The Extracellular A-loop of Dual Oxidases Affects the Specificity of Reactive Oxygen Species Release*

Background: Dual oxidase (Duox)-Duox activator (DuoxA) complexes produce H2O2, not O2⨪, suggesting that specialized mechanisms convert O2⨪ to H2O2. Results: In comparison with Duox2, Duox1 prevents O2⨪ leakage more stringently. Conclusion: Duox A-loops function in reducing O2⨪ release by promoting the stabilization and maturation of Duox-DuoxA complexes. Significance: The mechanism underlying H2O2 production by Duoxes has been clarified. NADPH oxidase (Nox) family proteins produce superoxide (O2⨪) directly by transferring an electron to molecular oxygen. Dual oxidases (Duoxes) also produce an O2⨪ intermediate, although the final species secreted by mature Duoxes is H2O2, suggesting that intramolecular O2⨪ dismutation or other mechanisms contribute to H2O2 release. We explored the structural determinants affecting reactive oxygen species formation by Duox enzymes. Duox2 showed O2⨪ leakage when mismatched with Duox activator 1 (DuoxA1). Duox2 released O2⨪ even in correctly matched combinations, including Duox2 + DuoxA2 and Duox2 + N-terminally tagged DuoxA2 regardless of the type or number of tags. Conversely, Duox1 did not release O2⨪ in any combination. Chimeric Duox2 possessing the A-loop of Duox1 showed no O2⨪ leakage; chimeric Duox1 possessing the A-loop of Duox2 released O2⨪. Moreover, Duox2 proteins possessing the A-loops of Nox1 or Nox5 co-expressed with DuoxA2 showed enhanced O2⨪ release, and Duox1 proteins possessing the A-loops of Nox1 or Nox5 co-expressed with DuoxA1 acquired O2⨪ leakage. Although we identified Duox1 A-loop residues (His1071, His1072, and Gly1074) important for reducing O2⨪ release, mutations of these residues to those of Duox2 failed to convert Duox1 to an O2⨪-releasing enzyme. Using immunoprecipitation and endoglycosidase H sensitivity assays, we found that the A-loop of Duoxes binds to DuoxA N termini, creating more stable, mature Duox-DuoxA complexes. In conclusion, the A-loops of both Duoxes support H2O2 production through interaction with corresponding activators, but complex formation between the Duox1 A-loop and DuoxA1 results in tighter control of H2O2 release by the enzyme complex.


that the A-loop of Duoxes binds to DuoxA N termini, creating more stable, mature Duox-DuoxA complexes. In conclusion, the A-loops of both Duoxes support H 2 O 2 production through interaction with corresponding activators, but complex formation between the Duox1 A-loop and DuoxA1 results in tighter control of H 2 O 2 release by the enzyme complex.
Dual oxidases (Duoxes 3 ; Duox1 and Duox2) are members of the NADPH oxidase (Nox) family proteins (Nox1-5 and Duoxes) that produce reactive oxygen species (ROS) (1-3). Duox1 (4) and Duox2 (5) are functional only in combination with maturation factors known as Duox activators (DuoxAs; DuoxA1 and DuoxA2) (6). Although DuoxAs were first described as factors required to permit Duoxes to exit the endoplasmic reticulum, it was later reported that Duox and DuoxA proteins form stable heterodimers and co-translocate to the plasma membrane (7). Both Duoxes were first characterized as thyroid oxidases supporting thyroid hormone biosynthesis, although Duox2 is the dominant form, with an expression level five times higher than that of Duox1 in thyroid tissue (8). Moreover, mutations or deficiencies in Duox2 or DuoxA2 have been reported to cause congenital hypothyroidism in mice (9,10) and humans (8), whereas deficiency in Duox1 has no effect on thyroid hormone levels in Duox1 knock-out mice (11). Bi-allelic mutations in Duox2 reportedly cause transient congenital hypothyroidism, suggesting that some compensation occurs by Duox1 (12). More recently, a patient with transient congenital hypothyroidism was described; this patient was compound heterozygous for a large deletion comprising DUOX2, DUOXA2, and DUOXA1 and a nonfunctional missense muta-tion of DUOXA2 in the other allele, suggesting compensation by the remaining Duox1-DuoxA1 complex or by the mismatched Duox2-DuoxA1 complex (13).
