Solution structure of the tandem Src homology 3 domains of p47phox in an autoinhibited form.

The phagocyte NADPH oxidase is a multisubunit enzyme responsible for the generation of superoxide anions (O(2).) that kill invading microorganisms. p47(phox) is a cytosolic subunit of the phagocyte NADPH oxidase, which plays a crucial role in the assembly of the activated NADPH oxidase complex. The molecular shapes of the p47(phox) tandem SH3 domains either with or without a polybasic/autoinhibitory region (PBR/AIR) at the C terminus were studied using small angle x-ray scattering. The tandem SH3 domains with PBR/AIR formed a compact globular structure, whereas the tandem SH3 domains lacking the PBR/AIR formed an elongated structure. Alignment anisotropy analysis by NMR based on the residual dipolar couplings revealed that the tandem SH3 domains with PBR/AIR were in good agreement with a globular module corresponding to the split half of the intertwisted dimer in crystalline state. The structure of the globular module was elucidated to represent a solution structure of the tandem SH3 domain in the autoinhibited form, where the PBR/AIR bundled the tandem SH3 domains and the linker forming a closed structure. Once PBR/AIR is released by phosphorylation, rearrangements of the SH3 domains may occur, forming an open structure that binds to the cytoplasmic proline-rich region of membrane-bound p22(phox).

dem SH3 domains, a polybasic region/autoinhibitory region (PBR/AIR), 1 and a proline-rich region in this order (Fig. 1A). In resting cells, the tandem SH3 domains of p47 phox are masked through an intramolecular interaction with PBR/AIR, resulting in an autoinhibited form (6, 7, 9 -11). Upon cell stimulation, a number of serine residues in PBR/AIR are phosphorylated (12)(13)(14)(15). Phosphorylation of p47 phox induces conformational changes that subsequently lead to rearrangements in intramolecular interactions and the exposure of the tandem SH3 domains that enable interaction with the p22 phox subunit in flavocytochrome b 558 (6,10,11). Anionic amphiphiles such as arachidonate and SDS, activators of the NADPH oxidase in vitro, facilitate a conformational change of p47 phox , exposing its SH3 domains as well as inducing phosphorylation (6,16,17).
Recently, the structures of the autoinhibited and the activated forms of the tandem SH3 domains of p47 phox (Protein Data Bank codes 1NG2 and 1OV3) were reported (18). We also determined the structure of the tandem SH3 domains in the autoinhibited form (p47 phox (151-340); Protein Data Bank code 1UEC) independently (19,20). The structure of the autoinhibited form of the tandem SH3 domains reveals that elongated monomers are related by a crystallographic 2-fold axis at the hinge, forming an intertwisted dimer with a dumbbell-like shape (89 ϫ 64 ϫ 64 Å). The strand-exchanged region where ␤A, ␤B, and ␤C of the N-terminal moiety of one monomer were intertwined with ␤D, the 3 10 helical region, and ␤E of the other (Fig. 1, B and C) takes a typical SH3 fold. This strand-exchanged and intertwisted region was identified as the N-terminal SH3 domain, although the distal loop of the canonical SH3 fold was extended to form the hinge. The split half of the intertwisted dimer in the crystal structure has been assumed to be physiologically relevant and to represent the structure of the tandem SH3 domains in the autoinhibited form (18,20), which we called the globular module (Fig. 1B).
However, a crucial issue remains to be clarified: whether the globular module exists in solution and if it represents the structure of the autoinhibited form of the tandem SH3 domains. This prompted us to investigate the structural characterization of the tandem SH3 domains in the autoinhibited form in solution. Herein, we report the structural properties of p47 phox (151-340) in solution elucidated by NMR spectroscopy and small angle x-ray scattering (SAXS) analysis.

EXPERIMENTAL PROCEDURES
Sample Preparation-The truncated form of p47 phox , residues 151-340 (p47 phox (151-340)) including PBR/AIR, was expressed and purified as previously reported (19). The truncated form of p47 phox , residues 151-286, corresponding to the tandem SH3 domains without PBR/AIR (p47 phox (151-286)), was cloned into a pGEX-2T vector (Amersham Biosciences), transformed in Escherichia coli BL21(DE3), and expressed, as previously described (17). The cells were disrupted by sonication at 4°C in 25 mM phosphate-buffered saline buffer at pH 7.4. The protein was applied to a glutathione-Sepharose 4B column (Amersham Biosciences) equilibrated with phosphate-buffered saline buffer at pH 7.4, and the bound protein was eluted using 25 mM Tris buffer, pH 8.0, and 25 mM reduced glutathione. The N-terminal glutathione S-transferase tag of p47 phox (151-286) was then removed by incubation with thrombin protease (Amersham Biosciences) for 12 h at 25°C, and the digested protein was dialyzed against 2 liters of 25 mM Tris buffer at pH 8.0. The protein was purified by anion exchange chromatography on a resource Q 6-ml column (Amersham Biosciences) equilibrated with 25 mM Tris buffer at pH 8.0, and the bound protein was eluted using a gradient of NaCl from 0 to 500 mM in the same buffer. Further purification was carried out by gel filtration chromatography on a Superdex 75 column (Amersham Biosciences) and eluted with 25 mM BisTris buffer at pH 6.5 and 150 mM NaCl. The protein was concentrated using Centriprep YM-10 (Millipore) to ϳ10 mg/ml.
