Architecture of the p40-p47-p67 phox Complex in the Resting State of the NADPH Oxidase

The phagocyte NADPH oxidase is a multiprotein enzyme whose subunits are partitioned between the cytosol and plasma membrane in resting cells. Upon exposure to appropriate stimuli multiple phosphorylation events in the cytosolic components take place, which induce rearrangements in a number of protein-protein interactions, ultimately leading to translocation of the cytoplasmic complex to the membrane. To understand the molecular mechanisms that underlie the assembly and activation process we have carried out a detailed study of the protein-protein interactions that occur in the p40-p47-p67 phox complex of the resting oxidase. Here we show that this complex contains one copy of each protein, which assembles to form a heterotrimeric complex. The apparent high molecular weight of this complex, as observed by gel filtration studies, is due to an extended, non-globular shape rather than to the presence of multiple copies of any of the proteins. Isothermal titration calorimetry measurements of the interactions between the individual components of this complex demonstrate that p67 phox is the primary binding partner of p47 phox in the resting state. These findings, in combination with earlier reports, allow us to propose a model for the architecture of the resting complex in which p67 phox acts as the bridging molecule that connects p40 phox and p47 phox .

The phagocytic NADPH oxidase is a multiprotein enzyme that catalyzes the reduction of molecular oxygen to superoxide in response to invasion of the body with bacterial, fungal, and viral pathogens. Superoxide anions are precursors of a variety of reactive oxygen species that are used for killing of the microorganisms (reviewed in Refs. [1][2][3][4]. The importance of the NADPH oxidase in host defense is exemplified by the inherited disorder chronic granulomatous disease in which patients suffer from recurring infections due to a defect in oxidase activity. The NADPH oxidase consists of six subunits. Four of these, p40 phox , p47 phox , p67 phox , 1 and the small GTPase Rac are cytosolic in unstimulated cells, whereas p22 phox and gp91 phox form a heterodimeric, membrane-bound flavocytochrome, also known as cytochrome b 558 . In resting cells, p40 phox , p47 phox , and p67 phox exist as a tight cytosolic complex of undefined stoichiometry that can be purified by gel filtration chromatography with an apparent molecular mass of 250 -300 kDa (5)(6)(7)(8). Activation of the NADPH oxidase is initiated by phosphorylation, which is believed to induce conformational changes that subsequently lead to rearrangements in intra-and intermolecular interactions in the p40-p47-p67 phox complex (9 -14). These events culminate in translocation of this complex to the membrane and association with both Rac-GTP and cytochrome b 558 to form the active enzyme.
p40 phox , p47 phox , and p67 phox are multidomain proteins that contain SH3 protein-protein interaction domains (see Fig. 1). Additionally, p40 phox and p47 phox each contain a PX domain, which has recently been shown to bind to phosphatidylinositols and seems to act as a membrane-targeting module (15)(16)(17)(18). Most PX domains identified so far also contain a Pro-X-X-Pro motif, which is the consensus target sequence for SH3 domain binding modules (19). This suggests that PX domains are bifunctional with the potential to coordinate membrane localization as well as protein assembly during signal transduction events. In support of this idea, it has been shown that the isolated PX domain of p47 phox binds to the second SH3 domain of p47 phox with an equilibrium dissociation constant (K d ) of ϳ50 M (20).
p47 phox plays a central role in the activation process due to its ability to bind to the cytoplasmic region of p22 phox , an interaction that is necessary for oxidase activity (21)(22)(23). p47 phox contains a PX domain, tandem SH3 domains, and a polybasic region followed by a proline-rich sequence within its C-terminal region (Fig. 1). Studies by various groups suggest that p47 phox exists in an autoinhibited conformation in the resting state. In this model, the tandem SH3 domains are masked due to an intramolecular interaction with a C-terminal segment whereas the PXXP motif present in the PX domain simultaneously interacts with the second SH3 domain (24,25). In vivo phosphorylation of multiple serine residues within the C terminus of p47 phox liberates the N-terminal SH3 domain and allows it to interact with a proline-rich region in p22 phox . This, in turn, initiates translocation of the cytoplasmic complex to the membrane and activation of the oxidase (10, 12, 21, 24 -28).
p67 phox is complexed with p47 phox in the cytosol of resting neutrophils, and this interaction is absolutely required for translocation of p67 phox to the membrane. Evidence for this is provided by the observation that neither p67 phox nor p40 phox are able to translocate in chronic granulomatous disease neutrophils lacking p47 phox (29 -31). Once at the membrane, p67 phox interacts with Rac and possibly cytochrome b 558 to support catalysis by a mechanism that is not fully understood (32)(33)(34)(35)(36)(37). On the other hand, no clear function has yet been attributed to p40 phox , and it can apparently act as an activator or inhibitor depending on the experimental system. Furthermore, there are suggestions that it might stabilize p67 phox and thus act as a general modulator of NADPH oxidase activity (38 -40).
