Dual Oxidase-2 Has an Intrinsic Ca2+-dependent H2O2-generating Activity*

Duox2 (and probably Duox1) is a glycoflavoprotein involved in thyroid hormone biosynthesis, as the thyroid H2O2 generator functionally associated with Tpo (thyroperoxidase). So far, because of the impairment of maturation and of the targeting process, transfecting DUOX into nonthyroid cell lines has not led to the expression of a functional H2O2-generating system at the plasma membrane. For the first time, we investigated the H2O2-generating activity in the particulate fractions from DUOX2- and DUOX1-transfected HEK293 and Chinese hamster ovary cells. The particulate fractions of these cells stably or transiently transfected with human or porcine DUOX cDNA demonstrate a functional NADPH/Ca2+-dependent H2O2-generating activity. The immature Duox proteins had less activity than pig thyrocyte particulate fractions, and their activity depended on their primary structures. Human Duox2 seemed to be more active than human Duox1 but only half as active as its porcine counterpart. TPO co-transfection produced a slight increase in the enzymatic activity, whereas p22phox, the 22-kDa subunit of the leukocyte NADPH oxidase, had no effect. In previous studies on the mechanism of H2O2 formation, it was shown that mature thyroid NADPH oxidase does not release \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{2}^{{\bar{{\cdot}}}}\) \end{document} but H2O2. Using a spin-trapping technique combined with electron paramagnetic resonance spectroscopy, we confirmed this result but also demonstrated that the partially glycosylated form of Duox2, located in the endoplasmic reticulum, generates superoxide in a calcium-dependent manner. These results suggest that post-translational modifications during the maturation process of Duox2 could be implicated in the mechanism of H2O2 formation by favoring intramolecular superoxide dismutation.

Duox2 (and probably Duox1) is a glycoflavoprotein involved in thyroid hormone biosynthesis, as the thyroid H 2 O 2 generator functionally associated with Tpo (thyroperoxidase). So far, because of the impairment of maturation and of the targeting process, transfecting DUOX into nonthyroid cell lines has not led to the expression of a functional H 2 O 2 -generating system at the plasma membrane. For the first time, we investigated the H 2 O 2 -generating activity in the particulate fractions from DUOX2-and DUOX1-transfected HEK293 and Chinese hamster ovary cells. The particulate fractions of these cells stably or transiently transfected with human or porcine DUOX cDNA demonstrate a functional NADPH/Ca 2؉ -dependent H 2 O 2 -generating activity. The immature Duox proteins had less activity than pig thyrocyte particulate fractions, and their activity depended on their primary structures. Human Duox2 seemed to be more active than human Duox1 but only half as active as its porcine counterpart. TPO co-transfection produced a slight increase in the enzymatic activity, whereas p22 phox , the 22-kDa subunit of the leukocyte NADPH oxidase, had no effect. In previous studies on the mechanism of H 2 O 2 formation, it was shown that mature thyroid NADPH oxidase does not release O 2 . but H 2 O 2 . Using a spin-trapping technique combined with electron paramagnetic resonance spectroscopy, we confirmed this result but also demonstrated that the partially glycosylated form of Duox2, located in the endoplasmic reticulum, generates superoxide in a calcium-dependent manner. These results suggest that post-translational modifications during the maturation process of Duox2 could be implicated in the mechanism of H 2 O 2 formation by favoring intramolecular superoxide dismutation.
Reactive oxygen species (ROS) 1 have emerged as important molecules involved in regulating essential cell functions, such as growth and differentiation (1). NAD(P)H oxidases are a major source of ROS. Phagocyte oxidase is the oxidase that has been investigated most thoroughly (2). It catalyzes the production of superoxide by the one-electron reduction of oxygen, using NADPH as the electron donor. The catalytic moiety of the phagocyte NADPH oxidase is gp91 phox , a plasma membraneassociated flavohemoprotein. Recently, it was discovered that gp91 phox belongs to a family consisting of several very similar oxidases. Seven NOX (NADPH oxidase) and DUOX/ThOX (dual oxidase/thyroid oxidase) genes have been identified that encode different NADPH oxidases with differing mRNA tissue expression. The Nox family comprises gp91 phox , now known as Nox2; Nox1, which is predominantly expressed in the colon (3); Nox3, cloned from fetal kidney (4); Nox4 found in the kidney cortex (5,6); and Nox5, expressed in the testis, spleen, and lymph nodes (7). In addition to the basic structure of gp91 phox , Nox5 has a long intracellular N-terminal domain with four calcium binding sites implicated in its Ca 2ϩ -dependent activation (8).
