Structure, regulation and evolution of Nox-family NADPH oxidases that produce reactive oxygen speciesHideki Sumimoto Medical Institute of Bioregulation, Kyushu University, Fukuoka CREST,
Trang 1Structure, regulation and evolution of Nox-family NADPH oxidases that produce reactive oxygen species
Hideki Sumimoto
Medical Institute of Bioregulation, Kyushu University, Fukuoka
CREST, Japan Science and Technology Agency, Tokyo, Japan
Keywords
Duox; Nox; Noxa1; Noxo1; p22phox; p40phox;
p47 phox ; p67 phox ; Rac; Rboh
Correspondence
H Sumimoto, Medical Institute of
Bioregulation, Kyushu University, 3-1-1
Maidashi, Higashi-ku, Fukuoka 812-8582,
to four Ca2+-binding EF-hand motifs are present at the N-termini inseveral subfamilies, such as the respiratory burst oxidase homolog (Rboh)subfamily in land plants (the supergroup Plantae), the NoxC subfamily insocial amoebae (the Amoebozoa), and the Nox5 and dual oxidase (Duox)subfamilies in animals (the Opisthokonta), whereas an SH3 domain isinserted into the ferredoxin–NADP+reductase region of two Nox enzymes
in Naegleria gruberi, a unicellular organism that belongs to the supergroupExcavata Members of the Nox1–4 subfamily in animals form a stable hete-rodimer with the membrane protein p22phox, which functions as a dockingsite for the SH3 domain-containing regulatory proteins p47phox, p67phox,and p40phox; the small GTPase Rac binds to p67phox (or its homologousprotein), which serves as a switch for Nox activation Similarly, Rac acti-vates the fungal NoxA via binding to the p67phox-like protein Nox regula-tor (NoxR) In plants, on the other hand, this GTPase directly interactswith the N-terminus of Rboh, leading to superoxide production Here Idescribe the regulation of Nox-family oxidases on the basis of three-dimen-sional structures and evolutionary conservation
Abbreviations
AIR, autoinhibitory region; Duox, dual oxidase; FNR, ferredoxin–NADP+reductase; Fre, ferric reductase; FRO, ferric-chelate reductase; Noxa1, Nox activator 1; Noxo1, Nox organizer 1; NoxR, Nox regulator; PI3K, phosphatidylinositol-3-kinase; PKC, protein kinase C; PMA, 4b-phorbol 12-myristate 13-acetate; PPII, polyproline II; PRR, proline-rich region; PtdIns(3)P, phosphatidylinositol 3-phosphate; PtdIns(3,4)P2,phosphatidylinositol 3,4-bisphosphate; PX domain, phagocyte oxidase domain; Rboh, respiratory burst oxidase homolog; ROS, reactive oxygen species; TPR, tetratricopeptide repeat.
Trang 2Reactive oxygen species (ROS) are conventionally
regarded as inevitable deleterious byproducts of
aero-bic metabolism On the other hand, there exist
enzymes dedicated to ROS production The first
exam-ple of such enzymes is an NADPH oxidase expressed
in mammalian professional phagocytes [1–10] During
engulfment of invading microbes, the phagocyte
NADPH oxidase becomes activated to reduce
molecu-lar oxygen to superoxide anion (O2)), a precursor of
microbicidal ROS, in conjunction with oxidation of
NADPH As the rapid increase in oxygen consumption
during phagocytosis is known as the respiratory burst,
this enzyme is also called respiratory burst oxidase
The significance of the phagocyte oxidase in host
defense is exemplified by recurrent and life-threatening
infections that occur in patients with chronic
granu-lomatous disease, whose phagocytes genetically lack
the superoxide-producing activity [11,12]
The catalytic core of the phagocyte NADPH oxidase
(phox) is gp91phox, a membrane-integrated glycoprotein
with an apparent molecular mass of about 91 kDa
gp91phox contains two hemes in the N-terminal membrane region, and NADPH-binding and FAD-binding domains in the C-terminal cytoplasmic region(Fig 1A), forming a complete apparatus that trans-ports electrons from NADPH via FAD and two hemes
trans-to molecular oxygen In the mid-1990s, homologs ofthe flavocytochrome gp91phox were discovered in landplants; these have been designated respiratory burstoxidase homolog (Rboh) [13–15] Subsequent searches
in genome databases led to the identification of novelhomologs of gp91phox in animals, which are presentlyknown as Nox (NADPH oxidase) or Duox (dual oxi-dase) [1–10] The human genome contains seven genesencoding gp91phox homologs: Nox1–Nox5, wheregp91phox is renamed Nox2, and the distantly relatedoxidases Duox1 and Duox2 It is currently known that
a wide variety of eukaryotes express ducing NADPH oxidases that harbor a gp91phox-likeelectron-transferring system; the enzymes constitute theNox family Recent studies on Nox-family enzymeshave increasingly clarified the importance of deliberateROS production in various biological events, includingsignal transduction, development, and hormone bio-synthesis, in addition to well-established roles in hostdefense [1–10] It is likely that individual Nox enzymeshave developed regulatory systems, according to theirrespective special functions, during the course ofeukaryote evolution by inserting a regulatory domain(or motif) into their own sequences or by obtaining atightly associated protein as a regulatory subunit Reg-ulation by these proteins includes multiple protein–protein and protein–lipid interactions In this review, Idescribe post-translational regulation of Nox-familyenzymes on the basis of three-dimensional structuresand evolutionary conservation
superoxide-pro-Structure of Nox-family enzymes
The phagocyte NADPH oxidase gp91phox⁄ Nox2 (570amino acid residues) exists in not only the plasmamembrane but also the membrane of the specific gran-ule in neutrophils: the latter contains a higher amount
of Nox2 [1–10] Although the phagosomal membrane
is considered to primarily derive from the plasmamembrane, the specific granule is fused to the phago-some during phagocytosis, and thus gp91phox⁄ Nox2 isfurther enriched in the phagosomal membrane Forkilling engulfed microbes, superoxide (and microbicidalROS derived from superoxide) must be produced frommolecular oxygen within the phagosome The intra-phagosomal production requires electrons to be trans-ported from the cytoplasmic NADPH across themembrane into the interior of the phagosome Such
Fig 1 (A) A model for the structure of gp91 phox Cylinders
repre-sent six transmembrane a-helices (B) Bis-heme ligation in
gp91phox Heme-coordnating His residues are numbered according
to their localization in gp91 phox (C) Intramembrane bis-heme motifs
in various cytochromes The numbers of intervening amino acids
that separate a pair of His residues in a transmembrane segment
are indicated.
