PROTEINS: Structure, Function, and Genetics 25:112-119 (1996) Purification, Stabilization, and Crystallization of a Modular Protein: Grb2 J.P. Guilloteau,' N. Fromage,' M. Ries-Kautt? S. Reboul,' D. Bocquet,' H. Dubois,' D. Faucher? C. Colonna,' A. Ducruix? and J. Becquart' 'Service de Biochimie, RhGne-PoulencRorer SA, 94403 VitrylSeine, France; 'Laboratoire de Bwlogie Structurale, CNRS, 91198 GiflYuette, France; 3Seruice de Microbiologie MolCculaire, RhGne-Poulenc Gencell, 94403 VitrylSeine,France ABSTRACT We report here the purification and the crystallization of the modular protein Grb2. The protein was expressed as a fusion with glutathione-S-transferaseand purified by affinity chromatography on glutathione agarose. It was apparent from reverse phase chromatography that the purified protein was conformationallyunstable.Instability was overcome by the addition of 100 mM arginine to the buffers. Because Grb2 appeared to be extremely sensitive to oxidation, crystallization experiments were performed with a dialysis button technique involving daily addition of fresh DTT to the reservoirs. The presence of 8 to 14% glycerol was necessary to obtain monocrystals. These results are discussed in relation with the modular nature of Grb2. 0 1996 WiIey-Liss, Inc. Key words: Grb2, modular proteins, purification, crystallization, stabilizing agents, arginine, DTT, glycerol INTRODUCTION Many large proteins are combinations of several, clearly identifiable, autonomously folding modules.* They have recently been the object of tentative classification~.~ These - ~ modular proteins play various fundamental biological roles, including cell adhesion, clotting, fibrinolysis, and signal transduction. Despite the considerable interest in their tertiary structure, very few structures of intact modular proteins, containing more than two modules, have been reported so far (a monoclonal antibody6 and the protein Grb27).The three-dimensional structural determination of modular proteins is hampered by the difficulty of crystallizing such protein^.^ We present here the strategy that we used for the crystallization of Grb2. Some steps are extensively detailed and are given as experimental elements that may document the crystallization of modular proteins and could be of interest for other projects. *Modules are autonomously folded homologous structures. They are defined by consensus sequences of 40-100 conserved residues, encoded on discrete exons and bordered by introns of identical phase.',' 0 1996 WILEY-LISS, INC. Grb2 (Growth factor Receptor Bound protein 2) is a 25 kDa intracellular monomeric adaptor protein,' made up of one Src homology 2 (SH2) module flanked by two Src homology 3 (SH3) modules. SH2 and SH3 domains are protein modules (100 and 50 amino acids respectively) involved in protein-protein interactions. Grb2 plays a key role in the mitogenic signal transduction pathway linking tyrosine kinase receptor to Ras a c t i ~ a t i o n . ~ Considerable work has been devoted to the study of SH2 and SH3 modules. Numerous three-dimensional structures of isolated SH2 and SH3 modules have been published. In particular, the structures of the Grb2 N-terminal SH3I0-l3 and C-terminal SH314 domains complexed with peptides were recently resolved. The regulatory domain of the Srcfamily tyrosine kinase Lck15 was the first published structure with two juxtaposed SH2 and SH3 domains. In addition to the interest due to its important role in the mitogen signal transduction pathway, Grb2 constitutes a good model with which to study the orientations of different SH2 and SH3 domains in an entire protein. EXPERIMENTAL PROCEDURES Protein Expression vector Grb2 was expressed as a glutathione-s-transferase fusion protein, using the vector pGEX-TT (de- Abbreviations: DTNB, 5,5'-dithiobis(2-nitrobenzoicacid); DTT, D-L dithiothreitol; EDTA, ethylenediamine tetraacetic acid; EGFR, epidermal growth factor receptor; FMOC, 9-fluorenyl-methoxycarbonyl; Grb2, growth factor receptor bound protein 2; GSH, reduced glutathione; GST, glutathione S-transferase; IRS-1, insulin receptor substrate 1; MPD, Z-methyl-2,4-pentanediol;PDGFR, platelet-derived growth factor receptor; PEG, polyethylene glycol; RPLC, reverse phase