The biological functions of these Nox proteins are now under investigation. They are involved in signal transduction related to cell growth and cancer (9 -11) and to angiogenesis (12).
Duox1 and Duox2 are large homologues of Nox2 with an N-terminal extension comprising two EF-hand motifs, an additional transmembrane helix, and a peroxidase homology ectodomain (see Fig. 4A) (13,14). DUOX genes have been identified in the thyroid gland, where they are strongly expressed (13,14). However, the DUOX are also expressed on the mucosal surfaces of the trachea and the bronchi (15) and in the airway epithelial cells (16,17), where it has been suggested that Duox1 is the isoform responsible for acid production and secretion in airways (16) and plays a critical role in mucin expression (18). DUOX2 was also expressed throughout the digestive tract, where it was found to be functional (19,20), in addition to the salivary gland and rectum (15).
It has been suggested that Duox2, which was identified by purifying thyroid NADPH oxidase, may constitute the catalytic core of this enzyme and generate the H 2 O 2 used by Tpo to catalyze the biosynthesis of thyroid hormones at the apical surface of the thyrocytes (13). Although no functional Duoxbased H 2 O 2 -generating system has yet been reconstituted (21), this proposal is corroborated by a recent report of permanent and severe congenital hypothyroidism in a patient with a bial-lelic inactivating mutation in the DUOX2 gene (22). Because Duox2 is also co-expressed with lactoperoxidase in the salivary gland and rectum, it has been suggested that it may be the source of H 2 O 2 for lactoperoxidase-catalyzed reactions implicated in a host defense mechanism (15). A Duox-based enzyme is also involved in stabilizing the cuticle of Caenorhabditis elegans by the peroxidase-catalyzed formation of tyrosine-tyrosine bonds (23). This means that Duox2 must generate H 2 O 2 for peroxidase reactions catalyzed either by distinct peroxidases, such as Tpo and lactoperoxidase, or by the peroxidaselike ectodomain of Duox2 itself (23). On the basis of their homology with gp91 phox /Nox2, the Duox proteins are believed to generate superoxide anions that could dismute into H 2 O 2 (24).
Here, using particulate fractions from DUOX2-transfected nonthyroid cells, we provide data showing for the first time that the partially N-glycosylated form of the Duox2 protein generates H 2 O 2 via dismutation of the superoxide anion (O 2 . ) in a NADPH/ Ca 2ϩ -dependent manner. Interestingly, we found that the level of H 2 O 2 formation activity depended on the primary structure of the Duox proteins, since human Duox1 appeared to be less active than human Duox2, and porcine Duox2 was twice as active as its human counterpart. Tpo, found recently to be associated with Duoxs in human thyrocyte (25), had little effect on the H 2 O 2generating activity. On the other hand, p22 phox , the 22-kDa subunit associated with the glycosylated 91-kDa subunit (gp91 phox / Nox2) of the leukocyte NADPH oxidase and recently found interacting with Nox1 (26,27) and Nox4 (27), had no effect. These results highlight the key impact of the maturation of Duox on H 2 O 2 -generating activity.

MATERIALS AND METHODS
Isolation of Pig Thyroid Follicles-Pig thyroid follicles were prepared by the method previously described (28), except that 20 M hemin was added to the culture medium.
Stable and Transient Cell Transfection-The stable human DUOX1and DUOX2-expressing cell lines were established using the Flp-In system (Invitrogen). The protocol accompanying the kit was used without modification. Human DUOX1 and DUOX2 cDNAs were subcloned in pcDNA5/FRT vector (Invitrogen) designed for use with the Flp-In system. When cotransfected with the pOG44 Flp recombinase expression plasmid into an Flp-In-293 cell line, the pcDNA5/FRT vector containing DUOX1 or DUOX2 cDNA was integrated in an Flp recombinasedependent manner into the genome. Stable cell lines were established by exposure to 100 g/ml hygromycin.
