|Home | About | Journals | Submit | Contact Us | Français|
Dual oxidase (DUOX) enzymes support a wide variety of essential reactions, from cellular signaling to thyroid hormone biosynthesis. In Caenorhabditis elegans, the DUOX system (CeDUOX1/2) plays a crucial role in innate immunity and in stabilizing the cuticle by forming tyrosine cross-links. The current model suggests that superoxide generated by CeDUOX1 at the C-terminal NADPH oxidase domain is rapidly converted to H2O2. The H2O2 is then utilized by the N-terminal peroxidase-like domain to cross-link tyrosines. We have now created a series of mutations in the isolated peroxidase domain, CeDUOX11–589. One set of mutations investigate the roles of a putative distal tyrosine (Tyr105) and Glu238, a proposed covalent heme-binding residue. The results confirm that Glu238 covalently binds to the heme group. A second set of mutations (G246D and D392N) responsible for a C. elegans blistering cuticle phenotype was also investigated. Surprisingly, although not among the catalytic residues, both mutations affected heme co-factor binding. The G246D mutant bound less total heme than the wild type, but a higher fraction of it was covalently bound. In contrast, the D392N mutant appears to fold normally but does not bind heme. Molecular dynamics simulations of a CeDUOX11–589 homology model implicate displacements of the proximal histidine residue as the likely cause. Both enzymes are structurally stable and through altered heme interactions exhibit partial or complete loss of tyrosine cross-linking activity, explaining the blistering phenotype. This result argues that the CeDUOX peroxidase domain is primarily responsible for tyrosine cross-linking.
Members of the NADPH oxidase/dual oxidase (DUOX)2 family are found throughout eukaryotic species, including invertebrates, insects, nematodes, fungi, amoeba, algae, and plants (1–7). The reactive oxygen species (ROS) produced by members of this family are utilized for a wide variety of functions, including cellular signaling, innate immunity, and thyroid hormone biosynthesis. In nonmammalian or lower species, the DUOX primary function appears to be stabilization of extracellular matrices through oxidative tyrosine cross-linking (2, 8). Lower organisms have been the focus of recent DUOX research to help further clarify protein function. Analysis of the zebrafish dual oxidase enzyme revealed its role in leukocyte recruitment to the site of wounding by a cell signaling process based on a hydrogen peroxide gradient produced by a DUOX enzyme (9). A mosquito DUOX investigation has revealed a unique network of protection that results in survival of the malaria parasite by peroxidase/DUOX-mediated formation of a dityrosine cross-linked chitin matrix (10). The semi-permeable matrix, produced upon feeding, prevents an immune response to bacteria and parasites, allowing the Plasmodium parasite to develop for transmission back to humans. Similar results suggesting a role for DUOX in regulation of gut immunity have been obtained through studies with Drosophila. RNAi studies in Drosophila melanogaster targeted at ROS production by DUOX in the intestines demonstrated a higher mortality rate than targeting the main immune system tissues (hemocytes) (11, 12). This suggests that host protection from foreign ingested materials may be common among lower organisms. Finally, research into Caenorhabditis elegans DUOX1 has revealed two residues (Gly246 and Asp392) that, when mutated, give rise to a blistering phenotype. This physiological response has been shown to be independent of ROS production, because the C. elegans DUOX1 enzyme produces comparable amounts of hydrogen peroxide in WT and mutant organisms. It thus appears that the peroxidase domain of the nematode DUOX enzyme is alone responsible for tyrosine cross-linking but is not directly involved in ROS production (13).
Two highly conserved DUOX proteins are encoded in the C. elegans genome: CeDUOX1/BLI-3 (1497 amino acids) and CeDUOX2 (1313 amino acids). The CeDUOX2 isoform, however, has a stop codon that should eliminate the extreme C-terminal portion of the protein, including a segment of the pyridine nucleotide-binding site. This protein may therefore contain intact peroxidase and calmodulin-like domains but fail to encode a functioning NADPH-oxidase domain, rendering the protein inactive. The inference that CeDUOX2 is nonfunctional is supported by the finding that a CeDUOX2 knockout strain demonstrated no differences in ROS production after pathogen infection or developmental abnormalities relative to wild-type nematodes (13).