Trang 3transmembrane electron transport very often involves
di-heme membrane proteins
gp91phox⁄ Nox2 can be divided into two parts of
sim-ilar size (Fig 1A) The C-terminal half is a cytoplasmic
domain homologous to ferredoxin–NADP+ reductase
(FNR), bearing the NADPH-binding and
FAD-bind-ing sites [16,17], whereas the N-terminal moiety
com-prises six predicted a-helical transmembrane segments
(Fig 1A) The bipartite structure is common not only
to all the Nox-family enzymes but also to the family
of fungal ferric reductases (Fre) [18] Fre enzymes,
which are expressed in the plasma membrane, reduce
Fe3+ and Cu2+ for iron and copper uptake, but fail
to use molecular oxygen as a substrate [19,20]
Among the conserved transmembrane segments of
gp91phox⁄ Nox2, the third and fifth helices each contain
two invariant His residues, which are considered to
provide the axial and distal ligands for binding to the
irons of two nonidentical hemes, thereby placing one
heme towards the cytoplasmic face and the other
towards the outer face (Fig 1B) As the hemes are
ori-ented perpendicular to the surface of the membrane,
electrons are transferred from the cytosolic NADPH,
through FAD, and across the membrane via the hemes
to molecular oxygen, leading to superoxide production
Thus, transmembrane electron transport in gp91phox⁄
Nox2 is considered to occur in the N-terminal
bis-heme-containing region [18,21]
This model was initially proposed on the basis of a
similar motif consisting of two pairs of spaced His
resi-dues that was predicted to be linked to heme
coordina-tion in a class of organelle and bacterial b-type
cytochromes, such as the bis-heme cytochrome b of the
mitochondrial cytochrome bc1 complex (complex III)
and cytochrome b6 of the chloroplast cytochrome b6f
complex [22–24] (Fig 1C) The model for two
bis-hist-idyl heme ligation was subsequently verified by
determi-nation of crystal structures of protein complexes
containing these cytochromes [25–28] In cytochrome b
of the cytochrome bc1complex, containing eight
a-heli-cal transmembrane segments, the two b-type hemes are
bound within a four helix bundle formed by the first
four segments: His residues ligated to both hemes are
located in the second and fourth helices [25,26,29], and a
pair of the His residues in each helix are separated by 13
intervening amino acids (Fig 1C) A similar
coordina-tion occurs in cytochrome b6 of the cytochrome b6f
complex in cyanobacteria and chloroplasts [27,28]:
cyto-chrome b6 comprises four a-helical transmembrane
segments [30,31], and the two hemes are
bis-His-coordi-nated by imidazole side chains separated by 13 and 14
residues in the second and fourth helices, respectively
(Fig 1C) In the fungal Fre-family enzymes, both His
pairs are separated by 13 amino acids, as in chrome b of the cytochrome bc1complex (Fig 1C).Although gp91phox⁄ Nox2, as well as other Nox-familyenzymes, contains a pair of His residues in the thirdtransmembrane a-helix with 13 intervening amino acids(His101 and His115 in human gp91phox⁄ Nox2), the otherpair in the fifth helix (His209 and His222 in humangp91phox⁄ Nox2) are separated by 12 amino acids(Fig 1) It should be noted that the imidazoles sepa-rated by 12–14 amino acids are likely to face the sameside of the helix Substitution of any of these four Hisresidues results in disrupted insertion of hemes intogp91phox⁄ Nox2 [32], supporting the view that the twobis-histidyl heme ligation also occurs in gp91phox⁄ Nox2.Reduction of molecular oxygen to superoxide ingp91phox⁄ Nox2 requires both heme groups to be in thelow-spin (hexacoordinate) state, which implies thatelectron transfer to oxygen occurs via the outer heme
cyto-in a pocket near the heme edge, rather than throughdirect coordination of oxygen to the heme iron [33,34].This outer sphere (or peripheral) mechanism is consis-tent with the ‘two bis-histidyl heme ligation’ structure,and explains well why gp91phox⁄ Nox2-catalyzed super-oxide production is not inhibited by cyanide or carbonmonoxide Taken together, electron transfer fromNADPH to molecular oxygen occurs in a module des-ignated the Nox superdomain The Nox superdomaincomprises two moieties, the N-terminal bis-heme cyto-chrome b, composed of six a-helical transmembranesegments, and the C-terminal FNR, which containsFAD-binding and NADPH-binding domains Thus,gp91phox⁄ Nox2 and its relatives (the Nox family) areflavocytochromes [16,17,35]
It is known that members of the Duox subfamily inanimals, in contrast to oxidases of the other subfa-milies, release H2O2 without forming detectableamounts of superoxide [36] However, they are alsoexpected to produce superoxide as an initial product,
as Duox has the superoxide-producing Nox domain that comprises the bis-heme-containing trans-membrane region and the FNR-like moiety Indeed, ithas been reported that Duox in an immature form iscapable of producing superoxide [37] In mature Duox,superoxide produced by the Nox superdomain may berapidly converted to H2O2 via intramolecular dismuta-tion, possibly by a peroxidase-like ectodomain; thismodule is located on the outer surface of the mem-brane, where superoxide is expected to be released
super-In addition to the Nox and Fre families, both thebis-heme transmembrane segment and FNR-relatedmoiety are present in ferric-chelate reductase (FRO) ofland plants [38] To acquire iron from soils of low ironavailability, land plants such as Arabidopsis thaliana
Trang 4reduce Fe3+ to Fe2+ by FRO in the plasma
mem-brane of root epithelial cells Four His residues in
FRO that lie on two predicted, similarly orientated,
transmembrane a-helices are in equivalent locations to
the His residues in Nox enzymes that coordinate the
two hemes: the 13 and 12 amino acids separating
heme-liganding His residues exist in the helices
(Fig 1C) The FRO-family enzymes contain eight or
10 transmembrane segments, which is different from
the situation in members of the Nox and Fre families;
the precise membrane topology of FRO enzymes
remains controversial [39,40]
Origin of Nox-family enzymes
There is no evidence for the presence of Nox, Fre or
FRO in prokaryotes: a superfamily of
flavocyto-chromes that transport electrons across membranes On
the other hand, members of this superfamily are
pres-ent in a variety of eukaryotes The shared bis-heme
binding motif raises the possibility of an evolutionary
and functional relationship between the eukaryotic cell
surface membrane proteins Nox, Fre and FRO and the
prokaryotic (or organelle) b-type cytochromes On the
other hand, the C-terminal moiety of the
flavocyto-chrome superfamily is homologous to FNR, a
prokary-otic (organelle) protein that is made up of two
structural domains, each containing about 150 amino
acids: the C-terminal region includes most of the
resi-dues involved in NADPH binding, whereas the large
cleft between the two domains accommodates the FAD
group [41,42] It is tempting to postulate that a gene
encoding a protein containing two di-heme
transmem-brane helices was fused to an FNR gene in eukaryote
evolution, leading to a common ancestor of the Nox
and Fre families (Fig 2) In this context, it seems
inter-esting that FNR directly interacts with the chrome b6f complex of plant chloroplasts, albeit in anoncovalent manner, and participates in electron trans-fer [43]; cytochrome b6 in the complex has two b-typehemes across the membrane, as expected for Nox
cyto-A similar fusion of the FNR gene with a geneencoding an electron-transporting protein is also con-sidered to have occurred during evolution: eukaryoticdiflavin reductases such as NADPH–cytochrome P450reductase (CPR or P450R), methionine synthase reduc-tase and novel reductase 1 probably arose from thefusion of the ancestral genes for FNR and flavodoxin,
a prokaryotic FMN-containing protein that transferselectrons in a variety of photosynthetic and nonphoto-synthetic reactions in prokaryotes [43–48] (Fig 2) Inturn, a diflavin reductase is likely to be the precursor
of further fusion products such as nitric oxide tase, which consists of a C-terminal CPR-like domainand an N-terminal, heme-containing oxygenase domain[48,49]
reduc-Distribution of Nox-family enzymes
in eukaryotes
In contrast to the absence of Nox in prokaryotes,genes encoding Nox-family enzymes are found in awide variety of eukaryotes Eukaryotes can be dividedinto several major supergroups, including the Opi-sthokonta, the Amoebozoa, the Plantae, the Excavata,the Rhizaria, and the Chromalveolata (the Hetero-konta plus the Alveolata): animals and fungi belong tothe Opisthokonta; the social amoeba Dictyostelium dis-coideum is a member of the Amoebozoa; land plantsand red algae belong to the Plantae; and diatoms andoomycetes are members of the Heterokonta, a groupthat belongs to the Chromalveolata (Fig 3) [50–52]
Fig 2 A putative common ancestor of the Nox family CPR, NADPH–cytochrome P450 reductase.