Analytical Size Exclusion Chromatography-Size exclusion chromatography was carried out at 25°C using a Superose 12 HR 10/30 column (Amersham Biosciences) attached to an Ä KTA system (Amersham Biosciences). Sample containing 30 -120 g of purified p47 phox (151-340) was passed over the Superose 12 column equilibrated with running buffer containing 25 mM BisTris buffer at pH 6.5 and 150 mM NaCl. The samples were eluted at a flow rate of 0.5 ml/min, and the fractions were monitored by absorbance at 280 nm. The column was calibrated using the following molecular mass standards: RNase A (13.7 kDa), chicken egg white ovalbumin (43 kDa), bovine serum albumin (67 kDa), (Amersham Biosciences), and bovine carbonic anhydrase (29 kDa) (Sigma-Aldrich). A standard curve for molecular mass was constructed by plotting molecular mass against elution volume. Molecular weight for injected p47 phox (151-340) was estimated in comparison with the standard curve.
Analytical Ultracentrifugation-Sedimentation equilibrium experiments were carried out using a Beckman Model XL-I analytical ultracentrifuge (Beckman Coulter, Inc.) equipped with both absorbance and interference optical detection system with an An-60 Ti rotor at 25°C. Loaded p47 phox (151-340) sample concentrations were 1.5, 2.8, and 4.2 mg/ml in 25 mM BisTris buffer at pH 6.5 and 150 mM NaCl. The protein samples were 110 l with 130 l of reference buffer in six-channel centerpiece. The data were collected at equilibrium for three different angular velocities: 12,000, 16,000, and 20,000 rpm. An interference fringe was measured using the Rayleigh interference optics. After 12 h of centrifugation, displacement of the interference fringes was compared at 2-h intervals to ensure that the sedimentation equilibrium was reached. Data analysis was carried out using the Beckman Optima XL-A/XL-I software, version 4.1(Beckman Corlter, Inc.) based on the Origin software (Microcal, Inc.).
Small Angle X-ray Scattering-SAXS data were collected on both p47 phox (151-340) and p47 phox (151-286) within a concentration range of FIG. 1. Overall structure of the intertwisted dimer of p47 phox in an autoinhibitory form. A, domain organization of p47 phox showing the phox homology (PX) domain, N-SH3, C-SH3, PBR/AIR, and the proline-rich region (PRR). The constructs investigated in this work (N-SH3 and C-SH3 with and without PBR/ AIR) correspond to the sequences 151-340 and 151-286, respectively. B, ribbon diagram representation of the intertwisted dimer of p47 phox (151-340) (Protein Data Bank code 1UEC). One monomer is colored in red, and the other is blue. C, topology diagram of the secondary structure of the intertwisted dimer of p47 phox (151-340). The color coding is the same as that used in B. The arrows represent ␤ strands, and the rectangles represent helices. The secondary structural elements are named alphabetically starting from the N terminus. the proteins from 2 to 16 mg/ml to estimate the possible effects of protein concentration on the determination of structural parameters. All of the measurements were made using a SAXS diffractometer in BL-10C installed at the Photon Factory in Tukuba, Japan (21,22). The wavelength of the x-ray was 1.488 Å. The sample cell had a volume of 50 l and a 1-mm path length with quartz windows. The data acquisition time was 600 s for each measurement. The identical buffer solution to the sample was recorded to measure solvent scattering. Protein scattering was obtained by subtracting the solvent scattering as the background trace.
The scattering data were analyzed using the Guinier approximation I(Q) ϭ I(0) exp(ϪR g 2 Q 2 /3), where Q, R g and I(0) were the momentum transfer, the radius of gyration, and the intensity at a zero scattering angle, respectively (23). Q was defined as Q ϭ 4 sin/, where 2 and were the scattering angle and wavelength of the x-rays, respectively. The I(0) and R g values were calculated using the intensity of zero angle (ln(I(0)) and slope (Ϫ R g 2 /3) by linear extrapolation of the Guinier plots in the range of Q ϫ R g Ͻ 1.8 (22)(23)(24). To estimate the relative molecular weight of scattering species, the zero angle intensity I(0) was scaled to the relative molecular weight using the scattering data for bovine carbonic anhydrase (Sigma-Aldrich), a monomeric protein with a molecular mass of 29 kDa.