The predicted amino acid sequence of CeDUOX1 shows ~30% identity with the human proteins hDUOX1 and hDUOX2 and preserves roughly the same size and domain organization (2). The extreme N-terminal 21 amino acids of CeDUOX1 contain a secretory signal peptide sequence, consistent with localization of the N-terminal peroxidase domain in an extracellular compartment. The residues involved in calcium ligation (EF-hand calcium-binding motifs) are well conserved in hDUOX1 and hDUOX2 but are poorly conserved in CeDUOX1. This suggests that the function of this region(s) may have evolved away from calcium binding in nematodes. Unlike many of the human NADPH oxidase/DUOX family members, no interacting proteins that are required for activity have been identified in the C. elegans system. Interestingly, the C. elegans genome does not appear to encode a homolog of p22phox, and an ancient homolog to duoxa is found on a separate chromosome from the duox genes, suggesting that the CeDUOX functions in the absence of activator proteins (8, 14). The current model for ROS generation by CeDUOX1 proposes that superoxide is generated through reduction of oxygen by two electrons provided by NADPH at the C-terminal NADPH oxidase domain. This superoxide, generated at the extracellular surface, is rapidly converted to hydrogen peroxide and is utilized by the N-terminal peroxidase domain to facilitate tyrosine cross-linking. This model for CeDUOX1 activity is supported by the recent identification of two point mutations within the peroxidase domain of CeDUOX1: G246D and D392N (13). Both of these mutations caused a loss of tyrosine cross-linking activity, yet neither mutant demonstrated a significant decrease in ROS production. These results argue that the peroxidase-like domain is directly involved in enzymatic tyrosine cross-linking but does not participate in ROS production.
In vitro investigations into the ability of the CeDUOX1 N-terminal domain to act as a peroxidase demonstrated that cell lysate from expression in C. elegans and E. coli had tyrosine cross-linking activity (2). In our previous work, the peroxidase domain of C. elegans DUOX1 (CeDUOX11–589) was expressed and purified with a bound heme co-factor (15). Evaluation of this protein demonstrated that CeDUOX11–589 produces tyrosine cross-links. To complement this research, a sequence alignment was done to compare the residues within the DUOX peroxidase domain of several lower organisms (see Fig. 1A). The partial alignment highlights residues within the heme-binding region, including a distal tyrosine (Tyr105) residue and a proximal iron-coordinated histidine (His330), as well as those found to be responsible for a blistering phenotype in C. elegans. Residues that were found to be well conserved included the catalytic Arg residue (CeDUOX1 Arg235), proximal histidine (CeDUOX1 His330), and the aspartic acid blister phenotype residue (CeDUOX1 Asp392). These residues may be important indicators of activity and/or peroxidase domain ability to bind heme, perhaps making possible future DUOX heme binding predictions based on sequence. Here, we explore mutations of the CeDUOX1 truncated peroxidase domain to investigate (a) the possible role of the distal tyrosyl residue in heme binding, (b) whether introduction of a distal His to better approximate a mammalian peroxidase increases peroxidase activity, and (c) the mechanism of the two point mutations that cause severe cuticle defects without affecting ROS production.
Sf9 cells (Invitrogen) were grown in ExCell 420TM medium (SAFC Biosciences) supplemented with glutamine (2.7 g/liter). High FiveTM cells were grown in Express FiveTM medium (Invitrogen) supplemented with glutamine (2.7 g/liter) and 10% FBS. Both cell lines were kept in suspension at 27 °C (100 rpm) and maintained at densities between 0.5 × 106 and 2 × 106 cells/ml. TFA, formic acid, H2O2 (30% w/w), 2,2′-azino-bis(3-ethylbenzothiazoleline-6-sulfonic acid) (ABTS), l-tyrosine ethyl ester, LPO (lactoperoxidase, bovine), and CaCl2 were purchased from Sigma-Aldrich. T4 DNA ligase and restriction endonucleases were obtained from New England Biolabs. DNA sequencing was performed by Elim Biopharmaceuticals. The entire gene insert was completely sequenced for each plasmid construct. Spectrophotometric measurements were performed on a Cary 50 Bio UV-visible spectrophotometer (Varian). LC-MS was performed on a Waters Micromass ZQ coupled to a Waters Alliance HPLC system (2695 separations module equipped with a Waters 2487 dual λ absorbance detector). For experiments utilizing H2O2, the concentrations were determined spectrophotometrically at 240 nm by using the molar extinction coefficient ϵ = 43.6 m−1 cm−1 (16). All of the experiments were performed at room temperature unless otherwise stated.
