Jeanette H. W. Leusen^1,2, Arthur J. Verhoeven¹ and Dirk Roos^1,3

¹ Central Laboratory of the Netherlands Red Cross Blood Transfusion Service and Laboratory for Experimental and Clinical Immunology, University of Amsterdam, Plesmanlaan 125, 1066 CX Amsterdam, The Netherlands.

²Department of Pediatrics, Emma's Children Hospital, Academic Medical Center, Amsterdam, The Netherlands.

Received 04/15/96; Accepted 05/05/96; On-line 07/01/96

5.1. Cell-free system

Many of the recent advances in our understanding of the components of the NADPH oxidase have been the result of studies employing a cell-free activation system, first developed by Heyneman and Vercauteren (81) and Bromberg and Pick (82), and later extended by other investigators (83-88). The cell-free NADPH system is comprised of neutrophil membranes (containing cytochrome b₅₅₈), and three neutrophil cytosolic fractions containing p47-phox, p67-phox and rac (89). The system is activated by GTP or its non-hydrolysable derivative GTPgS and an amphiphilic agent (SDS or arachidonic acid). NADPH is needed as a substrate for the NADPH oxidase. Unlike in intact cells, phosphorylation of p47-phox is not required in this system: inhibitors of PKC have no inhibitory effect in the cell-free system.

5.2. Important domains within cytochrome b₅₅₈

CGD patients with a single amino-acid substitution in one of the subunits of cytochrome b₅₅₈ have provided useful information about the structure and function of the cytochrome. For the A22⁺ patient described in ref. (50) the functional defect was obvious: the ¹⁵⁶Pro--->Gln substitution in the proline-rich region of p22-phox disrupts a putative counter structure for SH3 domains in p47-phox. Indeed, we found no translocation of p47-phox and p67-phox to the plasma membrane in either intact neutrophils or in the cell-free system of this patient (50). We also showed a normal electron flow within the cytochrome b₅₅₈ of this patient when neutrophil membranes were tested in a cell-free system without p47-phox and p67-phox, in which negatively charged phospholipids were added as activators for cytochrome b₅₅₈ (90, 91). This proves that 1) the mutation in p22-phox in the neutrophil membranes does not affect the transfer of electrons from NADPH via FAD and heme to oxygen, and 2) the cytosolic proteins act only as activators of the NADPH oxidase, as was already stated by Koshkin and Pick (90, 91). This means that the superoxide-generating capacity is fully contained within the cytochrome b₅₅₈. Most likely, the cytosolic proteins p47-phox and p67-phox induce by their interaction with cytochrome b₅₅₈ a conformational change in the cytochrome that permits binding of NADPH and start of the electron flow in this flavocytochrome. The neutrophil membranes from all four X91⁺ patients tested by us in this system (with negatively charged phospholipids but without p47-phox and p67-phox as activators of the oxidase) were incapable of generating any superoxide, regardless of the ability of the cytosolic proteins to bind to cytochrome b₅₅₈ in the usual cell-free system.

Figure 4. Localization of X91⁺ CGD mutations in gp91-phox. Schematic representation of the gp91-phox protein. Indicated are the possible oritentation of the peptide in the membrane, the N and C termini, the possible glycosylation sites (Y), the putative binding regions for FAD and NADPH, and the positions of the mutations that induce the X91⁺ CGD phenotype (normal expression of gp91-phox protein, no oxidase activity).

This may implicate that any mutation in gp91-phox destroys the catalytic activity of the flavocytochrome under all conditions tested, suggesting that gp91-phox alone is the catalytic unit, and that p22-phox is only required for stabilization of gp91-phox. Indeed, sequence-homology studies between the C-terminal half of gp91-phox and the ferredoxin-NADP⁺ reductase flavoenzyme family imply that both FAD and NADPH can be bound by this part of gp91-phox, and the location of these putative FAD- and NADPH-binding domains within gp91-phox have been deduced (2729) (see Fig. 4).