The distance distribution function, P(r), defined by P(r) ϭ 1/2 2 ͐I(Q)Qr sin(Qr) dQ, corresponds to the distribution of distance, r, between the volume element. P(r) was calculated by the indirect Fourier transform algorithm using the GNOM program (25). The Q ranges used in the P(r) analysis were from 0.02 to 0.30 Å Ϫ1 for p47 phox (151-340) and from 0.03 to 0.19 Å Ϫ1 for p47 phox (151-286). The R g values, calculated as R g 2 ϭ ͐r 2 P(r) dr/(2 ͐P(r) dr), were estimated from the distance distribution function, P(r). The D max was a maximum dimension. Structural parameters were derived using both Guinier analysis and the distance distribution function, P(r). The distance distribution functions using the crystal structure coordinates (Protein Data Bank code 1UEC) were estimated from the frequency of distances among carbon, nitrogen, and oxygen atoms in a 1-Å interval (24). The R g was calculated from the coordinates. The maximum dimension, D max , was derived from the point at which P(r) approaches 1 ⁄100 of the maximum intensity.
Low resolution models of p47 phox (151-340) were generated from experimental scattering data by the ab initio shape determination program DAMMIN (26). DAMMIN calculates a volume of a protein filled with densely packed spheres (dummy atoms) to fit the experimental scattering data by a simulated annealing minimization procedure. The scattering data in the range of the momentum transfer with a Q value of 0.02-0.30 Å Ϫ1 were used for the fit. Ten independent fits were run with DAMMIN. The independent models were superimposed using the program SUPCOMB (27) and averaged by the program DAMAVER (28), highlighting common structural features.
NMR Spectroscopy-The sequence specific resonance assignments of p47 phox (151-340) were previously described (29). A NMR sample containing 1.0 mM 15 N-labeled p47 phox (151-340) in 25 mM BisTris buffer, 150 mM NaCl, pH 6.5, in 90% H 2 O, 10% D 2 O was utilized for the steady state heteronuclear 1 H-15 N NOE experiments. The NOE spectra were recorded on a Varian Unity plus 600 MHz NMR spectrometer at 25°C using a sensitivity enhanced technique with pulsed field gradients (30). Four sets of the spectra were acquired using a 3.0 s relaxation delay in the experiment.
The 15 N-labeled sample containing ϳ0.3 mM p47 phox (151-340) protein in 25 mM BisTris buffer (pH 6.5) and 150 mM NaCl was lyophilized from water and dissolved into 99.9% D 2 O. After leaving the sample for ϳ15 min to allow temperature equilibration, 1 H-15 N HSQC spectra were recorded under identical conditions at different time intervals. The total measurement time for each HSQC spectrum was ϳ30 min.
Two NMR samples were prepared to measure 1 H-15 N residual dipolar couplings in 500 l of 90% H 2 O/10% D 2 O (pH 6.5) in Wilmad sample tubes, the isotropic sample, and the aligned sample. The isotropic sample contains 1 mM protein dissolved in 25 mM BisTris and 150 mM NaCl. The aligned sample contains 0.3 mM protein dissolved in isotropic sample buffer with an additional 5% C12E5 polyethylene glycol/nhexanol mixture with a molar ratio of surfactant to alcohol of 0.96 (31). The weak alignment was established throughout the measurements because the deuterium signal was observed as a sharp doublet with ϳ24 Hz splitting at 25°C before and after the measurements. One-bond 1 H-15 N coupling constants were measured on a Varian Unity INOVA 800 MHz NMR spectrometer at 25°C using two-dimensional 1 H-15 N IPAP-type sensitivity enhanced HSQC (IPAP HSQC) spectra with inphase (IP) or anti-phase (AP) selections (32,33). The weighted sum and difference of the IP and AP IPAP HSQC spectra yielded spectra displaying only the up-field and down-field 15 N doublet components. The residual dipolar coupling (RDC) values were obtained by subtracting the observed coupling values of the isotropic sample from those of the aligned sample.
All of the two-dimensional NMR experiments were carried out using 256 and 1024 complex points in t 1 and t 2 , respectively. Final data sets comprised 1024 and 4096 real points with a digital resolution of 2.9 and 2.7 Hz/point in F 1 and F 2 , respectively. All of the pulse sequences were a modified version of the Varian Protein Pack (www.varianinc.com).
