Article pubs.acs.org/biochemistry
Structural Analysis of Streptococcus pyogenes NADH Oxidase: Conformational Dynamics Involved in Formation of the C(4a)Peroxyflavin Intermediate Jamie R. Wallen,† T. Conn Mallett,‡,∥ Takashi Okuno,‡,⊥ Derek Parsonage,‡ Hiroaki Sakai,§ Tomitake Tsukihara,§ and Al Claiborne*,‡ †
Department of Chemistry and Physics, Western Carolina University, Cullowhee, North Carolina 28723, United States Department of Biochemistry, Wake Forest School of Medicine, Winston-Salem, North Carolina 27157, United States § Institute for Protein Research, Osaka University, Suita, Osaka 565-0871, Japan ‡
ABSTRACT: In probing the oxygen reactivity of an Enterococcus faecalis NADH oxidase (Nox; O2 → 2H2O) C42S mutant lacking the Cys42-sulfenic acid (Cys42-SOH) redox center, we provided direct evidence of a C(4a)-peroxyflavin intermediate in the oxidative half-reaction and also described a conformational or chemical change that is rate-limiting for full reoxidation of the homodimer. In this work, the Nox from Streptococcus pyogenes (SpyNox) has been expressed and crystallized, and the overoxidized wild-type [Cys44SOH → Cys44-sulfinic acid (Cys44-SO2H)] and C44S mutant enzyme structures have been refined at 2.0 and 2.15 Å, respectively. We show that azide binds to the two-electron reduced wild-type (EH2) enzyme and to the mutant enzyme in solution, but with a significantly higher affinity for the mutant protein. The spectral course of the titration with the SpyNox EH2 form clearly indicates progressive displacement of the Cys44-S− → FAD charge-transfer interaction. An azide soak with C44S Nox crystals led to the structure of the complex, as refined at 2.10 Å. The active-site N3− ligand is proximal to the Ser44 and His11 side chains, and a significant shift in the Ser44 side chain also appears. This provides an attractive explanation for the azideinduced loss of charge-transfer absorbance seen with the wild-type EH2 form and also permits accommodation of a C(4a)peroxyflavin structural model. The conformation of Ser44 and the associated helical element, and the resulting steric accommodation, appear to be linked to the conformational change described in the E. faecalis C42S Nox oxidative half-reaction. ioinformatics analyses have classified flavoprotein disulfide reductases into three subgroups:1 the disulfide reductases that include glutathione reductase, alkyl hydroperoxide reductases that include the low-Mr thioredoxin reductase, and peroxidase-oxidase-reductases that include NADH oxidase (Nox; O2 → 2H2O), NADH peroxidase (H2O2 → 2H2O), and coenzyme A-disulfide reductase [(CoAS-)2 → 2CoASH]. In spite of clear similarities with regard to the reactions that they catalyze,2 Nox is mechanistically distinct from both NADH peroxidase and coenzyme A-disulfide reductase, in that the central conserved active-site cysteine [Enterococcus faecalis Nox (Ef Nox) Cys42] reacts with a C(4a)-peroxyflavin intermediate to regenerate the Cys-SOH redox center in catalysis.3 Elimination of H2O from the resulting C(4a)hydroxyflavin has been proposed to restore the oxidized enzyme, and the Cys-SOH center is reduced with equivalents from NADH, via the flavin cofactor, in the catalytic cycle. In E. faecalis NADH peroxidase (Ef Npx), Cys42 is essential for activity.4 C42S Ef Nox, however, catalyzes NADH-dependent reduction of O2 → H2O2, with a kcat of 2.3 s−1 at 5 °C.3 The stopped-flow diode-array analysis of the enzyme in turnover provides clear evidence of a C42S Nox-FADH 2 ·NAD + intermediate. A combination of diode-array, single-wavelength, and fluorescence stopped-flow analyses for the reaction of O2
B
© 2015 American Chemical Society
with this complex gives (1) clear evidence of the formation of a C(4a)-peroxyflavin intermediate in one active site (A) of the homodimeric enzyme and (2) a rate-limiting (2 s−1 ) reoxidation of the second FADH2·NAD+, in the B active site of the dimer. Some undefined conformational or chemical change linked to reoxidation of the A subunit has been proposed to control O2 reactivity of the B subunit. The appearance of the C(4a)-peroxyflavin intermediate distinguishes Nox from all other peroxidase-oxidase-reductase enzymes and flavoprotein disulfide reductases but is common, in particular, to the class A and B flavoprotein monooxygenases,5−7 such as p-hydroxybenzoate hydroxylase (PHBH) and cyclohexanone monooxygenase, respectively. With the class A enzyme, the C(4a)-peroxyflavin is generally stabilized (kinetically) by the bound substrate or an effector analogue. Azide ion (N3−) has also been demonstrated to kinetically stabilize the peroxyflavin and C(4a)-hydroxyflavin intermediates of the hydroxylase.8,9 The class B monooxygenase, unlike the class A enzyme, is a two-dinucleotide-binding domain flavoprotein,1 like Nox. Also similar to Nox, the oxidized Received: June 17, 2015 Revised: October 16, 2015 Published: October 27, 2015 6815