The model described above does not predict the position of the heme groups in cytochrome b₅₅₈, because ferridoxin reductases do not contain heme moieties. Also, the model only describes the C-terminal part of gp91-phox, and the hemes are most probably located in the N-terminal half of gp91-phox. In fact, thereare indications for two hemes for each cytochrome b₅₅₈ unit; one liganded by two histidines of gp91-phox (92) and another shared between the two cytochrome b₅₅₈ subunits (30), possibly involving histidine-239 of gp91-phox and histidine-94 of p22-phox (C.D. Porter et al., in press). A recent report on an X91⁺ CGD patient with an ⁵⁴Arg--->Ser substitution has substantiated the two-heme hypothesis (93, 94). This mutation abolishes the electron attraction of the positively charged arginine and perhaps disrupts a hydrogen bond between the arginine and a heme propionate side chain. As a result, the mutant form of the cytochrome contains two non-identical hemes with midpoint potentials of E_m7 = -220 and E_m7 = -300 mV. In the light of this information, the investigators reanalyzed the wild-type cytochrome b₅₅₈ and concluded that it also contains two separate heme centers, with midpoint potentials of E_m7 = -225 mV and E_m7 = -265 mV (94).

It is generally believed that the cytosolic components p47-phox and p67-phox have to interact with the membane-bound p22-phox and gp91-phox to induce NADPH oxidase activity. In the neutrophils of the patient with the ⁵⁴Arg--->Ser substitution in the Nterminal part of gp91-phox the cytosolic proteins translocated normally (93). Two other X91⁺ patients are known thus far with normal translocation of p47-phox and p67-phox: patient R.C. (42, 44, 64, 69) and a patient described in ref. (95). Patient R.C. carries a ⁴¹⁵Pro--->His substitution in a putative NADPH-binding region in gp91-phox (Fig. 4), which leads to decreased binding of NADPH to the oxidase (27). The second patient has a deletion of five nucleotides and an insertion of eight nucleotides in exon 12 of the CYBB gene, resulting in a replacement of ⁵⁰⁷Gln-⁵⁰⁸Lys-⁵⁰⁹Thr by His-Ile-Trp-Ala in gp91-phox. This mutation also resides in a putative NADPH-binding region (Fig. 4).

We have characterized three X91⁺ patients with a defective translocation of p47phox and p67-phox: patient D.S., described in ref. (44), patient C.E. and patient N.K. (J. Leusen et al., manuscript in preparation). All three patients had point mutations leading to an amino-acid substitution in the cytosolic, C-terminal part of gp91-phox. In two of the patients, the substitutions are situated in protein loops exposed to the cytosol, as predicted by a computer model of the C-terminal half of gp91-phox (96), viz. ⁵⁰⁰Asp--->Gly in patient D.S. and ³⁶⁹Cys--->Arg in patient C.E. (Figs. 4 and 5). We have established the domain around residue 500 as a binding region for either p47-phox or p67phox by blocking the translocation of both cytosolic proteins in a cell-free assay with a synthetic peptide with the sequence of this domain (44).

Previously, we have postulated (44) that p47-phox is the most likely candidate for interaction with the domain surrounding ⁵⁰⁰Asp of gp91-phox, because this protein has been described to translocate without p67-phox (69, 71), while the opposite (translocation of p47-phox independently of p67-phox) did not seem to occur (69). Because we found no translocation of either cytosolic protein to the neutrophil plasma membranes of patient D.S., it seemed logical to suppose a disturbed interaction of this gp91-phox domain with p47-phox. Furthermore, p47-phox contains several basic regions that might act as counter structures for the negatively charged ⁵⁰⁰Asp; it was observed that positively charged peptides also block NADPH oxidase assembly (97). However, upon neutrophil activation, p47-phox becomes phosphorylated at 6-8 serine residues, and only the two most acidic isoforms are associated with the membrane. Also, we found recently that p47-phox did not translocate in an A67⁰ patient (J. Leusen et al., submitted). Furthermore, there is good evidence that p47-phox interacts with p22-phox after activation of the oxidase, mediated via the SH3 domains of p47-phox (see next section) and the proline-rich region in p22-phox mutated in the A22⁺ patient (50). These observations render p67-phox a better candidate for binding to the domain in gp91-phox surrounding residue 500 (see also next paragraph). The loop containing residue 369 is involved in this interaction, because the two domains are quite close in the 3D model of gp91-phox (96) (Fig. 5).

Recently, it has been shown that p47-phox and p67-phox have distinct roles in the regulation of electron flow in cytochrome b₅₅₈ (98, 99). In the absence of p47-phox, p67phox alone can facilitate electron flow from NADPH to the flavin center, resulting in the reduction of FAD, whereas the presence of p47-phox is required for electron transfer to proceed beyond the flavin center to the hemes in cytochrome b₅₅₈ and then to oxygen. This fits with the idea suggested above that p67-phox is the cytosolic component that binds to the protein loops mutated in patients D.S. and C.E., because p67-phox would then be in close proximity to the NADPH-binding regions and the flavin center of gp91phox, enabling it to perform a regulatory function in this part of the protein.

The ⁴⁰⁸Gly, substituted for Glu in patient N.K., is localized between the two protein loops containing residue 500 and 369, and predicted to be buried in the protein (Fig. 5). Therefore, it is not a likely binding site for one of the cytosolic proteins. However, it is possible that the mutation changes the folding of the protein, perhaps leading to another orientation of the two loops with respect to each other. If this is true and one protein (p67-phox) binds to both loops, this interaction might be disturbed by the ⁴⁰⁸Gly--->Glu substitution.

A number of additional regions within gp91-phox have been identified as binding regions of p47-phox by inhibition studies with synthetic peptides in the cell-free system (oxidase activity or translocation) (70, 100) or by phage display and panning system with purified or recombinant proteins (101, 102). To substantiate the importance of these domains, synthetic peptides were used to assess inhibition of oxidase activation. However, the IC₅₀ of some of these peptides was sometimes relatively high (>10 µn;M), and scrambled peptides were not always used as a control. It is known that positively charged peptides are inhibitory under cell-free activation conditions with IC₅₀ values <2 µn;M (97, 103). Possibly, the effect of positively charged peptides is specific in the sense that they block the binding of p47-phox to domain ⁸⁶TRVRRQL⁹³ of gp91-phox (101). This domain, however, has been postulated by others to reside in an extracellular loop of the protein (see Fig. 4). Therefore, the specificity of these interactions remains to be proven, and genetic studies still provide the most convincing evidence for physiologically relevant interactions within the NADPH oxidase system.

5.3. Interactions between the cytosolic components

The cytosolic components of the NADPH oxidase appear to reside in a complex of 240-300 kD in the cytosol of resting neutrophils (104-106). Both p47-phox and p67-phox contain two SH3 domains, known to be involved in protein-protein interactions (47, 48, 107). P40-phox contains one SH3 domain (38). SH3 domains bind to proline-rich regions (49). Both p47-phox and p67-phox contain at least one such putative SH3-binding domain. As described in (108), p47-phox binds to p67-phox blotted onto nitrocellulose, and p67-phox binds to blotted p47-phox and p40-phox. We found that the proline-rich region in p67-phox is not required for interaction with either p47-phox or p40-phox. In contrast, the proline-rich region of p47-phox is essential for the binding to p67-phox, also when both proteins are in solution. In addition, the last 13 amino acids of p47-phoxseem to be involved in exposing this proline-rich region, because an antibody directed against the C-terminal 13 amino acids of p47-phox blocks the interaction between p47-phox and p67-phox, but only when p47-phox is in solution. A truncated p47-phox confirmed these findings. In Fig. 6, a hypothetical model is given of the interactions between the oxidase components, based on the data described in this section.

Figure 5. Model of the cytosolic part of gp91-phox in the activated, NADPH-bound state. The protein is modelled on Spinach ferrodoxin (157) as described by Taylor et al. (96). The figure was drawn with MOLSCRIPT (158), and kindly provided by Dr Nicholas Keep of prof. Segal's group (University College, London, G.B.). The position of the substitutions in patients C.E. (³⁶⁹Cys--->Arg), patient N.K. (⁴⁰⁸Gly--->Glu) and patient D.S. (⁵⁰⁰Asp--->Gly) are shown. Residues 369 and 500 are situated in protein loops exposed to the cytosol, but residue 408 is buried in the protein between these loops.

It is of interest to note that a synthetic peptide mimicking the last 13 amino acids of p47-phox becomes phosphorylated in vitro by PKC and other serine kinases (109). In combination with our findings, this indicates that the phosphorylation of the C-terminus could be an important switch for repositioning the cytosolic phox proteins, leading to their interaction with the membrane-bound oxidase components and thus to an active NADPH oxidase. Other investigators have shown that the C-terminal SH3 domain of p67-phox is the counter structure for the proline-rich region in p47-phox (110, 111) and that the SH3 domains of p47-phox can bind to proline-rich sequences in p47-phox itself and to the proline-rich sequence in p22-phox (111, 112). Addition of SDS or arachidonic acid, activators of the cell-free system, leads to exposure of the SH3 domains in p47-phox (112). In the cell-free system, the p47-phox without its proline-rich region (108) and p67phox without its SH3 domains (113) can still induce superoxide production. Hence, the interaction between p47-phox and p67-phox is not required for cell-free oxidase activity (108,113)). However, in intact cells (EBV transformed B-cell lines of CGD patients transfected with cDNA encoding the missing cytosolic component), the second SH3 domain of p67-phox that interacts with p47-phox is indispensable for oxidase activity (113), suggesting that a complex between p47-phox and p67-phox is required in intact cells.

With the use of the two-hybrid system the binding domains of p40-phox have also been mapped: the SH3 domain of p40-phox interacts with p47-phox, and its last 36 amino acids interact with p67-phox (114). The same method revealed that the C-terminal SH3 domain of p67-phox is crucial for the interaction with p40-phox and p47-phox (A. Fuchs, unpublished results). Thus far, there are no domains of p67-phox known to interact with cytochrome b₅₅₈ after activation.

Figure 6. Hypothetical model of the interactions between the NADPH oxidase components. The interactions between the cytosolic components are represented for the resting situation in panel A. For the activated state, the interactions between the cytosolic components and cytochrome b₅₅₈ are depicted in panel B. It is postulated that in the resting state, the C-terminus of p47-phox is folded backwards, binding or covering its SH3 domains and exposing the C-terminal proline-rich region, allowing it to bind to the C-terminal SH3 domain of p67-phox. The other two binding domains (one proline-rich region and one SH3 domain) of p67-phox are not involved in binding to one of the cytosolic proteins, but possibly interact with the cytoskeleton. It is known that the SH3 domain of p40-phox interacts with p47-phox (114), perhaps to its N-terminal proline-rich region, as is shown here. Upon cell activation, p47-phox becomes phosphorylated, and the phosphorylation of ³⁷⁹Ser has been shown to be crucial for activity and translocation (133). Therefore, we suggest here that phosphorylation of this site leads to uncovering of the SH3 domains of p47-phox, making them available for interaction with the proline-rich region of p22-phox (the idea that phosphorylation of p47-phox leads to exposure of SH3 domains was already suggested by Leto et al. (111)). This also disrupts the interaction with p67-phox. However, via binding to p40-phox, p67-phox is still retained in the complex. We also hypothesize that p67-phox binds to two protein loops containing residues 369 and 500 of gp91-phox, although the binding sites on p67-phox are still unknown. Rac can only bind after dissociation from rho-GDI and loading with GTP. To which domain of cytochrome b₅₅₈ rac binds, is unknown.

5.4. Role of phosphorylations in activation of the oxidase

Neutrophils stimulated with PMA or fMLP exhibit rapid phosphorylation of p47phox on multiple serine residues (115), which accompanies translocation of this protein to the submembraneous cytoskeleton. Protein kinase antagonists that inhibit phosphorylation of p47-phox (e.g. H-7, staurosporine or 1-O-hexadecyl-2-O-methylglycerol (AMG)) block translocation of this protein to the detergent-insoluble fraction (116, 117), and, concomitantly, block superoxide anion release (62, 118120). Six to eight isoforms of p47-phox can be observed in stimulated cells that result from different numbers of phosphorylated serine residues (64, 121). Only the most heavily phosphorylated species are observed in the cytoskeletal fraction, which suggests that extensive phosphorylation of this protein is required for superoxide anion production when PMA is the stimulus (122). Interestingly, p47-phox undergoes continuous cycling between phosphorylation and dephosphorylation in PMA-stimulated cells, with the phosphorylation reaction predominating. Interruption of the phosphorylation of p47-phox results in rapid dephosphorylation of p47-phox and in dissociation of this protein from the cytoskeletal fraction (117, 123, 124). Thus, the activity of p47-phox can be regulated by a phosphorylation-dephosphorylation cycle that governs association-dissociation of this protein with cytochrome b₅₅₈ (125). Protein kinase C may form a stable complex with p47-phox when PMA, but not when fMLP is the stimulus (117). Dephosphorylation of p47-phox is catalyzed by a phosphatase type 1 and/or 2A (124, 126).

All of the phosphorylations of p47-phox in neutrophils stimulated with PMA or fMLP occur in the C-terminal region between ³⁰³Ser and ³⁷⁹Ser (122). Edman degradation of ³²P-labeled p47-phox revealed that ³⁰³Ser, ³⁰⁴Ser, ³²⁰Ser, ³²⁸Ser, ³⁴⁵Ser, ³⁴⁸Ser and at least one of the three ³⁵⁹Ser, ³⁷⁰Ser, and ³⁷⁹Ser were phosphorylated. The first four serine residues listed are flanked by basic groups and exhibit consensus sequences for phosphorylation by PKC (123) or one or more novel protein kinases described in neutrophils (127, 128). These novel kinases appear to function in a stimulatory pathway downstream of the phosphatidylinositol (PI) 3-kinase (129). The sequence around ³⁴⁵Ser and ³⁴⁸Ser (PGPQSPGSP) constitutes a consensus sequence for proline-directed protein kinases (e.g. MAP kinases) (130, 131). However, we tested whether recombinant p47phox was a substrate for the ERK2 MAP kinase, immunoprecipitated from 3T3 fibroblasts transfected with the insulin receptor and stimulated with insulin, and observed no difference in labeling between wild-type p47-phox and a deletion mutant of p47-phox lacking the putative MAP kinase phosphorylation site. Moreover, a classical substrate for MAP kinase, myelin basic protein (MBP), was phosphorylated about 100 times better (J. Leusen et al., unpublished results). Besides, the putative MAP kinase phosphorylation site is not conserved in the mouse sequence of p47-phox (132).

Only recently, Faust et al. (133) found an answer to the question which residue(s) in p47-phox need to become phosphorylated for NADPH oxidase activation. This was performed by mutating all the putative serine phosphorylation sites of p47-phox (between ³⁰³Ser and ³⁷⁹Ser) and monitoring these mutants in intact, transfected cells (EBVtransformed B lymphocytes of a CGD patient deficient for p47-phox) for superoxide-inducing capacity and for translocation to the plasma membrane upon cell activation. In this study it was demonstrated that only the phosphorylation of ³⁷⁹Ser is essential for oxidase activity and membrane association (133). This is a very intriguing result, because apparently all phosphorylations of p47-phox can be reduced to one essential phosphorylation, and this phosphorylation happens to be in the C-terminal domain that we postulate to be an important switch for NADPH oxidase activation.

5.5. Role of rac and Rap1A in activating the oxidase

The absolute requirement for rac in the NADPH oxidase was shown by cell-free studies in which cytosol could be replaced with recombinant p47-phox, p67-phox, and rac1 or rac2 (134-136); similar studies in which the plasma membrane was replaced by purified cytochrome b₅₅₈ demonstrated that these four components were the minimum proteins required to obtain a fully active NADPH oxidase (28, 89). The cytosolic p40-phox and the cytochrome-associated Rap1A protein, absent from these purified preparations, appeared not to be necessary for oxidase function under the cell-free conditions utilized. Also, recombinant Rap1A was unable to support NADPH oxidase activity in the cell-free system (135). Transfection studies of dominant-negative (GDP-bound) and dominantpositive (GTP-bound) Rap1A in human B lymphocytes (137) or in differentiated HL60 cells (138) revealed that overexpression of either of these mutant forms of Rap1A suppressed oxidase activity. GTP-bound Rap1A is associated with purified cytochrome b₅₅₈ and GDP-bound Rap1A is not (56). Perhaps GTP-rap1A activates cytochrome b₅₅₈, by inducing a phosphorylation or a conformational change. As mentioned earlier, also phosphorylation of Rap1A disrupts the association with cytochrome b₅₅₈. Taken together, it is likely that Rap1A in its GTP-bound state activates and in its GDP-bound state deactivates the NADPH oxidase in a dynamic cycle in intact cells.

Rac is normally found in the cytosolic fraction of phagocytes, in a complex with the GDP dissociation inhibitor Rho-GDI (75-77, 139). Rac is maintained in its inactive form in this complex, but upon cell activation, biologically active lipids, such as arachidonic acid, phosphatidic acid and phosphatidylinositols, can disrupt the binding of rac to Rho-GDI (139, 140). In fibroblasts, it has been shown that PI 3-kinase acts upstream of rac to mediate PDGF-induced lamellipodia formation (141). In neutrophils, PI 3-kinase and the NADPH oxidase are simultaneously activated by several stimuli; furthermore, wortmannin, an inhibitor of PI 3-kinase, can prevent NADPH oxidase activation (142, 143). PI 3-kinase may therefore act upstream of rac in this system as well.

