Two Novel Proteins Activate Superoxide Generation by the NADPH Oxidase NOX1*

NOX1, an NADPH oxidase expressed predominantly in colon epithelium, shows a high degree of similarity to the phagocyte NADPH oxidase. However, superoxide generation by NOX1 has been difficult to demonstrate. Here we show that NOX1 generates superoxide when co-expressed with the p47 phox and p67 phox subunits of the phagocyte NADPH oxidase but not when expressed by itself. Since p47 phox and p67 phox are restricted mainly to myeloid cells, we searched for their homologues and identified two novel cDNAs. The mRNAs of both homologues were found predominantly in colon epithelium. Differences between the homologues and the phagocyte NADPH oxidase subunits included the lack of the autoinhibitory domain and the protein kinase C phosphorylation sites in the p47 phox homologue as well as the absence of the first Src homology 3 domain and the presence of a hydrophobic stretch in the p67 phox homologue. Co-expression of NOX1 with the two novel proteins led to stimulus-independent high level superoxide generation. Stimulus dependence of NOX1 was restored when p47 phox was used to replace its homologue. In conclusion, NOX1 is a superoxide-generating enzyme that is activated by two novel proteins, which we propose to name NOXO1 (NOX organizer 1) and NOXA1 (NOX activator 1).

Based on their primary structure all members of the NOX/ DUOX family should be flavo-heme electron transporters. However, it is not established whether all NOX enzymes transfer electrons to oxygen or whether some of them may use other electron acceptors as has been shown for a yeast homologue of gp91 phox that functions as a ferric reductase (13). Among NOX enzymes, only gp91 phox and NOX5 have appeared capable of generating large amounts of superoxide, both of them in a stimulus-dependent manner (1,9).
Based on data gained with NOX1-transfected NIH 3T3 cell clones NOX1 has been suggested to be a subunit-independent, low capacity superoxide-generating enzyme involved in the regulation of mitogenesis (4,14). However, we have not been able to measure any superoxide generation in NOX1-transfected cells, and the question arose whether NOX1 is a subunit-dependent enzyme or possibly not a superoxide-generating enzyme at all.

EXPERIMENTAL PROCEDURES
Cloning of Mouse p47 phox and p67 phox Homologues-Translated BLAST nucleotide searches were conducted in the mouse genome data base (www.ensembl.org/Mus_musculus/blastview) with the mouse p47 phox and p67 phox sequences. The exons of the genes identified were predicted with the GENSCAN software (genes.mit.edu/GEN-SCAN.html), and primers were designed to amplify the coding regions of both predicted transcripts (5Ј-catggcaagcccaagacacccagta-3Ј and 5Јctcaacggggaacccgagtccctt-3Ј for the p47 phox homologue and 5Ј-ccatgagctctctaggggatcag-3Ј and 5Ј-tgctagttctggtctcctggct-3Ј for the p67 phox homologue). Total RNA from mouse colon was purified with the TRIzol ® reagent (Invitrogen), and cDNA was synthesized with Superscript ® reverse transcriptase (Invitrogen) using random primers followed by PCR with Taq DNA polymerase using "Q buffer" (Qiagen).
Analysis of mRNA-Northern blot analysis and in situ hybridization were performed as described previously (9) using the entire coding region of NOXO1 and NOXA1 for probe generation in Northern blot experiments and using the region containing base pairs 1-360 of NOXO1 and the region containing base pairs 171-490 of NOXA1 cDNA for probe generation in the in situ hybridization experiments.
Superoxide Measurements-Luminol-enhanced chemiluminescence and SOD-inhibitable cytochrome c reduction were measured as described previously (9). For the microscopic NBT test, cells were seeded in 96-well plates and incubated at 37°C for 15 min in Hanks' balanced salt solution containing 0.5 mg/ml NBT. * This research was supported by Swiss National Foundation Grants 31-55805.98 and by United States Public Health Service, National Institutes of Health Grants AI20866 and AG19519. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM

RESULTS AND DISCUSSION
Activation of NOX1 by Cytoplasmic Subunits of the Phagocyte NADPH Oxidase-As NOX1 expression by itself did not lead to superoxide generation in transfected cells (Fig. 1), we considered that NOX1, like gp91 phox , might be a subunit-dependent enzyme. We therefore co-transfected HEK293 cells with NOX1 and the cytoplasmic subunits of the phagocyte NADPH oxidase, p47 phox and p67 phox . Cells transfected with NOX1 alone (Fig. 1) or with p47 phox and p67 phox alone (not shown) did not generate superoxide either with or without stimulation by the phorbol ester PMA. In contrast, HEK293 cells co-transfected with NOX1, p47 phox , and p67 phox generated superoxide but only after addition of PMA. These results suggested that NOX1 is a subunit-dependent enzyme. However, p47 phox and p67 phox are expressed mainly in myeloid cells. We therefore searched for homologues that may activate NOX1 in the colon.
