Analysis of mRNA Transcripts from the NAD(P)H Oxidase 1 (Nox1) Gene

Recent reports indicate that NAD(P)H oxidase 1 (Nox1) mRNA undergoes alternative splicing, producing a short transcript (NOH-1S) encoding a novel H+ channel. Although the H+ transport properties of NOH-1S-transfected cells resemble those of many cells, the production of a NOH-1S protein was never documented. We characterized Nox1 transcripts in colon-derived cells and present evidence that mRNA splicing does not produce NOH-1S; rather, NOH-1S appears to be an artifact of template switching during cDNA synthesis. The NOH-1S transcript was not observed by Northern blotting, despite claims of its abundance based on RNase protection assays. The shortened cDNA was generated by avian myeloblastosis virus reverse transcriptase, but not by thermally stable reverse transcriptase under conditions that produce full-length Nox1. Analysis of shortened cDNAs detected NOH-1S sequence and other variants that differ at the alleged splice junction site. Although no appropriate RNA splicing sites were found within Nox1 to account for NOH-1S formation, we found repetitive sequence elements bordering the deleted region, which could promote intramolecular template switching during cDNA synthesis. Template switching was confirmed in vitro, where the deleted cDNA was generated by avian myeloblastosis virus reverse transcriptase from a synthetic, full-length Nox1 RNA template. A survey of the expressed sequence tags database suggests that similar switching phenomena occur between repetitive elements in other Nox family transcripts, indicating such cloning artifacts are common. In contrast, genuine RNA splicing does account for another Nox1 transcript lacking the entire exon 11, which is abundant in colon cells but encodes a protein incapable of supporting superoxide production.

Although several studies have demonstrated ROS production by these Nox family gp91 phox homologs, their exact physiological functions remain unknown. Superoxide production is accompanied by a robust intracellular H ϩ production resulting from oxidation of NADPH to NADP ϩ and H ϩ and from the activation of the NADPH-regenerating hexose monophosphate shunt. Phagocytic cells have at least three mechanisms for removal of H ϩ : Na ϩ -H ϩ exchange, H ϩ -ATPase, and electrogenic H ϩ transport (10). The mechanism of electrogenic H ϩ transport in phagocytes is still debated; some evidence suggests that gp91 phox could act as a proton translocator (11)(12)(13), although the existence of other pathways is likely (14 -16). Henderson et al. (11,17) demonstrated that cells transfected with gp91 phox or the N-terminal portion of this protein acquire arachidonate-dependent H ϩ transport activity. This arachidonate-dependent activity appears to involve three histidine residues within the third transmembrane domain of gp91 phox ; these features are conserved in other Nox family homologs.
The notion that gp91 phox functions as a proton channel was supported further by studies on the gp91 phox homolog Nox1, where a putative alternatively spliced product (NOH-1S; NADPH oxidase homolog-1 short; GenBank accession no. NM013954) was shown to have H ϩ channel activity (18). That work, representing the first report of cloning of a mammalian H ϩ channel, suggested the H ϩ transport function could be dissociated from NADPH oxidase activity as a distinct molecular entity. The sequence encoding NOH-1S, originally derived from the database of expressed sequence tags (dbEST), was proposed to result from an unusual intra-exonic splicing event, involving only a portion of exon 5 spliced to the exon 14 sequence. The putative product, which was not detected at the protein level, consists of four transmembrane domains and a short cytoplasmic tail; the recombinant protein, when expressed ectopically in HEK 293 cells, exhibited the properties of voltage-gated H ϩ channels. These results agreed with pre-vious observations on the truncated, N-terminal fragment of gp91 phox (17).
In this study, we characterized Nox1 mRNA transcripts in normal colon and cancerous colon epithelial cells, where the Nox1 gene is expressed abundantly. We show that the mRNA transcript for NOH-1S is not synthesized from the Nox1 gene but is instead an artifactual product of cDNA synthesis. We attribute this result to the phenomenon of intramolecular template switching between repetitive sequence elements during cDNA synthesis. Our observations indicate that although the dbEST represents a rich source of novel transcript variants, these data should be viewed cautiously because of the frequent occurrence of such cloning artifacts. Furthermore, we demonstrate that the Nox1 gene encodes another genuine spliced variant of Nox1 that is abundant in colon cells but is not functional because of the absence of exon 11-encoded sequence.