DUOX proteins (H. sapiens, C. elegans, D. melanogaster, Anopheles gambiae, and Danio rerio) were truncated using the TMHMM transmembrane helix algorithm to identify the primary sequences that compose each N-terminal peroxidase domain (17). All of the DUOX sequences, along with that of bovine LPO, were submitted to ClustalW for alignment generation (18). A structural model of CeDUOX11–589 was built by sequence submission to the SWISS-MODEL program server for automatic modeling; the model with bovine LPO (Protein Data Bank code 3BXI) as the template was visualized with Pymol software (19). Heme co-factor placement in the model structure was achieved by overlay of the known bovine LPO structure and the CeDUOX11–589 homology model.
The original source of the C. elegans duox11–589 gene was a ProQuest C. elegans cDNA Library (Invitrogen). The wild-type peroxidase domain construct of CeDUOX1 (residues 1–589) was described previously (15). Mutations in C. elegans duox1–589 were introduced into pJLM022 by QuikChange site-directed mutagenesis (Clontech), according to the manufacturer's instructions. Primer sequences are included in the supplemental data (supplemental Table S1). The entire open reading frame was resequenced for each new plasmid. Each plasmid construct was subsequently digested with BamHI and PstI, and the gene(s) were inserted by ligation into pAcGP67-b (WT, pJLM023; Y105F, pJLM035; Y105H, pJLM029; E238A, pJLM038; G246D, pJLM061; and D392N, pJLM060).
CeDUOX11–589 and CeDUOX11–589 Y105F, Y105H, E238A, G246D, and D392N mutants were expressed and purified according to the previously reported procedure (15). Each purified protein was stored at −20 °C, and all stock concentrations were determined in triplicate by Bradford assay (20).
The ferrous absorbance spectra for CeDUOX11–589 mutants (6 μm) were collected in 100 mm phosphate buffer, pH 7.0, 80 μl of total sample volume. The proteins were reduced by the addition of sodium dithionite (0.8 μl, 100 mm). The cyanide (CN−) bound absorbance spectra for the CeDUOX11–589 mutant proteins (6 μm) were collected via titration with KCN in 100 mm phosphate buffer, pH 7.0. A stock solution of KCN (100 mm) was utilized to titrate each enzyme until saturation was achieved, and the Soret maxima at saturation were recorded.
All of the samples were analyzed by direct injection onto a 150 × 4.6-mm C4 reversed phase column (Vydac 214MS) on an Agilent 1200 series HPLC instrument. The protein (25 μl, 6 μm) was eluted with a linear gradient of 20–50% acetonitrile in water (0.1% TFA) over 30 min (1 ml/min) with detection at 215 and 400 nm. The proportion of covalently bound heme was determined by the percentage of the total 400-nm integration attributed to the peak(s) that co-eluted with the 215-nm absorbing protein peak (27–30 min).
Enzyme peroxidase activity was measured using tyrosine ethyl ester as a substrate (2.5 mm) in PBS buffer (pH 7.4, treated with Chelex-100 resin (Bio-Rad)) as previously described (15). Briefly, CeDUOX11–589 wild type and mutants (1 μm) were assayed; each reaction was initiated by the addition of H2O2 (100 μm). Initial reaction kinetics of tyrosine cross-linking were monitored using a Fluorolog 3 Spectrofluorometer (Horiba Jobin Yvon), with excitation at 295 nm and emission monitored at 414 nm; excitation and emission slits were set at 1 nm. Each enzyme was assayed in triplicate at room temperature. Bovine LPO was assayed as a control against the CeDUOX proteins by fluorescence and UV-visible spectroscopy at 315 nm in triplicate. The molar extinction coefficient for LPO dityrosine formation at 315 nm, ϵ = 4700 m−1 cm−1, was utilized to convert fluorescence assay slopes to turnover values for each protein (22). Wild-type and mutant CeDUOX reactions were also assayed for 1 h at 1.0 mm tyrosine ethyl ester, 100 μm H2O2, to maintain solubility and were injected onto a 50 × 2.1-mm C18 reversed phase column (Waters XTerraMS) on an Agilent 1200 series HPLC instrument for product analysis. The assays were characterized utilizing an isocratic gradient of 9% acetonitrile in water (0.1% formic acid), followed by a linear gradient of 9–10% acetonitrile in water (0.1% formic acid) over 10 min (0.2 ml/min) with detection at 280 nm and fluorescence detection by excitation at 295 nm and emission at 414 nm. The same gradient and column were utilized for analysis of the observed peaks by LC-MS.
