Independently cloned halves of cytomegalovirus assemblin, An and Ac, can restore proteolytic activity to assemblin mutants by intermolecular complementation.

Herpesviruses encode an essential serine proteinase called assemblin that is responsible for cleaving the precursor assembly protein during the process of capsid formation. In cytomegalovirus (CMV), assemblin undergoes autoproteolysis at an internal (I) site located near the middle of the molecule. I-site cleavage converts the enzyme to an active two-chain form consisting of the subunits An and Ac. We have recently shown that the recombinant An and Ac subunits can spontaneously associate within eukaryotic cells to yield active two-chain proteinase. This finding indicates that the subunits are able to independently assume their correct functional conformations and led us to test whether they are capable of intermolecular complementation. This was done by coexpressing inactive mutant (point, deletion, and insertion) forms of assemblin together with the wild-type subunit (either An or Ac) corresponding to the domain of assemblin that was mutated. Results of these experiments showed that both An and Ac are able to rescue the enzymatic activity of assemblin mutants. I-site cleavage of the mutated assemblin occurred during complementation but was not absolutely required, as shown by effective complementation of inactive assemblins with noncleavable I sites. We have also shown that intermolecular complementation can rescue the activity of an inactive mutant full-length proteinase precursor and can occur between different species of CMV (e.g., human CMV subunit can rescue activity of mutant simian CMV assemblin). These results indicate that assemblin is able to form active multimeric structures that may be of functional importance.     

Uncoupling the enzymatic and autoprocessing activities of Helicobacter pylori gamma-glutamyltranspeptidase

Gamma-glutamyltranspeptidase (gammaGT), a member of the N-terminal nucleophile hydrolase superfamily, initiates extracellular glutathione reclamation by cleaving the gamma-glutamyl amide bond of the tripeptide. This protein is translated as an inactive proenzyme that undergoes autoprocessing to become an active enzyme. The resultant N terminus of the cleaved proenzyme serves as a nucleophile in amide bond hydrolysis. Helicobacter pylori gamma-glutamyltranspeptidase (HpGT) was selected as a model system to study the mechanistic details of autoprocessing and amide bond hydrolysis. In contrast to previously reported gammaGT, large quantities of HpGT were expressed solubly in the inactive precursor form. The 60-kDa proenzyme was kinetically competent to form the mature 40- and 20-kDa subunits and exhibited maximal autoprocessing activity at neutral pH. The activated enzyme hydrolyzed the gamma-glutamyl amide bond of several substrates with comparable rates, but exhibited limited transpeptidase activity relative to mammalian gammaGT. As with autoprocessing, maximal enzymatic activity was observed at neutral pH, with hydrolysis of the acyl-enzyme intermediate as the rate-limiting step. Coexpression of the 20- and 40-kDa subunits of HpGT uncoupled autoprocessing from enzymatic activity and resulted in a fully active heterotetramer with kinetic constants similar to those of the wild-type enzyme. The specific contributions of a conserved threonine residue (Thr380) to autoprocessing and hydrolase activities were examined by mutagenesis using both the standard and coexpression systems. The results of these studies indicate that the gamma-methyl group of Thr380 orients the hydroxyl group of this conserved residue, which is required for both the processing and hydrolase reactions.     

Lysosomal neuraminidase. Catalytic activation in insect cells is controlled by the protective protein/cathepsin A

Lysosomal N-Acetyl-alpha-neuraminidase is active in complex with the protective protein/cathepsin A (PPCA) and beta-galactosidase. The interaction with PPCA is essential for the correct intracellular routing and lysosomal localization of neuraminidase, but the mechanism of its catalytic activation is unclear. To investigate this process, we have used the baculovirus expression system to co-express neuraminidase and PPCA precursors in insect cells, which resulted in high enzymatic activity of neuraminidase. Both the 34- and 20-kDa PPCA subunits were required for the activation. We further demonstrated that when expressed alone, the neuraminidase precursor remained dimeric (114 kDa) and had low enzymatic activity, but when co-expressed with PPCA and beta-galactosidase, it multimerized in a complex of approximately 1350 kDa, together with the other two proteins. The fully active neuraminidase co-precipitated with full-length PPCA and beta-galactosidase precursors. However, when co-expressed with the individual PPCA subunits, neuraminidase co-precipitated only with the small 20-kDa polypeptide, which therefore must contain a neuraminidase-binding site. Our finding suggests a model of activation of neuraminidase dependent on its oligomerization at acidic pH that is mediated by interaction with PPCA.     