Northern Blot Analysis-Total RNA was prepared from Caco2 and HT29 cells (15 g), separated electrophoretically on a 1% agarose formaldehyde gel, and transferred to nylon membranes. Membranes were probed at 65°C with a radiolabeled human Nox1 cDNA fragment (Amersham Biosciences) in Quickhyb hybridization solution (Stratagene) following standard hybridization protocols. The 32 P-radiolabeled probe was prepared with a Prime-It RmT random primer labeling kit (Stratagene). In other experiments, pre-made human gastrointestinal Northern blots (Clontech) were probed under the same conditions as detailed above.
In Vitro Transcription-A 2267-bp-long fragment of the human Nox1 (colon) cDNA was amplified with primers 1 and 3 and cloned into pCR4-TOPO TA-cloning vector. RNA was transcribed with 30 units T7 RNA polymerase (MBI Fermentas) using 1 g of linearized template DNA from this construct. After 1 h at 37°C, the template DNA was digested by DNase I (Roche Applied Science), and the RNA was purified by the RNeasy Mini Kit (Qiagen). RT-PCR was performed as detailed above, using the cDNA cycle TM kit (Invitrogen) for cDNA synthesis at 42°C and primers 1 and 2 for the PCR.
RNase Protection Assays-RNase protection assays were performed as described by Bá nfi et al. (18) with minor modifications. A radiolabeled 3Ј-terminal NOH-1S antisense probe was synthesized from a BanI-digested NOH-1S cDNA (I.M.A.G.E. clone 900560, Research Genetics) using T3 RNA polymerase (MBI Fermentas) and [␣-32 P]UTP (Amersham Biosciences). The resulting transcript produces a 407-base probe, of which 374 bases correspond to the 3Ј-portion of NOH-1S (starting at nucleotide 241 relative to the start codon). Synthetic, fulllength Nox1 RNA was transcribed and purified as above and used to anneal to the NOH-1S probe at 42°C for 16 h. The resulting hybrids were treated with RNase A and T1 mixtures for 30 min at 37°C using the RPA III TM RNase protection assay kit (Ambion). The protected products were analyzed by autoradiography after electrophoresis on 6% acrylamide/urea/TBE gels (Invitrogen).
Database Searches-Database searches were performed in the dbEST and the unfinished high throughput genomic sequences using the BLAST algorithm (www.ncbi.nlm.nih.gov/BLAST).

RESULTS
Exonic Organization of the Nox1 gene-Based on a genomic DNA database search, Bá nfi et al. (18) concluded that the NOH-1/Nox1 gene was located on Xq22 and consisted of 14 exons. They indicated that the first 13 exons are of comparable size and are homologous to the 13 exons of gp91 phox and that the 14th exon, containing 99 nucleotides, was unique to the Nox1 gene. They suggested that the generation of NOH-1S involved splicing of an intra-exonic site in exon 5 with the sequence in exon 14, which would encode a unique C-terminal portion of NOH-1S. Our reevaluation of genomic and corresponding cDNA sequences deposited in the human genomic database indicates that the Nox1 gene encompasses only 13 exons. Fig. 1a provides a revised exon map of Nox1, which was deduced from a BLAST-based alignment of the full-length cDNA sequence with a contiguous genomic Nox1 DNA sequence (GenBank TM accession number HsX_11808). In this revised genomic map, the last exon, exon 13, consists of 741 nucleotides and includes a 3Ј-terminal 99 nucleotides that previously represented exon 14 (18). These observations indicate that the putative NOH-1S transcript would involve the use of intra-exonic splice donor and splice acceptor sites within exon 5 and exon 13, respectively.
RNA splicing is a precisely regulated process involving strict adherence to the "GT-AG" rule (20, 21), even in those relatively rare examples where intra-exonic splicing occurs (22,23). We screened exon 5 and exon 13 of the Nox1 gene for suitable intra-exonic splice donor and acceptor sites corresponding to the proposed splice sites in NOH-1S. Using splice site predictions by neural network analysis (www.fruitfly.org/seq_tools) with low donor and acceptor site cutoff scores (0.40), we were unable to locate any appropriate splice sites corresponding to the proposed regions of NOH-1S.