Peroxidase activity was measured using 500 nm ABTS incubated with 500 nm CeDUOX11–589 wild-type or mutants in 50 mm phosphate buffer (pH 7.0, chelexed). The reactions were initiated by the addition of H2O2 (50 μm). Each reaction was monitored by recording the absorbance intensity at 414 nm as a function of time. The molar extinction coefficient for ABTS at 414 nm is ϵ = 36,800 m−1 cm−1 (23).
Far-UV CD spectra were collected on a JASCO J-715 spectropolarimeter using a cuvette path length of 1.0 mm and spectral collection in the range of 195–350 nm at 20 °C. All of the CD experiments were conducted in 10 mm phosphate buffer, pH 7.0, at a protein concentration of 4 μm. Raw ellipticity data were converted to mean residue ellipticity before plotting.
The molecular dynamics starting structure was the C. elegans DUOX1 homology structure generated by the SWISS-MODEL program. Residues missing from the homology model at the N and C termini were added using Modeler (24). Molecular dynamics were carried out using the Gromacs 4.0 simulation software package with the OPLS force field and explicit TIP3P water (25–27). The water box was created 10 Å from the protein boundaries, and periodic boundary conditions were used. Simulations were run at 300 K with 4-fs time steps, with constant temperature maintained by the V-rescale algorithm (28). Hydrogen bonds were constrained using the LINCS algorithm (29). Fourteen Na+ ions were added to achieve a neutral charge. Long range electrostatic interactions were treated by using the Particle-Mesh Ewald method (30). Previously published heme atom parameters such as partial charges were used, with the following changes: the distance between the proximal ligand, a histidine, and the iron was set to the crystal structure distance found in the bovine LPO structure used as a template for building the homology model (31). The simulation contained 96,366 atoms, including: 10,069 atoms in the protein, 28,761 waters, and 14 Na+ ions. Energy minimization was followed by a 200-ps equilibration under constant pressure maintained by coupling the system to a pressure bath at 1.0 atm (32). Following equilibration, the simulations were maintained at constant volume. The data are based on one 40-ns trajectory collected on eight processors in the UCSF QB3 Linux cluster.
To calculate the mutual information (MI) in the dynamics between pairs of residues, we represent the trajectory of each residue as a distance from a moving average. MI between two random variables was calculated as follows (33).
Specifically, the average position of the CA atom was calculated in a 500-ps window centered on the current time step, and the distance between the current position and the moving average position was determined.3 MI was normalized by the joint entropy so that MI between lower entropy residues was weighted equally with MI between higher entropy residues.
Normalized MI, MInorm, is reported throughout the manuscript.
Residues were chosen for mutation, based on sequence alignment and previous research, to investigate the heme-binding pocket and the blistering cuticle phenotype in the C. elegans DUOX1 N-terminal peroxidase-like domain. A homology model was generated to approximate the location of each mutation relative to the heme co-factor (Fig. 1B). The model was produced utilizing LPO (Protein Data Bank code 3BXI) as a target; both secondary structure prediction (supplemental Fig. S1) and similarity in far-UV CD spectroscopy support LPO as a template (15, 34). The structure generated nicely models the relevant conserved catalytic residues in relation to the heme prosthetic group, suggesting, in the absence of a crystal structure, that this model reliably approximates residue proximity and orientation to the heme molecule.
