective fluorescently labeled guanine nucleotides. The displacement of the fluorescent nucleotides was initiated by the addition of excess amounts of nonfluorescent nucleotides in the presence and absence of the DH-PH domain of Tiam1 (a Rac-specific GEF). Drastic increases in the intrinsic dissociation of mantGDP (26-fold), mantGTP (27-fold), and mantGppNHp (250-fold) from Rac1b compared with the very slow dissociation rates of the respective nucleotides from Rac1 were observed (Fig. 1B and Table II). Accordingly, the calculated Kd for nucleotide binding revealed that the 19-residue insertion dramatically affects the overall affinity for GDP, GTP, and particularly the GTP analog GppNHp (Table II). The reason for the reduced affinity of GppNHp compared with that of GTP is the disrupted hydrogen bond of the GTP-β,γ-bridging imino group to the main chain NH group of the P-loop residue Ala13. A similar observation has been reported for Ras·mantGppNHp (
Moreover, in contrast to the slow nucleotide dissociation of Rac1, which could be 50-fold accelerated in the presence of the DH-PH domain of Tiam1, the fast intrinsic mantGDP dissociation of Rac1b could not be further increased by the DH-PH domain of Tiam1 (Fig. 1C and Table II).
These in vitro results show that Rac1b does not require any GEF to get activated. It rather activates itself by a very fast nucleotide dissociation and by the subsequent binding of the cellular abundant GTP. Despite the drastically increased nucleotide dissociation, Rac1b exhibits a nucleotide binding affinity in the low nanomolar range, which is still high enough to act as a GTP-binding protein in cells. However, because the DH-PH domain of Tiam1 is a weak exchange factor in vitro (
Impaired Intrinsic GTP Hydrolysis Reaction of Rac1b—A second crucial function of small GTPases is their slow intrinsic GTP hydrolysis reaction, which needs to be stimulated by GAPs to switch off downstream signaling. Therefore, we measured the GTP hydrolysis reaction of Rac1b in direct comparison with that of Rac1 using a HPLC-based technique. Interestingly, we found that the intrinsic GTP hydrolysis reaction of Rac1b (0.0035 min-1) was about 30-fold reduced compared with that of Rac1 (0.11 min-1) (Fig. 1D and Table II). In contrast to our results it has been previously published that Rac1 and Rac1b show the same GTPase activity (
). This discrepancy can be explained by the method this group employed. The filter binding assay seems to be inappropriate for a protein with a fast nucleotide dissociation such as Rac1b. Unlike the constitutive active mutants of Rac1 (G12V in the P-loop and Q61L in the switch II region) that also have an impaired intrinsic GTP hydrolysis (
), the defective GTPase reaction of Rac1b can be restored by GAP proteins. As shown in Fig. 1D, the catalytic domain of p50GAP stimulated the GTPase reaction of Rac1b up to 55-fold (21-fold for Rac1), showing that GAP is able to stabilize the catalytic elements of Rac1b and thus accelerate the GTPase reaction.
High Level of Rac1b·GTP in COS-7 Cells—Considering the increased nucleotide dissociation and the decreased GTP hydrolysis, it was tempting to assume that Rac1b is GTP-bound in cells. To prove this assumption, Rac1, its constitutive active mutant Rac1(G12V) and Rac1b were overexpressed in COS-7 cells under serum-starved conditions for 48 h. The fact that wild-type Rac1b could be pulled down with GST-PAK-GBD verifies our hypothesis that Rac1b exists in an active conformation in serum-starved cells. Thereby it resembles the constitutive active Rac1(G12V) mutant (Fig. 2A). As expected wild-type Rac1 could not be detected under these conditions and obviously needs GEF proteins to be activated. A GTP-dependent Rac1b-PAK interaction was demonstrated by performing the pull-down assay with purified GDP- and GppNHp-bound Rac1b. As shown in Fig. 2B, GST-PAK-GBD selectively binds Rac1b·GppNHp but not Rac1b·GDP, similar to the Rac1 control experiment. These results reveal that Rac1b is, independent of external stimuli, GTP-bound in cells and can selectively interact with Rac effector proteins.
