Tài liệu Báo cáo khoa học: Expression and function of Noxo1c, an alternative splicing form of the NADPH oxidase organizer 1 doc

On the other hand, cell stimulation with phorbol 12-myristate 13-acetate PMA, an activator of Nox1–3, facilitates membrane transloca-tion of Noxo1c; as a result, Noxo1c is equivalent to

form of the NADPH oxidase organizer 1

Ryu Takeya1,2, Masahiko Taura1, Tomoko Yamasaki1, Seiji Naito3 and Hideki Sumimoto1,2

1 Medical Institute of Bioregulation, Kyushu University, Fukuoka, Japan

2 CREST, Japan Science and Technology Agency, Saitama, Japan

3 Department of Urology, Kyushu University Graduate School of Medical Sciences, Fukuoka, Japan

Members of the NADPH oxidase (Nox) family

pro-duce superoxide from molecular oxygen in conjunction

with oxidation of NADPH [1–10] Superoxide

gener-ated serves as a precursor of other reactive oxygen

spe-cies, which are currently considered to be involved

in various physiological processes The founder Nox

enzyme gp91phox, also termed Nox2, is predominantly

expressed in professional phagocytes, and plays a

cru-cial role in host defense; superoxide generation by

gp91phoxleads to subsequent formation of microbicidal

reactive oxygen species such as hydroxyl radical and

hypochlorous acid Nox1, the second member of the Nox family, is abundant in the colon and vascular tissues [11,12] and considered to participate in host defense at the colon and signaling to cell proliferation [11,13,14] Recent studies have revealed that Nox1, expressed heterologously, associates with the mem-brane-integrated protein p22phox to form a functional heterodimer [15,16]

Activation of gp91phox, also complexed with p22phox, absolutely requires the two cytosolic proteins p47phox and p67phox Nox organizer 1 (Noxo1) and Nox

Keywords

NADPH oxidase (Nox1); Nox organiser 1

(Noxo1); PX domain; phosphoinositide

Correspondence

H Sumimoto, Medical Institute of

Bioregulation, Kyushu University,

Fukuoka 812-8582, Japan

Fax: +81 92 642 6807

Tel: +81 92 642 6806

E-mail: hsumi@bioreg.kyushu-u.ac.jp

(Received 31 May 2006, accepted 12 June

2006)

doi:10.1111/j.1742-4658.2006.05371.x

Activation of the superoxide-producing NADPH oxidase Nox1 requires both the organizer protein Noxo1 and the activator protein Noxa1 Here

we describe an alternative splicing form of Noxo1, Noxo1c, which is expressed in the testis and fetal brain The Noxo1c protein contains an additional five amino acids in the N-terminal PX domain, a phosphoinosi-tide-binding module; the domain plays an essential role in supporting superoxide production by NADPH oxidase (Nox) family oxidases including Nox1, gp91phox⁄ Nox2, and Nox3, as shown in this study The PX domain isolated from Noxo1c shows a lower affinity for phosphoinositides than that from the classical splicing form Noxo1b Consistent with this, in rest-ing cells, Noxo1c is poorly localized to the membrane, and thus less effect-ive in activating Nox1 than Noxo1b, which is constituteffect-ively present at the membrane On the other hand, cell stimulation with phorbol 12-myristate 13-acetate (PMA), an activator of Nox1–3, facilitates membrane transloca-tion of Noxo1c; as a result, Noxo1c is equivalent to Noxo1b in Nox1 acti-vation in PMA-stimulated cells The effect of the five-amino-acid insertion

in the Noxo1 PX domain appears to depend on the type of Nox; in activa-tion of gp91phox⁄ Nox2, Noxo1c is less active than Noxo1b even in the presence of PMA, whereas Noxo1c and Noxo1b support the superoxide-producing activity of Nox3 to the same extent in a manner independent of cell stimulation

Abbreviations

CHO, Chinese hamster ovary; GST, glutathione S-transferase; HA, hemaglutinin; Nox, NADPH oxidase; Noxo1, Nox organizer 1; Noxa1, Nox activator 1; PMA, phorbol 12-myristate 13-acetate; PtdIns(3)P, phosphatidylinositol 3-phosphate; PtdIns(4)P, phosphatidylinositol 4-phosphate; PtdIns(3,5)P2, phosphatidylinositol 3,5-bisphosphate; PX, phox homology.

activator 1 (Noxa1), novel respective homologs of

p47phox and p67phox, have been identified by several

groups including ours [15,17,18] Noxo1 and Noxa1

are both required for activation of Nox1 [15,17–19]