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry NAD(P)+ product remains bound to the reduced enzyme; this dramatically stabilizes the nascent C(4a)-peroxyflavin intermediate of cyclohexanone monooxygenase,10 in the absence of substrate. Azide was earlier shown to act as a linear mixed-type inhibitor of wild-type Ef Nox in the standard assay.11 Studies of the oxidation of reduced, unliganded C42S Ef Nox (E-FADH2) in the presence of 0.1 M azide, however, led to the conclusion that azide does not kinetically stabilize a Nox C(4a)peroxyflavin intermediate.3 The Protein Data Bank, independent of this work, includes one Nox structure, as refined at 1.8 Å for the enzyme from Lactobacillus sanf ranciscensis (LsNox).12 Of mechanistic interest, the density for the oxidized Cys42 residue was modeled as the native Cys-SOH, with the side chain adopting two equally populated conformations. Less compelling is the conclusion that ADP remains bound throughout the catalytic cycle, redirecting the NADH substrate through an unprecedented mode of flavin access. This somewhat heterodox proposal has been examined previously, by us.13 Given that mechanistic work had focused on Ef Nox,3,14 we undertook efforts toward a structural study with this enzyme. All efforts to obtain diffraction-quality crystals were unsuccessful, but the enzyme from Streptococcus pyogenes (SpyNox) gave well-diffracting crystals in an initial screen of recombinant Nox enzymes from different sources. Here we describe the structures of the wild-type enzyme, with Cys44 modeled as an overoxidized (non-native) Cys-sulfinic acid (Cys-SO2H), and the C44S mutant. In combination with titration data for both twoelectron reduced wild-type (EH2) and C44S enzymes with azide and chloride, the structure of the C44S SpyNox·N3− complex, steady-state kinetic analyses, and modeling with the coordinates of a stable C(4a)-peroxyflavin analog, we present a detailed analysis of conformational dynamics in the oxidative half-reaction.
resuspended in 50 mM sodium phosphate buffer (pH 7.0) containing 2 mM EDTA and 2 mM DTT, and lysed using a French press (SLM/Aminco). All purification steps were conducted at 4 °C in the presence of either 2 mM DTT or 2-mercaptoethanol. After centrifugation, nucleic acids were removed by precipitation with 2.5% (w/v) streptomycin sulfate. The supernatant was loaded onto a 50 mL Q-Sepharose FF column equilibrated with 25 mM phosphate (pH 7.0) containing 0.1 M NaCl and 2 mM 2-mercaptoethanol. After washing, the SpyNox protein was eluted with a 0.1 to 1.0 M NaCl gradient. The fractions containing the yellow flavoprotein were pooled and brought to final concentrations of 0.4 M NaCl and 20 mM imidazole, and the pH was increased to 7.8−7.9 via titration with a 1.0 M K2HPO4 solution. The protein was loaded onto a 25 mL Ni-NTA agarose (Qiagen) column equilibrated with 50 mM sodium phosphate (pH 8.0) containing 20 mM imidazole, 0.5 M NaCl, and 2 mM 2mercaptoethanol. The column was washed until the A280 returned to 80% inhibition. A pattern of linear mixed-type inhibition gave Ki(N3−) and αKi(N3−) values of 5.7 and 26.5 mM for interactions with resting Ef Nox and with a second catalytic Nox·NADH complex, respectively. Azide binding perturbed the absorbance spectrum of active, oxidized Ef Nox, particularly over the range of 500−700 nm that represents Cys42-SOH → FAD charge transfer. Dithionite titration gave none of the EH2 spectral intermediate but still required 1.6−1.7 equiv of reductant/FAD and led to 20−30% neutral semiquinone formation at ∼1 equiv/FAD. We began this study by conducting spectral titrations with wild-type EH2 and C44S SpyNox forms with both N3− and Cl− and by measuring the effects of these monovalent anions (which may be compared with HSO3−) on both O2 → 2H2O and mutant O2 → H2O2 catalytic activities. The yield of C44S SpyNox is approximately 60−70 mg of pure enzyme from 3.3 L of Es. coli cultures, with a specific activity (O2 → H2O2) of 16 units/mg. This compares very favorably with the activity (16−20 units/mg) reported for C42S Ef Nox.3 Dithionite and NADH titrations with C44S SpyNox are complete with 1.0−1.1 equiv of reductant/FAD, as anticipated. As shown in Figure 2A, the starting EH2 spectrum compares very favorably with that in Figure 1A. Stepwise addition