The Cdc42/Rac Interactive Binding Region Motif of the Wiskott Aldrich Syndrome Protein (WASP) Is Necessary but Not Sufficient for Tight Binding to Cdc42 and Structure Formation*

Wiskott Aldrich syndrome is a rare hereditary disease that affects cell morphology and signal transduction in hematopoietic cells. Different size fragments of the Wiskott Aldrich syndrome protein, W4, W7 and W13, were expressed in Escherichia coli or obtained from proteolysis. All contain the GTPase binding domain (GBD), also called Cdc42/Rac interactive binding region (CRIB), found in many putative downstream effectors of Rac and Cdc42. We have developed assays to measure the binding interaction between these fragments and Cdc42 employing fluorescentN-methylanthraniloyl-guanine nucleotide analogues. The fragments bind with submicromolar affinities in a GTP-dependent manner, with the largest fragment having the highest affinity, showing that the GBD/CRIB motif is necessary but not sufficient for tight binding. Rate constants for the interaction with W13 have been determined via surface plasmon resonance, and the equilibrium dissociation constant obtained from their ratio agrees with the value obtained by fluorescence measurements. Far UV circular dichroism spectra show significant secondary structure only for W13, supported by fluorescence studies using intrinsic protein fluorescence and quenching by acrylamide. Proton and 15N NMR measurements show that the GBD/CRIB motif has no apparent secondary structure and that the region C-terminal to the GBD/CRIB region is α-helical. The binding of Cdc42 induces a structural rearrangement of residues in the GBD/CRIB motif, or alternatively, the Wiskott Aldrich syndrome protein fragments have an ensemble of conformations, one of which is stabilized by Cdc42 binding. Thus, in contrast to Ras effectors, which have no conserved sequence elements but a defined domain structure with ubiquitin topology, Rac/Cdc42 effectors have a highly conserved binding region but no defined domain structure in the absence of the GTP-binding protein. Deviating from common belief GBD/CRIB is neither a structural domain nor sufficient for tight binding as regions outside this motif are necessary for structure formation and tight interaction with Rho/Rac proteins.

Members of the Rho subfamily of GTP-binding proteins, namely Rho, Rac, and Cdc42, are involved in the regulation of the actin cytoskeleton. They cycle between a biologically inactive GDP-bound and an active GTP-bound state. The ratio of these states is regulated by GEFs, 1 guanine nucleotide exchange factors, and GAPs, GTPase-activating proteins. In the GTP-bound state they interact with so-called effectors or downstream targets that are defined as molecules that have a high affinity to the GTP-bound state as opposed to a low affinity to the GDP-bound state.
The Wiskott Aldrich syndrome is a genetically determined, X-linked recessive disorder that affects cells of the hematopoietic lineage and fibroblasts. Molecular symptoms include cytoskeletal abnormalities in T-cells and platelets as well as deficient chemotaxis in neutrophils. The three major phenotypic appearances are thrombocytopenia, eczema, and recurrent infections due to defects in the immune response toward polysaccharide antigens. The gene is allelic to X-linked thrombocytopenia, a rare recessive disorder with other clinical symptoms than WAS (1). Cloning of the gene (2) showed that the responsible protein WASP harbors several domains involved in signal transduction such as a proline-rich region binding to SH3 domains of adaptor proteins like Nck-and Src-type kinases. Other domains termed WH1 and WH2 (for WASP homology) are also found in other signaling molecules. A homologue called N-WASP has been cloned which shows 50% homology to WASP over the entire length (3).
Nothing is known about the structure of Rac effectors and their complexes. In the case of Ras, small Ras binding domains (RBDs) in the effectors Raf, Ral-GEF, (14 -16) and Byr2 (17) have been identified which fold into a stable domain. For Raf (18,19) and Ral-GEF (20,21), it has been shown that these domains fold into the ubiquitin superfold even though they are sequentially unrelated either to ubiquitin or to each other. It has also been shown by x-ray structural analysis that Ras/Rap interact with RafRBD by forming an intermolecular ␤-sheet close to the effector region of Ras (19). To study the interaction between Cdc42 and proteins containing the CRIB motif and as a first step toward understanding the molecular defects of WAS, we have investigated their interaction using biophysical techniques such as circular dichroism (CD), fluorescence spectroscopy, surface plasmon resonance, and NMR.