FIG. 2. Molecular mass of p47 phox (151-340) estimated from the size exclusion chromatography and the sedimentation equilibrium method. A, analytical size exclusion chromatography profile for p47 phox (151-340) at an initial loading concentration of 60 g. The inset shows the calibration curve for molecular mass using standard proteins: bovine serum albumin (dimer, 134 kDa, point 1; monomer, 67 kDa, point 3), ovalbumin (dimer, 86 kDa, point 2; monomer, 43 kDa, point 4), carbonic anhydrase (29 kDa, point 5), and RNase A (13.7 kDa, point 6). p47 phox (151-340) elutes at a position corresponding to a molecular mass of ϳ24 kDa B, sedimentation equilibrium profile for p47 phox (151-340) at an initial loading concentrations of 4.2 mg/ml. The samples in 25 mM BisTris buffer at pH 6.5 and 150 mM NaCl were centrifuged at 25°C for 12-16 h until equilibrium was achieved (see "Experimental Procedures") at 12,000, 16,000, and 20,000 rpm, respectively. Fringe displacements at sedimentation equilibrium are shown as a function of the radial cell position from the axis of rotation (lower panel). Nonlinear least squares fits to a model that assumes a single ideal species are overlaid onto these data sets (lower panel). Residuals for the nonlinear least square fits at three different rotor speeds are plotted versus the distance from the axis of rotation (upper panel). Fitting yielded apparent average molecular mass of 21,400 Ϯ 800 Da for p47 phox (151-340).
NMR spectra were processed using VNMR (Varian Instruments, Palo Alto, CA), or NMRPipe (34). TALOS software was used to predict backbone dihedral angles (35). Fitting of the dipolar couplings to the structure was made using the Module program (36). All of the structure figures were prepared using PyMOL (37).

RESULTS AND DISCUSSION
Molecular Mass in Solution-Because the tandem SH3 domains in the autoinhibited form of p47 phox (p47 phox (151-340)) exist as an intertwisted dimer in the crystalline state, we first investigated whether p47 phox (151-340) could form a dimer or not in solution. Analytical size exclusion chromatography of purified p47 phox (151-340) indicated that the protein forms a single species with no evidence of aggregation ( Fig. 2A). Calibration with molecular weight standards revealed that p47 phox (151-340) corresponded to a globular protein of ϳ24 kDa, roughly consistent with the molecular mass calculated by the primary amino acid sequence (ϳ22 kDa).
The sedimentation equilibrium experiments were subsequently performed, which give a molecular weight independent of a molecular shape. The sedimentation equilibrium data for We estimated an average molecular mass of 21400 Ϯ 800 Da for p47 phox (151-340), which is within an experimental error of the molecular mass of 21989 calculated for the p47 phox (151-340) monomer from its amino acid sequence. A higher order oligomer or molecular aggregate was not detected.
Molecular Mass and Average Size by Small Angle X-ray Scattering-The molecular masses and average sizes of p47 phox (151-340) and p47 phox (151-286) were estimated by SAXS analysis. The measured SAXS profiles showed obvious differences between p47 phox (151-340) and p47 phox (151-286) ( Fig. 3A and 3B). Because the SAXS profile is sensitive to the size and shape of scattering molecules, the difference was attributed to the difference in the structures of the two proteins. A more quantitative representation of the structural difference can be obtained from the analyses of the Guinier approximation of the scattering data, which provides two structural parameters, the radius of gyration (R g ) and the relative molecular mass (M r ) from the zero angle scattering intensity, I(0). The small angle regions of the Guinier plots (ln(I(Q)) versus Q 2 ) obtained from the scattering data were fitted to a single straight line (see "Experimental Procedures" for the definition of I(Q) and Q), as shown in Fig. 3 (C and D). These data sets for each protein were considered free of high molecular mass aggregates, because an upward shift in the lower angle region of the Guinier plots was not detected (22)(23)(24). Each of these data sets showed no concentration dependence in R g and I(0), indicating that no concentration dependent interaction existed. The relative molecular masses of p47 phox (151-340) and p47 phox (151-286) were scaled using the scattering data for bovine carbonic anhydrase, a monomeric protein with a molecular mass of 29 kDa. The molecular masses of p47 phox (151-340) and p47 phox (151-286) were estimated to be 22.5 and 15.4 kDa, respectively. These values are in good agreement with those calculated from the amino acid sequences. Considering the molecular mass estimated from the gel filtration analysis, the sedimentation equilibrium analysis, and the small angle x-ray scattering data, we concluded that both proteins exist as monomers in solution. Interestingly, R g of p47 phox (151-340) was found to be 19.3 Å in contrast to 25.4 Å of p47 phox (151-286), showing that the average molecular dimension of p47 phox (151-340) is much smaller than that of p47 phox (151-286) ( Table I).