To test the role of the distal tyrosine residue in the catalytic region, Tyr105 was mutated to either a histidine or phenylalanine (Y105H or Y105F). A comparison of available protein sequences for C. elegans DUOX1 can be found through NCBI for placement of a tyrosine and/or histidine at this position (accession numbers O61213 and AAF71303). In mammalian peroxidases, which are known to have greater peroxidase activity than observed for the isolated CeDUOX1 peroxidase domain, a histidine occupies this position (35–37). Interestingly, the DNA sequence encoding a tyrosine differs from that encoding a histidine by only one base, suggesting that it may be an evolutionary adaption. Replacement of the distal tyrosyl residue in CeDUOX11–589 by a histidine could yield higher peroxidase activity if the His → Tyr mutation alone is responsible for the much lower peroxidase activity of the CeDUOX1 peroxidase domain.
The glutamate residue responsible for a covalent bond to the heme co-factor (Glu238), demonstrated to be conserved in sequence alignment with mammalian peroxidases, is shown by our model to be in approximately the correct orientation to form a covalent bond with the heme. To perturb this interaction, the glutamate was mutated to an alanine, an amino acid incapable of covalent bond formation (E238A). This mutant protein should demonstrate less covalent character, and perhaps less activity, as is commonly seen when a heme-protein covalent bond is eliminated in other peroxidases (38–40).
Finally, mutations of Gly246 and Asp392, which are not located in the heme-binding pocket in the model structure, severely inhibit tyrosine cross-linking (G246D and D392N) (13). The effect of these two mutations on heme binding, structural stability, and catalytic activity will also be investigated. The location of Asp392 is especially interesting, because it is a solvent-exposed residue far from both the protein region that contacts the membrane and the heme-binding site. Mutation to an asparagine is not expected to perturb the charge of the region and is not predicted to introduce a glycosylated residue.
Each mutated C. elegans protein was stably overexpressed in baculovirus and purified by nickel-nitrilotriacetic acid affinity chromatography, as previously reported for the wild-type protein (15). No significant differences in heme association among the heme-binding pocket mutants were detected by absorbance spectral analysis (Table 1). Both CeDUOX11–589 Y105F and Y105H had a Soret maximum at 408 nm, with 408/280-nm ratios of 0.57 and 0.55, respectively (consistent with the WT value of 0.55). A pyridine hemochromogen assay established that this 408/420-nm ratio represented ~50% heme incorporation into the protein. This rate of heme co-purification, less than 1 to 1, is consistent with recombinant baculovirus expression of LPO and myeloperoxidase in the presence of 5-aminolevulinic acid, which proceed with 70 and 75% heme incorporation, respectively (41, 42). Small ligand binding was studied by titration with CN− to test for observable changes in its interaction with the heme iron at the distal position. No significant change was observed for any mutant in the Soret maximum at saturation (Table 1). A significant difference in the Soret was only noted when the covalent binding glutamate residue was removed (E238A). The Soret maximum for this protein is at 412 nm, but no change was observed in the amount of heme to co-purify with the recombinant protein (412/280 nm ratio of 0.55). This absorbance change is consistent with a change in the covalent character of the heme, as predicted for this residue. Further study of this interaction was conducted by tetramethylbenzidine staining of denaturing polyacrylamide gels and HPLC (Fig. 2). By tetramethylbenzidine staining, covalent heme binding was confirmed for all heme-binding pocket mutations, with an apparent decrease in the amount of heme co-migrating with the E238A protein. To confirm this difference, an HPLC system was utilized under acidic conditions that separate noncovalently bound heme from the apoprotein (43). As shown in Fig. 2B, only 20% of the prosthetic heme co-eluted with the protein peak (27–30 min, 215 nm trace not shown); in comparison the wild-type CeDUOX11–589 demonstrated 60% co-elution of the total heme with the protein. Both the Y105F and Y105H mutants were evaluated by this HPLC system and showed no difference from the wild type in covalent heme binding (data not shown). This significant decrease demonstrates that Glu238 participates in a covalent bond to the heme co-factor, as predicted by sequence alignment with mammalian peroxidases. Further investigation of this C. elegans DUOX mutant protein by enzymatic digestion to attempt identification of this second residue has not been successful because of instability of the protein under digestion conditions.