MATERIALS AND METHODS
Cloning and Purification of WASP Fragments-The plasmid pFu-sion3 (4) coding for residues 48 -321 of the human WASP was used as a template for polymerase chain reaction subcloning of fragments corresponding to residues 221-257 (W4) and 201-321 (W13). The oligonucleotides 5Ј-CGCGGATCCCCAGCTGATAAGAAACGCTCAGGG-3Ј and 5Ј-CAGAATTCATCCATTCTGGGGGTC-3Ј were used to amplify a 126base pair DNA fragment coding for the 4-kDa peptide and the oligonucleotides 5Ј-CGCGGATCCGACATCCAGAACCCTGACATCACG-3Ј and 5Ј-GTGAATTCATCGAGATGGCGGTG-3Ј to amplify a 380-base pair DNA fragment coding for the 13-kDa peptide. After hydrolysis with BamHI and EcoRI, the fragments were cloned into the glutathione S-transferase (GST) fusion vector pGEX-2T (Amersham Pharmacia Biotech), and the identity of the constructs was verified by sequencing using the cycle sequencing procedure. Plasmids were transformed in Escherichia coli BL21(DE3) (22) for production of the protein.
An overnight culture of BL21(DE3) cells containing the appropriate plasmid was diluted 100-fold into LB medium supplemented with 200 g/ml ampicillin. At an A 600 ϭ 0.8 synthesis of the GST fusion protein was induced by addition of 0.1 mM isopropyl-␤-D-1-thiogalactopyranoside. After further incubation at 37°C for 4 h cells were harvested by centrifugation. The bacterial pellet was resuspended at 4°C in 50 mM Tris/HCl, pH 7.5, 100 mM NaCl (buffer A), 1 mM phenylmethanesulfonyl fluoride. A Microfluidizer TM (Microfluids Corp., Newton, MA) operating at 600 bar was used to disrupt the cells, and the 30,000 ϫ g supernatant was applied to a glutathione-Sepharose (Amersham Pharmacia Biotech) column preequilibrated in buffer A. Unspecifically bound proteins were eluted by washing with buffer A containing 500 mM NaCl. Incubation of the bound protein with buffer A including 50 mM KCl, 10 mM MgCl 2 , and 2 mM ATP removed most of the contaminating E. coli DnaK protein. 2 The fusion protein was either eluted with 20 mM glutathione in buffer A or cleaved overnight with 10 units/ml thrombin.
Purification of Cdc42⅐Nucleotide Complexes-The plasmids coding for the GST fusions of the wild type (Cdc42wt) and G12V mutant (Cdc42V12) of the human Cdc42 protein were kindly provided by George Martin (Onyx Pharmaceuticals). After cleavage of the fusion proteins with thrombin, seven additional amino acid residues (GSKIISA) preceded the starting methionine at the N terminus of Cdc42wt, whereas the CdcV12 protein had only three additional amino acid residues (GSP) attached to its N terminus as determined by protein sequencing. Synthesis and initial steps of purification of Cdc42 were the same as described above for the WASP fragments except that the buffer used was buffer A with 5 mM MgCl 2 , 5 mM 1,4-dithioerythritol. The cleaved Cdc42 was in the GDP state as determined by high pressure liquid chromatography (see below). Exchange for both the fluorescent and non-fluorescent GTP analogue was achieved as described by John et al. (24). Complexation of Cdc42 with mGDP was possible by incubation with a 50-fold excess of mGDP in the presence of 10 mM EDTA for 2 h at ambient temperature. For all complexes further purification involved gel filtration on a G-75 Sephadex (Amersham Pharmacia Biotech) column in 20 mM HEPES/NaOH, pH 7.4, 1 mM MgCl 2 . Cdc42 concentrations were determined by the method of Bradford (25) or spectrophotometrically according to Hiratsuka (26) when complexed with mant nucleotides. Cdc42⅐nucleotide complexes were analyzed by isocratic high pressure liquid chromatography (27) on a C18 reversed phase column (Bischoff, Leonberg, Germany). Because of its higher stability in dilute solutions all measurements were done with the G12V mutant of Cdc42 (Cdc42V12), whereas for the NMR studies wild type Cdc42 (Cdc42wt) was used. Control experiments showed that the interactions between WASP fragments and Cdc42 are not influenced by the G12V mutation. However, due to the high intrinsic GTPase rate of Cdc42, it was assumed that the wild type protein might even hydrolyze Gpp-(NH)p to GppNH 2 in long term measurements and that the G12V mutation would prevent this.
Protein identities and purity were checked by SDS-polyacrylamide gel electrophoresis and Edman sequencing. Electrospray ionization mass spectroscopy on a Finnigan LCQ mass spectrometer was used to verify the molecular masses of the proteins and to determine the isotopic enrichment of 15 N-labeled proteins.