Disruption of a salt bridge dramatically accelerates subunit exchange in duck delta2 crystallin

Intragenic complementation is a unique property of oligomeric enzymes with which to study subunit-subunit interactions. Complementation occurs when different subunits, each possessing distinct mutations that render the individual homomutant proteins inactive, interact to form a heteromutant protein with partial recovery of activity. In this paper, complementation events between human argininosuccinate lyase (ASL) and its homolog, duck delta2 crystallin, were characterized. Different active site mutants in delta2 crystallin complement by the regeneration of native-like active sites as reported previously for ASL. The complementarity of the ASL and delta2 crystallin subunit interfaces was illustrated by the in vivo formation of active hybrid tetramers from inactive ASL and inactive delta2 crystallin mutants. Subunits of both ASL and delta2 crystallin do not dissociate and reassociate in vitro at room temperature, even after 6 days of incubation, indicating that the multimerization interface is very strong. However, disruption of a salt bridge network in the tetrameric interface of delta2 crystallin caused a drastic acceleration of subunit dissociation. Double mutants combining these interface mutants with active site mutants of delta2 crystallin were able to dissociate and reassociate to form active tetramers in vitro within hours. These results suggest that exchange of subunits may occur without unfolding of the monomer. Intragenic complementation in these interface mutants occurs by reintroducing the native salt bridge interaction upon hetero-oligomerization. Our studies demonstrate the value of intragenic complementation as a tool for investigating subunit-subunit interactions in oligomeric proteins.     

Biosynthesis of soluble carnitine acetyltransferases from the yeast Candida tropicalis.

Soluble carnitine acetyltransferase from Candida tropicalis is synthesized as a 76 kDa precursor, which is monomeric and possesses no or very little carnitine acetyltransferase activity. Maturation of the enzyme begins with proteolytic processing of the 76 kDa precursor to 64 and 57 kDa subunits. The processed subunits subsequently associate into two kinds of active oligomers; the 57 kDa subunits are assembled into a tetramer and the 64 kDa subunits into an octamer. Formation of these oligomers depends apparently on growth conditions, since both oligomers were present in cells grown in continuous culture, but cells grown batchwise contained only the tetrameric form of carnitine acetyltransferase.     

Nedd8 Regulates Inflammasome-Dependent Caspase-1 Activation

Caspase-1 is activated by the inflammasome complex to process cytokines like interleukin-1β (IL-1β). Pro-caspase-1 consists of three domains, CARD, p20, and p10. Association of pro-caspase-1 with the inflammasome results in initiation of its autocatalytic activity, culminating in self-cleavage that generates catalytically active subunits (p10 and p20). In the current study, we show that Nedd8 is required for efficient self-cleavage of pro-caspase-1 to generate its catalytically active subunits. Nedd8 silencing or treating cells with the neddylation inhibitor MLN4924 led to diminished caspase-1 processing and reduced IL-1β maturation following inflammasome activation. Coimmunoprecipitation and mass spectrometric analysis of 293 cells overexpressing pro-caspase-1 (and CARD) and Nedd8 suggested possible neddylation of caspase-1 CARD. Following inflammasome activation in primary macrophages, we observed colocalization of endogenous Nedd8 with caspase-1. Similarly, interaction of endogenous Nedd8 with caspase-1 CARD was detected in inflammasome-activated macrophages. Furthermore, enhanced autocatalytic activity of pro-caspase-1 was observed following Nedd8 overexpression in 293 cells, and such activity in inflammasome-activated macrophages was drastically diminished upon treatment of cells with MLN4924. Thus, our studies demonstrate a role of Nedd8 in regulating caspase-1 activation following inflammasome activation, presumably via augmenting autoprocessing/cleavage of pro-caspase-1 into its corresponding catalytically active subunits.     