Extensive efforts have been made to identify domains involved in the multiple protein-protein interactions and conformational changes that take place during NADPH oxidase activation. Nevertheless, the precise nature of the protein-protein interactions that occur during the different stages of the activation process and how these are affected by phosphorylation are still a matter of debate. In particular the C-terminal proline-rich region in p47 phox has been variously suggested to bind to SH3 domains present in p67 phox and p40 phox , as well as to the tandem SH3 domains present within p47 phox itself (21-23, 25, 41-45). Furthermore, there are conflicting reports concerning the architecture of the p40-p47-p67 phox complex in resting cells. In one scenario, p40 phox acts as an adaptor that holds p47 phox and p67 phox together whereas an alternative model assumes that p67 phox acts as a bridging molecule between p40 phox and p47 phox (1,29,39,41,44,46,47). A likely explanation for the detection of multiple binding partners for a particular region is that many previous studies have made use of isolated domains. It is possible that such an approach may exclude additional regions that contribute to specificity and/or affinity. In addition, many of the techniques used do not allow a precise quantitative assessment of the affinity of any particular interaction. For these reasons, the relative significance of one interaction with respect to another becomes difficult to assess. Nevertheless, it is possible that all interactions detected so far do take place at various stages of the activation process but occur in a sequential not a simultaneous manner.
Here we describe the thermodynamic and hydrodynamic characterization of protein-protein complexes of the cytosolic components of the NADPH oxidase. Using a combination of isothermal titration calorimetry, analytical ultracentrifugation, and gel filtration chromatography we demonstrate that p40 phox , p47 phox , and p67 phox form a trimeric protein complex with a 1:1:1 stoichiometry. Furthermore, we show unequivocally that p67 phox is the adaptor that links p40 phox and p47 phox in the resting complex, providing a solid framework upon which the activation process can be further investigated.
Dynamic Light Scattering-Dynamic light scattering (DLS) measurements were performed on a DynaPro-801 dynamic light-scattering instrument (Protein Solution). Samples were filtered through 0.02-m filters, and experiments were performed at sample concentrations of 0.5 to 2 mg/ml in the same buffer used for analytical ultracentrifugation measurements. Data were analyzed by autocorrelation, and the resulting autocorrelation function was then evaluated using single-exponential cumulant analysis. From this the translational diffusion coefficient was determined, which can subsequently be used to derive the hydrodynamic radius and thereby the molecular weight using an empirically derived relationship between the radius and molecular weights for globular proteins. All calculations were carried out using the software package supplied with the machine.
Analytical Ultracentrifugation-The protein partial-specific volume, solvent density, and viscosity were calculated using the SEDNTERP program (John Philo). Sedimentation equilibrium and velocity studies were carried out using a Beckman Optima XLA analytical ultracentrifuge. Prior to centrifugation protein samples were dialyzed exhaustively against a buffer containing 50 mM HEPES, 100 mM NaCl, 1 mM EDTA, and 2 mM ␤-mercaptoethanol adjusted to the pH used for purification. Equilibrium experiments were carried out on 110-l samples in an An-60ti rotor using six-channel centerpieces with protein absorbances of 0.8 OD, 0.5 OD, and 0.3 OD at ϭ 280 nm and three different rotor speeds (p67 phox at 12,000, 15,000, and 18,000 rpm; p40 phox and p47 phox at 10,000, 13,000, and 18,000 rpm; p40-p47 phox , p40-p67 phox , and p47-p67 phox at 7000, 8500, and 12,500 rpm; p40-p47-p67 phox at 6000, 8000, and 11,000 rpm). Radial scans of the absorbance at 280 nm (path length 1.2 cm) were measured at 0.001-cm intervals, at 15°C and 20-fold averaged. Scans indicating that equilibrium had been reached were used for analysis (typically around 20 h). Multiple data sets were analyzed by non-linear least-squares procedures provided in the Beckman Optima XL-A/XL-1 data analysis software, version 4.1.