In transient cell transfection experiments, HEK293 cells reaching 50 -60% confluence were transfected using the calcium phosphate precipitation procedure and the Invitrogen protocol. HEK293 cells were incubated overnight with calcium phosphate in the presence of 50 g of pcDNA3.1-human DUOX2 or pcDNA3.1-porcine DUOX2. 24 h after transfection, 20 M hemin was added to the medium. After 48 h, the cells were harvested. CHO cells were transiently transfected, using the FuGENE (Roche Applied Science) transfection reagent instead of the calcium phosphate procedure.
The pcDNA3.1-GFP vector was purchased from Invitrogen. Fulllength 3060-kilobase pair human TPO cDNA (kindly provided by B. Rapoport, Cedars-Sinai Research Institute, Los Angeles, CA) was cloned into the vector pcDNA3.1 as previously described (29 Preparation of the Cellular Particulate Fraction-Cells were washed with PBS and scraped into the same solution supplemented with a mixture of protease inhibitors (5 g/ml aprotinin, 5 g/ml leupeptin, 1 g/ml pepstatin, 157 g/ml benzamidine). After centrifuging at 200 ϫ g for 10 min at 4°C, the cell pellet was homogenized using a motor-driven Teflon pestle homogenizer in 2 ml of 50 mM sodium phosphate buffer containing 0.25 M sucrose, 0.1 mM dithiothreitol, 1 mM EGTA (pH 7.2), and the mixture of protease inhibitors. After centrifuging at 200,000 ϫ g for 30 min, the pellet was resuspended in 0.5 ml of 50 mM sodium phosphate buffer (pH 7.2) containing 0.25 M sucrose, 1 mM MgCl 2 , and the mixture of protease inhibitors. The same protocol was used to obtain follicular particulate fractions, except that follicles from two Petri dishes (about 2 ϫ 10 7 cells) were collected by centrifuging at 200 ϫ g for 7 min before being resuspended in 2 ml of 50 mM sodium phosphate buffer containing 0.25 M sucrose, 0.1 mM dithiothreitol, 1 mM EGTA (pH 7.2), and the mixture of protease inhibitors.
Preparation of Thyroid Plasma Membranes-The preparation of plasma membranes was as previously described (30). Fresh porcine thyroid glands (150 g) were obtained immediately after slaughter and transferred to the laboratory on ice. Fat and connective tissues were removed. The glands were cut into small pieces and homogenized, first with a Sorvall Omni mixer (30 s, 11,000 rpm) in 150 ml of 50 mM sodium phosphate, pH 7.2, containing 1 mM EGTA, 2 mM MgCl 2 , 1 mM dithiothreitol, and 16 mg/ml phenylmethylsulfonyl fluoride (buffer A) and then with an Ultra-Turrax (45 s, 60% V max /150 ml) after dilution to 600 ml with buffer A. The homogenate was filtered through six layers of cheesecloth and centrifuged at 4200 ϫ g for 15 min. The pellet was resuspended in 50 mM sodium phosphate, pH 7.2, containing 2 mM MgCl 2 and 0.25 M sucrose (buffer B) and washed twice by centrifugation at 4200 ϫ g for 15 min. The final pellet was resuspended in buffer B and mixed with 47 ml of a stock Percoll solution (1 volume of 2.5 M sucrose plus 9 volumes of Percoll); the volume was adjusted to 222 ml with buffer B. After centrifugation in a Sorvall SS34 rotor at 13,000 rpm (20,000 ϫ g) for 30 min, the turbid layer sedimenting at about 1 cm from the top of the gradient was collected and washed with 0.1 M KCl in buffer B by centrifugation at 5000 ϫ g for 15 min. The final pellet was resuspended in 15 ml of buffer B (membrane preparation) quick-frozen in liquid nitrogen, and stored at Ϫ80°C.
Western Blot Analysis-SDS-PAGE and immunoblot analyses were performed as described previously (31). Depending on the experiments, an anti-Duox antibody raised against a 14-amino acid peptide encompassing the Leu 410 -Thr 423 portion of human Duox2, which is exactly conserved in porcine Duox2, or raised against the first intracellular domain Glu 636 -Arg 1039 of human Duox2, were used to probe the immunoblots. Immune complexes were detected with an alkaline phosphatase-coupled, anti-rabbit IgG antibody (Promega). Human Duox1 was immunodetected using anti-Duox1 antibody kindly given by F. Miot, Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire (IRIBHM), Université Libre de Bruxelles, Brussels) (14). Human thyroperoxidase was immunodetected with a polyclonal rabbit anti-porcine Tpo antibody (dilution 1:2000) obtained as previously described (32). The anti-GFP antibody was purchased from Invitrogen. The anti-p22 phox antibody was purchased from Santa Cruz (Tebu-France).