Rac1 involvement in transcription and growth control and its requirement for Ras-induced malignant transformation is widely known (
). This knowledge is based on experiments using the expression of a constitutively active Rac1(G12V) mutant. A fast cycling mutant of Cdc42, Cdc42(F28L), has been shown to undergo spontaneous nucleotide exchange in the absence of a GEF while maintaining full GTPase activity (
). This mutant has an even greater cell-transforming potential in fibroblasts compared with the constitutively active Cdc42(G12V) mutant. However, our biochemical data clearly shows that Rac1b behaves as a self-activating GTPase that is predominantly GTP-bound in cells.
Low Affinity Binding of Rac1b to PAK-GBD—To characterize the effect of the insertion on effector interaction, we determined equilibrium dissociation constants (Kd) of PAK-GBD binding to GppNHp-bound Rac1b and Rac1 using the GDI assay (
). As shown in Fig. 3, increasing concentrations of PAK-GBD resulted in incremental inhibition of the mantGppNHp dissociation from Rac1b and Rac1. We obtained a Kd value of 0.49 μm for the PAK-GBD interaction with Rac1, which nicely corresponds to previous reports (
). For Rac1b we observed a 7-fold reduced binding affinity of PAK-GBD. The lower Kd of 3.55 μm can be most likely attributed to the 19-amino acid insertion. MantGDP dissociation from Rac1b was not inhibited under these conditions (data not shown), confirming the GTP-dependent interaction of Rac1b with PAK-GBD.
Furthermore, we examined the interaction of Rac1 and Rac1b with full-length PAK using a GST pull-down assay. Fig. 2B shows that we could not detect binding of Rac1b to full-length PAK, and hence Rac1b stands in contrast to Rac1. Compared with the high affinity binding of the isolated PAK-GBD domains to Cdc42, it has been recently shown that full-length PAK has a much lower binding affinity for Cdc42 (
). Assuming that this is also true for Rac1, our biochemical data suggest an extremely low affinity of full-length PAK for Rac1b.
Conserved Overall Structure of Rac1b·GDP and Rac1b· GppNHp—To gain insight into the structural impact of the 19-amino acid insertion of Rac1b, we determined the crystal structures of Rac1b in the GDP- and GppNHp-bound states. The crystals diffracted to 1.75 Å resolution (Table I). Size exclusion chromatography showed that Rac1b is monomeric in solution (data not shown), but it crystallized as a dimer with two molecules/asymmetric unit in a head to head fashion. The contact surface has a size of 1347 Å2 and is build up by β1 to β3, α1, and the switch I of both molecules. Both molecules in the asymmetric unit are very similar as can be derived from the low root mean square deviation of 0.45 Å for 165 common Cα atoms.
The ribbon representation in Fig. 4 shows the secondary structure elements of Rac1b·GDP and Rac1b·GppNHp superimposed on the GppNHp-bound Rac1 (
), of which three are well ordered in the Rac1b·GDP and Rac1b·GppNHp structures. The P-loop (13AVGKT motif) plays a crucial role in phosphate binding and strongly contributes to magnesium (Mg2+) coordination by the side chain of the Thr17 residue. The 115TKLD and 157CSAL motifs (Rac numbering), on the other hand, directly interact with the guanine base and thereby ensure specificity of guanine nucleotide binding.
The most critical motifs, the central region of switch I (32YIPT) and the N-terminal part of switch II (57DTAGQ), that govern the Mg2+ and γ-phosphate binding display the major structural difference between Rac1b and Rac1. The switch I region (amino acids 30-38) is drastically displaced from the nucleotide-binding site in the GDP- and GppNHp-bound Rac1b structures compared with the Rac1 structure (see below). We did not observe any electron density for the switch II region and the proximate 19-amino acid insertion (amino acids 76-94), indicating that this insert is highly mobile and also leads to a higher mobility of the switch II region (see below).