IL-1-converting enzyme requires aspartic acid residues for processing of the IL-1 beta precursor at two distinct sites and does not cleave 31-kDa IL-1 alpha

IL-1 converting enzyme (ICE) specifically cleaves the human IL-1 beta precursor at two sequence-related sites: Asp27-Gly28 (site 1) and Asp116-Ala117 (site 2). Cleavage at Asp116-Ala117 results in the generation of mature, biologically active IL-1 beta. ICE is unusual in that preferred cleavage at Asp-X bonds (where X is a small hydrophobic residue), has not been described for any other eukaryotic protease. To further examine the substrate specificity of ICE, proteins that contain Asp-X linkages including transferrin, actin, complement factor 9, the murine IL-1 beta precursor, and human and murine IL-1 alpha precursors, were assayed for cleavage by 500-fold purified ICE. The human and murine IL-1 beta precursors were the only proteins cleaved by ICE, demonstrating that ICE is an IL-1 beta convertase. Analysis of human IL-1 beta precursor mutants containing amino acid substitutions or deletions within each processing site demonstrated that omission or replacement of Asp at site 1 or site 2 prevented cleavage by ICE. To quantitatively assess the substrate requirements of ICE, a peptide-based cleavage assay was established using a 14-mer spanning site 2. Cleavage between Asp [P1] and Ala [P1']2 was demonstrated. Replacement of Asp with Ala, Glu, or Asn resulted in a greater than 100-fold reduction in cleavage activity. The rank order in position P1' was Gly greater than Ala much greater than Leu greater than Lys greater than Glu. Substitutions at P2'-P4' and P6' had relatively little effect on cleavage activity. These results show that ICE is a highly specific IL-1 beta convertase with absolute requirements for Asp in P1 and a small hydrophobic amino acid in P1'.     

Characterization of cyclic AMP-resistant Chinese hamster ovary cell mutants lacking type I protein kinase

A group of three mutants of Chinese hamster ovary cells (10260, 10265, and 10223) which are resistant to cyclic AMP (Gottesman, M. M., LeCam, A., Bukowski, M., and Pastan I. (1980) Somatic Cell Genet. 6, 45-61) have been characterized in this work. By genetic analysis, these mutants are all recessive and fall into two complementation groups. Cycl AMP-stimulated protein kinase activity in crude extracts of these mutants using histone as a substrate is decreased to 10 and 7% (complementation group I), and 31% (complementation group II), respectively, of the activity found in wild type extracts. The binding of cyclic [3H]AMP by extracts of all of these mutants is decreased to 30 to 50% of the binding found in wild type extracts. We have used the photoaffinity label 8-azidoadenosine 3':5'-[32P]monophosphate to label the regulatory subunits of type I and type II protein kinase in wild type and mutant extracts analyzed by DEAE-cellulose and Sephadex chromatography. We find that all three mutants lack type I cyclic AMP-dependent protein kinase and have reduced amounts of type II kinase activity. The regulatory subunits of type I and type II kinase are present in both complementation groups. We conclude that type I protein kinase is not needed for normal growth of Chinese hamster ovary cells. The defect in both classes of mutants appears to be in the failure of the catalytic subunit to associate normally with its regulatory subunits.     

Mechanisms for intragenic complementation at the human argininosuccinate lyase locus.

Argininosuccinate lyase (ASL) is a homotetrameric enzyme that catalyzes the reversible cleavage of argininosuccinate to arginine and fumarate. Deficiencies in the enzyme result in the autosomal, recessive disorder argininosuccinic aciduria. Considerable clinical and genetic heterogeneity is associated with this disorder, which is thought to be a consequence of the extensive intragenic complementation identified in patient strains. Our ability to predict genotype-phenotype relationships is hampered by the current lack of understanding of the mechanisms by which complementation can occur. The 3-dimensional structure of wild-type ASL has enabled us to propose that the complementation between two ASL active site mutant subunits, Q286R and D87G, occurs through a regeneration of functional active sites in the heteromutant protein. We have reconstructed this complementation event, both in vivo and in vitro, using recombinant proteins and have confirmed this hypothesis. The complementation events between Q286R and two nonactive site mutants, M360T and A398D, have also been characterized. The M360T and A398D substitutions have adverse effects on the thermodynamic stability of the protein. Complementation between either the M360T or the A398D mutant and the stable Q286R mutant occurs through the formation of a more stable heteromeric protein with partial recovery of catalytic activity. The detection and characterization of a novel complementation event between the A398D and D87G mutants has shown how complementation in patients with argininosuccinic aciduria may correlate with the clinical phenotype.     