The organizers p47phox and Noxo1 each contain two

SH3 domains, arranged tandemly In p47phox, the SH3

domains normally interact intramolecularly with the

autoinhibitory region, which prevents the domains

from binding to the target p22phox A conformational

change in p47phox, which can be induced by its

phosphorylation by protein kinase C, enables the

pro-tein to access p22phox, leading to superoxide

produc-tion On the other hand, Noxo1 lacks the

autoinhibitory region [15,17,18]; its SH3 domains are

capable of binding to p22phox even in a resting state

[15] This seems to explain why cell stimulation with

phorbol 12-myristate 13-acetate (PMA), a potent

acti-vator of protein kinase C, is not required for

Noxo1-dependent superoxide-producing activity of Nox1 [15]

In addition to the SH3 domains, Noxo1 and p47phox

harbor a phagocyte oxidase (phox) homology (PX)

domain in the N-terminus PX domains occur in a

variety of proteins involved in cell signaling,

mem-brane trafficking, and polarity establishment, and

func-tion as phosphoinositide-binding modules in the

assembly of proteins at membrane surfaces [20–22]

Through the interaction with phosphoinositides, the

PX domain of p47phox plays a crucial role in

mem-brane recruitment of the protein and subsequent

activation of the phagocyte oxidase [23] The

phospho-inositide-binding activity of the Noxo1 PX domain

seems also to be involved in activation of Nox1 [19]

The third oxidase, Nox3, is involved in otoconia

forma-tion in mouse inner ears [24], and appears to be

consti-tutively active even in the absence of an oxidase

organizer (p47phox or Noxo1) or an oxidase activator

(p67phoxor Noxa1) [25] Nox3, like gp91phoxand Nox1,

forms a functional complex with p22phoxin transfected

cells; and the organizers p47phoxand Noxo1 are capable

of enhancing the superoxide production by Nox3 via

the interaction of their SH3 domains with p22phox[25–

27] Although the SH3 domain of Noxo1 participates in

regulation of Nox3, the role of the PX domain in Nox3

activity remains unknown

In the process of cloning of human Noxo1, some

spliced transcripts of the NOXO1 gene have been

identified [15,17–19] They seem to be formed by

alternative splicing at two distinct sites, which results

in insertion of one amino acid at one site and⁄ or five

amino acids at another site in the PX domain

How-ever, little is known about the expression pattern of

the splicing variants, as they could not be

distin-guished in assays used in previous studies of

expression of the Noxo1 mRNA [15,17,18] In addi-tion, it remained to be elucidated whether each tran-script possesses the activity to support activation of the Nox enzymes In this study, we show the expres-sion of alternatively spliced transcripts of the NOXO1 gene, by PCR using variant-specific primers, and the roles of the protein products in activation of Nox oxidases

Results and Discussion Alternative splice forms of human Noxo1

We have previously identified a transcript (AB097667)

of human NOXO1 gene encoding 371 amino acids [15], which is identical with that reported by Ba´nfi

et al (AF539796) and Cheng & Lambeth (AF532984) [17,19] (Fig 1A) On the other hand, Geiszt et al reported the alternative transcript (AY255768) [18], which encodes a protein lacking Lys50 To investigate the relative abundance of spliced variants of Noxo1,

we performed PCR experiments using cDNA panels as template, and sequenced the PCR products (for details, see Experimental procedures) The sequencing analysis revealed that the transcript that we have previously reported (AB097667, AF532984, and AF539796), cur-rently referred to as Noxo1b [28], is the major mRNA form in various human tissues including the colon Another alternative transcript, Noxo1c (AF532985), is abundantly expressed in the testis This transcript is generated by the use of the alternative splice donor site

of the ends of exon 3 (Fig 1A) and thus contains five additional amino acids in the PX domain (Fig 1B) The five-amino-acid insertion is not expected to alter the overall PX structure of Noxo1, as the insertion is located in a loop between the polyproline II helix and a3 helix of the PX domain [29], where a considerable sequence divergence occurs among various PX domains (Fig 1C) On the other hand, the insertion in the loop may affect the affinity for phosphoinositides This loop is expected to play an important role, because the corresponding loop of p40phoxand p47phox

is directly involved in the interaction with phospho-inositides [30,31] The loop region in the PX domain

of p40phox faces the phosphoinositide-binding pocket

as shown by the crystal structure of the p40phox PX domain bound to phosphatidylinositol 3-phosphate [PtdIns(3)P]; Lys92 in the loop is critical for binding

to phosphoinositides [30] Lys79 in the loop of p47phox also seems to contribute to phosphoinositide binding [31] The five-amino-acid insertion in the loop of Noxo1 might alter the configuration of the phospho-inositide-binding region, affecting the affinity for

phosphoinositides The other two variants with

dele-tion of Lys50, Noxo1a (AY255768 and AF532983)

and Noxo1d (AY191359), have been deposited in the

GenBank database: the deletion in these variants is

generated by alternative splicing involved in a different

splice acceptor site of exon 3 In the present PCR experiments, Noxo1a was expressed in skeletal muscle and Noxo1d in the brain; these two variants were expressed to a much lesser extent than Noxo1b and Noxo1c (data not shown)