Trang 5The relationships between these supergroups, however,
remain to be determined, and thus the root of
eukary-otes is presently uncertain [53,54]
Recent expansion of information available in
gen-ome databases has revealed that Nox-family enzymes
are present in all the eukaryotic supergroups except
the Rhizaria (Fig 3) This suggests that a common
ancestor of Nox genes emerged at an early stage in the
evolution of eukaryotes; it diverged well in some
lin-eages (e.g in the Opisthokonta and Plantae), whereas
it was often lost in some other lineages The loss of
Nox genes appears to have occurred at multiple stages
in eukaryote evolution For example, in the
super-group Amoebozoa, the social amoeba Di discoideum
contains three Nox genes [55], whereas they are absent
in Entamoeba histolytica [56] In the supergroup
Excav-ata, the heterolobosa Naegleria gruberi has at least two
Nox genes, as found by the present search using the
database of the DOE’s Joint Genome Institute (http://
genome.jgi/psf.org/euk_home.html) (Fig 3); on the
other hand, no Nox gene has been found in the
kineto-plastid Leishmania major or Trypanosoma brunei, or
the diplomonad Giardia lamblia [56] In the
Chromal-veolata, Nox genes are present in genomes of the
oomycete Phytophthora sojae [56,57] and the diatom
Thalassiosira pseudonana [58], although they have not
been found in the genomes of Plasmodium falciparum
and Theileria parva, both of which belong to the same
supergroup [56] Even in the Opisthokonta, Nox genes
have been lost independently in several fungal lineages;
for example, a Nox gene is absent in budding andfission yeasts [56,57,59]
Regulation of Nox-family enzymes
by Ca2+
Superoxide production by Nox is regulated by variousmechanisms Several subfamilies of Nox enzymesappear to be directly regulated by Ca2+ It is wellestablished that mammalian thyroid oxidase and seaurchin NADPH oxidase, both of which belong to theDuox subfamily, are reversibly activated by Ca2+ Inaddition to the Nox superdomain comprising the bis-heme-containing transmembrane region and the FNR-homologous moiety, Duox-subfamily oxidases feature
an N-terminal peroxidase-like ectodomain that is rated from two EF-hands by an additional transmem-brane segment (Fig 4) Biosynthesis of thyroidhormones in humans requires Duox2, which is highlyexpressed in the thyroid gland: mutations in Duox2are associated with a loss of thyroid hormone synthesisand can lead to permanent and severe congenital hypo-thyroidism [60] With the H2O2 produced by Doux2,thyroid peroxidase catalyzes conjugation of iodide ions
sepa-to Tyr residues on thyroglobulin in the thyroid cles, an essential step for the synthesis of the activehormone [61] On the other hand, during fertilization
folli-of sea urchins, a rapid increase in H2O2 generationoccurs, which is catalyzed by the sea urchin Duoxhomolog Udx1, leading to formation of the fertiliza-
Fig 3 Distribution of Nox-family enzymes
in eukaryotes and animals Upper panel:
eukaryotes can be divided into six major
supergroups, including the Opisthokonta,
the Amoebozoa, the Plantae, the Excavata,
the Rhizaria, and the Chromalveolata All the
supergroups except the Rhizaria are known
to contain Nox genes Lower panel: M b.,
Mo brevicollis; N v., Ne vectensis; L g.,
Lot gigantea; C sp I, Capitella sp I; C e.,
Ca elegans; D m., Dr melanogaster; S p.,
S purpuratus; B f., B floridae; C i., Ci.
intestinalis; D r., Da rerio; X t., X
tropical-is; G g., Ga gallus; H s., Homo sapiens.
Trang 6tion envelope as the physical block to polyspermy [62].
It is also known that Duox1 of Caenorhabditis elegans
in the phylum Nematoda is involved in cross-linking
of Tyr residues of extracellular matrix proteins,
thereby facilitating cuticle formation [63], and that
Duox plays a critical role in innate immunity in the
gut of the fruit fly Drosophila melanogaster in the
phylum Arthropoda [64]
Regulation of Duox by Ca2+probably occurs via its
paired EF-hand motif It has been reported that
lim-ited proteolysis with a-chymotrypsin renders thyroid
NADPH oxidase fully and irreversibly active
indepen-dently of Ca2+ [65] This implies that the Ca2+
-bind-ing EF-hands of Duox serve as an autoinhibitory
domain, whereas those of Nox5 function as an
activa-tion domain [66] The inhibiactiva-tion of Duox by the
EF-hands might be released reversibly by physiological
Ca2+-induced conformational change and irreversibly
by proteolytic removal of the autoinhibitory domain
[37] A recent study has shown that ectopically
expressed Duox produces ROS without cell stimulants,
and the production is enhanced two-fold by the
addi-tion of ionomycin [67], suggesting that elevaaddi-tion of
cytoplasmic concentrations of Ca2+ is dispensable
Besides direct regulation by Ca2+, protein kinase C
(PKC) may modulate Duox in a Ca2+-independent
manner, as ROS production in thyrocytes is triggered
by 4b-phorbol 12-myristate 13-acetate (PMA), an
agent that activates PKC without elevating the
cyto-plasmic concentration of Ca2+[68] A PKC-dependent
pathway may also function in H2O2 generation atfertilization in the sea urchin [62]
In addition to the animal Duox subfamily, oxidases
of other two subfamilies have been shown to bedirectly regulated by elevations in cytoplasmic Ca2+concentrations: the Nox5 subfamily in animals[66,69,70], and the Rboh subfamily in land plants[71,72] (Fig 4) The regulation by Ca2+ appears to beconsistent with the presence of the Ca2+-bindingEF-hand motif in the cytoplasmic region N-terminal tothe Nox superdomain (Fig 4)
Human Nox5 is abundantly expressed in T and
B cells of spleen and lymph nodes, and also in the spermprecursors of testis [69] Although the role for mamma-lian Nox5 remains unknown, Drosophila Nox5 has beenreported to mediate smooth muscle contraction [70].Oxidases of this subfamily build on the basic structure
of the Nox prototype, adding an N-terminal extensionthat contains four EF-hands: three canonical motifsand one noncanonical motif [63] (Fig 4) Biochemicalanalysis has shown that activation of Nox5 is directlyregulated by Ca2+: superoxide production by Nox5-containing membrane fractions is dependent on thepresence of Ca2+ [66]; and cells ectopically expressingNox5 produce superoxide in response to the Ca2+iono-phore ionomycin [69] The Ca2+-binding domain ofNox5, in contrast to that of Duox, may function as anactivator module: the binding of Ca2+causes a confor-mational change, which leads to intramolecular interac-tion of the N-terminal Ca2+-binding domain with the
Fig 4 Models for structures of various types of Nox-family enzymes Cylinders rep- resent six transmembrane a-helices EF,
sub-Ca 2+ -binding EF-hand motif.