Determination of the Protein Content-The protein content was determined using the Bradford method (33).

Measurement of the H 2 O 2 Production of the Particulate Fractions-
The NADPH-dependent H 2 O 2 -generating activity of the particles was determined as previously described (34). Samples of cellular particulate fraction were incubated at 30°C in 0.5 ml of 200 mM sodium phosphate, pH 7.4, containing 1 mM NaN 3 , 1 mM EGTA, 1.5 mM CaCl 2 , and 0.1 M FAD. The reaction was started by adding 100 M NADPH. 100-l samples were collected at times up to 30 min and mixed with 10 l of 3 N HCl to stop the reaction by destroying the remaining NAD(P)H. The Ca 2ϩ dependence of the H 2 O 2 formation was determined by assaying parallel samples without Ca 2ϩ in the presence of 1 mM EGTA. The amount of H 2 O 2 in each sample was measured in 200 mM, pH 7.8, phosphate buffer by monitoring the decrease in 0.25 M scopoletin fluorescence in the presence of horseradish peroxidase in a Hitachi F-2000 spectrofluorimeter. In some experiments, the fluorescence (excitation 320 nm; emission, 420 nm) was measured in a Fluostar spectrofluorimeter (BMG LabTech) after adding to each aliquot 90 l of 200 mM sodium phosphate buffer, pH 7.4, containing 500 M homovanilic acid and 3 g/ml horseradish peroxidase. 10 l of 3 N NaOH was added to neutralize the medium after incubating for 15 min with HCl. The activity was measured using a standard curve obtained by incubating increasing amounts of H 2 O 2 (0 -10 M) in the same solution.
Both methods measured the H 2 O 2 produced either directly or through O 2 . dismutation and gave the same results. All superoxide anions that could have been generated during incubation spontaneously disproportionated to form H 2 O 2 before horseradish peroxidase and its substrate (scopoletin or homovanilic acid) were added.

Measurement of H 2 O 2 Release from Open Follicles and Cells-
The H 2 O 2 released by porcine cells and HEK293 transiently transfected cells in the medium was determined by measuring the oxidation of homovanilic acid into its fluorescent derivative in the presence of horseradish peroxidase (28). Cells and follicles from six wells were washed twice with Hepes-buffered Earle's solution and incubated for 45 min at 37°C in 0.8 ml of Hepes-buffered Earle's solution containing 0.44 mM homovanilic acid, 10 g/ml (2 units/ml) horseradish peroxidase, and 1 mM sodium azide, with or without 5 M ionomycin. At the end of the incubation, 200 l of medium from each well was collected, and the fluorescence level was measured (320-and 420-nm excitation and emission, respectively). The H 2 O 2 release was determined using a standard curve obtained by incubating increasing amounts of H 2 O 2 (0 -25 M) in the same solution.
Generation of Duox2 Constructs-Eight cysteine residues are present in the N-terminal domain of the pig Duox2, and their roles have been analyzed in a previous study (35). These cysteine residues were replaced by glycines by site-directed mutagenesis using the QuikChange site-directed mutagenesis kit from Stratagene. Deletion of the extracellular domain of Duox2 was also created using the same kit with the pig DUOX2-pcDNA3.1 vector as a template. The ⌬151-555 deletion was created using 5Ј-GACGTGGTGCTGCCCTTCCAGAGGTCCAGTGCCC-TGCAGCCCAATGTC-3Ј and 5Ј-GACATTGGGCTGCAGGGCACTGGA-CCTCTGGAAGGGCAGCACCACGTC-3Ј as mutagenic oligonucleotides. The constructs were confirmed by sequencing.