A

C

B

Fig 1 Alternative splice forms of human Noxo1 (A) The genomic organization of human NOXO1 gene Translated sequences are shown as black boxes, and untranslated sequences as open boxes In the lower panel, sequence around splice sites of the 3rd exon are shown Intron sequences are shown in lower case, and exon sequences in upper case A five-amino-acid insertion of Noxo1c is underlined (B) Schematic presentation of domain structures of Noxo1 and the location of the five-amino-acid insertion in the PX domain SH3, Src-homology 3 domain; PRR, proline-rich region (C) Sequence alignments of the PX domains of Noxo1, p47 phox , p40 phox , SNX3, and Vam7p The alignments take the secondary structure of the p47 phox PX domain into account [29] A consensus sequence is shown on the top, where # indicates hydro-phobic residues A five-amino-acid insertion of Noxo1c is highlighted Lys92 in p40phoxand Lys79 in p47phox, mentioned in the text, are underlined.

Expression of Noxo1b and Noxo1c in various

human tissues

To compare the distribution pattern of Noxo1b and

Noxo1c, we designed variant-specific primers as shown

in Fig 2A: the primers ‘a’ and ‘b’, which are specific

for Noxo1b and Noxo1c, respectively The specificity

of each primer was confirmed by PCR using control

plasmids (Fig 2B) With these specific primers, we

studied expression of the messengers of Noxo1b and

Noxo1c by PCR using the cDNA panel of various

human tissues As shown in Fig 2C, the mRNA for

Noxo1c was expressed substantially in the testis but

only slightly in the colon On the other hand, the

Noxo1b mRNA was relatively abundant in the colon

and also present in the testis, liver, thymus, and kidney

but to a lesser extent (Fig 2C) Among fetal organs

tested, the message of Noxo1c was most abundantly

expressed in the brain (Fig 2D) Similarly, the Noxo1b

mRNA was most abundant in the brain among the

fetal organs, although it was also present in the

thy-mus, liver, and kidney To compare the amounts of

the two variants expressed in the testis, we performed

the PCR using the primers ‘d’ and ‘c’, where the

cDNAs for both Noxo1b and Noxo1c are amplified as

the insert region is located between the pair of primers

(Fig 2A) The lengths of the PCR products of Noxo1b

and Noxo1c were 225 and 240 bp, respectively To

delineate the difference in the two products, we

subjec-ted the PCR fragments to PAGE; the two fragments

were clearly separated As shown in Fig 2E, Noxo1c

and Noxo1b were almost equally expressed in the tes-tis, whereas Noxo1b was the major form in the colon

To investigate the physiological relevance of Noxo1c expression in the testis, we examined expression of Nox1 and Noxa1 by PCR analysis and found a small but significant amount of the Nox1 and Noxa1 mRNAs in the testis (data not shown), which is consis-tent with the previous observation by Cheng et al [32]

It has also been shown that Nox1 is present in the androgen-independent prostate cancer LNCaP cells [33] RT-PCR analysis revealed that LNCaP cells abundantly expressed the mRNA for Noxo1c (Fig 2F) The Noxo1c mRNA also existed in several Nox1-expressing human cancer cell lines: the andro-gen-independent prostate cancer PC3 and DU145 cells and the testicular germ cell tumor NEC8 cells (Fig 2F) Noxa1, a protein that activates Nox1 in co-operation with Noxo1 [15,17–19], was also expressed (Fig 2F) in LNCaP, PC3, and DU145 cells, suggesting that Nox1 is regulated by Noxo1c and Noxa1 in these cancer cells The role of Nox1 in prostate tumors has been suggested: Nox1 seems to increase tumorigenicity

of DU145 prostate cancer cells [34]; and increased expression of endogenous Nox1 is observed in parallel with increasing tumor and metastatic potential in a series of cell lines developed from LNCaP cells [35] As the mRNA for Noxo1c was detected in fetal brain (Fig 2D), we also investigated its expression by PCR using the cDNA panel of the human fetal neural sys-tem As shown in Fig 2G, the Noxo1c mRNA as well

as the Noxo1b mRNA was expressed in the occipital

Fig 2 Expression of Noxo1b and Noxo1c in human tissues (A) Location of primers for PCR analyses The cDNA primers ‘a’ and ‘b’ are designed as specific primers for Noxo1b and Noxo1c, respectively The reverse primer ‘c’ is used in combination with sense primers ‘a’ and