Sedimentation velocity studies were carried out at protein concen- trations between 0.2 and 1 mg/ml and sedimentation boundaries were monitored at ϭ 280 nm. The data were analyzed using the program SVEDBERG, which directly fits sedimentation velocity profiles to give the sedimentation (s) and diffusion coefficient (D) (48). Values obtained in this way were then extrapolated to zero concentration and corrected to standard conditions to yield s 20,w and D 20,w . Hydrodynamic shape parameters, including frictional (f/f 0 ) and axial ratios (a/b) were calculated with the program SEDNTERP using the v-bar method.
Isothermal Titration Calorimetry-Isothermal calorimetric titrations were performed with a Microcal omega VP-ITC (MicroCal Inc., Northampton, MA). All proteins were dialyzed against ITC buffer (25 mM HEPES, 50 mM NaCl, 1 mM EDTA, 2 mM DTT; at the same pH used for purification), and experiments were performed at 15°C. Typically solutions of 10 -20 M of proteins or complexes in the cell were titrated by injection of a total of 290 l of 100-200 M of ligands (49). Heats of dilution of ligand into buffer were determined in control experiments and subtracted from the raw data of the binding experiment prior to data analysis. Data were fitted by least-squares procedures using the evaluation software, Microcal Origin version 5.0 provided by the manufacturer. The data were averaged over two to five ITC experiments. Titrations that followed a two-site binding isotherm were additionally carried out in the presence of 2 mM tris-(2-carboxyl)phosphine to ensure that the second binding site detected was not an artifact caused by the use of DTT.
Analytical Gel Filtration-Analytical gel filtration chromatography was carried out on a Superdex 200 HR 10/30 column (Amersham Biosciences, Inc.) equilibrated with 50 mM HEPES, pH 7.0, 100 mM NaCl, and 2 mM MgCl 2 at a flow rate of 0.4 ml/min.

Analytical Ultracentrifugation Studies
The proteins used in this study have been overexpressed in E. coli in a soluble form. They are monodisperse, as judged by dynamic light scattering, and can be highly concentrated apart from p40 phox , which shows a tendency to aggregate at higher concentrations. Similar constructs have been shown in previous studies to retain their biological activity as judged by cell-free NADPH oxidase reconstitution assays, indicating that the proteins are correctly folded upon expression in E. coli (9,12,27,33,37). Prior to studying complex formation between the cytosolic phox components, we investigated potential selfassociation of the individual proteins in sedimentation equilibrium experiments. Analysis of these data can be used to derive the true molecular weight, without making any assumptions about the form and shape of the molecule(s) under investigation. Nine data sets at three different speeds and three concentrations were collected for each protein. Global fitting of these data sets to a single species resulted in a random distribution of residuals indicating that all three proteins exist as mono-mers in solution ( Fig. 2A). The molecular masses determined for p40 phox , p47 phox , and p67 phox are 41,036, 43,892, and 61,621 Da, respectively, in good agreement with the formula molecular masses and those determined by electrospray mass spectrometry (39,966,44,963, and 60,015 Da) (Table I). p67 phox has previously been described as forming dimers based on gel filtration chromatography and small angle neutron scattering studies (45). To investigate if this discrepancy with our data might be due to a non-globular shape of the protein, we carried out sedimentation velocity studies to derive an estimate for the asymmetry of p67 phox (Table II). The velocity data yielded a sedimentation coefficient (s 20,w ϭ 3.2 ϫ 10 Ϫ13 s) and diffusion coefficient (D 20,w ϭ 4.7 ϫ 10 Ϫ7 cm 2 s Ϫ1 ), which are lower than would be expected for a globular protein of 60 kDa (Table II).