Spectroscopic Techniques-Fluorescence measurements were done on a Fluoromax TM spectrofluorimeter (Spex Industries) equipped with a thermostat set to 25°C using cuvettes with a 0.5 ϫ 1-cm 2 section. Proteins were diluted into 40 mM HEPES/NaOH, pH 7.4, 100 mM NaCl (standard buffer). In case of Cdc42, 5 mM MgCl 2 was added. Emission spectra were recorded after excitation at 280, 295, or 366 nm (1-nm bandwidth) with time constants of 0.1 or 1 s and 2-nm bandwidth. All spectra were corrected for buffer contributions. Titrations of Cdc42⅐mGpp(NH)p with WASP fragments were performed with an automatic titrator using a Hamilton syringe as reservoir for the ligand. The decrease in fluorescence at 435 nm was followed. Equilibrium dissociation constants were obtained by fitting the solution of a quadratic equation describing a bimolecular association model assuming a 1:1 stoichiometry to the data. For measurement of the dissociation of fluorescent nucleotides from Cdc42 and its complexes with WASP fragments a 1000-fold excess of unlabeled nucleotide was added, and the decrease in fluorescence at 435 nm (8 nm bandwidth) was followed. Single exponentials were fitted to the data using the program Grafit (Erithacus Software).
Fluorescence quenching measurements using acrylamide as the quenching agent were carried out as follows. To 1.2-ml sample normalized to 5 M tryptophan residues, aliquots of a 6 M acrylamide stock solution were added. After excitation at 295 nm (0.5-nm bandwidth) the fluorescence signal at 357 nm (2-nm bandwidth) was followed for 2 min and averaged. Buffer contributions and dilution effects were accounted for and the inner filter effect due to the absorption of acrylamide at 295 nm was corrected by multiplying the fluorescence intensity by 10 ϪA/2 , where A is the absorption of the solution at 295 nm measured in a 1-cm cuvette. The results were analyzed on the basis of the Stern-Volmer equation (28).
Far UV CD spectra were recorded in 1-mm quartz cuvettes on a J710 spectropolarimeter (Jasco) at room temperature. Measurements were done at a protein concentration 10 M in 20 mM potassium phosphate, pH 6.5, with a sensitivity of 20 mdeg, a time constant of 1 s, and a bandwidth of 1 nm. Each spectrum was accumulated 10-fold. All spectra were corrected for buffer contributions and converted to mean residue ellipticities according to Schmid (29).
For NMR measurements the samples were prepared by adding 50 l of 2 H 2 O (99.9%) to 450 l of a 1-2 mM solution of WASP fragment in 20 mM potassium phosphate, pH 6.5. Homogeneity of the sample was tested before and after recording of NMR spectra by SDS-polyacrylamide gel electrophoresis and electrospray mass spectroscopy. Spectra were recorded at 300 K on a Bruker DRX-500 spectrometer equipped with shielded z gradients. For assignment of resonances in the free WASP fragments the following spectra were recorded: DQF-COSY, NOESY, and Clean-TOCSY for W4, W7, and W13, ROESY for W4 and 2 J. Reinstein, personal communication.
W7, 15 N-gsHSQC for W4, W7, and W13, 15 N TOCSY-HMQC for W4 and W7, and 15 N NOESY-HMQC for W7. Spectra of WASP fragments complexed to Cdc42 were measured with 15 N-gsHSQC (W4 and W7), 15 N TOCSY-HMQC (W7), and 15 N NOESY-HMQC (W7). Sodium 2,2-dimethyl-2-silapentane-5-sulfonate was used as internal standard for calibration of proton resonances. For 15 N calibration the procedure of Wishart et al. (30) was applied. Processing and analysis of the two-and three-dimensional spectra were done using the NDEE program package (Biosymbiose Inc., Bayreuth, Germany). With exception of the DQF-COSY, spectra were recorded in the phase-sensitive absorption mode with the time proportional phase incrementation method for quadrature detection in f 1 . The time domain data for 1 H and 15 N frequencies were zero-filled once or twice prior to Fourier transformation. Twodimensional spectra were acquired with 2 to 8K data points in f 2 dimension and 512 to 1K data points in f 1 dimension. The water resonance was suppressed by applying the WATERGATE sequence (31) or by presaturation. Gradient-selected HSQC spectra were processed using a Blackman filter with a frequency cutoff at the water resonance and a down-sampling rate of 2. Prior to Fourier transformation all spectra were multiplied with a sine bell square window function shifted by /4. Definition of secondary structure was done by calculation of the chemical shift indexes according to Wishart et al. (32). Characterization of ␣-helical secondary structure was done using angle and distance constraints, namely NH-H␣ i, i ϩ 3 and NH-H␣ i, i ϩ 4 NOEs. Additionally, dihedral angles were included to analyze the structural elements.
Surface Plasmon Resonance-Association and dissociation reactions involving Cdc42⅐Gpp(NH)p and Cdc42⅐GDP without a fluorescence label were studied by surface plasmon resonance in a BIAcore TM system (BIAcore AB, Uppsala, Sweden). Since reaction of essential lysine residues can lead to partial inactivation of chemically coupled proteins, a sandwich assay (33) with anti-GST antibodies was used. These were coupled to the matrix following the protocol of the manufacturer (34). GST fusion proteins were incubated with the antibodies at a concentration of 0.2 mg/ml for 7 min, followed by incubation with free ligand. Regeneration with 20 mM glycine, pH 2.0, and 0.005% SDS resulted in complete dissociation of all noncovalently bound ligands, leaving the immobilized immunoglobulin at essentially full activity.