To examine the effect of each mutation on catalytic activity, the CeDUOX11–589 mutants were assayed with both ABTS and tyrosine ethyl ester as substrates (Fig. 3). Catalytic activity studies were first conducted by monitoring the absorbance change upon enzymatic incubation with ABTS and H2O2. As shown in Fig. 3A, mutation of the active site Tyr105 to a histidine does not cause a significant shift in the activity relative to the wild type with this classic peroxidase substrate (15.1 versus 14.1 mol oxidized per min/mol of enzyme, respectively). A modest decrease in activity was observed for the Y105F mutant, perhaps because of disruption of an important hydrogen bonding network in the distal heme pocket. The E238A mutation greatly abolished activity, consistent with removal of a covalent bond to the heme co-factor. The activity with tyrosine ethyl ester provided somewhat different results. A significant increase in dityrosine formation was observed when the distal tyrosine was replaced with a histidine residue, but this level did not approach that of the mammalian peroxidase LPO (Fig. 3B and supplemental Fig. S2). Initial reaction rates for tyrosine cross-linking place Y105H activity at seven times that of wild-type CeDUOX11–589; however, this increased turnover rate is still 170 times less than that of LPO. This suggests that although this residue exerts some control over activity with a natural substrate, other factors limit reactivity. This concept is supported by a previous study of myeloperoxidase in which mutation of two noncatalytic residues in the heme-binding region significantly altered the oxidizing potential of the enzyme (44). Product evaluation by HPLC, supported by fluorescent reaction rate studies (data not shown), demonstrated enzymatic inhibition of CeDUOX11–589 Y105H cross-linking activity over time. The dityrosine formed after 1 h for the proximal histidine mutant is only twice that of WT, suggesting that increased peroxidase activity is detrimental to the enzyme. Clearly, CeDUOX1 does not act on the peroxidase substrate ABTS in the same way as it does on tyrosine, showing that it is substrate-selective, either through filtering of substrate size or charge or through differential orientation in the active site.
The two C. elegans DUOX protein mutations known to produce a deficiency in tyrosine cross-linking were also overexpressed, purified, and characterized. Differences in heme association with each protein were noticed immediately; no heme appeared to co-purify with the D392N mutant, and a lower ratio of heme to protein was found for G246D mutant. These visual observations were verified by UV-visible spectroscopy, where a Soret maximum of 409 nm was observed for the G246D mutant. This slight red shift suggests a unique ratio of covalently to noncovalently bound heme for this mutant. The 409/280-nm ratio was found to be 0.42, demonstrating that less heme is associated with the G246D mutant than with the WT enzyme (Table 1 and Fig. 4A). No heme at all was bound to the D392N mutant, a surprising disruption in heme binding caused by a mutation that is located far from the heme-binding pocket. To ascertain whether the mutated protein(s) were correctly folded and structurally stable, CD measurements were performed for each mutant that were compared with those for wild-type CeDUOX11–589 (Fig. 4B). The far-UV spectra confirm that the protein constructs are folded with the same α-helical structure as WT, with a slight variation in the depth of the trough at 208 nm for the G246D mutant. To examine the character of the covalent heme bound to the G246D mutant, the previously described HPLC assay was utilized to promote dissociation of free heme from the apoprotein. Under these conditions, 76% of the total heme was found to be covalently bound to the protein, an increase over wild type and consistent with the observed shift in the Soret wavelength (Fig. 4C). The placement of Gly246 in proximity to Glu238 and the catalytic arginine residue Arg235 (Fig. 1) introduces a charged residue that may disrupt important interactions in the catalytic region. However, it is also possible that the noncovalently bound heme is held much more weakly, so that the ratio of bound to unbound heme increases artificially because of greater loss of noncovalently bound heme during purification.
To determine whether the changes in heme interaction go beyond covalent/noncovalent bonding to affect activity, the G246D mutant was assayed with both ABTS and tyrosine ethyl ester as substrates. The ABTS assay shows that the mutation decreases the ABTS turnover rate of the enzyme (3.6 ± 0.2 mol oxidized per min/mol of enzyme). A similar effect is also noted for tyrosine ethyl ester, with a net decrease in the ability to cross-link tyrosine residues at equal enzyme or heme concentrations (Fig. 4D and supplemental Fig. S3). This result is consistent with the phenotype observed for this mutation, in which a deficiency in tyrosine cross-linking results in blistering of the nematode cuticle. Heme concentration difference is not the sole factor contributing to the lower activity of the G246D mutant. Thus, this residue modification has effects other than simply decreasing heme binding.