In addition, Duox enzymes have been detected and believed to function on the epithelial cell surfaces of mucosal and exocrine tissues (2, 14 -16), including the airways and gastrointestinal tract. In these tissues, Duoxes also function in host defense against a broad spectrum of pathogens (17)(18)(19). Nox family proteins produce the primary product superoxide (O 2 . ) by directly transferring an electron to molecular oxygen (2,20). Duoxes produce O 2 . as an intermediate product (21), but the final product generated by mature Duoxes is H 2 O 2 , suggesting that intramolecular O 2 . dismutation or other mechanisms contribute to H 2 O 2 release (2). In tissues with high Duox expression, the enzymes accumulate on the apical plasma membrane, facilitating H 2 O 2 release from epithelial cell surfaces to support the activities of extracellular hemoperoxidases (22). In contrast to Nox1-5, Duoxes have an extended N-terminal extracellular domain called the peroxidase homology (PoxH) domain, followed by an additional transmembrane (TM) segment and an intracellular loop containing two calcium-binding EF-hand motifs (4,23) (Fig. 1A). Thus, Duoxes possess four extracellular regions: the PoxH domain (1-595 amino acids (aa) in human Duox2) and three loops (the A-loop, 1064 -1078 aa; C-loop, 1146 -1184 aa; E-loop, 1242-1251 aa in human Duox2) (Fig. 1A). The PoxH domain is a candidate for intramolecular O 2 . dismutation, although the isolated PoxH domains of both Duoxes demonstrate no O 2 . dismutation activity in vitro (24,25

EXPERIMENTAL PROCEDURES
Materials-A polyclonal antibody (pAb) against Duoxes, which preferentially detects Duox2 via the first intracellular loop (639 -1039 aa of human Duox2) but also detects Duox1, was previously described (15). A pAb against DuoxA1, which recognizes the extracellular N terminus of DuoxA1, was obtained from Santa Cruz Biotechnology, Inc. Unfortunately, no commercial Ab against DuoxA2 for immunoblotting or immunostaining is available. mAbs against HA(TANA2)-conjugated HRP, Alexa Fluor 488, and magnetic agarose were obtained from MBL International Corp. An mAb against FLAG(M2)-conjugated HRP was obtained from Sigma. Endoglycosidase H (Endo H) was obtained from New England Biolabs.
In Vitro Binding (Pulldown) Assays-Forward and reverse oligonucleotides for DuoxA N terminus (1-20 aa for DuoxA1 or 1-19 aa for DuoxA2) were annealed and cloned into the BamHI and EcoRI sites of pGEX-6P-1. Purified GST and GSTtagged DuoxA N-terminal proteins (DuoxA1(N), GST-MATL-GHTFPFYAGPKPTFP; DuoxA2(N), GST-MTLWNGVLPF-YPQPRHAAG) were obtained as described previously (27). Biotin-labeled Duox A-loop peptides (Duox1(Aloop), biotin-AAHHTGITDTTRV; Duox2(Aloop), biotin-ALPPSDIAQT-TLV) were synthesized by MBL International. GST-tagged DuoxA(N) was mixed with biotin-labeled Duox(Aloop) in 300 l of binding buffer (500 nM each) (27). After rotation for 2 h at 4°C, 40 l of streptavidin-coupled magnetic beads (Dynabeads M-280 Streptavidin; Invitrogen) were added to the solution, and the mixture was agitated for 90 min at 4°C. The precipitates were washed 3 times using a magnetic rack, and then the material absorbed to the beads was eluted in Laemmli sample buffer; the magnetic beads were then removed using a magnetic rack. The eluents were subjected to SDS-PAGE followed by immunoblotting using a polyclonal antibody against GST (Santa Cruz Biotechnology). Bound antibodies were detected with an HRP-conjugated secondary antibody using the ECL detection system (GE Healthcare).
Immunoprecipitation and Immunoblotting-Various pairs of 2ϫHA-Duox and FLAG-tagged DuoxA constructs were co-transfected into HEK293 cells plated on 10-cm dishes using FuGENE 6 (Promega). Forty-eight hours after transfection, the cells were lysed in 250 l of lysis buffer with a protease inhibitor mixture (27) by sonication. Total cell lysates were centrifuged at 800 ϫ g for 5 min at 4°C, and the supernatants were incubated with 10 l of magnetic agarose-conjugated HA mAb for 2 h at 4°C. The precipitates were washed 3 times, and aliquots of the precipitates were subjected to SDS-PAGE followed by immunoblotting using an HRP-conjugated FLAG mAb and detected using the ECL detection system.