Molecular Shapes and Dimensions of p47 phox (151-340) and p47 phox (151-286) Estimated from the Distance Distribution
Function-The distance distribution function, P(r) reveals the approximate histogram for the interatomic distances between carbon, nitrogen, and oxygen atoms in a molecule and directly gives information on the shape of the molecules. Therefore, P(r) provides a quantitative evaluation of the conformational properties of proteins in solution. P(r) functions of p47 phox (151-286) and p47 phox (151-340) were evaluated from a set of scattering data using the indirect Fourier transform method (25). The observed structural parameters from the P(r) analysis are summarized in Table I together with those estimated from the Guinier approximation. The P(r) functions of p47 phox (151-340) and p47 phox (151-286) are shown in Fig. 3 (E and F), respectively. The P(r) of p47 phox (151-340) showed a single Gaussianlike curve with a peak at 22 Å and a half-width of 28 Å. Because   FIG. 4. Ab initio shape analysis of p47 phox (151-340). A, experimental and calculated small angle x-ray scattering curves for p47 phox (151-340). The smooth curve in red corresponds to the scattering curve calculated from the dummy atom model derived using the program DAMMIN, which is superposed on the experimental curve (dots with error bars). B, low resolution model restored by the program DAMMIN, which is fit to the globular module of p47 phox (151-340) (ribbon diagram). The right model is rotated by 90°around the y axis, and the upper model is rotated by 90°around the x axis according to the arrows. These figures were prepared using PyMOL. 0.51 a R g is the radius of gyration, derived from the scattering data using Gunier approximation, the program GNOM, and the DAMMIN.
b The values in parentheses indicate standard deviations of the structural parameters estimated by SAXS analysis. c M r saxs and M r seq are molecular masses estimated from the scattering data and calculated from the primary sequence as a monomer. d D max is a maximum dimension. e Ab inito molecular shapes were calculated with the program DAM-MIN.
a Gaussian-like curve is characteristic to a spherical molecule, p47 phox (151-340) was considered a globular protein (Fig. 3E). Contrastingly, the P(r) of p47 phox (151-286) showed a curve with a peak located at nearly 25 Å, and the spread of the distribution curve extended to 80 Å (Fig. 3F), suggesting a relatively elongated structure. The R g values for p47 phox (151-340) and p47 phox (151-286) were also estimated to be 19.2 and 25.3 Å, respectively, based on analyses of the P(r) functions in good agreement with those calculated from the Guinier approximation (Table I). Thus, we concluded that p47 phox (151-340) takes a compact structure, in contrast to the extended structure of p47 phox (151-286).
Ab Initio Shape Analysis of p47 phox (151-340) by SAXS-The low resolution model of p47 phox (151-340) was determined by ab initio molecular shape analysis using the simulated annealing program DAMMIN (26) based on the scattering data of p47 phox (151-340). It is important to note that this shape analysis does not provide a unique solution but gives an ensemble of possible solutions consistent with the SAXS data. Ten independent runs of ab initio analysis were performed, all of which yielded similar results, considering the structural similarity of the calculated models. The models were superimposed, and an averaged model was constructed. Although all of the calculated models yielded nearly identical scattering curves, the best model provided a fit to the experimental data with 2 ϭ 0.51 for p47 phox (151-340), as shown in Fig. 4A. The averaged model shown in Fig. 4B roughly fits the globular module in the intertwisted dimer of the crystal structure, with respect to the molecular dimension and molecular shape. The R g and D max from ab initio analysis were in good agreement with those from the Guinier approximation and the distance distribution function analysis (Table I). Therefore, we concluded that p47 phox (151-340) is monomeric in solution and takes a compact and globular structure, consistent with the globular module in the crystalline state.

Comparison of Molecular Shapes for the Tandem SH3 Domains with PBR/AIR in Solution and in Crystal-The P(r)
functions for the intertwisted dimer and the globular module were approximately calculated as the pairwise distance distribution between carbons, nitrogens, and oxygens based on the crystal structure (Protein Data Bank code 1UEC). The P(r) calculated for the intertwisted dimer had two peaks at 19 and 55 Å, and the spread extended to more than 98 Å, whereas the P(r) calculated from the globular module showed a Gaussianlike curve with a single peak at 19 Å and a half-width of 24 Å, characteristic of a globular protein. Notably, the P(r) obtained from SAXS measurements for p47 phox (151-340) was similar to the Gaussian with a single peak at 22 Å (Fig. 3E). Considering the extreme difference in P(r) functions calculated from the intertwisted dimer and from the globular module, the SAXS data strongly supports the proposal that the molecular shape of p47 phox (151-340) in aqueous solution was quite similar to the globular module. We concluded that the globular module represents the structure of p47 phox (151-340) in aqueous solution. Furthermore, the intertwisted dimer in the crystalline state was not physiologically relevant but could be stabilized in the crystal lattice. This notion was further confirmed by NMR analyses of p47 phox (151-340).