The frictional ratio (f/f 0 ), which is an indication of the asymmetry of a protein, can be calculated from these data and has been found to be near 1.2 for globular proteins. The value calculated for p67 phox , however, is fairly high (f/f 0 ϭ 1.65), indicating that the shape of p67 phox significantly deviates from that of a globular protein. Modeling these results as a prolate ellipsoid yields an axial ration a/b of 12.1, suggesting that p67 phox adopts an elongated shape in solution (Table II). This conclusion is further corroborated by dynamic light scattering studies that showed p67 phox to be monodisperse (polydispersity indices 0.1-0.2) and yielded a translational diffusion coefficient  of 4.39 ϫ 10 Ϫ7 cm 2 s Ϫ1 , which was independent of the salt concentration from 50 to 500 mM NaCl. This value would be consistent with a molecular mass of 120 kDa in the case of a protein that holds a standard globular shape and hydration, close to that of a p67 phox dimer. Taken together, these hydrodynamic studies suggest that p67 phox adopts an extended and potentially flexible structure thus explaining its apparent dimeric behavior on gel filtration.

Formation of Dimeric Complexes between the Cytosolic Components of the NADPH Oxidase
The p40-p67 phox Complex-Isothermal titration calorimetry (ITC) allows direct measurement of the equilibrium binding constant K a (K d ϭ 1/K a ) and the enthalpy of complex formation (⌬H) without the need for producing fusion proteins that can be attached to a solid surface, or the introduction of radioactive or spectroscopic labels. Titration of p40 phox into p67 phox was exothermic (⌬H ϭ Ϫ5.4 kcal/mol) and resulted in the formation of a very tight complex with a dissociation constant of 10 nM and a stoichiometry of 1:1 ( Fig. 3A and Table III). This affinity is in reasonable agreement with that previously obtained by surface plasmon resonance studies (K d ϭ 43 nM (47)).
The p47-p67 phox Complex-The binding isotherm for the titration of p67 phox with p47 phox showed systematic deviations from a single-site binding model indicating that more than one binding event was taking place. Increasing the time between injections had no effect on this apparent biphasic behavior. A model assuming two independent binding sites resulted in a good fit and yielded dissociation constants of K d1 ϭ 20 nM and K d2 ϭ 150 nM and reaction enthalpies of ⌬H 1 ϭ Ϫ8.9 kcal/mol and ⌬H 2 ϭ Ϫ4.6 kcal/mol, respectively ( Fig. 3B and Table III). Only the high affinity binding site was fully occupied whereas the second binding site exhibited a stoichiometry of 1:0.25 (Ϯ0.1). The occurrence of a second binding site was rather surprising especially considering the low occupancy of this site. To investigate if p67 phox might bind more than one molecule of p47 phox we carried out gel filtration analysis of complexes between p47 phox and p67 phox at varying stoichiometries. Fig. 4 shows that a 1:1 mixture elutes as a single species with a retention time that is shorter than that of the individual proteins. In contrast, mixtures containing a molar excess of p47 phox or p67 phox elute as two species where the second species corresponds to the excess protein (data not shown). Because the ITC data indicated that the second binding site is only between 15 and 30% occupied, we tested if an excess of as little as 10, 20, or 30% of p47 phox might be incorporated into a p47-p67 phox complex. As is shown in Fig. 4, even these small amounts resulted in the appearance of an additional p47 phox peak, clearly indicating that p47 phox and p67 phox exist in a stoichiometric 1:1 heterodimeric complex. This conclusion is further corroborated by sedimentation equilibrium studies of a 1:1 complex of p47-p67 phox . Fitting of these data to a single species model resulted in a good fit, with residuals that were randomly distributed about the zero value, indicating that the complex is a monodisperse species. The molecular mass calculated from this fit was 105,160 Da, which is in very good agreement with its theoretically calculated molecular mass of 104,978 Da ( Fig.  2B and Table I). Taken together these observations suggest to us, that the second binding event that we have detected in the p47-p67 phox complex is not due to the binding of an additional molecule of p47 phox but must be an intrinsic feature of a stoichiometric p47-p67 phox complex (see "Discussion").