N-Deglycosylation Analysis-The deglycosylation assay was performed as previously described (7). Briefly, various pairs of 2ϫHA-Duox and DuoxA constructs were co-transfected into HEK293 cells plated in 10-cm dishes using FuGENE 6. Fortyeight hours after transfection, the cells were lysed in 250 l of lysis buffer with a protease inhibitor mixture. After centrifugation at 12,000 ϫ g for 10 min at 4°C, equal amounts of proteins were treated with 100 units/50 l of Endo H for 30 min at 37°C or left untreated and separated by SDS-PAGE. Immunoblotting was performed using an HRP-conjugated HA mAb.
Confocal Fluorescence Imaging Studies-A total of 2.5 ϫ 10 5 HEK293 cells were seeded in 35-mm glass-bottomed dishes (MatTek Corp.) 48 h before transfection and transfected using FuGENE 6. Thirty-two hours after transfection, the cells were fixed using 4% paraformaldehyde in 0.1 M PBS (pH 7.4) without permeabilization and stained using an Alexa Fluor 488-conjugated HA mAb (1:500) at room temperature for 2 h for visualization by confocal laser scanning fluorescence microscopy (LSM700; Carl Zeiss AG). All imaging experiments were performed in triplicate and were repeated in at least three independent transfection experiments (n Ն 9).
ROS Production Assay-HEK293 cells were seeded in 6-well dishes at 2.5 ϫ 10 5 cells/well 48 h before transfection. HEK293 cells were transfected using FuGENE 6 in complexes with various combinations of plasmids. The cells were fed 6 h posttransfection with complete medium and harvested using 0.02% EDTA solution (Nacalai Tesque). Thirty-two hours after transfection, 2 ϫ Red (Invitrogen) ϩ HRP for H 2 O 2 detection for 10 min using a luminometer (Mithras LB940; Berthold Detection Systems GmbH) (28) as previously described (7,29). O 2 . production from 5 ϫ 10 5 cells in 100 l of HBSS(Ϫ) stimulated by 0.2 M ionomycin ϩ 2 mM Ca 2ϩ was also measured based on the assay of cytochrome c (100 M, Sigma) reduction with a molar extinction coefficient of 21 mM Ϫ1 cm Ϫ1 at 550 nm using a monochrometer (Multiskan GO; Thermo Fisher Scientific) as previously described (7,30). O 2 . production was inhibited by the addition of 10 units/ml superoxide dismutase (Sigma) in the assay solution. Comparable expression of proteins was confirmed by immunoblotting using total lysates from the same number of cells. Mean oxidase activities were calculated from at least three independent transfection experiments. Statistical Analysis-All data are presented as the means Ϯ S.E. of mean. For comparisons of more than two groups, oneway analysis of variance was performed. Statistical analyses were performed using GraphPad Prism 5.0 software (GraphPad Software Inc.); p Ͻ 0.05 was considered statistically significant.    . production was used as a positive control. All the samples were subjected to the superoxide dismutase treatment, except for Duox2-DuoxA2. C, immunoblotting detects expression levels of Duox2 and Duox1 in various pairs as well as the expression of FLAG-and HA-tagged DuoxAs. but to a lesser extent than that by Duox2-DuoxA1 (Fig. 5B). Duox132(A:8aa) maintained capabilities to secret H 2 O 2 , which were detected by Amplex Red ϩ HRP, as in the case of Duox1 and Duox2 (Fig. 5B). We then made two additional Duox1 mutants: Duox132(A:8aaϩL) and Duox132(A: 8aaϩGL) (  . , and reactive oxygen species were measured by chemiluminescence assay using Diogenes and luminol ϩ HRP, respectively. Duox2 ϩ DuoxAs with the N-terminal extracellular region of DuoxA1 show O 2 . production in the following order: DuoxA(1-2) Ͼ DuoxA(1F-2) ϭ DuoxA(2-1S-2). Immunoblotting detected the expression of Duoxes and DuoxAs. A pAb against Duoxes preferentially reacts with Duox2 via its first intracellular loop but also faintly detects Duox1 (the last two lanes). A pAb for DuoxA1, which reacts with its extracellular N terminus, only detects wild-type DuoxA1. . and total reactive oxygen species were measured by chemiluminescence assay using Diogenes and luminol ϩ HRP, respectively. Pairs (red arrows) with the Duox2 portion after the first transmembrane segment (termed the Nox-like portion) ϩ DuoxA with the N-terminal extracellular region of DuoxA1 show O 2 . production. Immunoblotting detects the expression levels of various Duox and DuoxA pairs. A pAb against Duoxes faintly detected Duox2 chimeras in the two right-hand lanes. A pAb for DuoxA1 detects only wild-type DuoxA1.  . and reactive oxygen species production measured by chemiluminescence assay using Diogenes and luminol ϩ HRP, respectively, are shown. B, immunoblotting detects comparable expression levels of various Duox and DuoxA pairs. 8aaϩL); this was confirmed using PP/HP and PP/PH mutations in Duox132(A:8aaϩL) (data not shown). In addition, although the expression levels of Duox132(A:PPD), in which three critical residues (HHϩG) in the A-loop of Duox1 were exchanged for those of Duox2, were apparently low (ROS production detected by luminol ϩ HRP was 36.8 Ϯ 10.8%), it showed no O 2 . release (Fig. 5C). Taken together, these results suggest that these specific residues alone do not account for the function of the Duox A-loop in preventing O 2 . release. Comparable expression of constructs was confirmed by immunoblotting (Fig. 5, B and C). . and ROS production (Fig. 6A). Duox231(A:8aa)-DuoxA1 showed a kinetic curve of ROS production similar to that of Duox1-DuoxA1 (Fig. 6A). Duox2-DuoxA2(N-del) showed an increase in O 2 . release and a decrease in ROS production compared with Duox2-DuoxA2 (Fig. 6A). Duox231(A:8aaϩL)-DuoxA2 did not show any O 2 . release (Fig. 6A). In the Duox1-DuoxA1(N-del) pair, ROS production was severely impaired (Ͻ10% that of Duox1-DuoxA1), and no apparent plasma membrane targeting/localization of Duox1 or O 2 . release was observed (data not shown).

Tags at the N terminus of
Comparable expression of constructs was confirmed by immunoblotting (Fig. 6B). Finally, we assessed the involvement of the C-and E-loops in O 2 . release using chimeric mutants in which their sequences in Duox2 were replaced with those of Duox1 (Fig. 7, A and C). O 2 . release was not reduced in either the C-loop mutant, Duox231(C:10aa), or the E-loop mutant, Duox231(E:TY/RF) (Fig. 7, B and D). Comparable expression of constructs was confirmed by immunoblotting (Fig. 7, B and D). Taken together, we conclude that the A-loop, but not the C-or E-loop, of Duox2 is involved in O 2 . release. . and ROS were measured by chemiluminescence assays using Diogenes and luminol ϩ HRP, respectively. The Duox231(C:10aa) ϩ DuoxA pairs show no decreased O 2 . production. Immunoblotting (from the same membrane and image) detects comparable expression levels of various Duox and DuoxA pairs. C, alignment of the E-loops of Duox1 and Duox2 and illustrations of chimeric Duox2 mutants indicating the residues change in the E-loops from Duox2 to Duox1 (Duox231(E)). D, Duox231(E:TY/RF) and DuoxAs pairs were transfected into HEK293 cells. O 2 . and ROS were measured by chemiluminescence assay using Diogenes and luminol ϩ HRP, respectively. The Duox231(E:TY/RF) ϩ DuoxA pairs show no decreased O 2 . production. Immunoblotting detects comparable expression levels of various Duox and DuoxA pairs. . release even when matched with DuoxA1 (Fig. 8B). Comparable expression of constructs was confirmed by immunoblotting (Fig. 8B). Taken  Next, we investigated possible direct interaction between the biotin-labeled A-loop peptides of Duoxes and the GST-tagged N-terminal extracellular sequences of the DuoxAs by pulldown assays using streptavidin-conjugated magnetic beads. We detected the strongest interaction between Duox1(Aloop) and DuoxA1(N) among Duox1(Aloop)-DuoxA1(N), Duox2(Aloop)-DuoxA2(N), Duox1(Aloop)-DuoxA2(N), and Duox2(Aloop)-DuoxA1(N) pairs (Fig. 9).