Demonstration of Nox1 Expression by Northern Blot Analysis-The Nox1 gene is expressed abundantly in the colon and to a lesser extent in the uterus, the prostate, and proliferating vascular smooth muscle cells (3). To characterize the transcripts originating from the Nox1 gene, we studied Nox1 mRNA expression by Northern blot analysis on a gastrointestinal mRNA panel; various isolated portions of the colon were represented, as well as RNAs from two colorectal carcinoma cell lines (Caco2 and HT29). The cDNA probe encompassed a major 5Ј-fragment (409 nucleotides) of the Nox1 mRNA and included most of the NOH-1S sequence (Fig. 1A). Fig. 1B shows that one major band of ϳ2.5 kb, corresponding to the full-length Nox1 mRNA transcript, was recognized in all lanes, although no signal was detected at the expected size for NOH-1S (ϳ0.50 kb). We observed that Nox1 expression could be significantly induced by treating Caco2 cells with 1␣,25-dihydroxyvitamin D 3 or interferon-␥ (24); however, the NOH-1S messenger RNA was still not detected in induced Caco2 cells. These results indicate that the NOH-1S mRNA species was either absent or was below the detection limits of the Northern blot method, although the RNase protection experiments by Bá nfi et al. (18) suggested that NOH-1S and NOH-1L are expressed in comparable amounts in normal colon and colon carcinoma lines.
Detection of NOH-1S by RT-PCR-To examine the possibility of low level NOH-1S expression (i.e. below Northern analysis detection limits), we used RT-PCR to amplify NOH-1S transcripts from colon RNA. Using AMV reverse transcriptase and primers corresponding to the 5Ј-and 3Ј-region of the NOH-1S cDNA (primers 1 and 3), as described previously (18), we were able to amplify products corresponding to the expected size of NOH-1S mRNA ( Fig. 2A). We also amplified similar sized products from Caco2 and HT29 cells (data not shown). After randomly sequencing selected clones derived from these PCR products, we discovered several novel sequences with extensive variations within the alleged intra-exonic spliced region of NOH-1S (Fig. 2B), whereas the sequences flanking the alleged splice site were virtually identical. Specifically, we identified five unique sequence variations (Fig. 2B), including several clones that matched the EST clone AA493362 (originally described as NOH-1S). Repeated sequencing of both DNA strands excluded the possibility that the sequence variations were caused by sequencing errors. Interestingly, a nucleotide BLAST search in dbEST using NOH-1S as a query sequence identified an additional sequence variant (AW450431; Fig. 2B, sequence 6) that was identical to NOH-1S except for the region indicated in Fig. 2B. Note that several of these sequences (sequences 2-4) contain longer stretches of exon 5 sequence than was observed in the NOH-1S transcript (sequence 1), indicating that these sequence variations were not caused by low fidelity transcriptional errors during amplification of the NOH-1S cDNA.
These data, together with the negative Northern blot result, strongly suggest that NOH-1S is not a novel splice variant of Nox1 but represents an artifact of the cDNA cloning process. In these experiments, we performed the cDNA synthesis at 42°C using the AMV reverse transcriptase enzyme, as originally described (18). To examine the possibility that NOH-1S is a product of inaccurate cDNA synthesis, we used a different reverse transcriptase for cDNA synthesis which performs at higher temperatures. Using the heat-stable ThermoScript TM reverse transcriptase system for cDNA synthesis at 60°C, we could not amplify the NOH-1S from colon RNA preparations, although the full-length, 2.3-kb-long Nox1 cDNA was obtained (using two different primer pairs) in good yield despite its considerably larger size (Fig. 2C). Thus, NOH-1S cDNA appears to be a unique product of AMV reverse transcriptase. These experiments also establish that errors in the PCR amplification were not responsible for the generation of the NOH-1S product.