Trang 7C-terminal Nox superdomain, culminating in Nox5
acti-vation [66] On the other hand, the EC50for calcium of
about 1 lm, determined in a cell-free activation system
for Nox5 [66], is relatively high and unlikely to be
achieved in most cells treated with physiological
stimu-lants Two mechanisms for the elevation of the Ca2+
sensitivity have recently been proposed [73,74] First,
PKC phosphorylates Ser⁄ Thr residues in the
FAD-bind-ing domain of Nox5, which increases the Ca2+
sensitiv-ity of the Nox5 activsensitiv-ity regulated by the N-terminal
Ca2+-binding domain [73] Second, a consensus
calmod-ulin-binding site is present in the NADPH-binding
domain of Nox5; calmodulin interacts with the site at a
lower concentration of Ca2+, thereby elevating the
Ca2+sensitivity for Nox5 activation [74]
The Rboh subfamily of NADPH oxidases is
respon-sible for ROS formation associated with plant defense
responses, and also plays a crucial role in plant
devel-opment [13,14,75] Ten and nine members are present
in A thaliana [14] and the rice Oryza sativa [76],
respectively The Rboh-subfamily enzymes carry an
N-terminal extension with two EF-hand motifs
(Fig 4) It has been reported that only about a
two-fold to three-two-fold increase in superoxide production
occurs in the membrane fraction of tobacco and
tomato upon addition of Ca2+; in addition, the effect
requires high concentrations of Ca2+ (approximately
millimolar) [71] This observation suggests that the
direct effect of Ca2+may not contribute to Rboh
acti-vation to a large extent A recent study has shown that
elicitor-responsive phosphorylation of the N-terminal
region of Rboh is involved in superoxide production
[77,78], which is mediated by Ca2+-dependent protein
kinases [77] Activation of Rboh is also regulated by
plant homologs of the Rho-family small GTPase Rac
(also known as Rop for Rho-like protein) [79–81]; Rac
in the GTP-bound form functions by directly binding
to the N-terminal region of Rboh, and this is probably
inhibited by Ca2+ [76] Thus, regulation of Rboh is
more complicated than previously expected, as
described in detail in a later section
As in plant Rboh, two copies of EF-hands are
pres-ent in the N-terminal cytoplasmic region of the
NoxC-subfamily members in the Amoebozoa and Nox
enzymes in oomycetes of the eukaryotic supergroup
Chromalveolata [57], whereas enzymes in the fungal
NoxC subfamily contain a single EF-hand in the
N-terminal cytoplasmic region [56,57,59] These
sub-families of Nox enzymes may be regulated by Ca2+;
however, no experimental evidence for Ca2+-mediated
regulation has been obtained
It is presently unknown whether the
EF-hand-con-taining Nox subfamilies originated from a common
ancestor gene It seems rather likely that EF-handmotifs have been obtained independently several timesduring evolution The genomes of Monosiga brevicollis
in the choanoflagellates (a sister group of animals) andNematostella vectensis in the cnidarians (a basal group
of animals) contain solely Nox2-like enzymes, and notEF-hand-containing oxidases such as Nox5 and Duox,although these two families are found in a variety ofspecies of protostomes and deuterostomes (Fig 3).Thus, Nox5 and Duox may have evolved from Nox2-like prototype oxidases Similarly, in fungi, the NoxCsubfamily containing an EF-hand is found solely inmore evolved groups, including the Sordariomycetesand Dothideomycetes, whereas the Nox2-likeEF-hand-free subfamilies NoxA and NoxB are presentalso in relatively basal groups such as the Chytridi-omyceta and Basidiomycota [57] These features sug-gest that the NoxC subfamily emerged at a later stage
of fungal evolution Thus, the classification of the Noxfamily in eukaryotes into the two major groups,depending on the presence or absence of the EF-handmotif, does not seem to reflect molecular evolution
Regulation of Nox-family enzymes by protein–protein interactions
The genome of Na gruberi (http://genome.jgi/psf.org/euk_home.html), a member of the eukaryotic super-family Excavata, contains two genes encoding Nox-family enzymes of 627 and 630 amino acids, both ofwhich have an SH3 domain and thus are tentativelydesignated as NoxSH3 (Fig 5) The SH3 domain isinserted into a loop region in the NADPH-bindingdomain of the C-terminal FNR-homologous region(Fig 5) It is tempting to postulate that the NoxSH3enzymes in Naegleria are regulated by a protein har-boring a proline-rich region (PRR); SH3 domains aregenerally known as modules that recognize a PRR tomediate protein–protein interactions [82,83] Identifica-tion of a NoxSH3-binding protein will shed light onour understanding of Nox regulation
Nox1–Nox4 in animals form a heterodimer with thenonglycosylated integral membrane protein p22phox,which contains two (or possibly four) putative trans-membrane segments (Fig 4) The complex ofgp91phox⁄ Nox2 with p22phoxin phagocytes is known asflavocytochrome b558 In the C-terminal cytoplasmicregion of p22phox, there exists a PRR, which serves as
an anchoring site, thereby juxtaposing the catalyticcenter gp91phox and the SH3 domain-containing regu-latory proteins p47phox, p67phox, and p40phox; on theother hand, the small GTPase Rac functions in Noxactivation by interacting with p67phox or its homolo-
Trang 8gous proteins Detailed mechanisms for regulation by
these proteins will be described below Complex
forma-tion of p22phoxwith Nox also contributes to the
stabil-ization of each protein [1–10] Formation of themutually stabilizing complex appears to require thecorrect folding of Nox, because heme incorporationinto gp91phox⁄ Nox2 is essential for heterodimer forma-tion [84]
It is known that human Duox2 associates with aspecific maturating protein named DuoxA2, whichcontains putative five transmembrane helices [67,85](Fig 6) The membrane protein DuoxA2 allows thetransition from the endoplasmic reticulum to the Golgiapparatus, maturation and translocation to the plasmamembrane of functional Duox2 [67,85] Interestingly,the DuoxA2 gene is arranged head-to-head and coex-pressed with the Duox2 gene; the gene for DuoxA1, aparalog of DuoxA2, is similarly linked to the Duox1gene [85] Biallelic inactivation of the DuoxA2 genehas recently been reported as a novel cause of congeni-tal hypothyroidism [86], confirming its crucial contri-bution to function of Duox2, which is directlyinvolved in thyroid hormone synthesis [60]
The Nox subfamilies in animals
The Nox enzymes in animals can be divided into threesubfamilies: one containing Nox1–Nox4 (the Nox1–4
Fig 5 Structure of the SH3 domain-containing enzyme NoxSH3 in
Na gruberi The two NoxSH3 enzymes in Na gruberi, tentatively named NoxSH3-1 and NoxSH3-2, contain 627 and 630 amino acids, respectively The amino acid sequences of NoxSH3-1 and NoxSH3-2 are shown: the heme-coordinated His residues in transmembrane segment 3 (TM3) and TM5 are shown in red; residues of FAD-bind- ing motifs are shown in green; residues of NADPH-binding motifs are shown in blue; and residues of the SH3 domain are shown in magenta A model for the structure of NoxSH3 is also shown The SH3 domain is inserted into the C-terminal NADPH-binding domain Cylinders represent six transmembrane a-helices.
Fig 6 Models for the structure of Duox1 ⁄ 2 complexed with A1 ⁄ 2 Cylinders represent six and five transmembrane a-helices of Duox1 ⁄ 2 and DuoxA1 ⁄ 2, respectively EF, Ca 2+ -binding EF-hand motif.
Trang 9Duox-subgroup), which form a heterodimer with p22phox[87–
93]; the Nox5 subfamily; and the Duox subfamily
(Fig 4) Although the Nox5 and Duox subfamilies are
not found in the sea anemone Ne vectensis of the
Cnidaria (the basal group of animals)
(http://geno-me.jgi/psf.org/euk_home.html) [94] (Fig 3), they were
probably present at the protostome–dueterostome
diver-gence This is because the Nox5 and Duox subfamilies
exist in both extant protostomes and deuterostomes,
with the exception that Nox5 has been lost in the lineage
of the phylum Nematoda and in that of the order
Rod-entia in mammals: Nox5 is absent in the nematode
Ca elegansand the rodents mouse and rat (Fig 3)
On the other hand, Nox2, a member of the Nox1–4
subgroup, was present before the divergence of the
Choanoflagellata and Metazoa (equivalent to animals):
Nox2 is found not only in the Cnidaria but also in the
choanoflagellate Mo brevicollis (http://genome.jgi/
psf.org/euk_home.html) [94] (Fig 3) During the
evolu-tion of protostomes, one of the two major groups of
animals, Nox2 has been lost in the lineage of the clade
Ecdysozoa, including the phyla Arthropoda and
Nem-atoda: Nox2 is absent in the fruitfly Drosophila and
the nematode Ca elegans On the other hand, Nox2 is
present in the Lophotrochozoa, another major
proto-stomian clade, including the phyla Mollusca and
Annelida: this oxidase exists in the limpet Lottia
gigan-tea (Mollusca) and the leech Capitella species
(Annel-ida) (Fig 3) Thus, it is likely that Nox2 is the closest
to the ancestral Nox in animals In contrast, Nox1,
Nox3 and Nox4 are not found in protostomes or the
Echinodermata, a phylum that belongs to the
deuter-ostomes (a sister group of protdeuter-ostomes) These three
Nox enzymes appear to have diverged from Nox2 at
distinct stages of evolution of the phylum Chordata in
deuterostomes
The Chordata is divided into three subphyla: the
Vertebrata, Urochordata (also known as the Tunicata),
and the Cephalochordata Although tunicates were
long considered to be the earliest offshoot of the
chor-date lineage, and cephalochorchor-dates (such as
amphi-oxus) as the closest group to vertebrates, recent
analyses have reversed their positions: amphioxus is
now viewed as the ‘basal chordate’ [95], and tunicates
as the sister group, or closest relatives, of the
verte-brates [96] (Fig 3) Thus, our understanding of the
evolution of a certain protein in the Chordata requires
information on the corresponding protein in the
sub-phylum Cephalochordata The present search for the
database of the amphioxus (lancelet) Branchiostroma
floridae (http://genome.jgi/psf.org/euk_home.html)
rev-ealed that, in addition to Nox2, Nox5, and Duox,
Nox4 exists in the Cephalochordata It is known that
Nox4 is present in Ciona intestinalis (the Urochordata)[97] but not in the sea urchin Strongylocentrotus purpu-ratus of the phylum Echinodermata, another majorgroup of the Chordata [90] Therefore, Nox4 appears
to have branched from a root close to Nox2 with theemergence of the Chordata Nox1 is found from fishes
to mammals, but not in the Urochordata or chordata, suggesting that this oxidase probably arosewith the emergence of the Vertebrata Nox3 hasemerged most recently, probably from a commonancestor of birds and mammals, because it is foundsolely in mammals and birds, but not in fishes oramphibians [98] (Fig 3)
Cephalo-The membrane protein p22phox has been strated to be complexed with mammalian Nox1–Nox4.Consistent with this, p22phox is absent in the Ecdyso-zoa, where the Nox1–4 subfamily is absent, but widelydistributed in species that have Nox2, including thechoanoflagellate Mo brevicollis and the cnidarian
demon-Ne vectensis The known exceptions are the leech itella species (the Annelida) and the sea urchin S pur-puratus (the Echinodermata) Nox2 of these species orgroups might be stable without p22phox, leading to loss
Cap-of the p22phox gene It may also be possible that thep22phox gene escaped cloning or correct sequencing forunknown reasons, which is known to often occur invarious genome projects
Regulation of the Nox1–4 subfamily
in animals
The phagocyte oxidase gp91phox⁄ Nox2 in mammals isdormant in resting cells, but becomes activated duringphagocytosis to produce superoxide, a precursor ofmicrobicidal ROS The oxidase activity is spatially andtemporally restricted to the phagosome, as inappropri-ate or excessive production of ROS results in damage
to surrounding cells and severe inflammation tion of gp91phox⁄ Nox2 requires stimulus-induced mem-brane translocation of p47phox, p67phox, p40phox, andRac, i.e formation of the active oxidase complex atthe membrane (Fig 7A) The essential role of theseregulatory proteins is evident from the following twolines of evidence First, the phagocyte NADPH oxi-dase activity can be reconstituted in a cell-free systemwith gp91phox, p22phox, p47phox, p67phox, and Rac, using
Activa-an Activa-anionic amphiphile, e.g arachidonic acid, as Activa-an
in vitro stimulant Second, defects in any of the fourgenes encoding gp91phox, p22phox, p47phox and p67phoxcause the primary immunodeficiency chronic granu-lomatous disease [11,12]
In the cytoplasm of resting cells, p47phox, p67phoxand p40phox form a ternary complex, whereas Rac is
Trang 10complexed with Rho GDP dissociation inhibitor.
Upon cell stimulation, the three phox proteins are
en blocrecruited to the membrane; on the other hand,
Rac translocates independently but without Rho GDP
dissociation inhibitor, which remains in the cytoplasm
Although activation of gp91phox⁄ Nox2 complexed with
p22phoxin cells absolutely requires p47phox, p67phoxand
Rac as cytosolic regulators, p47phox is dispensable for
cell-free activation in the presence of excess amounts
of p67phox and Rac [99,100] It is thus considered that
p47phox functions as an organizer, whereas p67phox
serves as an activator that directly participates in
gp91phox⁄ Nox2 activation
Nox1 is abundantly expressed in colon epithelial
cells and also in vascular smooth muscle cells
[101,102], and seems to be involved in angiotensin
II-mediated hypertension [103–105] Nox1, as well as
Nox2, forms a complex with p22phox [88,106]
Although Nox1 is also inactive without an organizer
or an activator, it generates superoxide in the presence
of the p47phox paralog Noxo1 (Nox organizer 1) and
the p67phoxparalog Noxa1 (Nox activator 1) [106–109]
(Fig 7B) Rac is directly involved in Nox1 activation
as well [110–112]
In the inner ear of mice, Nox3 plays a crucial role in
formation of otoconia, tiny mineralized structures that
are required for perception of balance and gravity
[113] Although Nox3 also forms a functional
heterodi-mer with p22phox, this oxidase is capable of producing
superoxide in the absence of an organizer or an
activa-tor [88] The superoxide-producing activity can bestrongly enhanced by p47phox, Noxo1, and p67phox[89,110,114,115] In the presence of p67phoxor Noxa1,Nox3 activity is upregulated by Rac [110,116]
Although it is well known that Nox4 is highlyexpressed in epithelial cells of the adult and fetal kid-ney [117,118] and vascular endothelial cells [91,119], itsfunction remains to be elucidated Nox4 is complexedwith p22phox as well [80,82,84], and constitutively gen-erates superoxide in an NADPH-dependent manner[120] The mechanism of Nox4 regulation is largelyunknown at present: Nox4-mediated superoxide gener-ation appears to be independent of p47phox, Noxo1,p67phox, or Noxa1 [89,117,118], whereas the role ofRac remains controversial [79,121]
Nox2 activation
The oxidase organizer p47phoxis a 390 amino acid tein that contains, from the N-terminus, a phagocyteoxidase (PX) domain, tandem SH3 domains, and aPRR (Fig 7) The two SH3 domains cooperativelyinteract with the PRR in the C-terminal cytoplasmicregion of p22phox, an interaction that is essential forboth membrane translocation of p47phox and oxidaseactivation [122–124] The tandem SH3 domains sand-wich a short PRR of p22phox (amino acid resi-dues 151–160), containing a polyproline II (PPII) helix[125–127] (Fig 8) Pro152, Pro156 and Arg158 in thehuman p22phox PRR are indispensable for the inter-action with p47phox: Pro152 and Pro156 are recognized
pro-by the N-terminal SH3 domain, whereas Arg158directly contacts with the C-terminal one [127] (Fig 8)
On the other hand, Pro151, Pro155, Pro157 andPro160 are also involved in binding to p47phoxbut to alesser extent [127]
The gene encoding p22phoxexists in a wide variety ofanimals and also in the Choanoflagellata, a sistergroup of the Metazoa (Animalia) [128,129]; the p22phoxgene is absent in the Ecdysozoa, which is consistentwith the absence of the Nox1–4 subfamily in this clade(Fig 9) The p22phox region comprising the N-terminalcytoplasmic region and the two transmembrane seg-ments is functionally important [130] and well con-served in the Metazoa (animals) and Choanoflagellata,whereas the C-terminal cytoplasmic region is highlyvariable except for the PRR (Fig 8) The three resi-dues indispensable for binding to p47phox (Pro152,Pro156 and Arg158 in human p22phox) are invariant inall known animal p22phox proteins (Fig 8), althoughidentifiable p47phoxexists solely in the phylum Chorda-
ta, as described later (Fig 9)
Fig 7 Activation of gp91 phox ⁄ Nox2 and Nox1 (A) Interactions
required for activation of gp91 phox ⁄ Nox2 Interactions in a resting
state are indicated by blue arrows, stimulus-induced interactions by
arrows in magenta, and constitutive interactions by green arrows.
(B) Interactions required for activation of Nox1 Interactions in a
resting state are indicated by blue arrows; and stimulus-induced
interactions by arrows in magenta T1, T2, T3 and T4,
tetratricopep-tide repeats 1, 2, 3 and 4, respectively; AD, activation domain.
Trang 11In the resting state, the two SH3 domains of p47phox
are inaccessible to the target protein, as they are masked
via an intramolecular interaction with the C-terminally
flanking region called the autoinhibitory region (AIR)
[125,131–133] (Fig 7) During phagocytosis of invading
microbes or with soluble stimuli such as N-formyl
che-motactic peptide and PMA, p47phox undergoes
phos-phorylation at multiple Ser residues, several of which
are present in the AIR [134,135] Simultaneous
phos-phorylation of Ser303, Ser304 and Ser328 in the AIR, in
cooperation with other agonists such as arachidonicacid, induces unmasking of the SH3 domains [136]; as aresult, the bis-SH3 domain interacts with the PRR ofp22phox [125–127] The SH3-mediated interaction withp22phoxalso participates in p47phox-dependent enhance-ment of Nox3 activation [88]
The domain architecture of p47phox (PX, bis-SH3,AIR, and a p67phox-binding PRR) is conserved fromfishes to mammals (Fig 10) In addition, the genome
of the amphioxus B floridae, which belongs to the phylum Cephalochordata, a basal group of the Chor-data [96], contains the gene for p47phox (http://genome.jgi/psf.org/euk_home.html), in which proteinthe whole domain structure is duplicated (Fig 10) Thepresence of the p47phox gene in the Cephalochordataindicates that it emerged early in chordate evolution
sub-On the other hand, a typical p47phox is absent in theascidian Ci intestinalis of the Urochordata, suggestingthat the p47phoxgene has been lost in the ascidian line-age This organism contains a typical p22phox carryingthe conserved PRR [97], which may imply the presence
of its interacting protein Although Ci intestinalis hasbeen reported to possess a protein harboring a PXdomain and three SH3 domains, but lacks an AIR and
a PRR [97] (Fig 10), there is no experimental evidenceshowing that this protein participates in oxidase activa-tion The residues in the AIR that strongly interactwith the SH3 domains are all conserved in chordatep47phox proteins (Fig 11): in human p47phox, the Pro–Pro–Arg sequence at the AIR N-terminus adopts aPPII helix conformation, which is lined in the grooveformed between the tandem SH3 domains Amongthese residues, Pro199 and Pro200 undergo hydro-phobic interactions as addition to forming hydrogenbonds via their backbone carbonyl groups; Arg201contacts Asp243 and Glu244 via electrostatic inter-actions In addition, Ser303 forms a hydrogen bondwith the side chain of Glu241 (Fig 11) The PPRRSregion, however, is not sufficient to make the SH3domains inaccessible to p22phox [131]; its C-terminalextension is required for efficient masking of the SH3domains [131] Besides Ser303, Ser310 and Ser328 formhydrogen bonds with the side chains of Glu211 andArg267, respectively (Fig 11) It is known that phos-phorylation of Ser303 and Ser328 plays a crucial role
in disruption of the intramolecular interaction to vate the bis-SH3 domain [131,136] In the p47phoxautoinhibited structure, furthermore, Arg318 under-goes an electrostatic interaction with Asp261 [132],whereas Ile305 is located in a hydrophobic pocket ofSH3(N), and Tyr324 probably points into a hydro-phobic pocket formed by the linker region and SH3(C)(Fig 11) Intriguingly, the amino acids of the AIR and
acti-Fig 8 (A) A model for the structure of p22 phox and alignment of
the PRR Large asterisks indicate residues crucial for interaction
with the p47 phox bis-SH3 domain, and small asterisks denote
resi-dues involved in the interaction Hs, H sapiens; Mm, Mus
muscu-lus; Tr, Ta rubripes; Dr, Da rerio; Ci, Ci intestinalis; Bf, B floridae;
Lg, Lot gigantea; Nv, Ne vectensis; Mb, Mo brevicollis (B) The
complex structure of the p47 phox bis-SH3 domain with the p22 phox
PRR Crucial residues in the p22 phox PRR (Pro152, Pro156, and
Pro158) are drawn as red sticks Secondary structures in the
p22 phox PRR complexed with the p47 phox bis-SH3 domain are
indi-cated below the sequence of human p22 phox The figure was
drawn using PYMOL software (http://www.pymol.org) and the
Protein Data Bank coordinates 1WLP.
Trang 12SH3 domains, which are involved in the intramolecular
interaction as described above, are invariant among
vertebrate and amphioxus p47phox: the exceptions are
that Ser328 is replaced by Thr in Branchiostoma and
that Glu241 is replaced by Gln in the puffer fish
Takifugu rubripes and by Ile in amphioxus Thus,
autoinhibition of the p47phox SH3 domains and
phos-phorylation-mediated regulation are probably
con-served in p47phoxof chordates
It was recently proposed that the p47phox paralog
Noxo1 in the medaka fish Oryzias latipes retains an
AIR [94] However, it seems unlikely that this region is
functional, as fewer than a half of the crucial residues
described above are conserved in this fish Noxo1 In
addition, an AIR is absent in Noxo1 of the zebrafish
Danio rerio, which belongs to a more basal group than
medaka in the teleost fish [137], suggesting that the
AIR was lost at an early stage after branching from
the p47phoxgene
The homology between the N-termini of p47phoxand
p40phox was recognized when the p40phox cDNA was
cloned [138] It was subsequently pointed out that
p47phox exhibits a similarity to a region of the yeast
polarity protein Bem1 [139] This module of about 120
amino acids was later found to exist in a variety of
pro-teins [140,141], and was named the PX domain [140]
The PX domain of p47phoxis capable of binding to phoinositides such as phosphatidylinositol 3,4-bisphos-phate [PtdIns(3,4)P2], albeit with a relatively low affinityand specificity [142,143] In neutrophils, PtdIns(3,4)P2ispredominantly formed from phosphatidylinositol3,4,5-trisphosphate, a product of type I phosphatidyl-inositol-3-kinase (PI3K), which is activated upon cellstimulation The binding requires Arg43 and Lys90 inhuman p47phox[144–146] (Fig 12) Intriguingly, in con-trast to the complete conservation of Lys90, Arg43 isreplaced by Lys in the mouse and by Thr in the pufferfish [147] In contrast, the neighboring residue Arg42 iswell conserved; this residue is required for stability ofp47phox [12] but is not directly involved in binding tophosphoinositides [145] Like the p22phox-interactingactivity of the SH3 domains, the lipid-binding activity ofthe PX domain is negatively regulated under restingconditions [144,145] The phosphorylation-induced con-formational change of p47phox also renders the PXdomain accessible to membrane phosphoinositides Thisinteraction, in cooperation with the binding to p22phox,allows p47phox to be targeted to flavocytochrome b558,which is crucial for phagocyte oxidase activation [144]
phos-In addition to the phosphoinositide-binding pocket, the
PX domain may have a binding site for acidic lipids such as phosphatidic acid and phosphatidylserine
phospho-Fig 10 Domain structures of p47 phox and its related proteins The total numbers of amino acid residues of each protein are indi- cated on the right: human, H sapiens; zebrafish, Da rerio; ascidian, Ci intestinalis; amphioxus, B floridae.
Fig 9 Distribution of Nox2, p22 phox , p47 phox , p67phoxand p40phoxin animals and choano- flagellates M b., Mo brevicollis; N v.,
Ne vectensis; L g., Lot gigantea; C sp I, Capitella sp I; C e., Ca elegans; D m.,
Dr melanogaster; S p., S purpuratus; B f.,
B floridae; C i., Ci intestinalis; D r.,
Da rerio; X t., X tropicalis; G g., Ga gallus;
H s., H sapiens Note that it remains unclear whether p22 phox , p47 phox , p67 phox and p40 phox are present or absent in the annelid Capitella sp., because sequencing of the whole gen- ome has not been completed Amino acid sequences of p22 phox , p47 phox , p67 phox and p40phoxin the Cephalochordata B floridae are shown in supplementary Fig S1.
Trang 13(Fig 12), which facilitates membrane binding of the PX
domain [145,148] Another lipid-binding activity may
explain well why the p47phoxPX domain participates in
membrane translocation even in the absence of the
PI3K pathway products such as PtdIns(3,4)P2 [144]
The two residues Lys55 and Arg70 in the proposed
second binding site (Fig 12) are well conserved in
chordates, suggesting a role in oxidase activation
Noxo1, the organizer for Nox1
activation
Noxo1 plays an essential role in Nox1 activation This
organizer exhibits a domain architecture similar to that
of p47phox, except that it lacks an AIR (Fig 10).Noxo1 probably tethers the activator Noxa1 to Nox1:Noxo1 functions via its PRR by interactingwith Noxa1 and via its SH3 domains by binding to theNox1 partner p22phox [106,110,111,149] In cellsexpressing Noxo1 and Noxa1, Nox1 generates super-oxide without stimulants such as PMA, a potent
in vivoactivator of gp91phox⁄ Nox2, although the oxide production is further enhanced by PMA [106].The constitutive activity of Nox1 appears to be at leastpartially due to the absence of AIR in Noxo1; henceits SH3 domains are accessible to p22phox even in theresting state, in contrast to the p47phox SH3 domains[106,115] Regulation of the Noxo1 SH3 domains may
super-be more complicated than previously expected, as theyappear to interact intramolecularly with the N-terminalPRR, albeit with low affinity [149] It is likely that thebis-SH3 domain is partly accessible to the p22phoxPRR, and this intermolecular binding is facilitated bydisruption of the intramolecular interaction with thep47phoxPRR [149]
The PX domain of Noxo1 exhibits a tide-binding activity, which is also crucial for activa-tion of Nox1 [109,150,151] In addition to its essentialrole in Nox1 activation, Noxo1 is capable of enhanc-ing Nox3-catalyzed superoxide production[88,110,114,115,152]; the enhancement requires the PXdomain [150] and the SH3-mediated interaction withp22phox[88,157]
The oxidase activator p67phox of 526 amino acidstranslocates upon cell stimulation to the membrane in
Fig 11 Intramolecular interaction of the bis-SH3 domain with the
AIR in p47phox Residues crucial for the intramolecular interaction
with the bis-SH3 domain are highlighted (upper panel) The
sequence of the AIR in human p47 phox and the consensus AIR
sequence among p47phoxproteins derived from various species are
shown (middle panel) Large asterisks indicate residues crucial for
the intramolecular interaction with the bis-SH3 domain; small
aster-isks denote residues that play a role in the interaction Secondary
structure elements of the p67phox-binding region in p47phoxare
indi-cated below the sequence The three Ser residues Ser303, Ser310
and Ser328 directly interact with residues in the bis-SH3 domain
(Gle241, Glu211, and Arg267, respectively) (lower panel) The
figures were drawn using PYMOL software (http://www.pymol.org)
and the Protein Data Bank coordinates 1UEC.
Fig 12 Lipid-binding sites in the p47 phox PX domain Residues in the phosphoinositide-binding and second anion-binding pockets are shown in blue The sulfates bound in the two pockets (in a crystal
of the p47 phox PX domain [145]) are colored magenta The figure was drawn using PYMOL software (http://www.pymol.org) and the Protein Data Bank coordinates 1O7K.
Trang 14a manner dependent on p47phox p67phox contains two
SH3 domains (Fig 7A), both of which play a role in
activation of the phagocyte NADPH oxidase [153]
Although the N-terminal (the first) SH3 domain is the
most conserved region in p67phox [154], it is unknown
how this module functions The C-terminal SH3
domain participates in membrane translocation of
p67phox by specifically binding to the C-terminal PRR
of p47phoxwith high affinity [155–157] The high
affin-ity and specificaffin-ity are achieved by the following
mecha-nism: in addition to the canonical PxxP binding site of
p67-SH3(C), which is occupied by the eight PRR
resi-dues of p47phox (Gln362–Pro369), which adopt a PPII
helix conformation (Fig 13), p67-SH3(C) makes direct
contacts with the remaining C-terminal segment of the
p47phox tail, which comprises two a-helices (Asp372–
Asn376 and Glu380–Ser386), linked by a turn of
Arg377–Ser379 [157] (Fig 13) Structural analysis and
comparison of p47phox proteins from various species
reveal the consensus sequence xPxøPxRP(S⁄
A)xxøIL-xRC(S⁄ T)xxT(K ⁄ R)(R ⁄ K)xØ, where x is any amino
acid and Ø is a hydrophobic residue (Fig 13) The
res-idue adjacent to the invariant Cys is a Ser or a Thr;
the corresponding residue in human p47phox(Ser379) isknown to undergo phosphorylation during cell stimu-lation [158] This phosphorylation attenuates the bind-ing to p67phox [158], and thus negatively regulatesoxidase activation [159]
The consensus sequence is also conserved in theC-terminal region of Noxo1, which is capable of inter-acting with p67phox as well as with Noxa1 [106] Theinteraction of the Noxa1 SH3 domain with the Noxo1PPR (Fig 7B) plays an important role in both mem-brane localization of Noxa1 and activation of Nox1[110,111]
Rac, a member of the Rho-family small GTPases,plays an essential role in gp91phox⁄ Nox2 activation[160–162] Among phagocytes, human neutrophils pre-dominantly express Rac2, whereas both Rac1 andRac2 are present in monocytes⁄ macrophages Apatient with abnormal neutrophil function was shown
to have an inhibitory (dominant-negative) mutation inRac2, resulting in decreased oxidase activity and otherneutrophil function defects [163,164] Upon cell stimu-lation, Rac is recruited to the membrane independently
of p47phox or p67phox The recruitment requires nylgeranylation of Cys189 at the C-terminus At themembrane, Rac is converted to the GTP-bound activeform via the function of guanine nucleotide-releasingfactors Rac in the GTP-bound form directly interactswith the p67phoxN-terminal region of about 200 aminoacids, which is crucial for activation of gp91phox[165,166] On the other hand, Cdc42 neither binds top67phoxnor activates gp91phox
gera-In the Rac-binding region, four tetratricopeptiderepeat (TPR) motifs are found (Fig 14): TPR motifsare degenerate 34 amino acid repeats that are present
in a variety of organisms, ranging from bacteria tohumans, and are involved in a variety of protein–pro-tein interactions [167] The TPR domain of p67phoxcontains an insertion of 19 amino acids between thea-helices of TPR3 and TPR4: the insertion containstwo short antiparallel b-strands and a 310helical turn[168,169], with the consensus sequence of LRGNxøID-YxxLGøx(Y⁄ F)KLx, where x is any amino acid and Ø
is a hydrophobic residue (Fig 14) The b-hairpin tion and the loops connecting TPR1 and TPR2, andTPR2 and TPR3, create the binding surface for Rac[168] (Fig 14) The invariant Arg in the b-hairpin(position 32 of TPR3; Arg102 in human p67phox) plays
inser-a key role in complex forminser-ation with Rinser-ac inser-and oxidinser-aseactivation, and the side chain of this residue undergoesdirect hydrogen bonding interactions with main and
Fig 13 Interaction of the p67phoxC-terminal SH3 domain with the
p47 phox C-terminus (A) Ribbon diagram of the structure of the
com-plex of the p67 phox C-terminal SH3 domain with the p47 phox
C-ter-minal region Residues involved in binding to the SH3 domains are
drawn as red sticks (residues in the PPII helix) or magenta sticks
(residues in the a-helices) The figure was drawn using PYMOL
software (http://www.pymol.org) and the Protein Data Bank
coordi-nates 1K4U (B) Secondary structure elements of the p67 phox
-bind-ing region in p47 phox are indicated below the sequence Asterisks
indicate residues crucial for interaction with p67 phox ; the dot
denotes Ser379, which becomes phosphorylated upon cell
stimu-lation.