We then focused on the interaction between full-length Duoxes and DuoxAs at the cellular level because our previous paper established that Duox maturation, reflected in N-glycosyl modifications and stable interactions with DuoxAs, affects the type of ROS produced in that fully processed and stable complexes do not leak O 2 . (7). The relationship between O 2 . release and Duox binding to DuoxAs was examined by immunoprecipitation assays using lysates from HEK293 cells transfected with various Duox and DuoxA combinations. The order of the strength of Duox2 binding to FLAG-tagged DuoxAs was 3NϫFLAG-DuoxA2 Ͻ DuoxA1-3CϫFLAG Ͻ DuoxA2-3CϫFLAG (Fig. 10A). This order inversely correlated with the amount of O 2 . release by the Duox2 complex: 3NϫFLAG- 10A). The order of the strength of DuoxA1 binding to HAtagged Duoxes was HA-Duox132(A:8aaϩL) Ͻ HA-Duox2 Ͻ HA-Duox1 ϭ HA-Duox231(A:8aa) (Fig. 10B). This order also inversely correlated with the amount of O 2 . release: HA-Duox132(A:8aaϩ L) Ͼ HA-Duox2 Ͼ HA-Duox1 ϭ HA-Duox231(A:8aa) (Fig. 10B) (Fig. 10C,  . release was also susceptible to Endo H treatment (Fig. 10C

DISCUSSION
Duox enzymes serve as dedicated H 2 O 2 generators at the plasma membrane, where they support the activities of extracellular hemoperoxidases (22). We (7) and another group (26) previously reported a switch in the type of ROS released, from H 2 O 2 to O 2 . , with co-expression of the mismatched Duox2 and DuoxA1 pair, but not with Duox1 and DuoxA2 (7). These findings suggest that DuoxA proteins contribute to Duox matura-tion and subcellular targeting; they also function as a part of the ROS-generating complex. Subsequently, Hoste et al. (31) showed leakage by supporting the stabilization and maturation of the . production. C, various HA-tagged Duox and DuoxA pairs were transfected into HEK293 cell. Forty-eight hours after transfection lysates were treated with Endo H, and immunoblotting was performed using an HRP-conjugated HA mAb. Representative data (n Ն 3) show the presence of Endo H-sensitive bands in O 2 . producing pairs (HA-Duox2 ϩ DuoxA1 and HA-Duox132(A:8aaϩL) ϩ DuoxA1).
Duox-DuoxA complex. Although the A-loop of Duox1 is more effective at preventing O 2 . leakage than that of Duox2, the A-loops of both the Duox isozymes function in preventing O 2 .
release (Fig. 8). A recent study (33)  conserved in Duox1 in many species (Fig. 11), and the corresponding PP residues in the A-loop of Duox2 are also well conserved in many species. . , also lack HH residues and are even more divergent, consistent with their effects in inducing or enhancing O 2 . release by chimeric Duox1 or Duox2 proteins, respectively (Fig. 8). Moreover, the compound heterozy-gous mutation of Duox2(L1067S) in the A-loop combined with other Duox2 mutations was identified in patients with transient congenital hypothyroidism (12), further supporting the importance of the A-loop structure to Duox function. We found direct binding of the A-loops of Duoxes to the N termini of DuoxAs in vitro (Fig. 9). In addition, the strength of the interaction between the Duox and DuoxA pairs at the cellular level inversely correlated with the amount of O 2 . releasedweaker interactions correlated with greater O 2 . leakage (Fig. 10,   A and B). Furthermore, the A-loop of Duox1 induced complete Endo H-resistant glycosyl modifications of Duoxes (Fig. 10C).
To determine whether glycosyl modifications of Duoxes affect ROS production and O 2 . leakage, we created glycosylation-defective mutants of untagged and HA-tagged human Duox1 and Duox2 in which all five putative glycosylation sites (4) (Asn-94, -342, -354, -461, and -534 in Duox1; Asn-100, -348, -382, -455, and -537 in Duox2) were replaced by Gln. In Duox-reconstituted HEK293 cell systems, these mutants showed severely impaired glycosylation (almost no band shift in Duoxes upon immunoblotting), plasma membrane targeting/localization of the Duox-DuoxA complex, and ROS production or O 2 . leakage detected by luminol ϩ HRP or Diogenes (data not shown). Taken together, these results suggest that 1) binding of the A-loop of Duox1 to the N terminus of DuoxA1 contributes to the enhanced interaction between Duox1 and DuoxA1 observed at the cellular level, thereby inducing more stable, mature Duox-DuoxA complexes than those involving Duox2, and 2) glycosyl modifications of Duoxes are essential for the targeting/localization of the Duox-DuoxA complex to the plasma membrane, increasing the stability of the Duox-DuoxA complex and developing capabilities for ROS production (including O 2 . leakage) on the plasma membrane. The formation of more stable Duox-DuoxA heterodimers or structural changes associated with maturation, which are probably acquired at the final destination, the apical cell surface, may create an environment that more efficiently supports ROS conversion and prevents O 2 . leakage. In