Hetero- and autoprocessing of the extracellular metalloprotease (Mpr) in Bacillus subtilis

Most proteases are synthesized as inactive precursors which are processed by proteolytic cleavage into a mature active form, allowing regulation of their proteolytic activity. The activation of the glutamic-acid-specific extracellular metalloprotease (Mpr) of Bacillus subtilis has been examined. Analysis of Mpr processing in defined protease-deficient mutants by activity assay and Western blotting revealed that the extracellular protease Bpr is required for Mpr processing. pro-Mpr remained a precursor form in bpr-deficient strains, and glutamic-acid-specific proteolytic activity conferred by Mpr was not activated in bpr-deficient strains. Further, purified pro-Mpr was processed to an active form by purified Bpr protease in vitro. We conclude that Mpr is activated by Bpr in vivo, and that heteroprocessing, rather than autoprocessing, is the major mechanism of Mpr processing in vivo. Exchange of glutamic acid for serine in the cleavage site of Mpr (S93E) allowed processing of Mpr into its mature form, regardless of the presence of other extracellular proteases, including Bpr. Thus, a single amino acid change is sufficient to convert the Mpr processing mechanism from heteroprocessing to autoprocessing.     

Complementation in vitro between guaB mutants of Escherichia coli K12

Guanine auxotrophs of Escherichia coli were isolated following mutagenesis by N-methyl-N'-nitro-N-nitrosoguanidine or ethyl methanesulphonate. The mutants were classified according to growth properties and absence of IMP dehydrogenase or GMP synthetase activity. Mutations in guaB (IMP dehydrogenase-less) were analysed by reversion and suppression tests; all were of the base substitution missense type except for one possible frameshift and one polar nonsense mutation. GuaB mutants were examined for protein (CRM) that cross-reacts with monospecific antibodies to IMP dehydrogenase; approximately half were CRM+. Enzyme complementation in vitro was detected in mixed denatured and renatured cell-free extracts of any CRM+ guaB mutant and PL1138 (guaB105, CRM+); CRM- mutants did not complement. GuaB105 maps distal to all other guaB mutations except guaB86 (CRM-). Two hybrid enzymes produced by complementation were less stable to heat than native IMP dehydrogenase, although kinetic constants were similar. These observations indicate interallelic complementation between guaB mutants and are consistent with the demonstration of identical subunits for IMP dehydrogenase (Gilbert et al., 1979). Only the subunits supplied by PL1138 are catalytically active in the hybrid enzymes suggesting that this mutant may produce a repairable polypeptide whereas the enzymes of complementing mutants may be defective at the active site.     

Intersubunit location of the active site of ribulose-bisphosphate carboxylase/oxygenase as determined by in vivo hybridization of site-directed mutants

Ribulose bisphosphate carboxylase/oxygenase from Rhodospirillum rubrum is a homodimer of 50.5-kDa subunits with two substrate binding sites per molecule of dimer. To determine whether each subunit contains an independent active site or whether the active sites are created by intersubunit interactions, we have used a novel in vivo approach for producing heterodimers from catalytically inactive, site-directed mutants of the carboxylase. When the alleles encoding these mutant proteins are placed separately into compatible plasmids and coexpressed in the same Escherichia coli host, activity is observed at about 20% of the wild-type level. Analysis of the carboxylase purified from these cells reveals the presence of heterodimers of the two mutant proteins. This interallelic complementation demonstrates that domains from each of the subunits interact to form a shared active site.     

Characterization of qa-2 mutants of Neurospora crassa by genetic, enzymatic, and immunological techniques.

Genetic and complementation mapping studies using 20 qa-2 mutants defective for catabolic dehydroquinase indicate that the qa-2 gene encodes a single polypeptide chain and is the structural gene for catabolic dehydroquinase, a 220,000-molecular-weight protein composed of identical 10,000-molecular-weight subunits. Many qa-2 mutants are capable of reversion, but no evidence has yet been obtained for nonsense mutations in this gene. The biochemical consequences of the mutations in two complementing qa-2 strains (M239 and M204) have been determined. Both mutants have extremely low levels of catalytic activity and form a heterocaryon with about 4% of the wild-type activity. As assayed by immunological cross-reactivity, mutant M239 and the heterocaryon have nearly wild-type levels of native-molecular-weight catabolic dehydroquinase protein, whereas M204 has no detectable amount of this protein. Thus it is concluded that M239 has a mutation at or near the catalytic site which reduces the activity 10,000-fold but has little or no influence on the formation of the native multimeric structure. In contrast, M204 apparently has a mutation that severely inhibits aggregation and may have only a minor effect on the inherent potential for catalytic conversion at the reactive site. The heterocaryon would appear to form a mixed multimer with the monomeric subunits from M239 providing the aggregated structure and those from M204, the catalytically active moiety.     