A more pronounced deficiency is found for the CeDUOX11–589 D392N mutant, because it has no activity with either substrate. This was to be expected, because no heme co-purifies with this protein. To rationalize the lack of heme association with this protein, which has a mutation in a solvent-exposed residue distant from the heme, we turned to our CeDUOX11–589 model structure (Fig. 1B). Initial manipulation of the model demonstrated that Asp392 is located on the proximal side of the heme in a highly charged loop region below the helix supporting proximal His330. We speculated that this residue might interact with residues from the α-helix, causing a shift or increased mobility of the proximal histidine upon mutation that leads to an inability of the heme to bind properly. To test this hypothesis, we ran two independent computational models: coarse-grained anisotropic network models of dynamics and all-atom molecular dynamics simulations. Coarse-grained model results implicate correlated motion between the proximal histidine region and the conserved aspartic acid in our CeDUOX1 model (supplemental Fig. S4, Asp392) and in that of the known crystal structure of LPO (Protein Data Bank code 3BXI, Asp527; supplemental Fig. S5) (45). Based on this initial study, further validation was sought through all-atom molecular dynamics simulation, a technique that provides more detail than coarse-grained modeling (for example, including side chain and solvent atoms, and more sophisticated force fields). We generated a 40-ns all-atom molecular dynamics trajectory to examine the interactions between Asp392 and the residues in the proximal α-helix. Because Asp392 does not interact directly with these residues, we calculated mutual information to quantitate the correlation between the dynamics of the two regions (see “Experimental Procedures” for details; Fig. 5). Correlated motion between the two regions would imply that they influence each other despite the lack of direct interactions. There are two regions with highly coupled motion to the Asp392 loop: the helical residues under the heme, Ser331–Pro334, and the neighboring loop Phe80–Glu83 (Fig. 5A). Conversely, the helical region under the heme is highly coupled to these same regions: Gly82–Ser84 and Gln386–Ile394 (Fig. 5B). Collectively, these data demonstrate that residues not found in direct contact with the heme co-factor control important interactions crucial to its stability and orientation in the binding pocket, and in a greater sense that there is a complexity to enzymes that extends beyond sequence alignments and simple model predictions.
The catalytic proficiency of mammalian peroxidases depends on their heme coordination and environment (34, 40, 46–48). The C. elegans peroxidase domain has been shown to bind heme and display a structural fold similar to that of the mammalian peroxidases (11). However, the activity toward peroxidase substrates differs greatly between these enzymes, because the CeDUOX1 peroxidase domain displays a much lower catalytic turnover that may be a unique characteristic of at least some DUOX enzymes. This difference can be attributed to differences in the heme-binding pocket, although no crystal structure is available as yet to shed light on the exact nature of the heme environment. To further investigate this region, we prepared a series of mutant proteins, each of which was expressed and purified in the recombinant baculovirus system used for wild-type CeDUOX11–589.
Replacement of the distal tyrosine residue with either a histidine or phenylalanine did not increase the peroxidase activity when measured with the classic peroxidase substrate ABTS. Not surprisingly, substitution by a phenylalanine residue caused a modest decrease in overall activity. Although significant, the decrease was not as high as that observed upon the loss of a covalent bond to the heme. The phenylalanine placement may maintain steric interactions found with tyrosine, a structurally similar amino acid, while disrupting hydrogen-bonding interactions that facilitate efficient turnover. Mutation of the distal tyrosine to a histidine, the usual residue in the distal site of conventional peroxidases, increases tyrosine cross-linking activity. However, the level of turnover still does not approach that of a mammalian peroxidase. Clearly, although the active site now includes the major catalytic residues found in mammalian peroxidases (i.e. a distal and proximal histidine, a catalytic arginine, and two covalent bonds to the heme), they are not deployed in an optimal manner to promote enzymatic turnover. In fact, a low rate of peroxidase activity may be desirable in this organism, because it may enable specific, localized tyrosine cross-linking without giving rise to a high level of deleterious peroxidative stress. In this context, it is worth remembering that no regulatory factors have been identified that modulate the peroxidase activity.