Measurement of Superoxide Formation by Spin Trapping and Electron Paramagnetic Resonance Spectroscopy (EPR)-
The spin trap 5-tert-butyloxycarbonyl 5-methyl-1-pyrroline N-oxide (BMPO) was synthesized using the procedure described by Zhao et al. (36). A typical incubation mixture contained 0.1 mM NADPH, 1 mM NaN 3 , 1 mM EDTA, 1.5 mM CaCl 2 , 0.1 M FAD, and 50 mM BMPO in 200 mM sodium phosphate, pH 7.4. Proteins from particulate fractions of HEK293 cells stably transfected with human NOX5b (25 g) or from particulate fractions of HEK293 cells transiently transfected with porcine Duox2 (370 g) were mixed with the previous mixture and incubated for 30 min at 37°C (total incubation volume 700 l). The reaction mixture was quenched by freezing in liquid nitrogen and thawed and centrifuged for 5 min at 10,000 rpm at 4°C just before EPR analysis. The supernatant was rapidly transferred into the Aqua-X sample cell fitted in an shq001 cavity (Bruker, Wissembourg, France), and data accumulation was started immediately. To analyze superoxide production in the thyroid cell, pig thyroid follicles were used for the EPR experiment. Open follicles (ϳ2 ϫ 10 6 cells) were incubated for 30 min at 37°C in 0.8 ml of Hepes-buffered Earle's solution (pH 7.2) containing 1 mM sodium azide, 1 mM CaCl 2 , 0.1 mM EDTA, 1 M ionomycin, and 50 mM BMPO. The reaction mixture was analyzed as indicated above.
All measurements were carried out at 20°C in a Bruker EPR Elexsys 500 spectrometer operating at X-band frequency (9.82 GHz). The following instrument settings were used: field modulation frequency, 100 kHz; field modulation amplitude, 0.2 milliteslas; time constant, 0.081 s; microwave power, 5.1 milliwatts; scan time, 41.9 s; averaged scans, 9. BMPO-OOH and BMPO-OH spectra were identified by comparison with incubations performed in the presence of xanthine/xanthine oxidase (for BMPO-OOH) as previously described (36). As expected, the BMPO-OOH spin adduct spontaneously decomposed to the BMPO-OH spin adduct with a half-life of about 15 min (36), and its formation was completely abolished in the presence of SOD. The relative amounts of BMPO-OOH and BMPO-OH adducts observed in the incubation mixtures were estimated by Xsophe software simulations (Bruker). The second peaks of the BMPO-OOH and BMPO-OH adducts are superimposed (maximum, 348.8 milliteslas; minimum, 349.2 milliteslas), and the amplitude of this peak was used to quantify the amounts of superoxide generated in our incubations. The term "BMPO-OOH spin adduct" will be used instead of "BMPO-OOH and BMPO-OH spin adducts" throughout this work for brevity's sake.

NADPH/Ca 2ϩ -dependent H 2 O 2 Generation Catalyzed by
Duox2 and Duox1-Using the Flp-In system, two stable, isogenic HEK293 cell lines were generated expressing human Duox1 and Duox2, respectively. This system has the advantage of making it possible to integrate DUOX1 and DUOX2 into the cells at the same specific genomic location, thus generating two cell lines with comparable expression of DUOX1 and DUOX2. Two antibodies directed against the median domain of each of these proteins were used to evaluate their respective levels of expression by Western blot analysis (Fig. 1A). The antibodies revealed Duox1/2 as two bands with apparent molecular masses of 150 and 165 kDa, corresponding to the nonglycosylated and partially glycosylated form, respectively (21,28). The highly glycosylated 175-kDa form of Duox, thought to be the mature protein involved in the active NADPH oxidase at the plasma membrane (21,28), was not detected in this experiment. Duox2 and Duox1 expression levels appeared to be similar in the two cell lines (Fig. 1A).
We measured the ability of the particulate fraction from DUOX1/2-transfected cells to generate H 2 O 2 in the presence of NADPH. Thyroid NADPH oxidase requires the presence of micromolar concentrations of calcium to acquire a functional conformation and to generate H 2 O 2 (34,37,38). Duox1 and Duox2 contain two Ca 2ϩ -binding motifs (EF-hand), which could be involved in the direct activation of the H 2 O 2 generator by calcium. Consequently, the NADPH-dependent H 2 O 2 -forming activities were measured in the presence and in the absence of calcium. Fig. 1A shows that the particulate fractions of human DUOX1-or DUOX2-transfected cells incubated with NADPH generated H 2 O 2 in a Ca 2ϩ -dependent manner, as did the thyroid particulate fraction (34). Interestingly, the activity of the particulate fraction from DUOX1 stably transfected cells was much lower than that found for their DUOX2 counterpart. In contrast, the NADPH/Ca 2ϩ -dependent H 2 O 2 -generating activity of the particulate fraction from cells transiently transfected with porcine DUOX2 cDNA was double that found for the particulate fraction from cells transiently transfected with human DUOX2 (Fig. 1B). This did not seem to be due to differences in the levels of expression of the proteins, which appeared to be similarly expressed in both fractions (Fig. 1B), but rather to structural differences resulting from slight variations in the primary sequence in the human and porcine forms of Duox2. Similar results were obtained when CHO cells were used for the transfection experiments instead of HEK293 cells (Fig. 1C).