Molecular Cloning of Putative NOX1 Cytoplasmic Subunits-A search for p47 phox homologues yielded a full-length human sequence deposited as colon cancer antigen NY-CO-31 (GenBank TM accession number AAC18046). Subsequently, we found the open reading frame of the mouse p47 phox homologue in chromosome 17 and cloned the cDNA of the coding region by reverse transcription PCR from mouse colon RNA (GenBank TM accession number AF539797). The homologue had an expected molecular mass of 39 kDa and displayed 25% sequence identity with p47 phox (Fig. 2A). A search for motifs located a phox homology domain, which targets proteins to the phosphoinositide groups of membranes (15), two Src homology SH3 domains, and a C-terminal proline-rich region that is crucial in p47 phox for interaction with p67 phox (16) (Fig. 2A). Interestingly, the homologue was missing the proline-proline-argininearginine-containing region of p47 phox , which is involved in autoinhibition (17) through binding to the N-terminal SH3 domain, as well as the adjacent serine phosphorylation sites (18), which relieve autoinhibition when phosphorylated (19). This suggested a stimulus-independent activity of the homologue.
A search of the mouse genome for p67 phox homologues yielded an open reading frame on chromosome 2. Based on the genomic sequence we cloned the mouse p67 phox homologue by reverse transcription PCR (GenBank TM accession number AF539798). The predicted protein had a molecular mass of 49 kDa and displayed 30% sequence identity with p67 phox (Fig.  2B). A computer search for motifs revealed tetratricopeptide repeats, which are important for Rac binding in p67 phox (20) (Fig. 2B). The so-called activator domain of p67 phox , which interacts with gp91 phox , was also found in the homologue (21) and it conserved the C-terminal but not the N-terminal SH3 domain of p67 phox . The p40 phox -binding PB1 domain of p67 phox was weakly conserved (22). In addition the homologue had a C-terminal hydrophobic stretch, which may represent a transmembrane tail (Fig. 2B).
After discussions with Dr. Ruth Lovering from the Human Genome Organization Nomenclature Committee and with colleagues working in the field, the novel factors have been named NOXO1 (NOX organizer 1) and NOXA1 (NOX activator 1). NOXO1 is the p47 phox homologue, and NOXA1 is the p67 phox homologue. The terms NOXO2 and NOXA2 have been intro- duced as aliases for p47 phox and p67 phox , respectively. NOXO1 and NOXA1 mRNAs Are Expressed Predominantly in Colon Epithelium-We next investigated the tissue distribution of NOXO1 and NOXA1 by Northern blotting (Fig. 3A). The probe derived from NOXO1 labeled a ϳ1.5-kb mRNA, while the NOXA1 probe labeled a ϳ1.7-kb mRNA. Interestingly, both mRNAs showed a relatively similar tissue distribution with a predominant expression in the colon. Low level expression was seen in uterus, small intestine, and stomach for both transcripts. Testis expressed NOXO1 but not NOXA1 (the increased length of the testis NOXO1 mRNA is due to differences in the non-coding regions, not shown). In situ hybridization with NOXO1 and NOXA1 antisense probes labeled colon epithelial cells strongly, while neither the NOXA1 sense probe (Fig. 3B, control) nor the NOXO1 sense probe (not shown) hybridized. Thus, NOXO1 and NOXA1 were expressed in the same tissue and same cell type as NOX1.
NOXO1 and NOXA1 Enable Superoxide Generation by NOX1-To investigate the function of NOXO1 and NOXA1, we transfected HEK293 cells with NOX1, NOXO1, and NOXA1, or with the empty expression vector (control). The cells transfected with all three constructs generated superoxide, while control-transfected cells did not (Fig. 4, A and B). Addition of a flavoprotein inhibitor, diphenylene iodonium (DPI), inhibited superoxide production (Fig. 4A). The block of NOX1 activity by DPI was slower than that observed for other NOX proteins. The half-time of DPI inhibition of superoxide generation was 239 Ϯ 16 s in NOX1-transfected HEK293 cells, while it was 40 Ϯ 1.5 s in NOX5-transfected HEK293 cells (not shown). This is, to our knowledge, the first clear indication of pharmacological variability among different NOX enzymes.
Individual transfection of NOX1, NOXO1, or NOXA1 did not lead to any detectable superoxide generation (Fig. 4B). Similarly, no superoxide generation was observed upon co-transfection of NOX1 with only one of the novel homologues or cotransfection of NOXO1 and NOXA1 without NOX1 (Fig. 4B).
The co-transfection of NOX1, NOXO1, and NOXA1 led to superoxide generation not only in HEK293 cells but also in Chinese hamster ovary, HeLa, and COS-7 cells (not shown), although we detected the highest activity in HEK293 cells.
Next we verified the NOX1-dependent superoxide generation using the SOD-inhibitable cytochrome c reduction assay. As shown in Fig. 4C, cells transfected with NOX1, NOXO1, and NOXA1 reduced cytochrome c in an SOD-inhibitable manner. The observed activity was continuous and did not require any stimulus. The rate of superoxide generation was 0.69 Ϯ 0.04 nmol of superoxide/min/10 7 cells.