The p40-p47 phox Complex-p40 phox has been proposed to be the adaptor that links p47 phox to p67 phox in the cytoplasmic complex (41,44,45). However, titration of p40 phox with p47 phox did not result in any significant heat change suggesting that the two proteins either do not or only weakly interact, that the heat capacity of complex formation ⌬C p is such that ⌬H is very small at the experimental temperature, and/or that the interaction is mainly entropy driven. To discriminate between these possible explanations, further studies were carried out using a variety of techniques. In surface plasmon resonance experiments using the BIAcore system, p40 phox was directly coupled to a Ni-NTA chip via its C-terminal hexahistidine tag, and binding of p47 phox was monitored. Although there was clear net binding of p47 phox to p40 phox the data were not of sufficient quality to be able to determine an equilibrium constant from kinetic "on" and "off " rates or equilibrium binding analysis (data not shown). This poor data quality is likely due to the fact that the interaction of the hexahistidine tag of p40 phox with the Ni-NTA chip does not seem to be sufficiently tight, resulting in the protein being partially washed off the chip during the experiment. Sedimentation equilibrium runs of a 1:1 mixture of p40 phox and p47 phox and fitting of the data to a single species model yielded a molecular mass of 70,842 Da providing further evidence that the two proteins interact in solution. However, analysis of these data assuming a monomer-dimer equilibrium is complicated by the fact that the two proteins have different extinction coefficients at 280 nm (⑀ ϭ 39,400 M Ϫ1 cm Ϫ1 for p40 phox and ⑀ ϭ 57,750 M Ϫ1 cm Ϫ1 for p47 phox ), which makes it impossible to correctly convert the association constant, which is calculated in absorbance units into molar concentrations. Nevertheless, because the molecular weights of the two proteins are very close and there is only a 1.4-fold difference in extinction coefficients, we assumed a mean value of ⑀ 280 ϭ 48,575 M Ϫ1 cm Ϫ1 per monomer to estimate a lower limit of 4 M for the dissociation constant. This value is in very good agreement with a previous estimation for the affinity of this complex, which has been made based on small angle neutron scattering studies (45).

Formation of the Trimeric p40-p47-p67 phox Complex
To investigate whether the occurrence of a two-site binding isotherm upon interaction of p67 phox with p47 phox is a particular feature of this complex or whether it also occurs if p67 phox is already complexed with p40 phox , we titrated a purified p40-p67 phox complex with p47 phox . Data from this titration could, again, only be fitted to a two-site binding model with dissociation constants of 14 and 128 nM and reaction enthalpies of Ϫ7.5 and Ϫ4.6 kcal/mol, respectively (Fig. 3C). As observed before, the second binding site exhibited a stoichiometry of 1:0.25 (Ϯ0.1). The similarity between results from this titration and that of p67 phox with p47 phox suggests that complex formation between p47 phox and p67 phox takes place independently of the presence of p40 phox . In addition, it confirms that p40 phox and p47 phox do not share a common binding site on p67 phox . As observed for the p47-p67 phox complex, a 1:1:1 mixture of p40 phox , p47 phox , and p67 phox elutes as a single species on gel filtration and addition of an excess of any of the components results in the appearance of an additional peak (data not shown). Sedimentation equilibrium analysis of a 1:1:1 complex showed that it behaves as a single, monodisperse species with a molecular mass of 144,546 Da in good agreement with its formula molecular mass of 144,944 Da (Fig. 2C). DLS measurements of this complex confirmed its monodispersity (polydispersity index 0.3), however, a diffusion coefficient of 3.44 ϫ 10 Ϫ7 cm 2 s Ϫ1 determined by single-exponential cumulant analysis, which would be consistent with a molecular mass of 240 kDa in the case of a globular protein indicates that the shape of this complex deviates significantly from that of a globular protein. This observation is substantiated by sedimentation velocity studies from which a frictional ratio f/f 0 ϭ 2.02 and an axial ratio a/b ϭ 20.6 were calculated (Table II). These values are comparatively high for a protein but can be rationalized by the fact that all three components of this complex are multidomain proteins made up of individual domains that are con-nected by linkers that are likely to be rather flexible. This results in a heterotrimeric complex that adopts a highly elongated shape due to the asymmetry of one of its components (p67 phox ) and due to its flexible organization.

Protein-Protein Interactions in the p47-p67 phox Complex
To further characterize the structural features of complex formation between p47 phox and p67 phox and to understand the reason for the occurrence of a biphasic binding isotherm, we have produced various N-and C-terminally truncated p47 phox and p67 phox fragments and assessed their behavior in ITC titrations in comparison to the full-length proteins.
The proline-rich region in the C terminus of p47 phox has invariably been suggested to be the target for the SH3 domain in p40 phox as well as C-terminal SH3 domain in p67 phox . To investigate if the tight, stoichiometric binding we detect in titrations of p67 phox with p47 phox is due to this interaction, we  measured complex formation between the isolated C-terminal SH3 domain of p67 phox and full-length p47 phox by ITC. This titration follows a single site binding isotherm with a 1:1 stoichiometry as would be expected for complex formation between an SH3 domain and its proline-rich target. Binding occurred with an affinity of 21 nM and a reaction enthalpy of Ϫ7.4 kcal/mol, indicating that this interaction indeed constitutes the high affinity binding site detected in titrations using the fulllength proteins. This conclusion is further substantiated by the fact that removal of the proline-rich region (p47 phox 1-354) completely abrogated complex formation. Interestingly, further C-terminal truncation of p47 phox (fragment 1-295) to remove the polybasic region, which is thought to be responsible for the intramolecular masking of the tandem SH3 domains, restored binding and resulted in the formation of a 1:1 complex with a dissociation constant of 6 M. This interaction results from a favorable enthalpy change, which is of similar size to that observed for the interaction between full-length p47 phox and p67 phox . However, in contrast to that interaction the entropy change exhibited here is unfavorable, resulting in a 300-fold lower affinity (Table III). To investigate if the N-terminal region in p67 phox , including the TPR domain and the proline-rich region around amino acids 219 -234 might contribute to the apparent biphasic behavior of p47-p67 phox complex formation, we deleted this region (p67 phox 234-end) and measured binding to full-length p47 phox by ITC. This construct behaved almost identically to full-length p67 phox . The titration had to be fitted to a two-site binding model and exhibited a tight, stoichiometric binding site with a K d of 8 nM and ⌬H of Ϫ8.8 kcal/mol, whereas the second binding event again showed an apparent stoichiometry of 1:0.25 (Ϯ0.1). Removal of the N-terminal portion of p47 phox , including the PX domain (p47 phox 155-end) had a similar effect, resulting in a two-site binding isotherm with a K d of 12 nM and reaction enthalpy of Ϫ9.8 kcal/mol for the first site and a low stoichiometry occupancy for the second. The similarity of these titrations with those of full-length p47 phox and p67 phox but also p67-SH3 B suggests to us that tight, stoichiometric complex formation between p67-SH3 B and the proline-rich region of p47 phox occurs independently of the remainder of the two proteins and is not influenced by the second binding event detected in these titrations.

DISCUSSION
The NADPH oxidase is a multiprotein enzyme whose activity is regulated by the reversible formation of multiple proteinprotein interactions between the cytosolic and membrane components as well as between the cytosolic proteins themselves. Many of these interactions involve SH3 domains and their target proline-rich sequences, which are present in p22 phox , p40 phox , p47 phox , and p67 phox . Activation of cells by appropriate stimuli leads to phosphorylation of p47 phox , which in turn induces intra-and intermolecular rearrangements within a number of these protein-protein interactions. This study was aimed at carrying out a detailed quantitative characterization of the protein-protein interactions, which occur in the resting state of the NADPH oxidase to establish a basis upon which the activation process can be investigated.
p40 phox , p47 phox , and p67 phox exist as a tight complex in the cytosol of resting neutrophils from which it can be purified by gel filtration with an apparent molecular mass of 250 -300 kDa, twice that expected for a complex containing one copy of each of the proteins. Sedimentation equilibrium studies presented here show that the individual proteins exist as monomers, which associate to form a monodisperse, 1:1:1 complex. Addition of an excess of any of the individual proteins to this complex does not result in additional binding. The shape of this complex significantly deviates from that of a globular molecule as evidenced by dynamic light scattering as well as sedimentation velocity studies from which a fractional coefficient of 2.02 was derived. Thus the large molecular weight observed by gel filtration is not due to the existence of multiple copies of any of the proteins in this complex but is rather due to an extended shape and possibly flexible organization of the heterotrimer.