liquid chromatography; SDS-PAGE, sodium dodecyl sulfatepolyacrylamide gel electrophoresis; SH2(3), Src homology domain 2(3);TFA, trifluoroacetic acid. Received September 26,1995; revision accepted November 3, 1995. Address reprint requests to Jean-Pierre Guilloteau, RhBnePoulenc Rorer SA, CRVA, BLtiment Monod, Service de Biochimie, 13 quai J. Guesde, 94403 VitrylSeine, France. Grb2 CRYSTALLIZATION rived from pGEX-2T, Pharmacia, France) to ensure a more efficient cleavage of the fusion by thrombin. The particular feature of this vector lies in the sequence encoded between GST and the fusion partner that contained two thrombin cleavage sites separated by a 9 amino acid linker.16 The complete Grb2 cDNA sequence was subcloned into this vector. Expression Transformed cells were grown in 7-liter fermentors under fed batch conditions at 37°C in a medium composed of yeast extract, glucose, mineral salts, oligoelements, and 100 pg/ml ampicillin. IsopropylP-D-thiogalactopyranoside was added to 1mM and 2 h later cells were recovered by centrifugation a t 10,OOOg for 10 min and frozen a t -80°C. Purification E . coli cells from 1liter of culture were disrupted by homogenization with a minilab 6 (Rannie, France) at a pressure of 1,000 bars at 4°C in one volume of lysis buffer pH 7.3 (16 mM Na,HPO,, 5 mM EDTA, 150 mM NaC1, 1 mM Pefabloc), then centrifuged at 10,OOOg for 20 min to remove insoluble cellular debris. GSH agarose gel (150 ml of gel Sigma, France, G 4510) was suspended in the supernatant (crude extract) and gently agitated for 2h a t 4°C. The lysate was then removed by filtration on glasfilter (por. 4).The agarose beads were washed three times with 600 ml of lysis buffer. Grb2 was then eluted by cleavage with thrombin (human plasma thrombin, Sigma T3010). The agarose gel was washed with one volume of thrombin cleavage buffer (50 mM phosphate pH 8.0) for 15 min at room temperature. After filtration on glasfilter, the agarose beads were suspended in two volumes of cleavage buffer to which 5 pg of thrombin per mg of fusion protein were added (the amount of fusionprotein bound to the gel was previously determined with a 0.5 ml aliquot of gel). The gel was then gently shaken at room temperature for 30 min. The eluate was removed by filtration on glasfilter (por. 4). The beads were washed twice with 1.5 bead volume of cleavage buffer for 15 min, and the eluates were pooled. The thrombin in the Grb2 eluate was removed through the addition of antithrombin I11 agarose (Sigma A 8293,l ml of gel for 100 pg of thrombin). After 45 min a t 4"C, the antithrombin gel was removed by filtration on glasfilter (por. 4). Then DTT was added to the Grb2 solution to a final concentration of 10 mM. The last purification step consisted of Mono Q ion exchange chromatography (HR10/10). The column was equilibrated with 20 mM phosphate buffer pH 8.0, and the sample was loaded at 3 ml/min and eluted with an NaCl gradient. The Grb2 monomer was eluted with 0.125 M NaCl and the dimer with 0.22 M NaC1. To stabilize the protein, 1M arginine in Tris buffer pH 8.0 was added to a final concentration of 0.1 M and the solution was stored at 4°C. 113 Peptides The following phosphotyrosine-containing peptides were obtained from Neosystem (Strasbourg, France): L-10-S (EGFR Tyr1068: Leu-Pro-Val-ProGlu-Tyr(P03H2)-Ile-Asn-Gln-Ser), G-8-F (IRS-1Tyr 895: Gly-Glu-Tyr(P03H2)-Val-Asn-Ile-Glu-Phe), S11-E (PGDF Tyr 977: Ser-Val-Leu-Tyr(P03H2)-ThrAla-Val-Gln-Pro-Asn-Glu). Proline-rich peptides (G-20-AhSOS-1: 1143-1162) were synthesized in the solid phase on 0.1 mM [(hydroxymethyl) phenoxymethyll polystyrene resin by using FMOC chemistry, on an Applied Biosystems (USA) Model 431A peptide synthesizer. All the solvents and reagents used were purchased from Applied Biosystems. After synthesis, protecting groups were removed and the peptides cleaved and precipitated by the addition of tert-butyl methyl ether. Peptides were then purified by preparative HPLC on a BioRad, France, RSL C18 100 A column, eluted with an increasing linear acetonitrile gradient containing 0.07% TFA in water and lyophilized. Protein-PeptideInteractions Protein-peptide interactions were studied using a BIAcore'" (Pharmacia, France, Biosensor) that