Binding of Cdc42 to binary complexes of antibody and GST-WASP fragments was analyzed in a concentration-dependent manner. To eliminate the contribution of nonspecific binding, equivalent controls with Cdc42⅐GDP were used to calculate specific signal changes. Experiments were performed at 20°C in 10 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM MgCl 2 , 0.0005% (w/v) Igepal CA-630 (Sigma), a buffer system stabilizing the immobilized anti-GST antibodies. Binding and dissociation curves were fitted to the sum of a single exponential and a linear function to take into account the association and dissociation of GST complexes.

Expression of Recombinant WASP Fragments-It has been
shown previously that small fragments of Rac/Cdc42 effectors containing the CRIB motif, also called the GTPase binding domain, GBD, fused to GST are able to bind to Cdc42 in an overlay assay and that the binding is dependent on Rac/Cdc42 being loaded with GTP (4,5). In the case of human WASP the smallest fragment that retained this property is a peptide of 34 amino acids comprising residues 235-268 (4). Binding of an 81-residue human WASP fragment (amino acids 215-295) to Cdc42 has also been investigated by the ability of the fragment to inhibit the interaction of Cdc42-GTP and Cdc42GAP (35). In order to define a Cdc42 binding domain for structural studies, to see whether any such fragments are necessary and sufficient for high affinity binding and to develop methods for the direct measurement of these interactions, we have produced various fragments from the human WASP in E. coli as GST fusion proteins (Fig. 1A). SDS-polyacrylamide gel electrophoresis analysis of the purified fragments (Fig. 1B) and mass spectroscopy show that the WASP fragments of 4 kDa (residues 221-257), 7 kDa (201-268), and 13 kDa (201-321) molecular mass, termed W4, W7, and W13, respectively, can be isolated as soluble non-GST fusion proteins without degradation and in reasonable quantities.
WASP as a Nucleotide Dissociation Inhibitor of Cdc42-The different fragments were tested for their ability to bind to Cdc42 in a nucleotide-dependent fashion. We have shown previously (36,37) that the Ras binding domains of c-Raf1 and Ral-GEF act as a GDI protein by inhibiting nucleotide dissociation from Ras in a saturable manner and that this effect can be used to measure the dissociation constant in solution for this interaction. When Cdc42 is loaded with the fluorescent nonhydrolyzable GTP analogue mGpp(NH)p, the dissociation of the fluorescent analogue, as measured by the decrease in fluorescence in the presence of excess unlabeled Gpp(NH)p, is 0.4⅐10 Ϫ3 s Ϫ1 , 2.6-fold slower than the dissociation of unlabeled Gpp(NH)p (not shown), similar to what has been found for the dissociation of mGDP versus GDP for Cdc42 (38). In the presence of increasing amounts of W13, the dissociation of nucleotide is inhibited (Fig. 2). With different WASP fragments different degrees of guanine nucleotide inhibition can be achieved, and it appears that the dissociation rate of mGpp(NH)p from the different Cdc42⅐mGpp(NH)p⅐WASP complexes is different (Table I). This indicates that in principle the affinity between the WASP fragment and Cdc42 can be measured by the GDI assay and that WASP affects nucleotide binding of Cdc42 in a similar manner as Ras effectors have an effect on nucleotide binding on Ras (36, 37). Equilibrium Fluorescence Titration and Nucleotide-specific Binding-Since, however, the difference between the dissociation rate constants in the presence and absence of WASP is very small as compared with the Ras system, and the amplitude of the kinetics diminishes with increasing concentration of W13, a more direct approach to measure the equilibrium dissociation constant was developed. As both the WASP fragments and Cdc42 contain a tryptophan residue, intrinsic fluorescence was not used as a spectrometric probe. In Fig. 3A it is shown that the fluorescence of the nucleotide analog mGpp(NH)p bound to Cdc42 decreases on addition of W13 by a factor of more than 2 under saturating conditions and almost approaches the fluorescence of the unbound nucleotide. This change in signal can thus be used to measure the interaction directly by equilibrium titration. Following the decrease of fluorescence (excitation and emission wavelength 366 and 435 nm, respectively) with increasing concentrations of W13 and fitting a binding equation with a 1:1 stoichiometry to the data, a dissociation constant of 77 nM was obtained (Fig. 3B). The stoichiometry was verified by an active site titration where increasing amounts of W13 were added to 1 M Cdc42⅐mGpp(NH)p, a concentration much larger than the K D of the interaction (not shown). The smaller WASP fragments W4 and W7 also bind with high affinity. However, the dissociation constants for both are approximately 6-fold higher (Table II). Thus the GBD/CRIB region seems to be necessary but not sufficient for tight binding. An indication that the extra sequences of larger WASP fragments supply additional interaction sites can be seen from the maximal change of the mant fluorescence in the three complexes; the decrease induced by W4, W7, and W13 is very different (Fig. 3A).