Comparison of the H 2 O 2 -generating Activity of Mature and Immature Porcine Duox2-Although porcine DUOX2-transfected HEK293 cells expressed an active Ca 2ϩ -dependent H 2 O 2 generator, it had a specific activity that was only one-tenth that of the NADPH oxidase present in the particulate fraction from porcine thyroid follicles (Fig. 2A). The low specific activity of the recombinant porcine Duox2 in HEK293 cells could be related to the absence of expression of the highly glycosylated, 175-kDa mature form of Duox in these cells rather than to differences in Duox expression levels, which appeared to be similar in both particulate fractions (Fig. 2B).
The ability of porcine follicles and of porcine or human DUOX2 transiently transfected HEK293 cells to release H 2 O 2 into the medium was measured in the presence or absence of the Ca 2ϩ ionophore, ionomycin, because H 2 O 2 production by the thyroid gland is stimulated by calcium (39). As previously shown (28), pig thyroid follicles released H 2 O 2 (Fig. 2C), and this activity was increased when ionomycin and calcium were added to the medium. In contrast, porcine or human DUOX2transiently transfected HEK293 cells containing the partially glycosylated 165-kDa form of Duox2 located in the endoplasmic reticulum (35) did not release H 2 O 2 (Fig. 2, C and D). These observations reinforce the suggestion that the 175-kDa protein could be the mature, active form of Duox located at the plasma membrane of thyrocytes (21,28).
Role of Tpo and p22 phox in the Duox2 Basal H 2 O 2 -generating Activity-A recent study has shown that thyroperoxidase and p22 phox co-immunoprecipitated with Duox in human thyrocytes (25). These findings indicate that these two proteins could be involved in the formation of a multimeric complex implicated in a H 2 O 2 -generating system. Co-transfection experiments with Tpo and p22 phox did not allow the Duox to mature, and no extracellular H 2 O 2 production was detected under these conditions (25). To investigate the role of Tpo or p22 phox in the H 2 O 2 -generating activity of Duox2 under our conditions, transfections and co-transfections were carried out. A slight stimulation of the H 2 O 2 -generating activity of human Duox2 was seen with Tpo alone (Fig. 3A). Western blot analysis showed that this stimulation was not related to an increase in the Duox2 protein content or to a change of the glycosylation status (Fig. 3B). Thus, it appears that the interaction between the two proteins could occur in the endoplasmic reticulum. In contrast, transfection of p22 phox alone (Fig. 3A) or in association with TPO (data not shown) had no impact on the enzymatic activity.
Role of the N-terminal Domain in the Enzymatic Activity-In a previous study, we have shown that substitution of 4 of the 8 cysteines (Cys 351 , Cys 370 , Cys 568 , and Cys 582 ) located in the ectodomain of Duox2 plays a critical role in its maturation (35) (Fig. 4A). In the present study, we analyzed the effect of the substitutions of these cysteines on the H 2 O 2 -generating activity of immature Duox2. The results obtained showed that neither of them inhibited the H 2 O 2 -generating activity (data not shown). A deletion between the amino acids Ser 151 and Ser 555 in the N-terminal domain of Duox2 was undertaken to evaluate the role of this domain in the enzymatic activity. As shown in Fig. 4B, the deletion induced a 50% reduction in the NADPHdependent H 2 O 2 -generating activity. As shown in the Western blot analysis (Fig. 4C), this reduction was associated with a decrease in the expression of the truncated Duox2 protein compared with its normal counterpart. In addition to the truncated Duox2 band, other bands with lower molecular weights were detected, suggesting that it could be more vulnerable to degradation. The co-transfection experiment with GFP did not show less transfection efficiency in the truncated Duox-trans- fected cells, suggesting that the inhibition of activity associated with the truncated protein was probably due to a lower level of expression rather than to the modification of the N-terminal domain. This would imply that the N-terminal part of Duox could stabilize the partially glycosylated protein rather than play any role in the transfer of electrons from NADPH to O 2 .