Repetitive Sequence Elements and Their Role in NOH-1S cDNA Synthesis-Although exons 5 and 13 do not contain suitable splice sites that account for NOH-1S formation by RNA splicing, we observed more than one repetitive sequence element at the boundaries of the deleted region, which would promote intramolecular template switching during cDNA synthesis (25) and can account for the heterogeneity of the Nox1derived short forms shown in Fig. 2B. The basis for such a template switching mechanism is illustrated in the cases of two deleted Nox1 transcripts that were already deposited in dbEST (Fig. 3A). To explore directly the possibility of intramolecular template switching, we cloned a 2.2-kb portion of the fulllength Nox1 cDNA, produced a synthetic mRNA template in vitro with T7 RNA polymerase, and then synthesized the cDNA at 42°C using AMV reverse transcriptase. After PCR amplification, we identified two high yield products (Fig. 3B): the original full-length Nox1 cDNA copy and a product that corresponds to the size of NOH-1S. Sequencing of the short PCR product revealed a NOH-1S sequence variant corresponding to sequence 2 (Fig. 2B). This experiment shows that NOH-1S can be derived spontaneously from the full-length Nox1 transcript in the absence of any splicing machinery and proves that NOH-1S cDNA formation involves intramolecular template switching between repetitive sequence elements within the RNA template.
GenBank TM searches of the dbEST revealed that internally deleted forms of other gp91 phox homologs were also deposited in this database. Using the cDNAs of mouse NADP(H) oxidase 4/renal oxidase (Nox4/Renox) and the rat dual oxidase 2/thyroid oxidase 2 (Duox2/Thox2) as query sequences, we identified internally deleted forms of both sequences (AW107048, AW012126, and AW259939 for NOX4 and AW920088 for Duox2). Because a partial sequence of the mouse Renox (Nox4) gene was already deposited in the high throughput genomic sequences database, we confirmed that these deletions in the Nox4 cDNA also occurred within exons and in the absence of appropriate splice signals, arguing against an alternative RNA splicing mechanism. We examined the specific deleted regions in full-length Nox4 and Duox2 cDNA sequences, and, interestingly, both sequences also contain repetitive elements within the boundaries of the deleted regions (Fig. 3C, boxed  sequences).
RNase Protection Assays Using Synthetic RNA-RNase protection was the only other experimental approach used besides RT-PCR to support the model of RNA splicing as the basis for NOH-1S synthesis (18). Because we showed that cDNA synthesis by AMV reverse transcriptase spontaneously generates deleted Nox1 products from full-length templates (Fig. 3B), we reevaluated the RNase protection assay described by Bá nfi et al. (18). In these experiments, we prepared a radiolabeled riboprobe from the 3Ј-end of NOH-1S, similar to this previous study, and examined the RNase protection patterns resulting from hybridization to the synthetic, full-length RNA transcribed in vitro from Nox1 cDNA. Fig. 4 shows that hybridization of this single, undeleted transcript provides RNase protection resulting in a mixture of two major fragments: one that corresponded to hybridization of the probe up to the boundary of the deletion site (235 nucleotides), which was predicted for detection of the full-length transcript, and another larger fragment (309 nucleotides) corresponding to the intact NOH-1S sequence, which would require formation of a stable looped structure allowing hybridization between complementary terminal sequences (Fig. 4). These results indicate that the NOH-1S probe used in earlier RNase protection studies does not distinguish between the full-length and the deleted (spliced) Nox1 transcripts and, therefore, cannot be used to detect this deleted variant.