Folding pathway mediated by an intramolecular chaperone: dissecting conformational changes coincident with autoprocessing and the role of Ca(2+) in subtilisin maturation.

Subtilisin is produced as a precursor that requires its N-terminal propeptide to chaperone the folding of its protease domain. Once folded, subtilisin adopts a remarkably stable conformation, which has been attributed to a high affinity Ca(2+) binding site. We investigated the role of the metal ligand in the maturation of pro-subtilisin, a process that involves folding, autoprocessing and partial degradation. Our results establish that although Ca(2+) ions can stabilize the protease domain, the folding and autoprocessing of pro-subtilisin take place independent of Ca(2+) ion. We demonstrate that the stabilizing effect of calcium is observed only after the completion of autoprocessing and that the metal ion appears to be responsible for shifting the folding equilibrium towards the native conformation in both mature subtilisin and the autoprocessed propeptide:subtilisin complex. Furthermore, the addition of active subtilisin to unautoprocessed pro-subtilisin in trans does not facilitate precursor maturation, but rather promotes rapid autodegradation. The primary cleavage site that initiates this autodegradation is at Gln19 in the N-terminus of mature subtilisin. This corresponds to the loop that links alpha-helix-2 and beta-strand-1 in mature subtilisin and has indirect effects on the formation of the Ca(2+) binding site. Our results show that the N-terminus of mature subtilisin undergoes rearrangement subsequent to propeptide autoprocessing. Since this structural change enhances the proteolytic stability of the precursor, our results suggest that the autoprocessing reaction must be completed before the release of active subtilisin in order to maximize folding efficiency.     

Characterization of Helicobacter pylori gamma-glutamyltranspeptidase reveals the molecular basis for substrate specificity and a critical role for the tyrosine 433-containing loop in catalysis

Helicobacter pylori gamma-glutamyltranspeptidase (HpGT) is a member of the N-terminal nucleophile hydrolase superfamily. It is translated as an inactive 60 kDa polypeptide precursor that undergoes intramolecular autocatalytic cleavage to generate a fully active heterodimer composed of a 40 kDa and a 20 kDa subunit. The resultant N-terminus, Thr 380, has been shown to be the catalytic nucleophile in both autoprocessing and enzymatic reactions. Once processed, HpGT catalyzes the hydrolysis of the gamma-glutamyl bond in glutathione and its conjugates. To facilitate the determination of physiologically relevant substrates for the enzyme, crystal structures of HpGT in complex with glutamate (1.6 A, Rfactor = 16.7%, Rfree = 19.0%) and an inactive HpGT mutant, T380A, in complex with S-(nitrobenzyl)glutathione (1.55 A, Rfactor = 18.7%, Rfree = 21.8%) have been determined. Residues that comprise the gamma-glutamyl binding site are primarily located in the 20 kDa subunit and make numerous hydrogen bonds with the alpha-amino and alpha-carboxylate groups of the substrate. In contrast, a single hydrogen bond occurs between the T380A mutant and the remainder of the ligand. Lack of specific coordination beyond the gamma-glutamyl moiety may account for the substrate binding permissiveness of the enzyme. Structural analysis was combined with site-directed mutagenesis of residues involved in maintaining the conformation of a loop region that covers the gamma-glutamyl binding site. Results provide evidence that access to this buried site may occur through conformational changes in the Tyr 433-containing loop, as disruption of the intricate hydrogen-bond network responsible for optimal placement of Tyr 433 significantly diminishes catalytic activity.     