Extensive studies have been carried out to characterize the domains that mediate the protein-protein interactions within the p40-p47-p67 phox complex, but despite this wealth of data it is still unclear at which stage during the activation process a particular interaction occurs. Our data show that p40 phox and p67 phox form a very tight complex with a dissociation constant in the low nanomolar range. This interaction has been previously mapped to a region between the two SH3 domains in p67 phox , which contains the recently identified BP1 domain and its target sequence, the PC motif (phox and Cdc; also known as the OPR motif)), which resides in the C-terminal part of p40 phox (50,51). p40 phox has often been described as the primary binding partner for p47 phox in the resting state due to an interaction of its SH3 domain with the Pro-rich region in the C terminus of p47 phox . Consequently, many models for NADPH oxidase architecture assume that p40 phox is the link between p47 phox and p67 phox . However, this Pro-rich region in p47 phox has also been described to bind to p67 phox and possibly, in an intramolecular fashion, to its own tandem SH3 domains. Because a single, short proline-rich motif can only bind to a single SH3 domain at any one time, we carried out ITC studies to determine the affinities of the individual oxidase components for one another. We show that the affinity of p40 phox for p47 phox is comparatively low, in the micromolar range, in accordance with neutron scattering data, which estimated the affinity to be around 4.0 M (45). In contrast, p47 phox binds to p67 phox with nanomolar affinity. To our surprise this titration did not follow a single site binding isotherm but could only be fitted assuming two non-identical sites. Despite this behavior, sedimentation equilibrium analyses and gel filtration chromatography of p47 phox and p67 phox at various ratios showed convincingly that p47 phox and p67 phox form a 1:1 complex. This clearly indicates that the second binding event detected by ITC does not reflect an additional binding site. Interestingly, binding of p47 phox to a preformed complex of p40-p67 phox exhibited a similar behavior. The agreement, within experimental error, between the K d values and ⌬H values for the stoichiometric binding site determined in these two experiments implies that the presence of p40 phox has no influence on complex formation between p47 phox and p67 phox in the resting state. By carrying out ITC titrations using a C-terminally truncated p47 phox as well as the isolated C-terminal SH3 domain of p67 phox we could show that the tight interaction that we detected between full-length p47 phox and p67 phox is due to binding of the C-terminal Pro-X-X-Pro motif in p47 phox to p67-SH3 B . Taken together, these results clearly demonstrate that p40 phox can not be the link between p47 phox and p67 phox in the resting state, because the affinity of p67 phox for the Pro-rich region in p47 phox is about 1000-fold higher than that of p40 phox .
Our ITC data for complex formation between full-length p47 phox and p67 phox indicated that a second process is taking place in addition to interaction of the two proteins via their respective C termini. This process has no influence on the C-terminal SH3 domain-proline-rich region interaction. ITC titrations using N-terminally truncated versions of either protein indicated that neither the PX domain of p47 phox nor the TPR domain of p67 phox is responsible for this process. At present we are not able to explain the reason for this second event and can only conclude that it seems to require the presence of the tandem SH3 domains of p47 phox that are masked by an intramolecular interaction with the subsequent polybasic region (24,25) as well as the C-terminal half of p67 phox . Further studies are currently underway to explore this phenomenon.
Based on the data presented here combined with data from other reports we suggest the following model for the NADPH oxidase architecture in the resting state (Fig. 5). p67 phox is the bridge that connects p40 phox and p47 phox . No direct interaction takes place in this complex between p40 phox and p47 phox . Furthermore, association of p40 phox with p67 phox appears not to induce a conformational change in the p47 phox binding site of p67 phox , because binding of the latter to the C-terminal Pro-rich region of p47 phox follows a similar behavior in the absence or presence of p40 phox . p47 phox exists in an autoinhibited conformation in this complex in which its SH3 domains are masked by intramolecular interactions with a polybasic region in the C-terminal portion of the protein and possibly with the PX domain. Upon activation, phosphorylation induces a conformational change in p47 phox , which unmasks the tandem SH3 domains, thereby allowing binding to p22 phox and translocation to the membrane. Babior et al. (26) have shown that phosphorylation of Ser-359 or Ser-370 is absolutely required for translocation as well as oxidase activity. Interestingly, both of these serines are adjacent to the poly-proline motif, which is the target of the C-terminal SH3 domain of p67 phox , suggesting that phosphorylation might interfere with this interaction and may potentially induce the release of p67 phox rendering this sequence accessible for p40 phox . Our data show that a second binding site exists between p47 phox and p67 phox that is only accessible once the tandem SH3 domains of p47 phox have been unmasked. This site might become the primary interface between the two proteins if phosphorylation is capable of disrupting the "end-to-end" p47-p67 phox interaction. Further studies are now needed to describe in molecular detail the rearrangements that take place during the activation and assembly process.