Microscopic analysis by the NBT reduction assay showed that ϳ10% of the transiently NOX1 ϩ NOXO1 ϩ NOXA1transfected cells but none of the control-transfected cells stained NBT-positive (Fig. 4D). We infer that the superoxide  B, statistical analysis of peak superoxide production by HEK293 cells transfected with the indicated constructs. Superoxide was detected as in Fig. 1. C, measurement of superoxide generation with the SODinhibitable cytochrome c assay by HEK293 cells transfected with NOX1, NOXO1, and NOXA1 or with vector (control). Note that the seemingly different kinetics shown in A and C are caused by the two different types of assays, e.g. cytochrome c reduction measurement yields cumulative values, while the chemiluminescence measurement yields instantaneous ones. D, NBT reduction was assessed on HEK293 cells transfected with NOX1, NOXO1, and NOXA1 or with vector (control) after incubation with 0.5 mg/ml NBT at 37°C for 15 min. E, statistical analysis of peak superoxide production by HEK293 cells co-transfected with NOX1 plus the indicated combinations of NOXO1, NOXA1, p47 phox , and p67 phox either without stimulus (black bars) or in the presence of 100 nM PMA (white bars). Data shown in A, C, and D are from a single experiment representative of three or more independent studies. Data shown in B and E represent the mean Ϯ S.E. of three or more independent experiments. RLU, relative light units. generation per cell expressing an active NADPH oxidase was ϳ10 times higher than the mean value given above, or ϳ7 nmol of superoxide/min/10 7 cells. This is in the range reported for phagocyte NADPH oxidase-expressing COS-7 cells (23), and NOX5-expressing HEK293 cells (9).
Stimulus Independence of NOX1 Activation by Cytoplasmic Factors-The superoxide generation by NOX1 ϩ NOXO1 ϩ NOXA1-transfected cells shown in Fig. 4 occurred in the absence of external stimuli. In contrast the phagocyte NADPH oxidase requires external stimuli even in reconstituted systems (23,24). To investigate whether external stimuli are able to enhance the NOX1 respiratory burst, we treated the transfected cells with agents known to activate the phagocyte NADPH oxidase (1). However, none of the following treatments increased significantly the superoxide production compared with nonstimulated cells (100%): 100 nM PMA (84 Ϯ 2%), 100 M arachidonic acid (108 Ϯ 2%), and 1 M ionomycin (107 Ϯ 18%).
Cofactor Requirement for Stimulus-dependent and Stimulusindependent NOX1 Activation-To understand further the activation of NOX1 by cofactors, we co-transfected NOX1 with the following combinations of cofactors: NOXO1 ϩ NOXA1, p47 phox ϩ p67 phox , p47 phox ϩ NOXA1, and NOXO1 ϩ p67 phox (Fig. 4E). Stimulus-independent activation of NOX1 occurred exclusively with the NOXO1 ϩ NOXA1 combination. When NOXO1 was replaced by p47 phox , stimulus-independent activation disappeared, but the system could be activated by PMA. Under this condition, the superoxide generation was lower than that observed with NOXO1 ϩ NOXA1 but similar to that observed with p47 phox ϩ p67 phox . Only minimal NOX1 activity was detected with the combination of NOXO1 ϩ p67 phox . These results suggested that the differences between p47 phox and NOXO1 are important for stimulus-dependent versus stimulus-independent NOX1 activity. This is substantiated by the different primary structures of the two proteins, i.e. the lack of the autoinhibitory region and the adjacent phosphorylation sites in NOXO1. However, as the combination of NOXO1 ϩ p67 phox did not display any stimulus-independent activation, some NOXA1specific element might also be involved in the stimulusindependent activity.
In summary, we have described two novel proteins, NOXO1 and NOXA1, that support superoxide generation by NOX1. As is the case for NOX1, these proteins are expressed predominantly in colon epithelium and are thus likely to be physiologically relevant partners of NOX1. It is an intriguing question whether other subunits are involved in NOX1 function. In particular p22 phox and Rac1 are of interest as both of them have wide tissue distribution and both of them were reported to be expressed in HEK293 cells (25,26).
The activation of NOX1 by NOXO1 and NOXA1 is stimulusindependent in the reconstituted system and probably also in colon. However, we cannot exclude that there are some distinct mechanisms of NOXO1 inhibition in the colon, which are not reproduced in the HEK293 system. Is the stimulus-dependent activation of NOX1 by p47 phox and p67 phox of physiological relevance? Superoxide generation by vascular smooth muscle, which contains NOX1, is stimulus-dependent and involves p47 phox (27). Thus, an interaction of NOX1 with phagocyte NADPH oxidase subunits may occur in some tissues and would make NOX1 a versatile enzyme that could change activation mechanisms depending on the subunits present in the cell type in which it is expressed.