Structural Properties for p47 phox (151-340) by NMR-NMR experiments were performed to evaluate the solution structure of p47 phox (151-340). The line width in the NMR spectrum suggested that p47 phox (151-340) existed as a ϳ22-kDa monomer rather than a ϳ44-kDa dimer, which is consistent with the results of the analytical size exclusion chromatography, the sedimentation equilibrium analysis and the SAXS analysis. The backbone 1 H and 15 N resonances of p47 phox (151-340) were assigned as previously reported (29) and were utilized as probes in the structural study of p47 phox (151-340) by NMR. According to the crystal structure, p47 phox (151-340) was divided into the following domains, N-SH3 domain (residues 159 -212), the linker connecting the N-SH3 and C-SH3 domains (residues 213-228), the C-SH3 domain (residues 229 -282), and PBR/AIR (residues 283-331) (Fig. 5A) (18,20). Backbone dihedral angles for p47 phox (151-340) were predicted using TALOS software (35). There is an appreciably good correlation between the backbone dihedral angles predicted by TALOS and those calculated from the globular module (Protein Data Bank code 1UEC), indicating that the overall structure in solution is similar to the globular module. In an effort to investigate the stabilities of the secondary structural elements of p47 phox (151-340) in solution, hydrogen-deuterium exchange measurements were applied using NMR. The 1 H-15 N HSQC spectra for monitoring the hydrogen-deuterium exchange rate of each amide proton enabled us to identify 63 amide protons with slow hydrogendeuterium exchange rates, which were derived from the stable secondary structural elements. These amide protons are located in both SH3 domains, whereas all of the amide protons in the linker and PBR/AIR in the autoinhibited form were completely exchanged with solvent deuterons before the NMR measurements started.
Backbone Dynamics of p47 phox (151-340)-In an effort to characterize the dynamics of each 1 H-15 N bond vector of p47 phox (151-340) in solution, the steady state heteronuclear 1 H-15 N NOE was measured using the uniformly 15 N-labeled protein. p47 phox (151-340) contains 10 proline residues. In addition, five residues, including 153, 189, 288, 324, and 326, overlapped considerably and were therefore excluded from NOE analysis. The 18 arginine residues are located distinctively in PBR/AIR of p47 phox (151-340), thus specifically 15 N-labeled protein at arginine residues was prepared to investigate the dynamic behavior of PBR/AIR. Fig. 5B FIG. 5. NMR characterization of the tandem SH3 domains with PBR/AIR in p47 phox . A, the domain organization and the secondary structure units of p47 phox (151-340) defined from the crystal structure (Protein Data Bank code 1UEC). The N-SH3 and C-SH3 domains are shown in blue and green. The arrows and the rectangles represent the ␤ strands and the helices, respectively. B, NOE values plotted as a function of the residue numbers. The blue bars indicate the NOE values for arginine residues observed for the specifically 15 N-labeled proteins at the main chain arginine residues. summarizes a plot of the NOE values for each amino acid residue, where those of the arginine residues are indicated with blue bars.
The residues with large NOE values (0.7-1.0) were expected to have lower flexibility and to be located in the rigid core, whereas those with small or negative NOE values (Ͻ0.6) were expected to be located at the loop or linker region with higher flexibility (24,38). The present measurements showed that N-SH3 (159 -212), C-SH3 (229 -282), and the residues 298 -330 of PBR/AIR had large, positive NOE values (more than 0.7), FIG. 6. Characterization of the alignment tensor for p47 phox (151-340). A and B, superposition of selected regions from the two-dimensional 1 H-15 N IP-AP HSQC spectra of p47 phox (151-340). The marked splitting in the isotropic spectrum (A) corresponds to 1 J NH , and the marked splitting in the partially aligned spectrum (B) corresponds to the sum of 1 J NH and the residual dipolar coupling. 1 H-15 N RDCs were determined by subtracting the splittings in the isotropic spectrum from those in the partially aligned spectrum. C and D, correlations between experimentally measured and the back-calculated 1 H- 15  demonstrating that in solution both SH3 domains and PBR/ AIR (298 -330) behaved as a folded structural core. In addition, the large NOE values were observed for linker residues 213-228, indicating that the linker was also involved in the structural core. However, the N-terminal region of PBR/AIR (residues 286 -295) had slightly smaller NOE values (ϳ0.3-0.6), indicating that this region was flexible. The N-terminal and C-terminal regions displayed smaller positive (Ͻ0.6) or negative NOE values, showing that these regions were flexible. In summary, the region of residues 159 -333 of p47 phox (151-340) comprising two SH3 domains, the linker, and PBR/AIR was a single folded unit in solution.
The Orientation of the SH3 Domains in the Autoinhibited Form Delineated by the Analysis of Residual Dipolar Couplings-In liquid crystalline media, a weak molecular alignment was induced by a static magnetic field, resulting in 1 H-15 N RDCs in the backbone amide groups (39). RDC is a sensitive probe for the structure in solution and has been utilized as a direct and simple tool to evaluate the consistency between solution structures and crystal structures. Recently, the analysis of the alignment tensor from RDC values has been successfully applied in an effort to determine the relative domain orientations of multidomain proteins (40 -44).
The globular module in the crystal structure was investigated by analysis of the alignment tensor for the RDC values of p47 phox (151-340) to determine whether it represents the solution structure. The highly flexible residues based on NOE experiments (NOE values Ͻ 0.65) were excluded from the analysis, including residues 150 -159 (the N-terminal region); 199 -201 (the distal loop in N-SH3); 269 -271 (the distal loop in C-SH3); 284 -297, 307-308, and 314 (PBR/AIR); and 334 -340 (the C-terminal region). The residues involved in the structural core were subsequently applied to the RDC analysis. The RDC values for 126 residues (40 residues for N-SH3, 46 residues for C-SH3, and 40 residues for the linker and PBR/AIR) were obtained by subtracting the coupling values of the isotropic sample (Fig. 6A) from those of the aligned sample (Fig. 6B), which ranged from Ϫ33 to 37 Hz. The experimentally measured RDCs of p47 phox (151-340) were then analyzed in an effort to search for the principal axes and the principal values of the alignment tensors.
The relative orientation of the two SH3 domains in solution might be different from that in the globular module because of the strand-exchanged dimer formation in crystal (Fig. 1, B and  C), therefore each SH3 domain was assumed to have an individual alignment tensor. The alignment tensor and the RDC values, back-calculated from the structure of each SH3 domain, were analyzed to fit the observed RDC values. The error function, 2, the measure of the agreement between the RDCs from the NMR experiments and those calculated from the model, were minimized to find the alignment tensor that fit the experimental data. The correlation between the experimental and the back-calculated RDC values for each SH3 domain is shown in Fig. 6C. Strong correlations between the experimental and the back-calculated RDCs for both the N-SH3 and C-SH3 domains were obtained. In addition, a close similarity between both alignment tensors was noticeable (Fig. 6, C and E, and Table II). These results indicate that the two SH3 domains have a common alignment tensor.
The single alignment tensor of p47 phox (151-340) was then searched, and the back-calculated RDC values were compared with the experimental RDC values for the globular module, as shown in Fig. 6D. The principal axes of the alignment tensor and the tensor parameters are also shown in Fig. 6F and Table  II. Low 2 values were obtained using the structure of the globular module including a linker and PBR/AIR. The experimental RDC values are in good agreement with the backcalculated RDCs, with a correlation coefficient of 0.967 for the globular module (Fig. 6D). The globular module was determined to represent a solution structure in which RDC values could be explained by a single alignment tensor.
Monomer-Dimer Conversion of p47 phox  in Solution-The molecular mass of p47 phox (151-340) derived from analyses for gel exclusion chromatography ( Fig. 2A), sedimentation equilibrium (Fig. 2B), and SAXS (Fig. 3C) was in good agreement with the molecular mass calculated from the amino acid sequence. However, both the monomer (ϳ95%) and dimer (ϳ5%) fractions were observed at concentrations of 5-10 mg/ml (25 mM BisTris buffer, pH 6.5, and 150 mM NaCl) when p47 phox (151-340) was applied to a gel filtration column in the purification process. To investigate the possibility of a conversion from monomer to dimer, the purified monomer fraction (concentration, ϳ25 mg/ml) was reapplied to the gel exclusion chromatography column. Conversion from the monomer to dimer was not observed, suggesting that p47 phox (151-340) was stable as a monomer, and the monomer-dimer conversion was quite slow under these conditions. Unexpectedly, p47 phox (151-340) formed the strand-exchanged and intertwisted dimer in the crystalline state (Protein Data Bank code 1UEC), although only monomeric species were applied for crystallization (19). The energy barrier for the monomer-dimer transition appeared to be relatively high, and the time scale for the conversion was long based on our observation; the intertwisted dimer of p47 phox (151-340) was crystallized over one month (19). The intertwisted dimer of p47 phox (151-340) was stable under crystallization conditions.