Superoxide Generation-Previous studies using intact thyroid cells (40) and thyroid plasma membrane (41) 5A). When incubated in the presence of the spin trap BMPO, NADPH, and calcium, particulate fractions of NOX5-transfected cells (25 g) (Fig. 5B) and of porcine DUOX2-transfected cells (370 g) (Fig. 5C) generated EPR spectra identical to those formed by incubations with xanthine and xanthine oxidase (data not shown). They contained a mixture of BMPO-OOH, the radical O 2 Ϫ adduct of BMPO, and BMPO-OH, resulting from the self-decomposition of the BMPO-OOH adduct (half-life about 15 min) (36). Simulations of the relative amounts of the BMPO-OOH and BMPO-OH spin adducts led to ratios close to 15:85, closely matching the known halflife of the BMPO-OOH adduct (36). The spectrum observed can thus be considered to result from the direct addition of  . produced by the proteins to BMPO and will be indicated below as BMPO-OOH. In contrast, the EPR signals detected with the corresponding control transfected HEK293 cells were weak (Fig. 5, B and C). The spectrum of the BMPO-OOH spin adducts generated in the presence of calcium (Fig. 5, D and E, line 1) were strongly reduced in incubations of Nox5and Duox2-containing particles performed in the absence of calcium (Fig. 5, D and E, line 2) and almost completely abolished in incubations performed in the presence of SOD (Fig. 5, B and C, line 3). These data confirmed that the partially glycosylated Duox2 was responsible for O 2 . production in a Ca 2ϩ -dependent manner, in the same way as Nox5.
Because this finding apparently conflicted with previous data showing that the thyroid NADPH oxidase in the plasma membranes does not liberate O 2 . (41), we undertook the same experiments using these membrane fractions. Since the specific activity of the H 2 O 2 -generating enzyme in these membranes was 600 Ϯ 33 nmol ϫ h Ϫ1 ϫ mg protein Ϫ1 , we performed spin-trapping experiments using 45 times less protein from the thyroid plasma membrane than from the porcine DUOX2-transfected ones. Under these conditions, we measured the same NADPH/Ca 2ϩ -dependent H 2 O 2 -generating activities (Fig. 6A). However, when a EPR spectrum was ob-served with porcine DUOX2-transfected cells under these conditions, only a weak EPR signal was generated with the thyroid plasma membrane. Using higher amounts of protein from the thyroid membrane fraction, we did not detect any EPR signal comparable with the one generated by the NOX5transfected cell fractions with the same H 2 O 2 production (Fig. 6C). In addition, the weak signal generated by the thyroid plasma membrane was not Ca 2ϩ -dependent.   (13). (ii) The association of inactivating mutations in the human DUOX2/ThOX2 gene with the complete absence of thyroglobulin iodination (22) and the abolition in Caenorhabditis elegans of the peroxidase-catalyzed formation of di-and trityrosine linkages in the cuticle of animals treated by DUOXdirected, double-stranded RNA suggest that Duox provides H 2 O 2 for peroxidase reactions (23). (iii) The release of H 2 O 2 by Duox1-expressing bronchial epithelial cells was significantly reduced following exposure of these cells to an antisense oligonucleotide (15). However, to date, no Duox-based functional NADPH oxidase has been reconstituted from the plasma membrane of transfected cells because of the incomplete maturation of Duox proteins, which accumulate in the endoplasmic reticulum instead of reaching the plasma membrane (21,28). In the present study, we demonstrate for the first time that transfecting DUOX cDNA into cells initially devoid of NADPH-oxidase activity allows the expression of a functional Ca 2ϩ /NADPH-dependent, H 2 O 2 -generating system. Although Duox1/2 proteins were only partially glycosylated and not targeted to the plasma membrane, we found a significant level of H 2 O 2 -generating activity associated with the intracellular membranes of transfected CHO and HEK293 cells in vitro. The level of activity depended on the primary structure of the Duox proteins, since porcine Duox2 was twice as active as its human counterpart. Interestingly, we observed that human Duox1 was less active than human Duox2 regardless of the cell types and conditions of transfection used. The activity found for porcine Duox2 was only 10% of that found for the pig thyrocyte particulate fractions, suggesting that a partner and/or a post-translational process was required for full activity. As previously reported (21), the lack of extracellular H 2 O 2 release from transfected cells was correlated to the absence of the most highly glycosylated 175-kDa form of Duox2, which has so far only been observed in porcine thyroid tissue and digestive tract, in forskolin-stimulated cultures of pig and human thyrocytes and in rat thyroid cell lines (20,21,28). By studying the targeting of the human and porcine Duox2 N-terminal region to the plasma membrane, we identified two regions responsible for their remaining in endoplasmic reticulum: a human retention signal encompassing residues 596 -685 and a porcine retention domain in the first intracellular loop that inhibit the Golgi apparatus-specific modification of carbohydrate motifs (35). One or more additional component(s) involved in the maturation of Duox may be specifically expressed in tissues containing a functional Duox-based NADPH oxidase, but if they exist, they remain to be identified.