The Nox1 Gene Encodes an Exon 11-spliced Variant, Nox1v, Which Is Catalytically Inactive-As noted above, we detect one major band of ϳ 2.5 kb in Northern blots of colon RNA using a cDNA probe corresponding to the 5Ј-end of the Nox1 open reading frame (Fig. 1), although Northern blotting does not distinguish between transcripts that differ slightly in size. Therefore, we used RT-PCR to analyze large Nox1 gene transcripts. Using primers aligned to the 5Ј-and 3Ј-ends of the Nox1 open reading frame (primers 1 and 4), we detect two different PCR products that are close in size (1.7 and 1.55 kb) in normal colon and in two colon carcinoma cell lines (Fig. 5A). After cloning and sequencing the PCR products, we confirmed that the larger sequence is Nox1, whereas the smaller product, lacking 147 nucleotides, is identical to the sequence reported previously as NAD(P)H oxidase homolog 1 long variant, NOH-1Lv (GenBank accession no. NM013955; 18). Because the short variant of Nox1 (NOH-1S) represents a cDNA synthesis artifact, we suggest designating this long Nox1 variant as Nox1v (consistent with the new nomenclature system for NAD(P)H oxidases). The human Nox1 gene was already sequenced as a part of the human genome project, therefore we determined that the 147 nucleotides absent in Nox1v corresponds precisely with exon 11 of the Nox1 gene, which is flanked by appropriate RNA splice donor and acceptor sequences. Thus, unlike the NOH-1S product, Nox1v appears to represent a genuine spliced variant that lacks a complete exon. Protein sequence comparison of Nox1 and Nox1v revealed that the exon 11-encoded sequence (residues 433-482) contains a highly conserved motif ( 435 FYWICRE 441 ) shared by other Nox enzymes (Fig. 5B). In view of the abundance of the deleted Nox1v transcript in colon cells and the presumed importance of this conserved motif in NADPH ribose binding in gp91 phox and related oxidases (3, 18, . Numbers indicate nucleotide positions in these database files. B, a synthetic, full-length Nox1 RNA template generates both the fulllength and short (NOH-1S) cDNA products when using AMV reverse transcriptase (low temperature protocol). The Nox1-specific primers for PCR are 1 and 2. C, alignment of full-length Nox4 and Duox2 cDNA sequences with deleted versions deposited in dbEST reveals repetitive elements, suggesting that template switching during cDNA synthesis also occurs in these cases. The sequences are annotated as in A.

FIG. 4. RNase protection assay of NOH-1S.
A, schematic representation of the NOH-1S RNase protection assay, adapted from earlier studies (18) and applied to a synthetic Nox1 (undeleted) transcript target. The radiolabeled NOH-1S antisense riboprobe was prepared from the 3Ј-direction using T7 RNA polymerase. The synthetic Nox1 RNA (transcribed from the 5Ј-end) anneals to NOH-1S riboprobe along two separate aligned regions (upstream of exon 5 (E5) and downstream of exon 13 (E13)); unpaired intervening sequence shown as looped structure and repetitive elements are indicated as black boxes. Fragments corresponding to the whole riboprobe (407 nucleotides), and those protected up through exon 5 (235 nucleotides) or beyond the fusion with exon 13 (309 nucleotides) are shown. B, RNase protection assay using synthetic full-length Nox1 RNA results in protection of both the 235-and 309-nucleotide fragments. Left lane, molecular weight standards (sizes indicated in nucleotides); center lane, full-length Nox1protected fragments after RNase treatment; right lane, undigested NOH-1S riboprobe (407-nucleotide band indicated by an arrow on the right). The results of the RNase protection assay represent two experiments with identical results. 26,27), we examined the catalytic activities of Nox1 and Nox1v expressed in transfected HEK 293 and COS-7 cells (Fig. 6).
Recent studies indicate that Nox1 functions as a multicomponent oxidase complex similar to the phagocytic oxidase (19,28,29). Nox1 activity is relatively low when transfected into a variety of cell hosts, unless it is coexpressed in the presence of cofactors detected in colon epithelial cells (p41 nox1 /Noxo1 and p51 nox1 /Noxa1), which are homologous to the cytosolic phox proteins (19,28,29). Other work showed that p22 phox also contributes to higher Nox1 activity, consistent with the ability of p22 phox to interact with Noxo1 (29). Fig. 6A shows that HEK 293 cells transfected with the Nox1-based muticomponent system including Noxo1 and Noxa1 exhibit robust oxidative responses to phorbol 12-myristate 13-acetate stimulation, whereas cells transfected with Nox1v along with the same cofactors show no significant ROS release in response to stimulation, similar to cells transfected with control (empty) pcDNA3.1 vector. The Nox1-transfected COS-7 model, which included cotransfected Noxo1, Noxa1, and p22 phox vectors, exhibited considerably higher activity than the HEK 293 cells (Fig. 6A). Western blotting was performed on transfected COS-7 cells to determine whether the deleted Nox1v protein was detectable in these oxidase-deficient cells. Because neither Nox1 nor Nox1v was detected using several commercial and custom Nox1 antipeptide antibodies, both the Nox1 and Nox1v cDNAs were constructed to fuse their C-terminal coding sequences with the V5 peptide epitope. The epitope-tagged fusion proteins were produced in COS-7 cells by cotransfection with p22 phox -pcDNA3.1. Western blotting of the membrane (particulate fraction) proteins from these transfected cells confirmed that both the Nox1 and Nox1v isoforms were detectable with the expected molecular masses (Fig. 6B), despite the inability of Nox1v to support any detectable ROS generation. Both fusion proteins generated a similar pattern of breakdown products, although the Nox1v form appears to exhibit lower stability. These findings confirm a critical role for exon 11-encoded sequence in oxidase function, showing that in the absence of this sequence the expressed Nox1v spliced isoform is a nonfunctional protein.