allows quantitative analysis of molecular interactions in real The flow rate for all the experiments was 5 pl/min. Peptides were covalently bound to a CM5 sensor chip (Pharmacia Biosensor) equilibrated in HBS buffer (10 mM Hepes, 150 mM NaCl, 5 mM EDTA, 0.005% Tween 20, pH 7.4) with the following procedure: the carboxymethylated dextran matrix was activated with 40 p1 of a mixture of N-hydroxysuccinimide and N-ethyl-N'-(3-diethylaminopropy1)-carbodiimide (amine coupling kit, Pharmacia Biosensor). Next, 40 pl of a 5 mg/ml peptide solution in 50 mM borate buffer pH 8.5 containing 1 M NaCl were injected. Following peptide injection, 35 p1 of 1 M ethanolamine were injected to block unreacted activated groups. The immobilization of phosphotyrosine peptides was controlled with a monoclonal antibody directed against phosphotyrosine (UBI, Lake Placid, NY). Interaction of Grb2 with immobilized peptides was performed with 20 p1 of a 10 pg/ml Grb2 solution in HBS buffer. After Grb2 injection, 15 pl of 100 mM HC1 in water were applied to the matrix to remove bound material and to regenerate the matrix for the next injection. Crystallization Reagents The following reagents were purchased from the indicated sources: PEG (Fluka, France); MPD, sodium para-toluene-sulfonate (Merck, France, for synthesis); sodium chloride, potassium thiocyanate (Merck for analysis); arginine monohydrochloride (Merck for biochemistry); sodium-potassium tartrate, glycerol (Prolabo, France, for analysis); so- 114 J.P. GUILLOTEAU ET AL. dium citrate (Sigma molecular biology reagent); sodium acetate (Sigma ACS reagent); ammonium sulfate (Sigma ultra); DTT (Sigma). MilliQ deionized water (Millipore, France) was used to prepare the solutions. Protein concentration At the end of the purification process, the chromatographic fractions containing the monomeric protein were collected. In these fractions, Grb2 was a t a concentration between 0.1 and 1 mg/ml, in 20 mM pH 8.0 phosphate solution with 100 mM arginine, 10 mM DTT and 110-130 mM NaC1. Prior to any crystallization experiment, the protein was simultaneously dialyzed and concentrated a t room temperature with a MicroProdicon, Polyabo, France, device equipped with a 2-liter reservoir and a 10 kDa cut-off dialysis membrane. The dialysis solution was a 50 mM pH 8.0 Tris/HCl solution containing 100 mM arginine and 5 mM DTT. The final protein concentration, ranging from 5 to 40 mg/ml, was determined spectrophotometrically by measuring the A,,,. Crystallization Stock solutions of all crystallizing agents were prepared in 50 mM TridHCl, 100 mM arginine, pH 8.0 buffer. Further dilutions were performed with the same buffer. For the dilution of the protein solutions, this buffer contained additionally 5 mM DTT. The solutions were filtered with 0.22 pm cellulose acetate filter units. The crystallization screening experiments were achieved using the vapor diffusion method in Linbro boxes, with 4 to 8 p1 hanging drops and 1ml reserv o i r ~ at , ~4~and 19°C. Freshly concentrated DTT was added to the reservoir in order to obtain a 5 mM final concentration. The drops were prepared by mixing equal volumes of protein and reservoir solutions, such that a two-fold increase in concentration was expected for all components in the drop during the course of the experiment. Crystallizations by dialysis were performed in 10 pl microdialysis buttons with Spectrapor, France, 6 dialysis membranes (molecular weight cut-off = 6,000-8,000 Da). The buttons were soaked in 10 ml reservoirs. The crystallization experiments were stored at 19 ? 0.1"C in a thermal regulated chamber (Memmert, France) or a t 4 5 0.5"C in a refrigerator (Brandt, France). Space group determination Space group and unit cell dimensions were determined from data collected with synchrotron radiation (A = 0.901 A), on the wiggler beam-line DW32 at LURE (Orsay, France),' by using a MarResearch, Germany, Imaging plate. RESULTS AND DISCUSSION Purification The GST-Grb2 fusion protein was expressed a t a level of 10-15% of the total soluble protein in E. coli. It was purified with high yield (200-600 mg/liter of culture) to >95% homogeneity by affinity chromatography on GSH-agarose. After thrombin cleavage, Grb2 is 80%pure, containing Grb2 fragments but no GST. We finally obtained 30 mg pure (>98%) monomeric Grb2 after the polishing step on Mono Q (Fig. 1).The purification steps that appear to be critical for the crystallization of Grb2 are detailed below. Thrombin cleavage To facilitate the cleavage between GST and Grb2, we started from the construction described by Guan and Dixonl' composed of one thrombin cleavage site followed by a glycine-rich linker containing the sequence P-G-I-S-G-G-G-G-G.In order to avoid the addition of these nine amino acid residues to the N-terminal side of Grb2, we introduced another thrombin cleavage site between this linker and Grb2. Experiments showed that the cleavages were sequential: the thrombin site upstream to the linker being cleaved first. So, an insufficient quantity of thrombin resulted in contamination of Grb2 with molecules of Grb2 plus linker. On the other hand, the presence of two sites inside Grb2 susceptible to be cleaved by thrombin (Ar8l-Gly2, and Arg17'Glyl") could lead to proteolysis. It was thus crucial to optimize the thrombin concentration in order to obtain a compromise between efficient cleavage (with minor contamination by Grb2 plus linker) and slight degradation of Grb2 (Fig. 2). Elimination of thrombin Concentrating Grb2 was a prerequisite for the crystallization experiments. In this step remaining traces of thrombin were detrimental. To remove residual thrombin after cleavage, we tested gels specific for thrombin adsorption. Antithrombin I11 agarose gel was preferable to paraminobenzamidine agarose because of its better adsorption capacity, although a nonspecific adsorption of Grb2 onto the gel was observed. In this way 98% of the thrombin was removed. The remaining 2% contamination was eliminated with ion exchange chromatography on Mono Q. Electrophoresis of concentrated Grb2 solutions, that had been stored for several weeks, indicated that no degradation had occurred. Elimination of GST and GST-Grb2 Weak desorption of GST-Grb2 from the GSH-agarose gel during the cleavage at pH 8.0 led to contamination by GST in subsequent steps. As Grb2 and GST have approximately the same molecular weight, their separation could not be followed by SDS-PAGE. Experiments of spiking GST and GST- 115 Grb2 CRYSTALLIZATION Fig. 1. Coomassie blue-stained SDS-PAGE (14Y-Novex) of the purification of Grb2 from E. m/i.Lane 1: Molecular weight markers (kDa). Lane 2: Cells (10 pg). Lane 3: Total soluble protein. Lane 4: Supernatant after binding onto GSH agarose. Lane 5: Purified GST-Grb2 (bead suspension) (2 pg). Lane 6: Grb2 after cleavage with thrombin (2 pg). Lane 7: Mono Q flow through. Lane 8: Grb2 monomer (0.125 M NaCI) (5 pg). Lane 9: Grb2 dimer (0.22 M NaCI) (5 pg). Grb2 onto Mono Q allowed us to find the conditions for which the two proteins were separated from Grb2. A Western blot of the final Grb2 solution with anti-GST polyclonal antibodies confirmed that GST and GST-Grb2 had been eliminated. Polishing step A polishing step on Mono Q was introduced in the purification process to obtain “crystallization grade protein,” because contaminating molecules closely related to Grb2 (fragments, Grb2 plus linker, oligomers, and GST-Grb2) could later interfere with the crystal growth. GST and thrombin were eliminated in the flow-through. The first major eluted peak corresponded to monomeric Grb2, and the second one was identified as non-covalent dimeric Grb2. The use of a smooth gradient allowed the separation of Grb2 plus linker from Grb2, respectively in the rise and fall of the first peak. It has to be noticed that phosphate buffer, unusual for Mono Q , eliminated Grb2 fragments better than Tris buffer did. Characterization The purity of the protein in the first peak was checked by native and SDS-PAGE. Grb2 was characterized by Electrospray Mass Spectrometry (EMS) and amino-terminal sequencing. Titration of the sulfhydryl groups with DTNBZ1showed the absence of a disulfide bond between the only two cysteines of the protein (C3’ and C1’*). Size exclusion chromatography, analytical ultracentrifugation, and light scattering experiments indicated that the concen- trated solution was monodisperse and contained monomeric Grb2. Because Grb2 cannot be tested enzymatically, the binding activity of the protein was controlled using BIAcore’”. Plasmon resonance with BIAcore’” was ideally suited to analyzing the ability of the purified protein to bind various peptides known to interact with SH2 or SH3 domains. As described in the lite r a t ~ r e , 2we ~ ~observed ~~ specific interactions of very high affinity between Grb2 and short phosphotyrosine-containing peptides derived from a major insulin receptor substrate IRS-1, from the epidermal growth factor receptor (EGFR), and from the platelet-derived growth factor (PDGFR). Grb2 was able to bind L-10-S (EGFR) and G-8-F (IRS-1) either immobilized on the matrix or in solution. S-11-E (PDGFR) was not recognized. In the same way, Grb2 had a high affinity with the proline-rich peptide G-20-A. Stability Although Grb2 proved to be pure and monomeric at the end of the purification process, the protein stability i n solution vs. time had to be investigated prior to any crystallization attempt. Verification by electrophoresis, gel filtration chromatography, and reverse phase liquid chromatography (RPLC) showed: - the formation of covalent dimers due to cysteine oxidation, - the formation of non-covalent dimers when the solutions were stored deep-frozen, and 116 J.P. GUILLOTEAU ET AL. Fig. 2. Coomassie blue-stained SDS-PAGE (14%N -ovex) of thrombin cleavage of the purified fusion protein GST-Grb2. GST-Grb2 bound to glutathione-agarose beads was incubated with various quantities of thrombin at room temperature for 90 min in cleavage buffer. The molecular weight of GST-Grb2 and Grb2 is 50 kDa and 25 kDa, respectively. After thrombin cleavage, fragments corresponding to internal sequences are around 18 kDa, 20-22 kDa and 3-4 kDa; Grb2 plus linker is around 26 kDa. The thrombin to substrate ratios (w/w) are as follows: Lane 1: 1/50. Lane 2: 1/100. Lane 3: 1/200. Lane 4: 1/400. Lane 5: 1/800. Lane 6: 1/1600. The amino terminal was sequenced to confirm the presence of Grb2 plus linker in lanes 4 and 5. Lane 7: Molecular weights markers (kDa). - the progressive appearance of conformational heterogeneity, indicated by RPLC profiles,24 and unrelated to oligomer formation. The protein solution could be stabilized by: - addition of 10 mM DTT to prevent from covalent dimers, - storage at 4°C to avoid the formation of non-covalent dimers, and -addition of 100mM arginine as a stabilizing agent. In the case of Grb2, reverse phase analysis appeared to be a valuable tool to detect the change of protein conformation and to screen stabilizing agents. Such agents have been extensively described in the literature (for a review, see reference 25). For Grb2, the best results and the most prolonged effect were obtained with arginine. Some stabilization was also observed with lysine, taurine, and high concentrations of some salts (ammonium sulfate, sodium chloride). Surprisingly, widely used agents, such as glycerol and sugars, had no effect. Arginine is known to stabilize protein in solutions by increasing the surface tension and by interacting with negative charge^.'^ Grb2 has an experimental PI of 5.9 and is therefore negatively charged a t pH 8.0. The X-ray structure of the protein7 later proved that the juxtaposition of the two SH3 domains forms a continu- ous surface a t the bottom of the molecule which presents an alignment of negatively charged residues. These charges could create strong repulsive interactions between the two SH3. The surface of interaction between them is rather small and these domains could dissociate and adopt different orientations. So arginine may lower internal electrostatic repulsions, and thus stabilize a single conformation. No arginine molecule has been detected in the electronic density of Grb2, at 3 A resolution. Conformational heterogeneity may be rather frequent in modular proteins, and of course could be very detrimental to their crystallization. In such a case, screening of stabilizing agents to help freeze internal motions is recommended. Crystallization A variety of crystallization conditions and crystallizing agents, including various salts (ammonium sulfate, sodium phosphate, sodium chloride, potassium thiocyanate, sodium paratoluene-sulfonate), PEG (400,1,500, and 4,000 kDa), and MPD, was tested. Experiments were carried out a t 4 and 19°C in the presence of 100 mM arginine. The first crystals were obtained with 14 mg/ml Grb2, 0.75 M sodium citrate in the pH 8.0 buffer a t 19°C. They grew overnight. They were small (50 pm in length) and 117 Grb2 CRYSTALLIZATION twinned. In addition, some protein aggregation (spherulites) appeared 4 or 5 days after the beginning of the experiments (Fig. 3a). From a chemical point of view, citrate was the only carboxylate tested. Its efficiency to crystallize Grb2 appeared to be linked to the carboxylate functionality, as tartrate and acetate gave the same small twinned crystals and spherulites. Because citrate and tartrate are strong cation chelating agents that could later hamper soaking experiments with heavy atom solutions, all the further crystallization experiments were done with sodium acetate. Attempts t o reduce the nucleation rate by optimizing the crystallization conditions yielded larger crystals. The best crystals (350 pm length) were obtained with 15 mg/ml Grb2,1.7 M sodium acetate, a t pH 8.0 and 19°C. Nucleation was observed only 2 to 3 days after the beginning of the experiment but the growth was stopped by the formation of spherulites 2 days later. The overall shape of the crystals was improved (Fig. 3b). Cutting the twinned crystals in two and using synchrotron radiation allowed collection of the first X-ray diffraction data. We succeeded in preventing the formation of spherulites by maintaining a permanent reductive environment for Grb2. This was performed by switching from vapor diffusion to a dialysis technique. Dialysis allows the reservoir to be changed daily with renewal of DTT. Spherulites did not appear, even after several weeks (Fig. 3c), although crystals were still twinned. This demonstrated that the origin of the protein aggregation was due to Grb2 oxidation, once DTT had lost its efficiency. The use of dialysis buttons made it possible to maintain the necessary reductive potential for several weeks and to pursue the growth experiments over a longer time scale. In order to improve the shape, size, and quality of the crystals by decreasing their growth rate,26 we undertook another series of dialysis experiments where different percentages of glycerol were added to the reservoir solutions. Glycerol reduces protein diffusion in the crystallization solution by increasing viscosity. Indeed, monocrystals appeared after 2 to 4 days with a growth rate of 20 pndday with 6-22% glycerol, sodium acetate, a t pH 8.0 and 19°C. Nevertheless, most of them soon grew twinned. In a last optimization step, monocrystals with dimensions around 200 x 200 x 200 pm3 (Fig. 3d) were obtained with a carefully adjusted acetate concentration and 8 to 14% glycerol. These crystals belong to the tetragonal space group P4, with a = b = 90.0 8 and c = 97.7 8,and contain two molecules in the asymmetric unit. They were used for X-ray data collection and enabled the determination of the threedimensional structure of Grb2.7 Glycerol is increasingly being used as an additive in crystallization trials.27 It has two major positive effects. First, it prevents the protein from denaturating and aggregating, improving the reproducibil- ity of the experiments. Second, it improves crystal quality by decreasing the crystal growth rate and by reducing twinning. It is sometimes absolutely necessary for crystallization.2s In the case of Grb2, we found that it had no effect on the stability of the protein, but that it allowed a better crystallization. Without glycerol, Grb2 always crystallized in the form of two embedded twinned crystals. The linking zone between the two crystalline subunits was more or less clearly defined depending on the growth speed. The slower the crystals grew, the better was their shape. FINAL REMARKS SH2 and SH3 domains are conserved protein modules of 100 and 50 amino acids, respectively, found in a variety of cytoplasmic signaling proteins. The proteins involved in signal transduction often contain several copies of SH2 and SH3 modules and can thus be described as modular proteins. Considerable interest lies in their three-dimensional structure analysis. As modular proteins have been proved difficult to crystallize, and intractable for X-ray crystallographic studies, an alternative “dissect and build strategy in three steps has been developed for modular protein^.^' First, the structures of isolated modules are determined, then structural studies on module pairs provide insight into the way consecutive modules come together. Finally, this information, combined with other low resolution data such as that obtained by electron microscopy, is used to rebuild models of intact modular proteins. In the case of Grb2, starting with the known structure of the SH2-SH3 domain of Lck,15 the “dissect and build” method would have predicted relative orientations for the SH2 and SH3 domains different from the one observed in the 3-D X-ray s t r ~ c t u r e . ~ The crystallization of entire modular proteins is a difficult challenge. Detailed experimental data and observations concerning purification, characterization, and crystallization of Grb2 are given here as they are of more general interest, although the answers for Grb2 (use of DTT, arginine, glycerol) may not be universal. The strategy described below is recommended for proteins in general: (a) high purity is a prerequisite for successful crystallization; (b) loss of conformational homogeneity can be accurately checked by reverse phase c h r ~ m a t o g r a p h y ; ~ ~ (c) screening of stabilizing agents may allow to select a stable conformation necessary for crystallization; (d) slowing down crystal growth improves the quality of the crystal;26 and (e) a careful control of environmental parameters (redox, pH, T . . .) avoids shifts from their nominal values. Points (b) and (c) are specially relevant for rnodu- Fig. 3. Progressive improvement of the shape of the crystals of Grb2. a: Twinned microcrystals (50 pm in length) grown from 4 pl hanging drops in the presence of sodium citrate (see text). Most of the protein appears to be aggregated in the form of spherulites around the crystals. b: Twinned crystal (250 pm in length) grown from 8 pI, hanging drop in the presence of sodium acetate. c: Twinned crystals grown in a 10 pl dialysis button in the presence of sodium acetate, with daily addition of fresh D l T in the resewoir. No spherulites appeared, even after several weeks. d: Mono crystals grown in a 10 pl dialysis button in the presence of sodium acetate and glycerol. Grb2 CRYSTALLIZATION lar proteins, as their difficulty to crystallize is linked to conformational heterogeneity. ACKNOWLEDGMENTS We thank F. Schweighoffer for providing us with the construction of GST-Grb2. We are greatly indebted to G. Jung for the production of bacteria in fermentors. We thank B. Monegier for the EMS analysis. Sebastien Maignan performed the X-ray characterization of the crystals. This project was carried out as part of the BioAvenir program supported by Rh6ne-Poulenc with the participation of the Ministere de 1’EnseignementSuperieur et de la Recherche and the Ministere de YIndustrie. REFERENCES 1. Patthy, L. Intron-dependent evolution: Preferred types of exons and introns. FEBS Lett. 214:l-7, 1987. 2. Patthy, L. Modular exchange principles in proteins. Curr. Opin. Struct. Biol. 1:351-361, 1991. 3. Baron, M., Norman, D.G., Campbell, I.D. Protein modules. Trends Biochem. Sci. 16:13-17, 1991. 4. Bork, P. Mobile modules and motifs. Curr. Opin. Struct. Biol. 2:413-421, 1992. 5. Campbell, I.D., Downing, A.K. Building protein structure and function from modular units. Trends Biotech. 12:168172, 1994. 6. Harris, L.J., Larson, S.B., Hasel, K.W., Day, J., Greenwood, A., McPherson, A. The three-dimensional structure of an intact monoclonal antibody for canine lymphoma. Nature 360:369-372, 1992. 7. Maignan, S., Guilloteau, J.P., Fromage, N., Arnoux, B., Becquart, J., Ducruix, A. Crystal structure of the mammalian Grb2 adaptor. Science 268:291-293, 1995. 8. Chardin, P., Cussac, D., Maignan, S., Ducruix, A. The Grb2 adaptor. FEBS Lett. 369:47-51, 1995. 9. Rosakis-Adcock, M., Fernley, R., Wade, J., Pawson, T., Bowtell. D. The SH2 and SH3 domains of mammalian Grb2 couple the EGF receptor to the Ras activator mSosl. Nature 363:83-85, 1993. 