There is no change in fluorescence on addition of even very high concentrations (10 M) of W4 (Fig. 3C) and W7 (data not shown) to Cdc42 loaded with the GDP analogue mGDP, and only a slight change for 10 M W13 (Fig. 3D). Thus, binding of the fragments to Cdc42 depends on the conformational state of the latter. Assuming the maximum fluorescence change for the interaction of WASP fragments with Cdc42⅐mGDP to be similar to that of Cdc42⅐mGpp(NH)p, we estimate a lower limit of approximately 50 M for the equilibrium dissociation constant for the interaction between Cdc42⅐mGDP and W13. As a result a minimal 500-fold difference in affinity between the two conformational states can be proposed. Although the exact affinity difference is not known, it is in agreement with the Ras system (36,37) and Ran system (33) where a 1000-fold difference has been measured. Therefore it can be assumed that the WASP fragments isolated mimic the effector interaction between WASP and Cdc42.
Surface Plasmon Resonance Measurements-To measure quantitatively the kinetics of the interaction between WASP and Cdc42⅐Gpp(NH)p surface plasmon resonance was used. In a sandwich assay described earlier for the interaction of Ran with its effector RanBP1 (33), a monoclonal antibody against GST was coupled to the sensor chip and used to bind the GST fusion proteins GST-W4 and GST-W13. This system can be used to measure the interaction with Cdc42 in a GTP-and concentration-dependent manner in real time. Fig. 4A shows traces of the change in mass on the sensor chip upon binding of Cdc42 to immobilized GST-W4. Only Cdc42⅐Gpp(NH)p but not Cdc42⅐GDP nor nucleotide free Cdc42 binds tightly to the surface. An association rate constant of 1.9⅐10 5 M Ϫ1 s Ϫ1 (Fig. 4, B and C, Table II) was obtained for the W13/Cdc42⅐Gpp(NH)p system, which is fairly slow for a bimolecular association reaction. The dissociation rate constant measured by surface plasmon resonance (not shown) was found to be 11.9⅐10 Ϫ3 s Ϫ1 , the ratio of the rate constants resulting in a kinetically determined dissociation equilibrium constant of 63 nM. The reversed system with GST-Cdc42 immobilized and W13 as the freely diffusible ligand yields a value of 69 nM (data not shown). These results are in good agreement with the value of 77 nM obtained by equilibrium fluorescence titration taking into account the small differences in buffer composition and temperature (20 versus 25°C, 150 mM NaCl versus 100 mM NaCl). It supports observations made for the Ran interaction with RanBP1 using the sandwich coupling technique where good correlation between fluorescence and surface plasmon resonance data is found (33). Dissociation rate constants were also measured for the W4 and W7 fragments using this technique and are about 7-fold faster than those for the W13 fragment. Association rate constants for W7 and W4 were not evaluated quantitatively due to the small net signal change. Compared with the affinities measured by fluorescence titration (Table II), these data show that the difference in affinities between W13 on the one side and W4/W7 on the other is almost completely due to a difference in the dissociation rate constants.
Structural Differences between WASP Fragments-The difference in binding constants between W4, W7, and W13 could either be caused by the smaller fragment not being folded in a native-like conformation or due to the W13 supplying extra residues that are necessary for tight binding. From Fig. 5A it can be seen that the residue Trp-252 just outside the CRIB region in WASP exhibits different fluorescence emission properties in the W4, W7, and W13 fragments. W4 and W7 do not differ significantly in their fluorescence properties and show a fluorescence emission maximum at 357 nm. Thus, the tryptophan residue seems to be located in a polar environment. In contrast to this, a slight increase in fluorescence intensity and a blue-shift of 6 nm can be seen for W13 indicating a less polar environment for the tryptophan residue. Thus, not only the mant group in Cdc42⅐mGpp(NH)p bound to WASP fragments but also the tryptophan residue in uncomplexed WASP frag-  ments is located in different structural environments.
To address the solvent accessibility of the tryptophan residue in the WASP fragments quenching studies were performed. Fluorescence quantum yields are decreased by collisional quenching of the excited state of a fluorophor with certain small molecules such as acrylamide, iodide, or oxygen, where the magnitude of the decrease depends on the concentration of the quencher (28). Fig. 5B shows that the tryptophan fluorescence of W4, W7, W13, and Cdc42⅐nucleotide complexes decreases linearly with increasing concentrations of acrylamide, the degree of quenching, indicated as the Stern-Volmer constant (Table III), being lowest for W13 and highest for W4. The value for W4 is close to that for N-acetyltryptophanamide, a model compound for a totally solvent exposed tryptophan residue. The quenching data agree with the fluorescence emission spectra and can be interpreted as a further indication of gradually increasing structural complexity going from W4 to W13, monitored by different solvent accessibilities of the tryptophan residue in the three fragments.