Highly Mobile Switch II and the 19-Amino Acid Insertion—The switch II region and the adjacent 19-amino acid insertion (amino acids 59-92 in the GDP-bound and 61-93 in the GppNHp-bound structures) are not resolved and therefore not included in the crystal structures (Fig. 4). In most structures of GppNHp-bound GTPases, the switch II region is well ordered (
), indicating that the insertion in Rac1b contributes to a higher mobility of the switch II region and thus leads to an impaired GTPase reaction of Rac1b (Fig. 1D). The catalytic residue Gln61 in switch II (57DTAGQ motif) is crucial in this context, and its high flexibility is most likely the reason for the impaired GTPase reaction of Rac1b (Fig. 1D). This is due to the missing stabilization of the nucleophilic water leading to a decreased GTP hydrolysis rate of Rac1b. It has been shown before that the mutation of this key residue (Gln61) significantly affects the intrinsic and the GAP-stimulated GTP hydrolysis reactions (
). Interestingly, the fact that GAP is able to stimulate the GTP hydrolysis of Rac1b strongly indicates that the switch regions, naturally providing the GAP-binding site, can be stabilized in a GTPase-competent conformation. It has been proposed previously that the 19-amino acid insertion could form a new functional domain built of two β-strands connected by a turn (
), but our results make it unlikely that this region, because of its flexibility, has any secondary structure.
Displacement of the Switch I Region in Rac1b—The most obvious structural difference of Rac1b in comparison with Rac1 is observed in the region between Ala28 and Asn39 encompassing the switch I of Rac1b. The drastic displacement of the switch I region from the nucleotide-binding site is similar in the Rac1b·GDP and Rac1b·GppNHp structures (Fig. 5, A and B). Switch I exhibits a maximum distance of 6.5 Å to the nucleotide-binding site compared with 3.2 Å in Rac1, where it completely covers the nucleotide (Fig. 5C).
The reason for this open switch I conformation is most likely the disruption of the interaction between switches I and II. In the Rac1·GppNHp structure Phe37 of switch I lies in a hydrophobic cleft that is composed by the side chains of Thr58, Tyr64, Leu67, and Arg68 of switch II. Additionally, Val36 makes a hydrophobic interaction with Tyr64. Because of the high mobility of the switch II in Rac1b, this interaction cannot take place, resulting in a destabilization of the switch I region. In the crystal, the open conformation of switch I is stabilized by a hydrophobic interaction of the Phe37 side chains of the adjacent molecules. Although switch I is stabilized by the crystal packing, it still exhibits an increased B-factor of about 45 compared with 30 for residues in the core of the protein. This suggests that switch I is probably very flexible in solution.
High affinity nucleotide binding and GTPase activity of small GTPases are crucially dependent on the presence of an Mg2+ ion. The octahedral Mg2+ coordination is highly conserved throughout small GTPases (
) but shows significant differences in Rac1b. In the GDP-bound state, the magnesium is directly coordinated by an oxygen of the β-phosphate, the side chain of T17 (P-loop) and three water molecules (Wat2, 116 and 117), but lacks the coordination of Thr35. This key interaction is replaced by a fourth water molecule (Wat1; Fig. 5A). A similar arrangement is observed for Rac1b·GppNHp, except for the replacement of one water molecule (Wat2) by the γ-phosphate oxygen of GppNHp (Fig. 5B).
As a consequence of the open switch I, contacts of the invariant Thr35 with the γ-phosphate (main chain NH-group) and the Mg2+ ion (side chain OH group) are disrupted and provide an explanation for the rapid nucleotide dissociation of Rac1b. This situation is comparable with the T35A mutation in Ras, which drastically reduces the nucleotide affinity, because of the loss in Mg2+ coordination (
). The crystal structure of the nucleotide-free Rac1·Tiam1 complex has shown that the interaction of the DH domain of Tiam1 with Rac1 has shifted switch I and β2 (amino acids 27-45) up to 2.7 Å along the nucleotide-binding cleft (
). Thereby, Thr35 in Rac1 is displaced, and the Thr35-Mg2+ interaction is disrupted (Fig. 5C). Thus, Tiam1 binding to Rac1 prevents Thr35 from binding to the Mg2+ ion and allows GDP release from Rac1 (
). Because the P-loop contacts with either GDP or GppNHp are well conserved and the increased dissociation rates of GDP and GTP are rather similar, we propose that the 19-amino acid insertion in Rac1b induces similar effects on the switch I region as Tiam1.