10. Goudreau, N., Cornille, F., Duchesne, M., Parker, F., Garbay, C.. Roaues, B.P. NMR structure of the N-terminal SH3 domain of Grb2 and its complex with a proline-rich peptide from Sos. Nature Struct. Biol. 1:898-906, 1994. 11. Terasawa, H., Kohda, D., Hatanaka, H., Tsuchiya, S., Ogura, K., Nagata, K., Ishii, S., Mandiyan, V., Ullrich, A,, Schlessinger, J., Inagaki, F. Structure of the N-terminal SH3 domain of Grb2 complexed with a peptide from the guanine nucleotide releasing factor Sos. Nature Struct. Biol. 1:891-897, 1994. 12. Wittekind, M., Mapelli, C., Farmer 11, B.T., Suen, K.L., Goldfarb, V., Tsao, J., Lavoie, T., Barbacid, M., Meyers, C.A., Mueller, L. Orientation of peptide fragments from Sos proteins bound to the N-terminal SH3 domain of Grb2 determined by NMR spectroscopy. Biochemistry 33: 13531-13539, 1994. 13. Guruprasad, L., Dhanaraj, V., Timm, D., Blundell, T.L., Gout, I., Waterfield, M.D. The crystal structure of the N-terminal SH3 domain of Grb2. J . Mol. Biol. 248:856866, 1995. 119 14. Kohda, D., Terasawa, H., Ichikawa, S., Ogura, K., Hatanaka, H., Mandiyan, V., Ullrich, A., Schlessinger, J., Inaeaki. F. Solution structure and lieand-binding site of the-caiboxy-terminal SH3 domain ,f Grb2. Siructure 2:1029-1040, 1994. 15. Eck. M.J., Atwell. S.K., Shoelson, S.E., Harrison, S.C. Structure of the regulatory domains of the Src-family tyrosine kinase Lck. Nature 368:764-769, 1994. 16. Guan, K.L., Dixon, J.E. Eukariotic proteins expressed in Escherichia coli: An improved thrombin cleavage and purification procedure of fusion proteins with glutathione S-transferase. Anal. Biochem. 192:262-267, 1991. 17. Fagerstam, L. A non-label technology for real-time biospecific interactions analysis. Tech. Protein Chem. 2:65-71, 1991. 18. Johnsson, B., Lafhs, S., Lindquist, G. Immobilization of proteins to a carboxymethyldextran modified gold surface for biospecific interaction analysis in surface plasmon resonance. Anal. Biochem. 198:268-277, 1991. 19. Ducruix, A,, Giege, R. (eds.). Methods of crystallization. In: “Crystallization of Nucleic Acids and Proteins, A Practical Approach.” Oxford IRL Press, 1992:73-98. 20. Fourme, R., Dhez, P., Benoit, J.-P., Kahn, R., Dubuisson, J.-M., Besson, P., Frouin, J. Bent crystal, bent multilayer optics on a multipole wiggler line for an X-ray diffractometer with an imaging plate detector. Rev. Sci. Instrum. 63~982-987, 1992. 21. Riddles, P.W., Blakeley, R.L., Zerner, B. Ellman’s reagent 5,5’-dithiobis(2-nitrobenzoic acid)-a reexamination. Anal. Biochem. 94:75-81, 1979. 22. Skolnik, E.Y., Lee, C.-H., Batzer, A., Vicentini, L.M., Zhou, M., Daly, R., Myers, Jr., M.J., Blacker, J.M., Ullrich, A,, White, M.F., Schlessinger, J. The SH2/SH3 domaincontaining protein Grb2 interacts with tyrosine-phosphorylated IRSl and Shc: Implications for insulin control of ras signalling. EMBO J . 12:1929-1936, 1993. 23. Cussac, D., Frech, M., Chardin, P. Binding of the Grb2 SH2 domain to phosphotyrosine motifs does not change the affinity of its SH3 domains for Sos proline-rich motifs. EMBO J. 13:4011-4021, 1994. 24. Lin, S., Karger, B.L. Reversed-phase chromatographic behavior of proteins in different unfolded states. J . Chromatogr. 499539-102, 1990. 25. Timasheff, S.N. Stabilization of protein structure by solvent additives. In: “Stability of Protein Pharmaceuticals,” part B (series: Pharmaceutical Biotechnology 3). T.J. Ahera, M.C. Manning, eds. New York: Plenum Press, 1992:265-285. 26. Luft, J.R., Arakali, S.V., Kirisits, M.J., Kalenik, J., Wawrzak, I., Cody, V., Pangborn, W.A., DeTitta, G.T. A macromolecular crystallization procedure employing diffusion cells of varying depths as reservoirs to tailor the time course of equilibration in hanging- and sitting-drop vapor-diffusion and microdialysis experiments. J . Appl. Crystallogr. 27:443-452, 1994. 27. Sousa, R. The use of glycerol, polyols, and other protein structure stabilizing agents in protein crystallization. Acta Crystallogr. D 271-277, 1995. 28. Sousa, R., Lafer, E.M., Wang, B.C. Preparation of crystals of T7 RNA polymerase suitable for high-resolution X-Ray structure analysis. J. Cryst. Growth 110:237-246, 1991. 29. Williams, M.J., Campbell, I.D. Solution structures of modular proteins by nuclear magnetic resonance. Methods Enzymol. 245451-469, 1994.