The two crystal structures of the monomer and dimer of the SH3 domain of Eps8 were reported under different crystalliza- a The globular module is the half of the intertwisted dimer splitted at the hinge. The globular module comprises the residues 156 -199 of the N-terminal moiety from one monomer and residues 200 -336 of the C-terminal moiety from the other. The N-SH3 and the C-SH3 were defined as residues 157-212 and residues 229 -282, respectively. b denotes the alignment tensor rhombicity, Aa/Ar, where Aa and Ar denote the axially symmetric rhombic component of the traceless alignment tensor in the principal tensor frame.
c The Euler angles, ␣, ␤, and ␥ define the rotation that transforms the molecular frame into the principal tensor frame. d N is number of residues. e The dipolar coupling R factor (R dip ) was used to access the quality of the derived alignment tensor (47,48). The validity of the structural data obtained here is justified by the low values of R dip , indicating good agreement between the measured RDC values and the calculated RDCs from the structure. f R is Pearson correlation coefficient between the observed and back-calculated RDC values.
tion conditions (45,46). Both of the monomer and strandexchanged region of the dimer took a canonical SH3 fold, similar to p47 phox (151-340) (Protein Data Bank codes 1AOJ, 1I07, and 1I0C). However, in contrast to p47 phox (151-340), the n-Src loop of the SH3 domain in Eps8 was extended to form the hinge. Thus, the canonical SH3 fold tends to form the strandexchanged dimer by taking an extended conformation at the n-Src or distal loop. The Structural Properties of the Tandem SH3 Domains in the Autoinhibitory Form in Solution-The SAXS and NMR analyses revealed that the structure of p47 phox (151-340) in solution was consistent with that of the globular module in the crystal structure. Because of the extended structure of PBR/AIR in the globular module, the buried surface area with the two SH3 domains was as large as 3625 Å (18,19). The linker and the C-terminal region of PBR/AIR (residues 297-331) showed large positive NOE values with averages of 0.76 and 0.74, suggesting that these regions were involved in the structural core in solution. The linker seems to play a critical role in maintaining p47 phox (151-340) in the autoinhibited structure, where a number of interactions were observed among the SH3 domains and the linker in crystalline state (18,20). However, considering that p47 phox (151-286) shows a rather extended structure in solution, possibly because of the flexible nature of the linker, the linker-SH3 domain interactions are not thought to be tight enough to maintain the tandem SH3 domains in the autoinhibitory conformation. PBR/AIR bundles the two SH3 domains and the linker into the globular module, which significantly contributes to the stability of p47 phox (151-340), as shown in the crystalline state. The hydrogen-deuterium exchange experiments demonstrated that the globular module was not a static single folded unit because the amide protons on the linker and PBR/AIR were readily exchanged with deuterons. We were able to conclude that the globular module was maintained in solution by synergistic interactions between the SH3 domains, the linker, and PBR/AIR, but these interactions were dynamic with local fluctuations. The dynamic nature of such interactions is crucial for rearrangements of both of the SH3 domains to enable ligand exchange with p22 phox proline-rich region when p47 phox is activated by phosphorylation of specific serine residues in PBR/AIR. The phosphorylation of the serine residues in PBR/AIR could destabilize and sequentially release the intramolecular interactions and induce changes in the relative position and orientation of both SH3 domains, which would thereby enable efficient binding of p47 phox to p22 phox prolinerich region.
Conclusion-Both of the analyses on the molecular shape and size by SAXS and on the alignment anisotropy and the relaxation behavior by NMR have revealed the first solution structure of the tandem SH3 domains in the autoinhibited form. The structure is in good agreement with the globular module, the split half of the intertwisted dimer in the crystalline state (18,20). These results support the possibility that the physiologically relevant structure is similar to the globular module but not the intertwisted dimer. Considering that the isolated tandem SH3 domains without PBR/AIR did not form a compact globular structure but exhibited an extended and flexible structure, the PBR/AIR plays a crucial role in bundling the SH3 domains and the linker into a compact globular structure. The synergistic interactions between the SH3 domains, the linker, and PBR/AIR for maintaining the autoinhibited form were neither stable nor static but rather in dynamic equilibrium with local fluctuations. These structural properties in solution would endow the tandem SH3 domains with a critical role as a molecular switch controlled by phosphorylation of specific serine residues.