Mutation of Glu238, predicted by sequence alignment with LPO to covalently bind to the heme co-factor, confirms that it functions in this role, as demonstrated by gel separation, heme staining, and HPLC analysis. Loss of the covalent heme bond in this mutant decreases its ABTS peroxidase activity but does not greatly modify its tyrosine ethyl ester cross-linking efficiency. The finding that some of the heme is still covalently bound to the protein despite the loss of the covalent bond to Glu238 is consistent with our proposal that a second heme-protein covalent bond is present with an as yet unidentified residue (15). A similar shift in covalent character was noted upon mutation of Glu375 in LPO, one of two covalent heme-binding residues verified by crystal structure(s). Mutation of Glu375 to an aspartic acid residue resulted in a decrease in covalent heme association, from 67% bound in the recombinant wild type to only 30% in the mutant protein (38). This demonstrates that disruption of a single covalent bond in a multi-covalent heme-bound protein can result in greater free heme content, as seen for the CeDUOX11–589 E238A mutant.
Mutations of Gly246 and Asp392 give rise in C. elegans to a blistering physiological response associated with insufficient tyrosine cross-links (13). Both of these residues are located in the peroxidase domain, but outside the catalytic heme-binding pocket (Fig. 1B). Recombinant expression of these two physiologically identified mutant proteins demonstrated that they bind decreased levels of the heme prosthetic group. Mutation of Gly246 to an aspartic acid modestly decreased heme binding and activity toward both ABTS and tyrosine. Sequence alignments indicate that in DUOX enzymes this position is normally occupied by a small, noncharged residue, specifically an alanine or glycine (Fig. 1A). Introduction of a larger charged residue may interfere with substrate binding as a result of both steric and electrostatic effects. This possibility is supported by the model structure, in which Gly246 lies along a hydrophobic α-helix within a hydrophobic region just above the catalytic arginine on the distal side of the heme. Introduction of a charged amino acid into this region may constrict the heme distal binding pocket (49). This is reflected in the loss of activity and the smaller shift in the Soret band upon the binding of cyanide. The electronic and steric effects of the mutation may also, as observed, weaken the binding of heme groups that are not covalently linked to the protein.
The mutation of Asp392 to asparagine has a particularly unexpected effect, because the change in this solvent-exposed residue distant from the heme-binding region yields a protein to which heme was not detectably bound. The high degree of conservation of this aspartic acid residue in all DUOX proteins strongly suggests that it plays a crucial role in heme binding and/or activity (Fig. 1A). A 40-ns molecular dynamics simulation of our CeDUOX11–589 structural model demonstrated a strong correlated motion between the proximal histidine region and the α-helix containing Asp392. A linking of the motions of these two otherwise unrelated regions of the protein provides a reasonable mechanism for Asp392 mutations to modulate the interaction of the proximal histidine ligand and the heme.
Our data support the current model of C. elegans DUOX1 as a self-contained catalyst that mediates tyrosine cross-linking, with H2O2 generated by the NADPH oxidase domain being utilized by the N-terminal peroxidase domain to form the tyrosine cross-links. This mechanism of action contrasts with that of the human DUOX1 system, which, as isolated, does not bind heme and requires several regulatory proteins to control enzymatic function (14, 50). Thus, two different mechanisms may exist for DUOX proteins, depending on the processes in which the DUOX participates in addition to generating superoxide and/or H2O2. Permanent binding of the heme in a peroxidase domain that has low tyrosine cross-linking activity may be compatible with the generation of large amounts of H2O2 by the NADPH oxidase domain but may not be compatible with protein integrity in a DUOX that also has high peroxidase activity or other roles. Heme binding in these latter proteins may be triggered by association with other proteins such as DUOXA, because this would allow closer regulatory control of the peroxidative activity of the system. Organisms that utilize DUOX primarily for di- and tri-tyrosine formation through a peroxidase domain with low catalytic activity, such as those in C. elegans, D. melanogaster, and A. gambiae, may express a DUOX in which the heme is permanently bound. Differentiation between these two types of DUOX proteins may be signaled by the presence of the proximal histidine residue, which is found, as in C. elegans, in heme-binding DUOX proteins.
We thank the Brian Shoichet lab for use of the JASCO J-715 spectropolarimeter.
2The abbreviations used are: