| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Published online before print July 7, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Veterinary Molecular Biology, Montana State University, Bozeman
1Correspondence: Department of Veterinary Molecular Biology, Montana State University, Bozeman, MT 59717-3610. E-mail: mquinn{at}montana.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPHAGOCYTE NADPH OXIDASE...NONPHAGOCYTE NADPH OXIDASE...MODEL OF NADPH OXIDASE...TRANSCRIPTIONAL REGULATIONFUNCTIONAL ASPECTSSUMMARYREFERENCES |
|---|
Key Words: phagocyte NADPH oxidase • Nox • nonphagocyte NADPH oxidase • superoxide anion • oxidants • free radicals • chronic granulomatous disease
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPHAGOCYTE NADPH OXIDASE...NONPHAGOCYTE NADPH OXIDASE...MODEL OF NADPH OXIDASE...TRANSCRIPTIONAL REGULATIONFUNCTIONAL ASPECTSSUMMARYREFERENCES |
|---|
Neutrophils (a.k.a., polymorphonuclear leukocytes) are normally found circulating in the bloodstream (circulating half-life of 
7 h) and migrating through tissues (2–3 days) and devote their short lifetime to surveillance [3
]. However, during an infection, the neutrophil lifespan is increased, and large numbers of neutrophils are rapidly recruited to the site(s) of infection where they function to destroy invading pathogens. In this capacity, neutrophils serve as one of the body’s first lines of defense against infection. These cells use an extraordinary array of oxygen-dependent and oxygen-independent microbicidal weapons to destroy and remove infectious agents [4
]. Oxygen-dependent mechanisms involve the production of reactive oxygen species (ROS), which can be microbicidal [5
], and oxygen-independent mechanisms include most other neutrophil functions, such as chemotaxis, phagocytosis, degranulation, and release of lytic enzymes and bactericidal peptides (reviewed in ref. [4
]).
The generation of microbicidal oxidants by neutrophils results from the activation of a multiprotein enzyme complex known as the reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase, which catalyzes the formation of superoxide anion (O2·–; reviewed in ref. [6 ]). The classical NADPH oxidase was first described and characterized in neutrophils and other phagocytes, and it was originally thought that this system was restricted only to phagocytes and used solely in host defense. However, recent studies in the last 10–15 years indicate that similar NADPH oxidase systems are present in a wide variety of nonphagocytic cells of leukocyte and nonleukocyte origin (reviewed in refs. [7 , 8 ]). These oxidase systems are functionally distinct from the phagocyte oxidases, as they produce much lower levels of ROS and appear to play distinct roles in inter- and intracellular signaling events. Conversely, structural features of many nonphagocyte oxidase proteins do seem to be similar or even identical to those of their phagocyte counterparts. Therefore, although the primary focus of this review will center on the neutrophil/phagocyte NADPH oxidase, relevant details about structural and functional features of various nonphagocyte oxidase proteins will be included for comparison.
|
PHAGOCYTE NADPH OXIDASE COMPONENTS |
|---|
|
TOPABSTRACTINTRODUCTIONPHAGOCYTE NADPH OXIDASE...NONPHAGOCYTE NADPH OXIDASE...MODEL OF NADPH OXIDASE...TRANSCRIPTIONAL REGULATIONFUNCTIONAL ASPECTSSUMMARYREFERENCES |
|---|
It is now generally accepted that the core NADPH oxidase enzyme is composed of four oxidase-specific proteins (p22phox, p47phox, p67phox, and gp91phox) and a GTPase (Rac1/2). One other oxidase-specific protein (p40phox) and a second GTPase (Rap1A) have also been shown to play roles in regulating oxidase activity; however, their specific functions are still not well understood. Originally, the nomenclature for the various components differed throughout the literature; however, the generally accepted nomenclature for the phagocyte oxidase-specific components now includes the suffix phox, which refers to phagocyte oxidase [10 ]. The one exception is gp91phox, which has also been named NADPH oxidase 2 (Nox2) [11 ]. Overall, the phox proteins are highly conserved throughout the various species studied to date, confirming the absolute requirement for this system in mammalian host defense (e.g., see ref. [12 ]; Table 1 ). Details of each of the individual component are summarized below.
|
-band absorption maximum of 558–559 nm or cytochrome b–245 because of its unusually low midpoint reduction potential of –245 mV (reviewed in ref. [14
]). Initially, it was thought that phagocyte cytochrome b was comprised of a single protein; however, subsequent analyses demonstrated it was actually a heterodimer of a 91-kDa glycoprotein (a.k.a., ß-chain) and a 22-kDa nonglycosylated protein (a.k.a., -chain) [15
]. These proteins are now known universally as gp91phox and p22phox, respectively. Hydrodynamic analyses of purified cytochrome b were consistent with the complex being an -, ß-type heterodimer or an , ß, ß-type hetero-oligomer [16
], and early structural models of cytochrome b were developed around an -, ß-, ß-type complex [17
]. However, this paradigm has been subsequently revised, as later studies indicated cytochrome b was actually a 1:1 -, ß-type heterodimer of p22phox and gp91phox [18
]. Hydropathy analyses of p22phox and gp91phox indicated the presence of two to three and four to six transmembrane regions in these proteins, respectively [19 , 20 ]. Based on the experimentally determined heme content of purified cytochrome b, Parkos et al. [20 ] concluded that more than one heme was present per cytochrome b molecule. Subsequent studies of cytochrome b confirmed this conclusion, indicating a bihistidinyl, multiheme cytochrome with closely spaced hemes [21 22 23 ]. This concept was further refined when re-evaluation of the averaged heme potential of –245 mV, using higher resolution methods, showed that cytochrome b contained two nonidentical hemes with midpoint redox potentials of –225 and –265 mV, respectively [24 ]. Based on this and other evidence, it is now generally accepted that cytochrome b contains two hemes.
The exact location of the hemes associated with cytochrome b has been difficult to determine, and two basic models of heme placement are currently under consideration (Fig. 1 ). According to the shared-heme model, one heme is coordinated within gp91phox, and the second heme is coordinately shared between gp91phox and the invariant histidine of p22phox [25 ] (Fig. 1A) . In an alternative model, which is based on sequence similarities between gp91phox and yeast iron reductase Fre1, both hemes are stacked within gp91phox between transmembrane helices III and V and are coordinated by histidines 101, 115, 209, and 222 [26 ] (Fig. 1B) . Presently, there are direct and indirect experimental data to support both models.
|
The nonshared heme model (Fig. 1B) is also supported by substantial experimental evidence. For example, transgenic cell lines expressing gp91phox alone exhibit a heme spectrum very similar to neutrophil cytochrome b, whereas cells expressing p22phox alone lack a heme spectrum, indicating that at least in this system, gp91phox is able to coordinate both hemes [31 ]. In addition, replacement of gp91phox histidines 101, 115, 209, and 222 with leucine or arginine results in lost or significantly decreased heme spectrum [32 ], and missense mutations in any of these residues have been associated with X-linked CGD [33 ]. Furthermore, mutagenesis of the p22phox His94 suggested, in contrast to the CGD mutation described above, that a histidine at this position was not required for flavocytochrome b function and that heme was not shared between subunits [34 ]. Thus, the role of the two cytochrome b subunits in heme coordination remains a matter of debate, and further studies will be necessary to resolve this issue.
Initially, it was thought that the NADPH oxidase was composed only of cytochrome b and an unknown flavoprotein [35 ]. The search for the putative flavoprotein led to the identification of several candidate proteins (reviewed in ref. [6 ]); however, many discrepancies remained, and ongoing searches led to the consideration that the FAD-binding moiety might actually be cytochrome b itself. This issue was independently resolved by three groups, which concurrently provided data demonstrating that cytochrome b was indeed a flavocytochrome and led to the renaming of cytochrome b558 to the currently accepted nomenclature of flavocytochrome b558 (a.k.a., flavocytochrome b) [36 37 38 ]. Although sequence homology with the ferridoxin-NADP+ reductase (FNR) family of reductases suggested that the putative FAD-binding region involved gp91phox residues 214–246, 335–345, and 350–360 [36 37 38 ], more recent information indicates that one of these regions (residues 214–246) falls within an extracellular loop and cannot participate in FAD binding [39 ].
As NADPH is the electron source for the catalysis of O2 to O2·–, a NADPH-binding component must be present in or near the enzyme complex. As with the FAD-binding protein, the search for the NADPH-binding moiety also resulted in the identification of several candidates, including cytosolic proteins (reviewed in ref. [6 ]). Based on amino acid sequence comparison between gp91phox and FNR, it was proposed that flavocytochrome b might also contain a NADPH-binding domain [36 37 38 ], and this idea was subsequently confirmed by photoaffinity-labeling studies [40 ]. Furthermore, Koshkin and Pick [41 ] showed that relipidated flavocytochrome b alone could generate O2·–, providing direct evidence that flavocytochrome b was able to functionally bind FAD and NADPH. It should be noted, however, that several recent studies suggest flavocytochrome b may not be the sole NADPH-binding component and that p67phox may also contribute in some way to this process (see below).
Based on the information summarized above, it can be concluded that flavocytochrome b contains all redox components used by the NADPH oxidase for transmembrane electron transport (reviewed in ref. [14 ]). Conversely, flavocytochrome b cannot and does not function independently in the cell, thereby insuring against inappropriate activation. Instead, a number of oxidase protein cofactors (described below) are absolutely required for enzymatic activity, presumably to initiate or facilitate the electron transfer process. Regardless of whether the hemes are shared between subunits or not, the currently accepted electron transport pathway involves sequential and step-wise transfer of two electrons from NADPH via FAD and two hemes to ultimately reduce O2 (Fig. 2 ). It should be noted, however, that most evidence suggests the final step in this process occurs via a peripheral mechanism, whereby electrons are transferred from the outer heme to O2 in a pocket near the heme edge, without actual binding of O2 to the heme [14 , 42 ] (Fig. 2) . This type of scheme would be analogous to that of other heme proteins involved in long-range electron transfer [43 ].
|
|
Phosphorylation of p47phox is one of the key intracellular events associated with NADPH oxidase activation, and it has been proposed that phosphorylation causes a conformational change in p47phox and/or neutralizes the cationic domain of the protein (see Fig. 3 ) in vivo so that it can interact with the membrane and other oxidase proteins (reviewed in ref. [62
]). Indeed, Ago et al. [63
] found that phosphorylation of only three serines in p47phox could mimic this conformational change. In the cell-free assay system, amphiphilic agents such as sodium dodecyl sulfate (SDS) or arachidonic acid (AA) serve a similar role, enabling p47phox to undergo analogous, conformational changes in vitro that are likely to occur during phosphorylation in vivo [64
]. Recent studies by Peng et al. [65
], showing that p47phox mutants with unmasked SH3 domains are able to fully reconstitute oxidase activity in a cell-free assay lacking AA, and studies by Ago et al. [66
], showing that p47phox phosphorylation caused unmasking of phosphoinositide-binding domains, further support a requirement for activation-induced, conformational changes in p47phox. A number of kinases have been proposed to participate in p47phox phosphorylation, including protein kinase C (PKC; e.g., ref. [67
]), p38 mitogen-activated protein kinases [MAPK; and extracellular signal-regulated kinase (ERK)1/2; refs. 67
68
69
], p21-activated kinase (PAK) [70
], Akt [71
], casein kinase 2 [72
], and a phosphatidic acid-activated kinase [73
]. PKC plays a dominant role in this process, and it has been shown that PKC-, -ßII, -
, and -
isoforms can individually phosphorylate p47phox to varying degrees, resulting in oxidase activation [74
75
76
]. The physiological role of various other protein kinases in p47phox activation is not as well understood. Consensus MAPK substrate sequences have been identified in p47phox, although phosphorylation at these sites does not appear to be critical for O2·– generation in phorbol myristate acetate (PMA)-stimulated cells [67
, 68
]. In contrast, recent studies using physiological activators suggest that ERK1/2 and PKC actually cooperate in p47phox phosphorylation [69
]. PAKs have also been implicated in p47phox phosphorylation, and Knaus et al. [70
] suggested these kinases may participate in linking chemoattractant receptor stimulation with oxidase activation. Additionally, Akt is rapidly activated when neutrophils are treated with oxidase-activating agents, and p47phox phosphorylation is enhanced in cells expressing membrane-targeted phosphoinositide-3 kinase (PI-3K), which constitutively activates Akt [71
]. Recent studies showing that Akt can phosphorylate p47phox further support a role for Akt in mediating PI-3K-dependent phosphorylation of p47phox [77
, 78
]. Thus, depending on the nature of the activation conditions, a variety of kinases seems to cooperate in mediating p47phox phosphorylation events.
Recently, a novel phosphatidylinositol (PI)-targeting module was identified in p40phox and p47phox, and this conserved, proline-rich motif was designated as the PX domain because of its presence in these phox proteins [79
] (Fig. 3)
. However, it is now known that PX motifs serve as phosphoinositide-binding modules in many different proteins [80
]. The p47phox PX domain is located within an 120 amino acid module encompassing residues 4–125 [49
] and has been shown to bind phosphoinositides with apparent preference for PI 3,4-bisphosphate [PI(3,4)P2; refs. 81
, 82
], although the relative selectivity for various phosphoinositides is not completely resolved [83
]. The ability to bind phosphoinositides suggests that the PX domain plays a role in targeting p47phox to membranes, and PX domain-mediated targeting of this protein has been demonstrated [82
83
84
]. In addition, analysis of the crystallized p47phox PX domain by Karathanassis et al. [85
] showed that it actually contained two binding pockets, one for PI(3,4)P2 and the other for anionic phospholipids, such as phosphatidic acid or phosphatidylserine. They also confirmed previous studies of Hiroaki et al. [49
], who reported that the PX domain was masked by an intramolecular interaction with the C-terminal SH3 domain of p47phox. These findings suggest an additional role for PX modules in protein–protein binding, and this idea is supported by studies showing that binding p47phox to moesin is mediated by the PX domain, although in a phosphoinositide-dependent process [86
]. Furthermore, the identification of a putative phosphatidic acid-binding region is consistent with the ability of phosphatidic acid to activate the oxidase [87
].
p67phox
Like p47phox, p67phox was initially identified as a missing cystolic factor (NCF2) in patients with autosomal, recessive CGD [45
, 46
, 88
]. p67phox contains two SH3 domains (residues 245–295 and 458–517, respectively), a proline-rich domain (residues 219–231), and four N-terminal TPR motifs (within residues 6–154) [88
, 89
] (Fig. 3)
. Complementation studies with neutrophil fractions obtained from p67phox-deficient CGD neutrophils suggest that p67phox is the limiting oxidase cofactor [90
], and there appears to be two to three times less p67phox in neutrophil cytosol compared with p47phox [53
, 91
], indicating most or all of the p67phox must be complexed with p47phox. In support of this conclusion, addition of exogenous p67phox has been reported to enhance binding of p47phox to the membrane in vitro, possibly by associating with free p47phox [58
]. This would be expected if there was always more p47phox available for active complex formation at the membrane.
Most reports agree that in vivo translocation and membrane association of p67phox are dependent on cotranslocation of p47phox to the plasma membrane and prior interaction of p47phox with flavocytochrome b [56 57 58 ]. This conclusion is supported by recent work of Paclet et al. [92 ], who used atomic force microscopy to verify that p47phox preceded p67phox and enhanced the affinity of p67phox for binding to flavocytochrome b during oxidase activation. It should be noted, however, that in vitro reconstitution of NADPH oxidase activity in the absence of p47phox is possible when relipidated flavocytochrome b and high concentrations of p67phox and Rac are combined [93 , 94 ], supporting the idea that p67phox and Rac play more direct roles in electron transport and suggesting a possible direct interaction between p67phox and flavocytochrome b. Subsequent studies showing that chimeric proteins containing truncated p67phox fused to Rac were also able to support p47phox-independent oxidase activity in vitro [95 , 96 ] provide further support for this interaction. Ultimately, direct binding between p67phox and flavocytochrome b was demonstrated by Dang et al. [97 ], who also observed that binding was enhanced by the presence of Rac, which is consistent with the chimeric protein studies.
The possibility that p67phox is the NADPH-binding protein of the oxidase or at least participates in NADPH binding is still a matter of debate. Based on a number of studies, Smith et al. [98 ] concluded that p67phox was an NADPH-binding protein and that the oxidase actually contained two NADPH-binding sites, one low-affinity site in gp91phox and one of higher affinity in p67phox. Furthermore, Dang et al. [99 , 100 ] recently showed that p67phox was able to bind NADPH directly via the TPR domains and could catalyze pyridine nucleotide dehydrogenation. Thus, it is possible that p67phox and gp91phox may be involved in NADPH binding, perhaps to create a shared or cooperative NADPH-binding site involving both proteins. This possibility might explain the apparent contradictions between various reports.
The interaction between p67phox and Rac is essential for NADPH oxidase activation and appears to be mediated primarily by binding Rac to the N-terminal region of p67phox (residues 1–200) [101 ], although Faris et al. [102 ] found that the C terminus of p67phox may also help to stabilize Rac binding. Detailed analysis of the p67phox N terminus resulted in the identification of an array of four tandem TPR motifs (TPR1–4), which mediate Rac binding [89 ] (Fig. 3) . Furthermore, crystallization of the TPR domain bound to Rac demonstrated that this interaction represents a novel mode of TPR domain-mediated binding [103 , 104 ]. Analysis of the p67phox N-terminus also resulted in the identification of an activation domain encompassing residues 199–210, and Han et al. [105 ] showed that this domain was required for oxidase activation (Fig. 3) . Furthermore, this domain appears to play a role in regulating electron flow from NADPH to flavocytochrome b-associated FAD [106 ].
Although it is now evident that p67phox can be phosphorylated, the physiological role of this event in NADPH oxidase assembly/activation is still unclear. El Benna et al. [107 ] used immunoprecipitation to show that p67phox was phosphorylated in neutrophils stimulated with N-formyl-methionyl-leucyl-phenyalanine (fMLF) or PMA and that phosphorylation occurred by PKC-dependent and independent pathways. Indeed, the actual phosphorylation site was mapped to Thr233, which is in a consensus MAPK substrate sequence [108 ], and it was also found that p67phox phosphorylation occurred in the cytosol and was independent of any interaction with p47phox [109 ]. Recently, Dang et al. [110 ] showed that p67phox was also phosphorylated by ERK2 and p38 MAPK in vitro and in intact neutrophils. Phosphorylation occurred at several sites, with the primary ERK2 target localized to the N-terminal fragment and MAPK-mediated phosphorylation primarily in the C terminus. It is interesting that the C-terminal phosphorylation site(s) appears to be masked in the intact protein and may become accessible only after a conformational change [110 ], suggesting phosphorylation could be regulating an intramolecular interaction between the p67phox C terminus and the N-terminal TPR motifs.
p40phox
Compared with the other oxidase proteins, much less is known about the function of p40phox, which was identified in a fractionated, resting neutrophil cytosol via its binding to immunoprecipitated p67phox and contains a single SH3 domain (residues 175–226) [111
], an N-terminal PX domain (residues 24–143) [79
], and a C-terminal PC motif (residues 283–310) [112
] (Fig. 3) . p40phox resides within the cytosolic complex [111
] and seems to bind preferentially to p67phox [113
]. As with the other cytosolic phox proteins, p40phox is phosphorylated during NADPH oxidase activation, and mapping studies indicate phosphorylation occurs at Thr154 and Ser315 [114
].
p40phox has been proposed to play a role in stabilization of the cytosolic phox protein complex and can cotranslocate to the membrane during neutrophil activation [111 ]. However, the actual role of p40phox in oxidase function is still not well understood. Tsunawaki et al. [115 ] found that dissociation of p40phox from the cytosolic complex inhibited the cell-free assay, suggesting that p40phox was a positive regulator of the oxidase. In contrast, Sathyamoorthy et al. [116 ] reported that p40phox inhibited oxidase activity in vitro and in transfected K562 cells, albeit at relatively high p40phox concentrations, and proposed that it functioned in down-regulating the oxidase by competing with SH3 domain interactions between other oxidase components. More recently, Cross [117 ] reported that p40phox promotes oxidase activation by increasing the affinity of p40phox for flavocytochrome b by approximately threefold. In support of the latter observation, Kuribayashi et al. [118 ] recently used a number of approaches to show that p40phox is probably a positive regulator of the oxidase and enhances translocation of p47phox and p67phox in stimulated cells.
In contrast to the p47phox PX domain, the p40phox PX domain preferentially binds PI 3-phosphate [PI(3)P; refs. 81 , 82 ], suggesting differential membrane targeting of these proteins based on phosphoinositide specificity. Crystallization of the p40phox PX domain bound to PI(3)P showed that it contains a phosphoinositide-binding pocket similar to that of p47phox; however, unique binding constraints present in the p40phox PX domain influence its phosphoinositide-binding specificity [119 ]. The physiological role of the p40phox PX domain is not well understood; however, it is thought that participation in intracellular targeting is likely [120 ]. PI(3)P accumulates in phagosomal membranes [121 ], and Ellson et al. [122 ] suggested that PI(3)P could facilitate oxidase assembly by binding to the p40phox PX domain, thereby recruiting p67phox (and possibly other components) to the assembling oxidase complex via association with p40phox.
Analysis of the interaction between p40phox and p67phox resulted in the identification of a novel, protein–protein-binding motif (residues 282–309), and because of its presence in p40phox and Cdc24p (a guanine nucleotide exchange factor in yeast), this motif was designated as the PC motif [112 ]. The PC motif-mediated interaction is not dissociated by anionic amphiphiles in vitro, suggesting that this interaction might be maintained throughout the activation process [112 ]; however, further studies are necessary to verify this conclusion in intact cells. The PC motif target in p67phox also appears to be a novel modular domain (residues 345–427), which is located between the two SH3 domains and is designated the PB1 domain because of its presence in p67phox and Bem1p, a yeast scaffold protein involved in cell polarity [112 ] (Fig. 3) . PB1 domains exist in a variety of proteins and appear to provide a scaffold for PC motif binding, facilitating protein–protein interactions in a number of biological processes [123 ].
Rac
Early on, it was suggested that a cytosolic guanosine triphosphate (GTP)-binding factor participated in the NADPH oxidase, and this factor was concurrently identified by two groups as the small GTP-binding protein Rac (Rac1 or Rac2, depending on the cell type and species; reviewed in ref. [124
]). Furthermore, Dorseuil et al. [125
] showed that Rac antisense oligonucleotides caused dose-dependent inhibition of O2·– production in transformed B lymphocytes, demonstrating a requirement for Rac in vivo. Subsequently, analysis of Rac-deficient mice showed that neutrophils from these mice had diminished O2·– production; however, the defect could be partially corrected by treating with tumor necrosis factor (TNF-) prior to stimulation with PMA [126
]. Further studies in Rac2-deficient mice showed that the requirement for Rac2 in oxidase activation may be stimulus-specific and that Rac2 was essential when cells were activated with physiologically relevant agents such as fMLF or immunoglobulin G (IgG)-opsonized particles [127
]. As Rac1 and Rac2 are capable of reconstituting oxidase activity in cell-free assays, it was suggested that the ability to achieve partial oxidase activity in neutrophils from Rac2-deficient mice may be a result of substitution by Rac1. However, Rac1 cannot substitute completely for Rac2, as shown recently by Glogauer et al. [128
] and Gu et al. [129
], who found that oxidase activity was normal in Rac1-deficient cells but was markedly diminished in Rac2-deficient cells, concluded that Rac2 was physiologically critical for oxidase activation, and that Rac1 was more important in other neutrophil functions. The physiological importance of Rac2 in the human neutrophil oxidase was substantiated recently when a patient with abnormal neutrophil function was shown to have an inhibitory (dominant-negative) mutation in Rac2, resulting in decreased oxidase activity and other neutrophil-functional defects [130
, 131
].
In resting cells, Rac is maintained in a soluble, cytosolic complex with a guanine nucleotide exchange protein known as guanosine diphosphate (GDP) dissociation inhibitor (GDI; reviewed in ref. [124 ]). However, Rac dissociates from GDI during phagocyte activation, allowing GTP-bound (active) Rac to translocate to the membrane and/or interact with its downstream oxidase targets [60 , 132 ]. Rac translocation corresponds temporally and quantitatively with p47phox/p67phox translocation and with oxidase activation [60 , 132 ], although Rac seems to translocate independently of the other cytosolic oxidase components, suggesting it does not directly mediate phox protein translocation [91 , 133 ]. Nevertheless, recent studies suggest Rac activation can induce oxidase assembly [134 ], possibly through an indirect mechanism involving activation of PAK, thereby leading to p47phox phosphorylation [70 ]. Rac has been shown to interact with p67phox [101 ], as well as with flavocytochrome b [91 ]. Thus, it is likely that Rac can modulate the function of more than one NADPH oxidase protein. Furthermore, studies showing that components of the oxidase are associated with the actin cytoskeleton [135 ] suggest the possibility that Rac may play an additional role related to oxidase function via its ability to regulate cytoskeletal structure [136 ].
Rap1A
The first GTPase to be identified in association with the neutrophil NADPH oxidase was Rap1A, which is a member of the Ras superfamily of GTP-binding proteins and was shown to be associated with flavocytochrome b [137
]. The association between flavocytochrome b and Rap1A was confirmed using reconstitution procedures in vitro, and Rap1A was found to form stoichiometric complexes with flavocytochrome b, indicating a direct binding of Rap1A to flavocytochrome b [138
]. Although Rap1A is not required for cell-free oxidase reconstitution [36
], several studies suggest it may play an important regulatory function in intact cells. For example, transfection of transformed B lymphocytes or differentiated HL-60 cells with dominant inhibitory (17N) mutants of Rap1A resulted in significant inhibition of NADPH oxidase activity, directly supporting a role for Rap1A in the regulation of the oxidase [139
, 140
]. Nevertheless, further studies are necessary to define the exact role of Rap1A in this complex system.
|
NONPHAGOCYTE NADPH OXIDASE COMPONENTS |
|---|
|
TOPABSTRACTINTRODUCTIONPHAGOCYTE NADPH OXIDASE...NONPHAGOCYTE NADPH OXIDASE...MODEL OF NADPH OXIDASE...TRANSCRIPTIONAL REGULATIONFUNCTIONAL ASPECTSSUMMARYREFERENCES |
|---|
Nox
Based on the demonstrated presence of NADPH oxidase-like activity in a number of tissues [142
] and the key importance of gp91phox in electron transport [14
], the search for nonphagocyte oxidase homologs initially focused on identifying gp91phox-related species. The first of these homologs to be cloned was designated as mitogenic oxidase 1 because of its postulated role in regulating fibroblast growth and transformation [143
], although it is now thought that transformation was a result of contamination with mutated V12 Ras in these cells [8
]. Concurrent with these studies, Bánfi et al. [144
] identified the same protein and named it NADPH oxidase homolog 1 (NOH-1). To alleviate confusion, a common nomenclature was subsequently adopted, designating this protein as Nox1 [11
], which is 564 amino acids long and exhibits 56% sequence identity with gp91phox (a.k.a., Nox2), including strikingly homologous, functional domains [143
] (Fig. 4
). Conversely, the Nox1 message is not present in phagocytes but is found primarily in colon epithelium and at lower levels in prostate, uterus, and vascular smooth muscle cells [143
]. Alternative splice forms of Nox1 have been reported and were designated as NOH-1S and NOH-1L and NOH-1Lv [144
]. Whereas NOH-1L and NOH-1Lv messages code for Nox1, NOH-1S appeared to represent a significantly smaller species, missing several functional domains [144
]. However, recent studies suggest NOH-1S is actually a cloning artifact and does not exist in nature.
|
and -ß) found in adult spleen (), lymph nodes (), and testis (ß) [151
]. As summarized above, p22phox and gp91phox (Nox2) are required in phagocytes for normal flavocytochrome b function. Conversely, p22phox is expressed in a variety of nonphagocytic cells, some expressing Nox2 and others expressing different Nox homologs (reviewed in ref. [152 ]). Although it is still not clear if the p22phox interaction occurs with all Nox homologs, recent studies suggest that Nox1 can interact with p22phox in vascular smooth muscle cells [153 ] and transfected Chinese hamster ovary cells [154 ]. Conversely, obvious vascular defects have not been noted in CGD patients with p22phox mutations; thus, the physiological relevance of these interactions remains to be elucidated.
NOXO1/NOXA1
Based on analogy with the phagocyte NADPH oxidase, it seemed reasonable that other Nox-based systems might require the participation of additional cofactors for proper function. In support of this idea, cells transfected with Nox1 alone produced very low levels of O2·– [143
], whereas activity could be substantially enhanced by coexpression of p47phox and p67phox, indicating that these phox proteins could functionally associate with Nox1 [155
]. Nevertheless, it was clear that other Nox-related cofactors must exist, as various combinations of phox proteins are coexpressed with various Nox proteins in nonphagocytic cells, and some Nox-expressing cells apparently don’t express any classical phox proteins (reviewed in ref. [142
]). This issue was recently addressed by three groups, which concurrently identified homologs of p47phox and p67phox in mouse [156
] and human [154
, 157
] colons.
The p47phox homologue was originally designated as p41 or p41nox, based on its predicted molecular weight, but is now known as NOXO1 as a result of its putative role in organizing oxidase proteins during oxidase assembly [154 , 156 , 157 ]. NOXO1 is expressed primarily in colon and testis but is also found at low levels in pancreas, liver, thymus, and small intestine [154 , 156 , 157 ]. Although it exhibits only 23% amino acid sequence identity with p47phox, NOXO1 is structurally similar to p47phox and contains analogously spaced PX, tandem SH3, and proline-rich domains (Fig. 3) . In contrast, NOXO1 lacks the polybasic autoinhibitory domain, which becomes multiply phosphorylated during p47phox activation [154 , 156 , 157 ]. Thus, it appears that the NOXO1 may be constitutively active with its SH3 domains always accessible for binding to Nox1, which would be consistent with the constitutive nature of the Nox1-based enzyme [143 ].
The p67phox homologue was originally designated as p51 or p51nox, based on its predicted molecular weight, but is now known as NOXA1 because of its putative role in oxidase activation [154 , 156 , 157 ]. Like NOXO1, NOXA1 is highly expressed in the colon, although the relative levels vary widely between studies [154 , 156 , 157 ]. In addition NOXA1 seems to be expressed in a wider range of tissues than NOXO1 [154 ]. NOXA1 exhibits 28% amino acid sequence identity with p67phox, although it is significantly shorter as a result of the absence of the first SH3 domain and a small region near the second SH3 domain [154 , 156 , 157 ] (Fig. 3) . Conversely, NOXA1 contains an analogous TPR motif, activation domain, PB1 domain, and C-terminal SH3 domain [154 , 156 , 157 ].
|
MODEL OF NADPH OXIDASE ASSEMBLY |
|---|
|
TOPABSTRACTINTRODUCTIONPHAGOCYTE NADPH OXIDASE...NONPHAGOCYTE NADPH OXIDASE...MODEL OF NADPH OXIDASE...TRANSCRIPTIONAL REGULATIONFUNCTIONAL ASPECTSSUMMARYREFERENCES |
|---|
|
|
As discussed above, multiple phosphorylation events are associated with NADPH oxidase activation. Phosphorylation of p47phox, primarily in the polybasic domain and C terminus, induces conformational changes that induce release of the intramolecular, autoinhibitory interaction [63 , 159 , 166 ] (Fig. 5) . However, phosphorylation alone is apparently not sufficient to activate p47phox in vivo, as Dana et al. [167 ] showed that cytosolic phospholipase A2-generated AA was essential for complete activation of the oxidase in vivo. In support of this concept, Shiose and Sumimoto [160 ] found that p47phox activation in vivo required the synergistic action of phosphorylation and AA. Additionally, other lipids known to accumulate during oxidase activation, such as phosphatidic acid, may also contribute to this process [168 ]. In any case, unmasking of the p47phox autoinhibitory conformation is essential and results in exposure of both SH3 domains, the polybasic region and the PX domain [63 , 66 , 166 ]. Apparently, p47phox and p67phox remain associated throughout these conformational changes via association of the p47phox proline-rich C-terminus with the p67phox C-terminal SH3 domain [52 , 65 , 161 ] (Fig. 6 ). Phosphorylation may help to neutralize charge within the exposed polybasic region, thereby facilitating interactions with p67phox and possibly stabilizing the active, open conformation of the tandem SH3 domains [169 ].
During oxidase activation, cytosolic components translocate and eventually bind to flavocytochrome b (Fig. 6) ; however, it is not clear at what point in this process full activation of the cytosolic proteins is achieved. Phosphorylation occurs in a step-wise manner, and apparently, only the most highly phosphorylated forms of p47phox associate with the membrane [170 ]. Indeed, two of the most acidic forms (those containing the greatest amount of phosphorylation) are not present in X-linked CGD cells, indicating that some phosphorylation occurs at the membrane following translocation and/or requires flavocytochrome b interaction [170 , 171 ]. Thus, it is plausible that activation of p47phox and the other factors of the cytosolic complex is a cumulative process occurring throughout translocation and that arrival at the membrane is nearly coincidental with full exposure of the tandem p47phox SH3 domains, which are essential for binding to their proline-rich target in p22phox [172 , 173 ] and exposure of the p40phox and p47phox PX domains involved in binding to membrane phospholipids [65 , 66 , 118 ]. In addition, phosphorylation of p22phox during activation may contribute to this process [174 ]. The association of the p47phox SH3 domains with p22phox is a critical step in the assembly process and may facilitate initial docking of p47phox and associated oxidase proteins with flavocytochrome b (Fig. 6) . Indeed, p67phox seems to depend on p47phox for binding to the oxidase assembly [58 ], although membrane-targeted Rac can serve as a surrogate p67phox target in a modified, cell-free assay [175 ]. In any case, correct alignment of p47phox would then insure accurate proximity for binding of the remaining sites of interaction between p47phox/p67phox and flavocytochrome b, including binding of p47phox to multiple sites within gp91phox [176 , 177 ] and binding of the p67phox N terminus to gp91phox [178 ], possibly via the activation domain [105 ]. It is interesting that p47phox binding also appears to be enhanced by the presence of p67phox [58 ], suggesting the possibility that the binding of p47phox helps chaperone p67phox to a zone where it can interact with gp91phox but also that this subsequent interaction may form a "sandwich" that further anchors p47phox (Fig. 6) . During p47phox/p67phox binding with flavocytochrome b, regulatory input seems to be required to initiate release of p67phox from the cationic domain of p47phox so that this region is accessible for subsequent high-affinity interactions with flavocytochrome b [169 ].
The most likely candidate for regulatory input at this stage of oxidase activation appears to be Rac, and several models of oxidase regulation by Rac have recently been proposed [124 ]. As discussed above, Rac translocates independently of the other cytosolic factors [91 , 133 ] and can bind, via the switch I region, to the p67phox N-terminal TPR motifs [89 , 179 ]. Rac also appears to associate with flavocytochrome b [91 , 180 ], possibly via the Rac insert region [179 ]. The relative timing or sequence of Rac-binding interactions during oxidase assembly is unclear; however, Diebold and Bokoch [180 ] recently provided evidence that the Rac:flavocytochrome b interaction may occur prior to the Rac:p67phox interaction and that two distinct, Rac-dependent steps participate in the electron transfer process. Step 1 involves direct binding of the Rac insert domain to flavocytochrome b [180 ], which is independent of p67phox but seems to facilitate subsequent targeting or modulation of p67phox [179 , 181 ], initiating electron flow from NADPH to FAD. In step 2, interaction between the switch I domain of Rac and p67phox is proposed to induce a conformational change in p67phox, thereby allowing electron flow from FAD to the hemes. Note that this two-step model is consistent with kinetics of oxidase activation described previously [182 ].
Rac has been shown to disrupt the p40phox–p67phox interaction, and this event seems to be essential for efficient oxidase activation [90 , 183 ]. Thus, although p40phox is required early on to modulate recruitment of p47phox/p67phox to the membrane via PB1–PC interactions with p67phox and PX domain interactions with the membrane [118 ], it subsequently becomes displaced by Rac binding. The binding of Rac and/or release of p40phox from p67phox could then induce conformational changes in p67phox, thereby liberating the p47phox cationic region for gp91phox binding [169 ] and/or positioning the p67phox activation domain correctly for imminent binding to gp91phox [105 ]. Apparently, one or more of these events induces conformational changes in flavocytochrome b, thereby permitting completion of electron transfer from FAD to heme and ultimately, to O2 [182 ]. In support of this conclusion, atomic force microscopy has demonstrated measurable changes in flavocytochrome b during transition to the active state [92 , 184 ]. Furthermore, Foubert et al. [165 ] recently found that amphiphilic agents used in cell-free oxidase activation induced significant conformational changes in flavocytochrome b, as determined by monitoring resonance energy transfer from an external fluorescent probe to the heme, and suggested that these changes in structure may facilitate heme alignment for efficient electron transfer.
The nature of the event that induces flavocytochrome b conformational change directly is still under investigation. In vitro, it has been reported that p47phox and p67phox can induce conformational changes in flavocytochrome b [92 ]; however, only p67phox binding induced electron flow and O2·– production [92 , 106 , 181 ]. Although not currently included in the various models of oxidase assembly and activation, Rap1A has been shown to play a regulatory role in intact cells [139 , 140 ] and could conceivably be important in this process by virtue of its ability to associate with flavocytochrome b [137 ].
Assembly of nonphagocyte oxidases
Currently, not much is known regarding actual binding events during assembly of the nonphagocyte oxidase systems; however, the overall similarity in domain structure between Nox/NOXO1/NOXA1 proteins and their respective phox homologs suggests that some of the interactions described above may be relevant (Figs. 3
and 4) . Indeed, NOXO1 can interact with p22phox and p67phox via analogous SH3 domain-mediated binding, and NOXA1 can bind to p47phox and Rac through interactions similar to those with p67phox [154
]. In addition, NOXO1 and NOXA1 can substitute for and combine with their respective phox homologs in activation of gp91phox [154
, 155
]. There does appear to be some structural specificity required for the various cofactors to activate Nox1, however, as only NOXO1 and NOXA1 are fully competent in this respect, and substitution with one or both p47phox and p67phox results in minimal Nox1 activity [154
155
156
157
]. Conversely, recent studies indicate that Nox3 is more compliant and can be activated by phox and NOX regulatory proteins [185
]. Nox3 can also be activated by NOXO1 alone in the absence of activator subunits (NOXA1 or p67phox) [185
].
Nonphagocyte oxidases also appear to lack some key regulatory interactions characteristic of the phagocyte oxidase. For example, the absence of an analogous, autoinhibitory region in NOXO1 suggests that its SH3 domains are constitutively accessible for Nox binding and that phosphorylation may not play as important a role in NOXO1 activation. Indeed, Cheng and Lambeth [186 ] recently reported that NOXO1 colocalizes with Nox1 in the membranes of resting cells. NOXA1 does not appear to bind p40phox, which may be a result of the lack of the conserved Lys355 in the NOXA1 PB1 domain [187 ]. As p40phox stabilizes the cytosolic phox protein complex in phagocytes [90 ], one might also conclude that such a complex is unstable or may not exist in nonphagocytic cells using Nox1. Alternatively, an unidentified homologue of p40phox could be present in cells expressing NOXA1. Finally, the activation-induced, binding interaction involving the exposed p47phox polybasic region and p67phox [169 ] would be absent in the NOXO1–NOXA1 system. The absence of these regulatory interactions would be consistent with and might help to explain the constitutive oxidase activity observed in many of the nonphagocyte oxidases (reviewed in refs. [8 , 142 ]). Indeed, agonist-independent oxidase activity is observed in cells cotransfected with NOXO1/NOXA1 and Nox1 or Nox3, although this activity can be acutely up-regulated in an agonist-dependent manner [154 , 156 , 157 , 185 , 186 ].
|
TRANSCRIPTIONAL REGULATION |
|---|
|
TOPABSTRACTINTRODUCTIONPHAGOCYTE NADPH OXIDASE...NONPHAGOCYTE NADPH OXIDASE...MODEL OF NADPH OXIDASE...TRANSCRIPTIONAL REGULATIONFUNCTIONAL ASPECTSSUMMARYREFERENCES |
|---|
During neutrophil maturation, expression of genes coding for NADPH oxidase proteins is highly regulated at the transcriptional level. For example, studies in bone marrow granulocytes and promyelocytic cell lines demonstrated variable timing in gp91phox, p67phox, and p47phox expression patterns during differentiation, indicating a role for transcriptional regulation in this process [191 ]. However, a number of studies indicate that cytokines and inflammatory mediators can also modulate the level of oxidase protein expression in mature phagocytes via transcriptional regulation involving the coordinated activity of a variety of transcription factors (reviewed in ref. [192 ]; Table 2 ).
|
Ets factors have been shown to play critical roles in transcriptional regulation of CYBB. A consensus sequence for ets factor-binding is located in the proximal promoter of CYBB and is designated as the Pu box, as it binds the myeloid-specific ets factors PU.1 and Elf-1 [199 , 210 ]. In addition, it appears that PU.1 or Elf-1 may cooperatively bind to the CYBB promoter as a complex comprised of two IRFs, IRF-1 and ICSBP, which are known as the HAF-1 complex [211 ]. Binding HAF-1 to the CYBB promoter has been reported to recruit the transcriptional coactivator CBP, which is thought to be involved in IFN-induced activation of CYBB and other myeloid genes [196 ]. Furthermore, Kautz et al. [198 ] reported that the protein-tyrosine phosphatase (SH2-containing tyrosine phosphatase 1) inhibited gp91phox and p47phox expression in undifferentiated myeloid cell lines by keeping IRF-1 and ICSBP dephosphorylated, thereby inhibiting HAF-1 binding. It should be noted, however, that the relative roles of PU.1 and HAF-1 in regulating CYBB transcription are still uncertain, and it has recently been suggested that PU.1 binds with higher affinity and is dominant, and HAF-1 is supplementary for activation of CYBB [212 ]. Additionally, YY1 has been shown to bind and activate the CYBB promoter [200 ].
Transcriptional repression has also been proposed to participate in CYBB regulation, with potential involvement of multiple repressors and overlapping binding elements. For example, Skalnik and co-workers [201 ] showed that the transcriptional repressor CDP binds to multiple sites in the proximal promoter of CYBB to repress this gene in immature myeloid cells and that down-regulation of CDP during final maturation is required for gp91phox expression. In addition, it was found that CDP binding repressed transcription by blocking the binding sites for multiple transcriptional activators, including CP1, IRF-1, and IRF-2 [202 , 203 ]. One of the CDP-binding sites was also found to be a consensus target sequence for binding of the repressor element HoxA10, and phosphorylation of HoxA10 during myeloid differentiation decreased its binding, thereby releasing transcriptional repression of CYBB [204 ]. Recently, seven binding sites for the matrix attachment region-binding protein SATB1 were identified in the CYBB promoter, and it was found that SATB1 also repressed gp91phox expression [205 ]. Like CDP, SATB1 binding to the CYBB promoter was abundant in immature cells but was do