‘b’ in PCR analyses using human adult tissues (C) and fetal tissues (D) The primers ‘d’ and ‘c’ are used in the PCR analyses (E) where both Noxo1b and Noxo1c are simultaneously amplified (B) The specificity of the variant-specific primers With the indicated combination of prim-ers, PCR was performed using cDNAs for Noxo1b or Noxo1c as a template, and the PCR products were subjected to 2% agarose-gel elec-trophoresis, and stained with ethidium bromide (C) Expression of Noxo1b and Noxo1c in human adult tissues The expression levels of Noxo1b and Noxo1c were analyzed by PCR using Human Multiple Tissue cDNA panels (Clontech): sk muscle, skeletal muscle; small intest., small intestine The PCR products were subjected to 2% agarose-gel electrophoresis, and stained with ethidium bromide The experiments have been repeated more than three times with similar results (D) Expression of Noxo1b and Noxo1c in human fetal tissues The expres-sion levels of Noxo1b and Noxo1c were analyzed by PCR using Human Fetal Multiple Tissue cDNA panels (Clontech) The PCR products were subjected to 2% agarose-gel electrophoresis, and stained with ethidium bromide The experiments have been repeated more than three times with similar results (E) The DNA fragments amplified by PCR using primers ‘d’ and ‘c’ were subjected to 10% polyacrylamide gel electrophoresis For details, see Experimental procedures The experiments have been repeated more than three times with similar results (F) Expression of Noxo1b, Noxo1c, Nox1, and Noxa1 in various human cell lines: androgen-independent prostate cancer cells (LNCaP), androgen-independent prostate cancer cells (PC3 and DU145), and testicular germ cell tumor cells (NEC8) The expression levels were analyzed by RT-PCR using total RNA extracted from each cell line as a template The DNA fragments for Noxo1b and Noxo1c were subjected to 10% polyacrylamide gel electrophoresis (upper panel) as in (E), and the fragments for Nox1 and Noxa1 were subjected to 2% agarose-gel electrophoresis (middle and lower panels) The experiments have been repeated more than three times with similar results (G) Expression of Noxo1b and Noxo1c in human fetal neural tissues The expression levels of Noxo1b and Noxo1c were analyzed by PCR using Human Fetal Neural Tissue cDNA panels (Biochain Institute) The PCR products were subjected to 10% polyacrylamide gel electrophoresis, and stained with ethidium bromide as in (E) The experiments have been repeated more than three times with similar results.

B

C

lobe, parietal lobe, pons, cerebellum, and spinal cord,

suggestive of the role in neurons

Activation of Nox1 by Noxo1b and Noxo1c in

Chinese hamster ovary (CHO) cells

It is well known that the classical splicing form

Noxo1b is essential for activation of Nox1 [15,17–19]

To know the activity of Noxo1c to activate Nox1, we

transfected Chinese hamster ovary (CHO) cells with pcDNA3.0–Nox1, pEF-BOS–p22phox, pEF-BOS– Noxa1, and pEF-BOS–Noxo1b or pEF-BOS–Noxo1c

As shown in Fig 3A, Noxo1b and Noxo1c equally supported Nox1 activation on stimulation with PMA Without stimulants added, Noxo1c also activated Nox1 but to a lesser extent than Noxo1b (Fig 3A,B)

On the other hand, Noxo1a, a minor spliced tran-script, was much less active even in the presence of PMA (Fig 3B) We further investigated the stimulus-independent activity of Nox1 using CHO cells trans-fected at various amounts of the Noxo1b or Noxo1c cDNA As shown in Fig 3C, Noxo1c was less active than Noxo1b in resting cells, indicating the difference between Noxo1b and Noxo1c

In the above experiments, we used CHO cells detached from culture dishes using trypsin⁄ EDTA to measure superoxide production It is known that the modulation of cell–cell adhesion can activate certain intracellular signaling pathways including the small GTPase Rac [36]; Rac seems to participate in Nox1

A

B

C

D

Fig 3 Noxo1c-supported activation of Nox1 (A) CHO cells were cotransfected with pcDNA3.0–Nox1, pEF-BOS–p22 phox , pEF-BOS– myc-Noxa1, and simultaneously with BOS–HA-Noxo1b, pEF-BOS–HA-Noxo1c, or pEF-BOS–HA-Noxo1a The transfected cells (1 · 10 6