Interaction with Regulators and Effectors—For comparative structural analyses of Rac1b interaction with regulators and effectors, we used the following structures: the nucleotide-free Rac1·Tiam1 complex (
), indicating that the structural requirements for the Rac1b-Tiam1 interaction are not affected by the 19-amino acid insertion. However, in contrast to the Rac1b-GAP interaction, which resulted in stimulation of the GTP hydrolysis reaction of Rac1b (Fig. 1D), no further increase in the nucleotide dissociation rate in the presence of a 50-fold molar excess of the Tiam1 DH-PH domain was observed (Fig. 1C). It is generally accepted that both GEFs and GAPs require activation and membrane recruitment to fulfill their regulatory function in the cell. This and the fact that Rac1b is, without external stimuli, GTP-bound in COS7 cells (Fig. 2A) strongly indicate that, independent of its regulators, Rac1b intrinsically persists in the activated state.
Our biochemical data lead us to the assumption that PAK-GBD has to stabilize the highly mobile switch regions of Rac1b by the expense of a 7-fold lower affinity (Table II). Because of an even more reduced affinity, we could not observe binding of Rac1b to full-length PAK. This suggests that Rac1b is not able to interact with PAK in the cellular context. An interaction of Rac1b with the other major Rac effector, p67PHOX has not been tested and cannot be excluded, because the p67PHOX-binding site on Rac1b is largely conserved.
The fact that Rac1b does interact with GAP and PAK-GBD (Figs. 1D and 3) indicates (i) that the highly mobile switch regions of Rac1b can in principle be recognized by different binding domains and (ii) that the insertion does not necessarily abolish binding. However, a proper interaction requires a stabilization of the Rac1b switch regions that obviously causes an overall reduction of binding affinity, as shown here for PAK-GBD.
Conclusions—The current study clearly demonstrates how the 19-amino acid insertion affects two fundamental biochemical properties of small GTP-binding proteins. An impaired GTP hydrolysis coupled with an accelerated nucleotide dissociation lead to a predominantly GTP-bound protein in the cellular context. The insertion seems to resemble an intrinsic GEF function by modifying the structure and dynamics of the switch regions. The fact that Rac1b binds GDP and GTP with comparable affinities and that the P-loop contacts with the phosphate groups of both nucleotides are unaffected strongly supports the notion that the missing stabilization of Mg2+ and γ-phosphate binding as well as the highly mobile switch II are the reasons for the changed properties.
Our data together with the recent report of Matos et al. (
) enabled us to suggest a concept of Rac1b regulation and its interaction with downstream targets (Fig. 6). The missing regulation by GDI most likely keeps Rac1b constitutively membrane-bound. In consideration of the aberrant intrinsic activities of Rac1b presented in this study, it is tempting to assume that in contrast to Rac1, Rac1b regulation by GEFs and GAPs seems to be redundant. Activated GAPs can effectively down-regulate Rac1b, but it enters immediately a new activation cycle by a self-activating mechanism. GEFs may contribute to an even faster exchange of the bound nucleotide. We conclude that the self-activation, the impaired GTPase reaction, and the GDI insensitivity are the reasons for the accumulation of GTP-bound Rac1b in cells. Although the presented overall structures of both nucleotide-bound forms of Rac1b are remarkably similar, we cannot exclude the possibility that the missing switch II region exhibits conformational differences and therefore may be responsible for the exclusive recognition and binding of GppNHp-bound Rac1b by PAK-GBD. These observations strongly suggest that the very flexible switch regions of Rac1b may adopt a GTP-bound conformation upon effector binding. In this state Rac1b may interact with p67PHOX and certainly other yet unknown effectors. However, important questions concerning the involvement of Rac1b in signal transduction and in tumor progression remain to be elucidated.
We thank A. Wittinghofer for continuous support and E. Lengyel for providing the cDNA of human Rac1b. We thank I. Schlichting, W. Blankenfeld and the staff of the European Synchrotron Radiation Facility for data collection at the ID14 beamline, as well as O. Daumke and A. Ghosh for critical reading of the manuscript.
The atomic coordinates and structure factors (codes 1RYF and 1RYH) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).