A further indication for the different structure of WASP fragments and of their complexes with Cdc42⅐mGpp(NH)p is shown in Fig. 5C. Excitation of the unique Trp-97 residue in Cdc42⅐mGpp(NH)p at 295 nm results in an almost complete fluorescence energy transfer between the tryptophan and the mant group, since there is only a very small fluorescence emission peak visible around 330 nm but a strong emission peak at 435 nm. Saturating the complex with either W4 or W13 shows differences in fluorescence energy transfer efficiencies. If in complex with W4 the tryptophan residues are excited at 295 nm, there is no significant change in mant fluorescence. For Cdc42 in complex with W13 on the other hand, the energy transfer efficiency drops by a factor of around 1.5 thus indicating different structural contexts of the Cdc42 bound mant group when in complex with W4 and W13, respectively.
Finally, to gain insight into the overall secondary structure content of the fragments, CD spectra of W4, W7, and W13 were recorded. It can be seen from Fig. 5D that the spectra of W4 and W7 in the Ͻ210 nm range resemble those expected for a random coil conformation. In W13 a significant amount of secondary structure, presumably ␣-helix, can be detected, as indicated from the absorbance in the 210 -230 nm and Ͻ195 nm range. In the initial attempts to prepare W13 from a pGEX vector containing a thrombin site, an internal thrombin cleavage site between Arg-268 and Ser-269 was found. The fragments resulting from this thrombin cleavage, W6, which does not bind to Cdc42, and W7 were also investigated by CD spectroscopy. It turned out that the sum of the W6 and W7 spectra was different from the W13 spectrum (not shown) indicating a concomitant loss of structure when W13 is cleaved by thrombin.
NMR, Sequence-specific Assignments-In order to gain more structural insight on WASP, two-and three-dimensional nuclear magnetic resonance techniques were used. According to the data shown above, W13 is the most tightly binding fragment and shows significant secondary structural content as assessed by circular dichroism. Yet, it tends to aggregate under high concentrations as used for the NMR measurements. Thus, both the W4 and W7 fragments were used for NMR investigations. Here, the sequence-specific resonance assignment of W7 is reported. Fragment W13 was only partially assigned by comparing the chemical shifts from resonances of amino acids that were also present in W4 and W7. Resonance assignment of W7 was achieved according to the standard method developed by Wü thrich (39). Spin systems were obtained from two-dimensional DQF-COSY and two-dimensional Clean-TOCSY spectra. Three-dimensional 15 N-TOCSY-HMQC spectra were recorded to avoid ambiguous assignment of overlapping resonances. Se-quential connectivities were obtained from two-dimensional NOESY spectra with mixing times of 200 ms and from a threedimensional 15 N-NOESY-HMQC with a mixing time of 100 ms. The strong Arg-268 NH-H␣ as well as the NH-H␣ signal of the unique Thr-208 were chosen as starting points for the assignment of the C and N termini. The spin systems of the rings of Tyr-212 and Trp-252 were used to trace the chain to Lys-226 and Lys-230, respectively. The assignment of Arg-227 to Gly-229 was ambiguous in the case of W7 but not in W4. Two NH-H␣ resonances for residues 227-229 were found indicating existence of two conformations. Three additional signals left unassigned belong to spin systems of two alanine residues and one glutamine residue. These signals had integrals of approximately 10% of those peaks that could be assigned unequivocally. Proline assignments were done by using the connectivities of their H␦ resonances and the H␣ peaks of the preceding amino acid.
Secondary Structure of W4 and W7-Short and medium range NOEs were analyzed to characterize the secondary structure of W7. Surprisingly, considering the tight binding of W7 to Cdc42, amino acid regions in the N-terminal part of W7 including the GBD/CRIB motif itself do not show any obvious secondary structure. Only the region from Trp-252 to Asp-264 shows NOEs (NH i -NH iϩ2 ; H␣ i -NH i i ϩ 2 ; Ha i -NH i ϩ 3 ; H␣ i -NH i ϩ 4 ) typical of an ␣-helical conformation (Fig. 6). Although only a few of these could be assigned unambiguously, calculation of the coupling constants confirms the existence of an ␣-helix: four 3 J NH-H␣ couplings in the DQF-COSY define angles in this region that are significantly smaller than 6 Hz which are typical for ␣-helices. Thus the region of secondary structure is limited to the C-terminal part of W7. Considering a maximum length of 10 residues the helix content within W7 is approximately 14%. This coincides well with the ϳ10% helical content estimated from the CD spectrum using the procedure developed by Holzwarth and Doty (40). The underestimation of helical elements in peptides by CD spectroscopy is a well known phenomenon since short or transient helices do not give rise to a strong CD signal (41,42). With exception of the NH-H␣ resonances of the first and last amino acids of the chain, no coupling constants higher than 7.5 Hz could be found thus pointing out that no ␤-sheet structure occurs within the protein. Additionally, five slowly exchanging NH protons belonging to residues Gln-255 to Asp-259 indicate hydrogen bonding. Compared with the H␣ shifts of tripeptides mimicking the random coil state of a protein (43), H␣ resonances in the C-terminal part of W7 shift to lower ppm values, and together with NOEs and 3 J NH-Ha coupling measurements support the ␣-helical assignment.
The chemical shifts of all H␣ resonances of the W4 fragment indicate a random coil for this region, supporting the notion that the presumed ␣-helix of WASP is at the C-terminal end of the GBD/CRIB motif. Considering the absence of stable secondary structure in the GBD/CRIB motif, it is surprising that there are slowly exchanging NH protons in this region as well, indicating that the main chain is protected against solvent.
Complexes of W4 and W7 with Cdc42⅐Gpp(NH)p-To obtain structural information on the interaction between Cdc42 and WASP, HSQC spectra of free and Cdc42⅐Gpp(NH)p-bound W4 and W7 fragments were recorded, using a 2-fold excess of the  4. Surface plasmon resonance measurement of the Cdc42-WASP interaction. A, anti-GST antibodies were covalently attached to the surface of a BIAcore TM sensor chip. This affinity matrix was used to stably couple GST-W4. Injection of 1 M either nucleotide-free Cdc42, Cdc42⅐Gpp(NH)p, or Cdc42⅐GDP into the solution buffering the chip induces a resonance response as indicated. B, association reactions of (from bottom to top) 100, 200, 400, 600, 800, and 1000 nM Cdc42⅐Gpp(NH)p in solution onto GST-W13 bound to a sensor chip as above. C, plot of the association rate constants (k obs ) from B against the concentration of Cdc42 in the solution. The data were fitted to obtain k on using the BIAevaluation 2.1 software. GTP-binding protein. Resonances in the complex were either unperturbed or shifted. Some resonances were broadened or could not be observed any longer. Fig. 6 and Table IV (data not shown for W4) summarize the 15 N-1 H peaks that have shifted or that could not be analyzed. Both in W4 and W7, the shifting resonances are clustered in the CRIB region. No additional resonance shifts of NH-H␣ peaks in other regions were found. DISCUSSION We have shown here that fragments of WASP ranging from 4 to 13-kDa molecular mass can be stably expressed in E. coli as fusion proteins and cleaved from the GST part without losing stability or posing problems like aggregation or diminished solubility at higher micromolar concentrations. The fact that these fragments are not proteolytically digested during expression in E. coli and chromatographic purification seems to suggest that they have some amount of tertiary structure.
The WASP fragments bind to Cdc42 in the triphosphate form with reasonable high affinity ranging from 77 to 490 nM (Table  II), as determined both by fluorescence titration and surface plasmon resonance with GST fusion proteins bound to sensor chips via anti-GST antibodies. This confirms similar findings made on the interaction between Ran and its effector RanBP1, where also a good correlation between surface plasmon resonance and fluorescence techniques was found (33). Equilibrium measurements of the interaction between a 72-residue fragment of the protein kinase mPAK-3 (PBD) and Cdc42 (44), using as a readout the fluorescence of the mant group, gave a dissociation constant in the micromolar range, similar to that found for W4 and W7. Surprisingly, the fluorescence of Cdc42bound mGpp(NH)p was increased on addition of PBD, whereas with WASP the fluorescence decreases. This may indicate that the interface between Cdc42 and the CRIB/GBD region is somewhat different between WASP and PAKs, reflecting a different sequence context between the different putative effectors of Cdc42. Zhang et al. (35) have assessed the affinity between an 81-amino acid fragment (residues 215-295) of WASP and Cdc42 by measuring its inhibition of the Cdc42-Cdc42GAP interaction. Surprisingly, the dissociation constant of such a fragment, the size and location of which is intermediate between W7 and W13, was found to be 4.1 M, well above the 77-490 nM values found for W13, W7 and W4.
The affinity of the W13 fragment is in the same range or even higher than that found between Ras and its effectors Raf and Ral-GEF (15,36,37,45), and lower than the affinities between Ran and its effectors RanBP1, RanBP2, and importin which are in the nanomolar or subnanomolar affinity range (33,46,47). 3 The structure of RafRBD does not change appreciably on complex formation with Rap or Ras (15,19). In contrast W13 seems to become structured only on complex formation with Cdc42, and it is surprising that the affinity of the latter complex is similar to that of Ras/RafRBD which seems to indicate the formation of a tight interface, the energy of which is used to of 10 M W4 and W13 was excited at 295 nm, and the fluorescence emission was recorded between 300 and 500 nm. D, Far UV CD spectra of a 10 M solution of the WASP fragments in 20 mM KP i , pH 6.5.  induce a structural transition in the GBD/CRIB motif. Although the affinity to the GDP-bound form could not be determined due to the lack of a measurable spectroscopic signal (at the concentration tested), we have estimated the affinity of W13 to Cdc42⅐GDP to be at least 500-fold less tight, again in agreement with observations for interactions between GTPbinding proteins and their effectors, and with the assumption that WASP is a true effector for Cdc42. The dissociation rate constants, measured with surface plasmon resonance, are relatively fast, approximately 1 min Ϫ1 for W13. Assuming that full-length WASP has a similar dissociation rate, one can conclude that the Cdc42⅐GTP⅐WASP complex has a short half-life in the cell, similar to the Ras-Raf complex (45) 4 which may allow fast regulation of signaling via Ras or Rac/Cdc42. We have shown that the region encompassing the GBD/CRIB motif is necessary but not sufficient for high affinity binding to Cdc42 and that other regions of WASP are necessary for the interaction. W4 and W7 bind with similar low affinity, whereas W13 which contains additional residues C-terminal to this region increases the affinity approximately 6-fold, which corresponds to a difference in free enthalpy of binding of only 4.4 kJ/mol at 25°C. This small difference is nevertheless reflected in a different behavior of the corresponding Cdc42⅐Gpp(NH)p⅐ WASP complexes, one indication being the stronger inhibition of nucleotide dissociation associated with the binding of W13 to Cdc42⅐Gpp(NH)p, as compared with the W4 and W7 complex.
Surprisingly, even though W4, W7, and W13 seemed to be stable protein fragments which bound specifically and with high affinity to Cdc42 in the triphosphate form, they did not show pronounced secondary structure in either the CRIB/GBD motif itself or the region N-terminal to it, as determined by proton NMR. Thus, although the GBD/CRIB region is com-monly called a domain, it is not a domain in a structural sense, as an independent folding unit with a defined structure, in contrast to the Ras binding domain of both Raf kinase (15,19) and Ral-GEF (20,21). The C-terminal part of W7, however, has a short and/or flexible ␣-helical structure as detected by NMR and CD, and it is tempting to suggest that this structural feature is stabilized in W13 thereby giving rise to the stronger CD signal of this fragment. The sequence C-terminal to the CRIB region in WASP does not show conserved residues in comparison to other proteins containing this motif. We could thus imagine that these residues do not bind directly but rather stabilize the CRIB region and thereby raise the affinity toward Cdc42. Fluorescence studies on W4, W7, and W13 show differences in their intrinsic fluorescence and also in accessibility of the tryptophan residues as determined by quenching studies indicating different environments of the Trp residue. This apparent difference in structure is reflected in the affinities toward Cdc42, where W4 and W7 have similar affinities as opposed to W13. The structural properties of the Cdc42 effector are different to those of the Ras effectors c-Raf1 and Ral-GEF and Rlf, which contain small Ras binding domains with a defined ubiquitin-like fold, but show no apparent sequence homology (18 -21) 5 It is also different from the Ran binding domain of RanBP1 and RanBP2 which have conserved sequence elements and a defined structure. 6 The NMR data show that upon binding to Cdc42⅐Gpp(NH)p a chemical shift change of the 15 N-1 H resonances occurs only in the CRIB/GBD region of W4 and W7 indicating that the conserved sequence element experiences a reorganization of structure on binding to Cdc42⅐Gpp(NH)p, which is, however, not detectable via CD. In contrast, the structure of the Ras binding domain of c-Raf-1 is very stable and very similar between the bound and unbound conformation (18,19). 7 Whether or not the CRIB/GBD motif of WASP or PAK has defined structure cannot be decided conclusively at the current stage of investigation. However, for some 15 N-labeled nuclei and their corresponding hydrogen atoms, which reflect the conformation of the main chain, two different chemical shifts are found. There are also some NH protons in the GBD/CRIB motif which are slowly exchanging within 24 h, indicating some degree of folding in this area. There are no large CD spectral changes on complex formation between WASP fragments and Cdc42. Taken together this could mean that the CRIB/GBD domain populates two or more conformations in solution, only one of which strongly interacts with Rac/Cdc42 which in turn stabilizes the bound conformation. It is also possible that the structure of the GBD/CRIB motif and the flanking regions depends strongly on the presence of the rest of the WAS protein, although we find it unlikely that the structure of the motif is very much different from the one it has in the W13 fragment. Nevertheless, to gain more insight into the structural requirements for the interactions between WASP and Cdc42, the structure of the complete WASP molecule in complex with the GTP-binding proteins may have to be solved by x-ray crystallography to learn about possible domain boundaries in the molecule and how WASP may contribute to the effect of Cdc42 on the cytoskeleton. Our studies show that the analysis of fragments of WASP and of other GBD/CRIB containing proteins such as the various PAK isoforms have to be treated with caution as the requirements for a productive interaction appear to be different between the fragments and the full-length proteins.