The stimulating effect of calcium on recombinant, Duoxbased, NADPH oxidase strongly suggests that the two EF-hand motifs located in the first intracellular loop of Duox are functional. They could trigger the reversible activation of the thyroid H 2 O 2 generator by calcium observed in vitro with porcine and human thyroid membrane fractions (34,38) and in intact cells from thyroid follicles (39) and slices (43,44). Recently, it has been clearly demonstrated that Nox5, which contains four calcium-binding sites in its N-terminal domain, including three canonical EF-hand motifs, is also directly activated by calcium (8). It has been suggested that the N terminus of Nox5 may act as a calmodulin-like activator module. The binding of calcium induces a conformational change, resulting in intramolecular interaction between the N terminus and the C-terminal catalytic domain, leading to the activation of the NADPH oxidase (8). Since limited proteolysis of thyroid NADPH oxidase by ␣-chymotrypsin results in an H 2 O 2 generator that is fully and irreversibly active in the absence of Ca 2ϩ (45), we propose that Ca 2ϩ activation may act via the displacement or conformational change of an autoinhibitory domain rather than via an activator domain, as has been reported for Nox5. This means that measuring the Ca 2ϩ -dependent activity of recombinant Duox could open new ways of analyzing the mechanism by which it is activated by this cation.
Hydrogen peroxide is the final electron acceptor in the biosynthesis of thyroid hormone catalyzed by thyroperoxidase at the apical surface of thyrocytes (46). Recently, immunoprecipitation experiments have shown that Tpo interacts with the Duox in the human thyrocyte and in COS cells cotransfected with TPO and DUOX, suggesting that these proteins are involved in a multimeric complex implicated in the iodide organification process (25). Our present data show that co-transfection of TPO with DUOX2 induced a slight increase in the H 2 O 2 -generating activity, indicating that the resting form of Duox2 located at the endoplasmic reticulum may interact with Tpo.
NADPH oxidase cytochrome b 558 , which is expressed in phagocytic cells, consists of two integral membrane proteins: gp91 phox or Nox2 and p22 phox (2). Recent data have shown that p22 phox interacts not only with Nox2 but also with the Nox2 homologues Nox1 (26,27) and Nox4 (27). These interactions were found to be required for the formation of functionally active enzymes (26,27). p22 phox mRNA is widely expressed, and it was also found in the thyroid (21). Recently, p22 phox has been shown to co-immunoprecipitate with the Duox proteins in human thyrocytes, suggesting that it could also constitute a subunit of the thyroid H 2 O 2 -generating system (25). In the present study, co-transfecting p22 phox with Duox2 had no effect on the H 2 O 2 -generating activity. This finding and the observation that the same enzymatic activity was obtained with the HEK293 cells, known to endogenously express p22 phox (27), and with CHO cells, which reportedly lack endogenous p22 phox expression (26), suggests that p22 phox may not interact with the resting form of Duox2 at the endoplasmic reticulum.
Studies carried out with various exogenous electron acceptors have shown that only ferricyanide is clearly reduced in a Ca 2ϩdependent manner by thyroid NADPH oxidase (30) in a partially purified preparation of thyroid plasma membrane. Despite the ability of NADPH oxidase to catalyze this monovalent reduction reaction, no superoxide anion could be detected. Furthermore, the cytochrome c reductase activity present in the thyroid plasma membrane could be separated from the NADPH-dependent H 2 O 2 -generating activity by detergent extraction without any impairment of its own activity (41) (23), this peroxidase-like, N-terminal region, although not found in our study to be essential for the electron transfer from NADPH to O 2 , might confer on the Duox proteins their unique ability to release H 2 O 2 rather than superoxide, which has also been found to inhibit thyroperoxidase activity (48).