cells) were incubated for 10 min at 37 C, and then stimu-lated with PMA (200 ngÆmL)1) Chemiluminescence change was continuously monitored with Diogenes, and superoxide dismutase (50 lgÆmL)1) was added where indicated (left panel) Expression of variant Noxo1 proteins in the transfected cells was determined by immunoblot analysis with the monoclonal antibody to HA (right panel) (B) CHO cells were cotransfected with pcDNA3.0–Nox1, pEF-BOS–p22 phox , pEF-BOS–myc-Noxa1, and simultaneously with pEF-BOS–HA-Noxo1b, pEF-BOS–HA-Noxo1c, or pEF-BOS–HA-Noxo1a Superoxide production was assayed by chemilumines-cence using Diogenes in the presence or absence of PMA (200 ngÆmL)1) Each graph represents the mean ± SD of the peak chemiluminescence values obtained from three independent trans-fections (C) CHO cells were transfected simultaneously with pcDNA3.0–Nox1 (1 lg), pEF-BOS–p22 phox (1 lg), pEF-BOS–myc-Noxa1 (1 lg), and the indicated amount of pEF-BOS–HA-Noxo1b (left panel) or pEF-BOS–HA-Noxo1c (right panel) Superoxide pro-duction was assayed by chemiluminescence using Diogenes in the presence or absence of PMA (200 ngÆmL)1), and expressed as the percentage activity relative to that of 1 lg pEF-BOS–HA-Noxo1b (left panel) or pEF-BOS–HA-Noxo1c (right panel)-transfected cells in the presence of PMA (D) Superoxide production in adherent CHO cells undetached from culture dishes CHO cells were

cotransfect-ed with pcDNA3.0–Nox1, pEF-BOS–p22 phox , pEF-BOS–myc-Noxa1, and simultaneously with pEF-BOS–HA-Noxo1b, pEF-BOS–HA-Noxo1c PMA-independent superoxide production was assayed by chemiluminescence using Diogenes in the presence or absence of superoxide dismutase (50 lgÆmL)1) at 37 C The graph represents the mean ± SD of chemiluminescence values obtained from three independent transfections.

activation [37] To test the possibility that cell

detach-ment elicits activation of Nox1, we estimated

superox-ide production by adherent cells Adherent CHO cells,

coexpressing Nox1 with both Noxo1b and Noxa1,

pro-duced a considerable amount of superoxide (Fig 3D)

Under the same conditions, Noxo1c weakly supported

superoxide production compared with Noxo1b,

con-firming that Noxo1c is more effective than Noxo1b in

activating Nox1 in unstimulated cells

Intracellular localization of Noxo1b and Noxo1c

Noxo1 appears to exist in a constitutively active form

[15], and is reported to be located in the membrane of

resting cells [19] The present finding that Noxo1c

weakly supports a stimulus-independent activation of

Nox1 suggests that Noxo1c is less associated with the

membrane in a resting state than Noxo1b To test the

possibility, we examined intracellular localization of

the Noxo1 proteins ectopically expressed in CHO cells

Noxo1b was barely located in the cytoplasm but

concentrated in punctate intracellular structures (Fig 4A); they resemble fused endosomes on which Noxo1b is reported to be located [19] On the other hand, Noxo1c was located in the cytoplasm and not concentrated in any internal membrane structures within the cytoplasmic compartment Less association

of Noxo1c with the membrane-integrated protein p22phox, the partner of Nox1 (Fig 4A), may be consis-tent with the finding that Noxo1c weakly supports superoxide production by Nox1 in resting CHO cells more weakly than Noxo1c (Fig 3) We also attempted but failed to assess the intracellular localization of Noxo1b after treatment of the cells with PMA, as the cells became rounded without detaching from cover-slips To biochemically assess the localization of Noxo1s after treatment with PMA, we prepared the membrane fraction and tested the localization of Noxo1c As shown in Fig 4B, Noxo1c localized to the membrane only partly in resting cells, but was fur-ther targeted to the membrane after stimulation with PMA On the other hand, Noxo1b was constitutively

A

C

PMA: (–) (+)

Blot: anti-HA Blot: anti-p22 phox

PMA: (–) (+)

0

2

1

3

B

Merge

Fig 4 Intracellular localization of Noxo1b

and Noxo1c in CHO cells (A) Intracellular

localization of Noxo1b (upper panels) and

Noxo1c (lower panels) in quiescent CHO

cells In merged images (right panels),

local-ization of Noxo1b and Noxo1c is shown in

green, and p22 phox in red Scale bars,

20 lm (B) Membrane translocation of

Noxo1c Before or after cell stimulation with

PMA (200 ngÆmL)1), the cell lysates were

fractionated by centrifugation, and the

mem-brane fractions were analyzed by

immuno-blot with antibodies to HA or p22phoxas a

loading control These experiments have

been repeated more than three times with

similar results (C) The extent of membrane

localization of Noxo1b and Noxo1c The

intensities of immunoreactive bands for

HA-Noxo1b and HA-Noxo1c in (B) were

quantified using a LAS-1000plus (Fuji film)

image analyzer and expressed as the fold

increase relative to that of the band for

Noxo1c in the absence of PMA.

associated with the membrane fraction (Fig 4B) As

the extent of the membrane localization of Noxo1

cor-related well with that of the superoxide-producing

activity of Nox1 (Fig 4C), a weaker activity of

Noxo1c to support Nox1 activity in unstimulated cells

(Fig 3) may be due to the fact that Noxo1c fails to

fully localize to the membrane

The mechanism for the PMA-dependent membrane

recruitment of Noxo1c is at present unknown It is

well established that p47phox undergoes

phosphoryla-tion in response to PMA, which is essential for

mem-brane translocation of this protein As Noxo1 also has

several potential protein kinase C phosphorylation

sites, Noxo1 might become phosphorylated in

PMA-stimulated cells, leading to membrane translocation

Phosphoinositide-binding activity of the PX

domains of Noxo1b and Noxo1c

The membrane localization of Noxo1b is mediated in

part by binding of the PX domain to membrane

phos-pholipids [19] Less association of Noxo1c with the

membrane (Fig 3) raised the possibility that the

phos-pholipid-binding activity of Noxo1c may be impaired

In this context, it should be noted that Noxo1c contains

the five-amino-acid insertion in the PX domain (Fig 1)

To determine the effect of the insertion, we examined

the phosphoinositide-binding activity of the PX domain

of Noxo1c by an overlay assay, in which each

phospho-inositide was spotted on the membrane and overlaid

with glutathione S-transferase (GST)-fused PX

domains The PX domain of Noxo1b bound to

phos-phatidylinositol 3,5-bisphosphate [PtdIns(3,5)P2] with

the highest affinity, which is consistent with the recent

report of Cheng & Lambeth [19]; it also interacted with

PtdIns(3)P and phosphatidylinositol 4-phosphate [PtdIns(4)P], but to a lesser extent (Fig 5A) On the other hand, the PX domain of Noxo1c showed a weaker binding activity to the phosphoinositides under the same experimental conditions (Fig 5B) Moreover, a lipo-some-binding assay also showed that the Noxo1c PX domain interacted with phospholipids such as PtdIns(3,5)P2 more weakly than that of Noxo1b (data not shown) Thus the insertion in the PX domain decreases the affinity for phosphoinositides

Activation of gp91phoxand Nox3 by Noxo1c

We next investigated the ability of Noxo1c to activate gp91phox⁄ Nox2 and Nox3 In CHO cells expressing gp91phox⁄ Nox2 with p67phox, Noxo1c supported super-oxide production to a much lesser extent than Noxo1b (Fig 6A) When coexpressed with Noxa1, the Noxo1c-supported superoxide production by gp91phox⁄ Nox2 was severalfold less than the Noxo1b-supported one (Fig 6B) On the other hand, Noxo1c and Noxo1b showed the same ability to support Nox3 activation in the presence (Fig 6C) or absence (Fig 6D) of Noxa1; the activity was 10-fold higher than those obtained in cells expressing Nox3, with or without Noxa1, but not the Noxo1 proteins (data not shown) Thus the effect

of the five-amino-acid insertion in the Noxo1 PX domain depends on the type of Nox

Role of the interaction between Noxo1c and p22phoxin Nox1-dependent and Nox3-dependent superoxide production

It is known that Noxo1 functions via the SH3-mediated interaction with p22phox, which forms a heterodimer

PI PI3P PI4P PI5P PI(3,5)P 2

PI(4,5)P 2

PI(3,4)P 2

PI(3,4,5)P 3

PI PI3P PI4P PI5P PI(3,5)P 2

PI(4,5)P 2

PI(3,4)P 2

PI(3,4,5)P 3

Fig 5 Phosphoinositide-binding activity of the PX domains of Noxo1b and Noxo1c The GST-fusion proteins of Noxo1b-PX (amino acids 1– 153) (A) and Noxo1c-PX (amino acids 1–158) (B) were tested in an overlay lipid-binding assay using the PIP array, in which serial dilutions of indicated phosphoinositides (100, 50, 25, 12.5, 6.25, 3.13, and 1.56 pmol) were spotted: PI, phosphatidylinositol; PI3P, phosphatidylinositol 3-phosphate; PI4P, phosphatidylinositol 4-phosphate; PI5P, phosphatidylinositol 5-phosphate; PI(3,5)P2, phosphatidylinositol 3,5-bisphosphate; PI(4,5)P2, phosphatidylinositol 4,5-bisphosphate; PI(3,4)P2, phosphatidylinositol 3,4-bisphosphate; PI(3,4,5)P3, phosphatidylinositol 3,4,5-tris-phosphate For details, see Experimental procedures.

with Nox1, gp91phox⁄ Nox2, and Nox3 [25] It may be

possible that the interaction with p22phoxis blocked by

the five-amino-acid insertion in the PX domain, which

leads to impaired localization of Noxo1c to the

mem-brane To exclude this possibility, we performed an

in vitro binding assay using purified Noxo1b and

Noxo1c As shown in Fig 7A, Noxo1c-DC and

Noxo1b-DC bound to the C-terminus of p22phoxto the

same extent; the binding was completely abolished by

the P156Q substitution in p22phox, a mutation leading

to defective interaction with the SH3 domains of

Noxo1 [15] In addition, Noxo1c as well as Noxo1b

interacted with p22phoxin a similar manner in the yeast

two-hybrid system (Fig 7B) Thus the insertion in the

PX domain does not seem to affect the SH3-mediated

interaction with p22phox To confirm this, we investi-gated the dependence of Noxo1c-supported Nox activation on p22phox.It is known that superoxide pro-duction by Nox1 in CHO cells expressing Noxo1b is largely but not completely dependent on the cotrans-fection with the p22phox cDNA [15], whereas Nox3 activity requires p22phox expression under the same conditions [25] Similarly, Noxo1c-supported superox-ide production by Nox1 is partly dependent on p22phox (Fig 7C); on the other hand, the expression of p22phox was a requisite for the Noxo1c-supported Nox3 activity (Fig 7D) Thus Noxo1c probably binds to p22phoxin a manner similar to Noxo1b

Role of phosphoinositide-binding activity of Noxo1 in Nox3 activation

To study the role of the Noxo1 PX domain by itself,

we expressed a mutant Noxo1b lacking the PX domain, Noxo1b-DPX, in CHO cells The deletion of the PX domain resulted in complete loss of superoxide production by Nox1 (Fig 8A) and by gp91phox⁄ Nox1 (Fig 8B) The enhancement of Nox3 activity by Noxo1b [25] was also entirely dependent on the PX domain (Fig 8C) In contrast with the essential role of the PX domain, Noxo1c, containing a PX domain with a weak lipid-binding activity (Fig 5), is capable

of fully activating Nox3 (Fig 6)

To clarify the role of the lipid-binding activity in Nox3 activation, we examined the effect of substituting Gln for Arg40 in the PX domain, which completely abrogates the phosphoinositide-binding activity [19]

As shown in Fig 9A, a mutant Noxo1b carrying the R40Q substitution failed to support the superoxide production by Nox1 Thus the PX-mediated lipid binding is required for Nox1 activation The R40Q substitution in Noxo1c also abolished superoxide pro-duction by Nox1 (Fig 9A), supporting the conclusion that Noxo1c retains considerable lipid-binding activity (Fig 5B) On the other hand, in Nox3 activation, the mutant Noxo1 proteins were threefold less active than the wild-type one (Fig 9B), suggesting that PX-medi-ated binding to phosphoinositides is involved in, but not absolutely required for, Nox3 activity This is in contrast with the observation that the PX domain by itself is essential for Nox3 activation (Fig 8C) The partial dependence on the lipid-binding activity may explain why Noxo1c with a weak but significant lipid-binding activity (Fig 5) is equivalent to Noxo1b

in Nox3 activation (Fig 6) The idea may be suppor-ted by the observation that a part of Noxo1c as well

as Noxo1b was localized to ruffling membranes in the Nox3-transfected CHO cells (data not shown)

A

D C

B

Nox3 + Noxa1

5 3

0 1

4 2

gp91phox + Noxa1

0 0.4

1.6

0.8

gp91phox + p67phox

5

3

0

1

4

2

Noxo1 ββ Noxo1γγ

Nox3

1.2

0

0.4

0.8

PMA(–) PMA(+)

Noxo1 ββ Noxo1γγ Noxo1 ββ Noxo1γγ

Noxo1 ββ Noxo1γγ

Fig 6 Noxo1c-supported activation of gp91 phox ⁄ Nox2 and Nox3.

CHO cells were cotransfected with the following combination of

plasmids: pcDNA3.0–gp91 phox , pEF-BOS–p22 phox ,

pEF-BOS–myc-p67 phox , and simultaneously with pEF-BOS–HA-Noxo1b or

pEF-BOS–HA-Noxo1c in (A); pcDNA3.0–gp91phox, pEF-BOS–

p22 phox , pEF-BOS–myc-Noxa1, and simultaneously with pEF-BOS–

HA-Noxo1b or pEF-BOS–HA-Noxo1c in (B); pcDNA3.0–Nox3 and

pEF-BOS–p22 phox , and simultaneously with pEF-BOS–HA-Noxo1b

or pEF-BOS–HA-Noxo1c in (C); pcDNA3.0–Nox3 and pEF-BOS–

p22 phox , pEF-BOS–myc-Noxa1, and simultaneously with pEF-BOS–

HA-Noxo1b or pEF-BOS–HA-Noxo1c in (D) Superoxide production

was assayed by chemiluminescence using Diogenes in the

presence or absence of PMA (200 ngÆmL)1) Each graph represents

the mean ± SD of the peak chemiluminescence values obtained

from three independent transfections Protein levels of Noxo1b and

Noxo1c in the transfected cells were estimated by immunoblot

analysis with the monoclonal antibody to HA (lower panels).

Concluding remarks

In this study, we show that Noxo1c, a novel

alternat-ive splicing form of human Noxo1 containing an

addi-tional five amino acids in the PX domain, is expressed

in the testis and fetal brain (Fig 2) During the

revision of this manuscript, Cheng & Lambeth [28]

reported the expression and function of the four splice

forms of human Noxo1 The Noxo1c mRNA is also

expressed in several Nox1-expressing and

Noxa1-expressing human cancer cell lines [the

androgen-independent prostate cancer LNCaP cells, and the

androgen-independent prostate cancer PC3 and

DU145 cells (Fig 2)], indicating that Noxo1c regulates

Nox1 in co-operation with Noxa1 in a single cell In

PMA-stimulated cells, Noxo1c and Noxo1b support

Nox1 activation to the same extent (Fig 3) The PX

domain of Noxo1c shows a lower affinity for

phos-phoinositides than that of Noxo1b (Fig 5), which

seems to attenuate the membrane localization in resting

cells (Fig 4) Consistent with this, Noxo1c supports the stimulus-independent activity of Nox1 more weakly than Noxo1b (Fig 3) We also demonstrate that Noxo1c fails to fully activate gp91phoxeven in the pres-ence of PMA, whereas Nox3 activity enhanced by Noxo1c is almost equivalent to that by Noxo1b (Fig 7) The difference may be due to the fact that the significance of the PX-mediated lipid binding depends

on the type of Nox, although the PX domain of Noxo1 by itself is indispensable for supporting super-oxide production by all the three Nox enzymes

Experimental procedures Isolation of cDNA for splice variants of human NOXO1 gene

Based on the sequence of mRNA for human NOXO1 (GenBank accession number AB097667), we synthesized the two unique oligonucleotide primers 5¢-GCAGGATCCAT

4

2

0

3

1

Nox3 + Noxo1γγ

2.0

1.2

0 0.4

1.6

0.8 Nox1 + Noxa1 + Noxo1γγ

– p22phox + p22phox

p22phox-C (WT) p22phox-C (P156Q)

- Δ

MBP–p22phox-C

Noxo1 γ-ΔC

His (+) (–)

Noxo1 β-ΔC

p22phox-C (WT)

p22phox-C (P156Q)

His (+) (–)

- Δ

– p22phox + p22phox

Fig 7 Role of the interaction between Noxo1c and p22phoxin Nox1-dependent and Nox3-dependent superoxide production (A) Interaction between Noxo1 and p22 phox estimated by an in vitro pull-down assay using purified proteins GST–Noxo1b-DC (amino acids 1–292) or GST– Noxo1c-DC (amino acids 1–297) was incubated with MBP–p22 phox -C (amino acids 132–195) or MBP–p22 phox -C (P156Q) and pulled down with glutathione–Sepahrose 4B The precipitated proteins were subjected to SDS ⁄ PAGE, followed by immunoblot analysis with an antibody

to maltose-binding protein (MBP) (B) Interaction between Noxo1 and p22 phox estimated by the yeast two-hybrid system The yeast HF7c cells were cotransformed with recombinant plasmids pGBT9g encoding the C-terminus of the wild-type or a mutant p22 phox and pGADGH encoding Noxo1b-DC (amino acids 1–292) or Noxo1c-DC (amino acids 1–297) After the selection for Trp+and Leu+phenotype, its histidine-dependent (right) and inhistidine-dependent (left) growth was tested CHO cells were cotransfected with the following combination of plasmids: pcDNA3.0–Nox1, pEF-BOS–myc-Noxa1, pEF-BOS–HA-Noxo1c, and with or without pEF-BOS–p22 phox in (C); pcDNA3.0–Nox3, pEF-BOS– myc-Noxa1, pEF-BOS–HA-Noxo1c, and with or without pEF-BOS–p22 phox in (D) Superoxide production was assayed by chemiluminescence using Diogenes in the presence of PMA (200 ngÆmL)1) Each graph represents the mean ± SD of the peak chemiluminescence values obtained from three independent transfections.