Towards a model to explain the intragenic complementation in the heteromultimeric protein propionyl-CoA carboxylase

Mutations in the PCCA or PCCB genes coding for alpha and beta subunits of propionyl CoA carboxylase can cause propionic acidemia. To understand the molecular basis of the intragenic complementation previously reported at the PCCB locus, we now examine the complementation behaviour of four carboxy-terminal and 11 amino-terminal naturally occurring mutant alleles both using cell fusion and reconstructing the complementation event by transfecting the mutant cDNAs to generate multimeric hybrid proteins. Alleles carrying mutations p.R410W and p.W531X are able to complement with 10 out of 11 amino-terminal mutations assayed. Only the unstable p.R512C, p.L519P and p.G112D mutants fail to complement. The results analyzed in the framework of the crystal structure of the homologous 12S transcarboxylase from Propionibacterium shermanii show that all mutant alleles studied are located at beta subunits interfaces, complementing alleles at the inter-trimer interface, where the catalysis probably happens, and non-complementing alleles at the intra-trimer interface, probably disrupting the trimer formation. Our results also show a remarkable stabilization effect when p.R410W is cotransfected with p.G246V. We propose a model for intragenic complementation requiring the production of two different beta subunits carrying carboxy and amino-terminal mutations that allow regenerating functional active sites and in which a stabilization effect between subunits could be relevant to ameliorate the biochemical phenotype of each mutation separately.     

Intermediates in the formation of mouse 20S proteasomes: implications for the assembly of precursor beta subunits.

The assembly of individual proteasome subunits into catalytically active mammalian 20S proteasomes is not well understood. Using subunit-specific antibodies, we characterized both precursor and mature proteasome complexes. Antibodies to PSMA4 (C9) immunoprecipitated complexes composed of alpha, precursor beta and processed beta subunits. However, antibodies to PSMA3 (C8) and PSMB9 (LMP2) immunoprecipitated complexes made up of alpha and precursor beta but no processed beta subunits. These complexes possess short half-lives, are enzymatically inactive and their molecular weight is approximately 300 kDa. Radioactivity chases from these complexes into mature, long-lived approximately 700 kDa proteasomes. Therefore, these structures represent precursor proteasomes and are probably made up of two rings: one containing alpha subunits and the other, precursor beta subunits. The assembly of precursor proteasomes occurs in at least two stages, with precursor beta subunits PSMB2 (C7-I), PSMB3 (C10-II), PSMB7 (Z), PSMB9 (LMP2) and PSMB10 (LMP10) being incorporated before others [PSMB1 (C5), PSMB6 (delta), and PSMB8 (LMP7)]. Proteasome maturation (processing of the beta subunits and juxtaposition of the two beta rings) is accompanied by conformational changes in the (outer) alpha rings, and may be inefficient. Finally, interferon-gamma had no significant effect on the half-lives or total amounts of precursor or mature proteasomes.     

Nox5 forms a functional oligomer mediated by self-association of its dehydrogenase domain

Nox5 belongs to the calcium-regulated subfamily of NADPH oxidases (Nox). Like other calcium-regulated Noxes, Nox5 has an EF-hand-containing calcium-binding domain at its N-terminus, a transmembrane heme-containing region, and a C-terminal dehydrogenase (DH) domain that binds FAD and NADPH. While Nox1-4 require regulatory subunits, including p22phox, Nox5 activity does not depend on any subunits. We found that inactive point mutants and truncated forms of Nox5 (including the naturally expressed splice form, Nox5S) inhibit full-length Nox5, consistent with formation of a dominant negative complex. Oligomerization of full-length Nox5 was demonstrated using co-immunoprecipitation of coexpressed, differentially tagged forms of Nox5 and occurred in a manner independent of calcium ion. Several approaches were used to show that the DH domain mediates oligomerization: Nox5 could be isolated as a multimer when the calcium-binding domain and/or the N-terminal polybasic region (PBR-N) was deleted, but deletion of the DH domain eliminated oligomerization. Further, a chimera containing the transmembrane domain of Ciona intestinalis voltage sensor-containing phosphatase (CiVSP) fused to the Nox5 DH domain formed a co-immunoprecipitating complex with, and functioned as a dominant inhibitor of, full-length Nox5. Radiation inactivation of Nox5 overexpressed in HEK293 cells and endogenously expressed in human aortic smooth muscle cells indicated molecular masses of ～350 and ～300 kDa, respectively, consistent with a tetramer being the functionally active unit. Thus, Nox5 forms a catalytically active oligomer in the membrane that is mediated by its dehydrogenase domain. As a result of oligomerization, the short, calcium-independent splice form, Nox5S, may function as an endogenous inhibitor of calcium-stimulated ROS generation by full-length Nox5.