DISCUSSION
In this study, we characterized mRNA transcripts derived from the Nox1 gene and show that Nox1 does not encode a H ϩ channel described previously as NOH-1S. Several independent results were obtained that refute all previous evidence suggesting NOH-1S is a natural product of mRNA splicing. First, NOH-1S was not detected by Northern blotting of colon cells, a site of high Nox1 expression, despite the suggestion of its relative abundance based on RNase protection assays (18). Second, the deleted NOH-1S could only be generated from colon cell RNA when using AMV reverse transcriptase at low temperatures, whereas only the large Nox1 transcripts were synthesized when using a reverse transcriptase (ThermoScript TM ) that performs at higher temperatures. Third, the shortened Nox1-derived cDNAs produced by AMV reverse transcriptase were heterogeneous at the alleged splice site. More important, the shortened cDNA was also generated artifactually using full-length, synthetic Nox1 RNA as a template. Thus, it appears that NOH-1S is derived by a template switching process that is unique to low temperature cDNA synthesis by AMV reverse transcriptase resulting in a deletion in the transcribed sequence. Finally, we showed that hybridization with fulllength Nox1 RNA provides protection to a deleted NOH-1S antisense probe against RNase degradation; thus, the probe does not distinguish between full-length and deleted transcripts. This result, as well as those showing spontaneous deletions during cDNA synthesis from the full-length template cast considerable doubt on the validity of the evidence presented for a NOH-1S mRNA splicing mechanism. These results were consistent with the formation of hybrids between the full-length and the complementary, deleted (NOH-1S) strands, which are likely stabilized by repetitive sequences present on the boundaries of the deleted sequence.
In addition to these findings, several structural features of the Nox1 gene are incompatible with an mRNA splicing mechanism. First, based on the corrected Nox1 genomic map, the proposed NOH-1S splicing requires intra-exonic sites at both the splice donor (exon 5) and acceptor sites (exon 13). Second, there are no suitable splice sites within the proposed intraexonic splice boundaries that conform to the GT-AG rule. A recently compiled database of known mammalian splice site sequences, with more than 22,000 entries (21), confirms that 98.71% contain the canonical GT-AG junctions; the remainder have one of two other splice site pairs (GC-AG and AT-AC), which were also not observed near the proposed splice boundaries of NOH-1S. Third, in general, the Nox family of gp91 phox homologs is highly conserved with minimal interspecies variations within coding sequences. For example, the human and the rat Nox1 proteins are 80% identical, whereas the 3Ј-untranslated regions are poorly conserved. In the case of NOH-1S, this protein would appear to be unique to the human species because we could not identify any "hidden" reading frame in the rat Nox1 mRNA which encodes an amino acid sequence homologous to the C-terminal portion of human NOH-1S. Fourth, the Nox1 transcript, like most eukaryotic mRNA molecules, contains an AAUAAA polyadenylation signal 17 nucleotides upstream from the polyadenylated 3Ј terminus. In the proposed spliced NOH-1S transcript of Bá nfi et al. (18), this polyadenylation signal is used instead as part of the C-terminal coding region of NOH-1S, whereby the UAA within this signal represents the proposed stop codon. These theoretical considerations based on our revision of the Nox1 genomic map also argue strongly against the feasibility of a NOH-1S RNA splicing mechanism.