Two potential targets for rac1 have been discovered: p67-phox and the serine kinase p68-pak, related to the brain p65-pak (144,145). The GTP-bound rac activates p68-pak, which in turn phosphorylates p47-phox on ³²⁸Ser (146). However, according to the findings of Faust et al. (133), this phosphorylation is not essential for oxidase activation. In contrast to p47-phox, p67-phox is recovered predominantly in the Triton X100 insoluble fraction of both stimulated and unstimulated neutrophils (116, 117). This property is consistent with p67-phox being bound to the cytoskeleton (147). Activated rac is known to be involved in cytoskeletal reorganization and could thus play a role in the function of p67-phox or mediate the interaction between p67-phox and other components of the NADPH oxidase. Some investigators have shown that rac associates with the plasma membrane upon cell activation (148, 149). However, we and others (150) could not reproduce these results. Rac translocation has also been reported to occur independently of p47-phox and p67-phox (151, 152). However, we have demonstrated with neutrophils from a CGD patient bearing a deletion of a lysine at position 58 in p67-phox that p47-phox and p67-phox failed to translocate in intact cells, whereas they did interact with cytochrome b₅₅₈ in the cell-free system, but still without inducing superoxide production. The disease in this patient was caused by a disrupted interaction of p67phox-delta⁵⁸Lys with rac (J. Leusen et al, submitted). This indicates that the binding of rac to p67-phox is essential for the activity of the NADPH oxidase in both intact cells and the cell-free system; it is possible that rac induces a conformational change or phosphorylation of p67-phox. Furthermore, a rac/p67-phox interaction is needed for the translocation of p47-phox and p67-phox in intact cells. Most likely, rac functions as a 'shuttle protein', carrying p67-phox to the membrane in its GTP-bound state, and dissociating in its GDP-bound state and returning to the cytosol. Conversely, rac might induce a change in the submembraneous cytoskeleton (like membrane ruffling (153)) and create a docking site for p67-phox. Thus, rac probably mediates translocation of p67-phox and, as a consequence, also the translocation of p47-phox.

5.6. NADPH oxidase: cell-free system vs intact cells

It is evident from several observations discussed above that there are some important differences between the cell-free system and intact cells concerning the activation of the NADPH oxidase. In intact cells, the NADPH oxidase is a dynamic enzyme that needs incessant stimulation to remain active (continuous phosphorylation of p47-phox, GTP loading of rac and cycling of Rap1A). In contrast, once the NADPH oxidase is assembled under cell-free conditions with SDS or arachidonic acid and with GTPgS, it remains active. For instance, the sulfhydryl reagent N-ethylmaleimide (NEM) inhibits ongoing oxidase activity in activated neutrophils, whereas in the cell-free system it has no effect once the oxidase has been assembled (154). Also synthetic peptides that interfere with the assembly of the NADPH oxidase are only effective in the cell-free system when added prior to activation (44, 70, 97, 101, 102, 155, 156).

Another difference between the cell-free system and intact neutrophils is that in the cellfree system the oxidase components p40-phox and Rap1A are not required, but Rap1A is responsible for at least a four-fold increase in superoxide production when overexpressed in differentiated HL-60 cells (138). Also, a complex between p47-phox and p67-phox is not mandatory for cell-free oxidase activity, as shown with mutated p47-phox proteins in (108). In addition, interaction of rac with p67-phox is not necessary for translocation of the cytosolic components in the cell-free system (but it is for oxidase activity in this system) (J.Leusen et al.,submitted). We can conclude from these data that although the cell-free system is very useful for a number of applications, results obtained in this system cannot be directly extrapolated to intact cells.

To recapitulate: the human NADPH oxidase is a very intriguing enzyme; although its catalytic unit is retained within cytochrome b₅₅₈, various additional proteins are required for activity of the NADPH oxidase. In the past few years, substantial progress has been made to elucidate the protein-protein interactions and the activation events involved. It has become evident that 1) activation of rac and subsequent interaction with p67-phox is crucial for the interaction of p67-phox with cytochrome b₅₅₈, probably with gp91-phox, 2) p47-phox interacts with p22-phox, and phosphorylation of ³⁷⁹Ser of p47-phox is obligatory for this event, and 3) p47-phox and p67-phox regulate each others translocation in a positive sense (see also ref. (71)). To put it differently: it is vital to gain insight in the intrigues within the phox family and associated characters to fully understand NADPH oxidase activation.