Database searches with the seven Nox family cDNAs indicate that spontaneous deletions within cDNA sequences depos-ited in unedited sequence tag databases (i.e. dbEST) are not unique to Nox1 and are more common than is generally appreciated. These searches revealed internal deletions in the cDNAs of mouse Nox4/Renox and rat Duox2/Thox2. In both cases, the generation of these deleted products can be explained by template switching during cDNA synthesis by reverse transcriptase because these cDNAs share several remarkable structural features with NOH-1S and other well characterized transcripts that undergo artifactual template switching (25). These features include the occurrence of deletions within intra-exonic sites, the absence of appropriate splice signals, the production of multiple, but imprecise, splice variants (sometimes resulting in frameshifts or missense mutations), as well as the presence of repetitive sequence elements flanking the deletion sites. Our analysis of multiple products derived by this process suggests that the repetitive sequence elements have a critical role in promoting the template switching process by enabling the newly synthesized strand to reanneal upstream to other homologous sequence elements. In cases where multiple deleted cDNAs were detected (as with Nox1), the presence of multiple, minor repeats or imperfect repeats accounted for these heterogeneous transcripts. These observations suggest that sequences derived from these rapidly growing sequence databases, such as dbEST, should be interpreted cautiously because these sequences frequently contain internal deletions resulting from template switching, in addition to other cloning, transcription, PCR amplification, and sequencing errors. One consequence of such errors is that calculations of protein diversity based on EST analysis could lead to overestimations of the frequency of alternative splicing in the human genome. Furthermore, template switching could generate unrecognized artifacts in RT-PCR protocols frequently applied in the diagnostic analysis of aberrant mRNAs.
Although RNA splicing does not account for NOH-1S synthesis, we confirmed that RNA splicing does generate significant amounts of another Nox1 transcript variant detected in normal colon and in two colon carcinoma cell lines (Fig. 5A), which we refer to as Nox1v (formerly NOH-1Lv (18)). The Nox1v transcript lacks the entire 11th exon, which encodes exactly 49 amino acids (residues 433-482). A BLAST search using this sequence revealed that the N-terminal portion of this sequence ( 435 FYWICRE 441 ) is highly conserved in the Nox family (Fig.  5B). The corresponding sequence in gp91 phox/nox2 ( 441 YWL-CRD 447 ) was proposed to represent a NADPH ribose binding motif, based on similarities with sequences of other flavoenzymes, notably cytochrome P450 reductase and nitric-oxide synthase (26,27). However, these features were not evident in the crystallographic structure of ferrodoxin NADP ϩ reductase (FNR; 30). Thus, the absence of crystallographic information on the NADPϩ-bound form of FNR, together with the lack of similarities between these regions of FNR and Nox proteins (Fig. 5B), have made the assignment of the NADP/ribose-binding residues in Nox proteins less certain than other, more conserved nucleotide binding motifs shared with FNR-related flavoproteins (26,27,31). Because of these uncertainties, we directly compared the abilities of Nox1 and Nox1v to support ROS production in two transfected cell models that coexpressed p22 phox and cytosolic factors (Noxo1 and Noxa1) shown to support maximum activity of Nox1 (19,28,29). These experiments showed that the exon 11-deleted Nox1v transcript does not encode a functional oxidase, despite its abundant expression in colon cells. These observations confirm a critical role for exon 11-encoded sequence in the oxidative function of Nox1 and provide further support for the proposed role of this conserved motif in NADPH ribose binding.
H ϩ -extruding mechanisms have an important role in the pH homeostasis of phagocytic cells because phagocytes produce a significant amount of H ϩ during superoxide production (10). Voltage-operated H ϩ channels represent one major route of H ϩ removal; however, a genuine H ϩ channel protein has not been identified at the molecular level. One proposed candidate for the H ϩ channel function in phagocytic cells is gp91 phox (11)(12)(13); however patch clamp experiments on gp91 phox -deficient neutrophils (16), monocytes (14), and COS-7 cells expressing the complete oxidase enzyme (32) question the contribution of gp91 phox to whole cell H ϩ currents. Thus, while the precise roles of gp91 phox and related Nox proteins in proton transport are debated (33)(34)(35)(36), the hunt for additional genes encoding genuine H ϩ channels should continue.