of NaN3−, with 2−3 min allowed for equilibration, leads directly to dramatic elimination of Cys44-S− → FAD charge-transfer absorbance. The increase in A446 and slight red-shift in λmax, with isosbestic points at ∼430 and ∼498 nm, confirm the monophasic process. The Kd(N3−) of 2.1 mM calculated from ΔA522 is consistent with the 80% inhibition of Ef Nox seen at 0.1 M and is within the range of the 5.7 mM Ki(N3−) value determined from the early steady-state inhibition data with Ef Nox. Most importantly, the working conclusion from this result is that the Cys44-S− → FAD interaction and N3− binding are mutually exclusive. A very similar result was obtained in a parallel titration of SpyNox EH2 with KCl, except that the spectral changes with Cl− are saturable with a Kd(Cl−) value 7− 8-fold higher (16.2 mM as determined from ΔA540) than that for N3−. C44S Nox lacks absorbance over the range of 500−700 nm, as both the weak oxidized Cys44-SOH → FAD and reduced EH2 charge-transfer bands require the redox-active Cys. Nonetheless, N3− titration gives a very clear shift in the absorbance spectrum (Figure 2B), as λmax is red-shifted from 438 → 446 nm with isosbestic points at 388 and 446 nm. Chloride induces the same qualitative spectral change. The Kd(N3−) value of 0.33 mM for the mutant reflects a 6−7-fold higher affinity for C44S SpyNox than for the wild-type EH2 form, as analyzed by ΔA482, and Kd(Cl−) is decreased from 16.2 mM (EH2) to 5.8 mM (C44S). Conversions to free energies of binding indicate that N3− binding is 1.2−1.7 kcal/mol more favorable than Cl− binding, and binding of either monovalent anion to C44S SpyNox is favored by 0.6−1.1 kcal/mol relative to that of EH2. The loss of the favorable Cys44-S− → FAD interaction is one significant contributing factor in the latter ΔΔG comparison. Figure 3A gives the spectral course for a representative enzyme-monitored turnover experiment with C44S SpyNox. NADH was the limiting substrate in all experiments. Oxidized
Figure 2. Azide titrations of wild-type and C44S SpyNox at pH 7.0. (A) The wild-type enzyme [54.9 μM, in 0.85 mL of 50 mM potassium phosphate (pH 7.0), containing 0.5 mM EDTA] was titrated with a 0.3 M solution of NaN3. Spectra shown, in order of decreasing absorbance at 540 nm, correspond to addition of 0 (black), 0.47 (red), 1.86 (green), 5.51 (orange), and 33.3 mM azide (blue). The inset shows the absorbance change at 522 nm vs the azide concentration. Kd(N3−) = 2.1 mM. (B) The C44S enzyme (60.3 μM) was titrated with NaN3. Spectra shown, in order of increasing absorbance at 482 nm, correspond to the addition of 0 (black), 0.12 (red), 0.35 (green), 0.92 (orange), and 6.4 mM azide (blue). The inset shows the absorbance change at 482 nm vs N3− concentration. Kd(N3−) = 0.33 mM.
enzyme is very rapidly reduced to the E-FADH2·NAD+ complex, as indicated by the decrease in A438 coupled with the appearance of a new charge-transfer band centered at 725 nm. By focusing initially on C44S SpyNox as a simpler redox system, we compared the effects of N3− and Cl− on the approach to steady state and on steady-state properties and initial velocity, using the enzyme-monitored turnover method. When we inspect the ΔA438 time course (Figure 3B) for the mutant in the absence of N3−, ∼45% of the enzyme is reduced in the dead time. The enzyme steady state in this experiment, which is largely reduced (E-FADH2·NAD+), persists to approximately 10−12 s, beyond which point NADH depletion promotes full reoxidation. In contrast, in the presence of 2.5 mM N3−, only ∼5% of the enzyme is reduced in the dead time, and the approach to steady state is quite slow, extending beyond 0.1 s. The enzyme steady state during turnover in the presence of 2.5 mM N3− is largely but not as reduced (EFADH2·NAD+), compared to that for enzyme without N3−. However, the return from steady state to oxidized enzyme occurs fastest with 2.5 mM N3−, in a concentration-dependent manner, relative to the absence of N3−. 6819
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry
Figure 4. Contrasting effects of azide on the activity of C44S and wildtype SpyNox. Enzyme turnover was measured as described in Figure 3B, and initial rates of NADH oxidation were measured at 340 nm. (A) C44S SpyNox (7.7 μM after mixing) turnover was measured in the presence of 0 (black; v0/e = 1.1 s−1), 0.2 (purple; v0/e = 1.5 s−1), 0.5 (blue; v0/e = 1.7 s−1), 1.25 (green; v0/e = 2.4 s−1), and 2.5 mM azide (red; v0/e = 3.1 s−1). (B) Wild-type enzyme (9.1 μM after mixing) turnover was measured in the presence of 0 (black; v0/e = 46 s−1), 0.5 (purple; v0/e = 44 s−1), 1 (blue; v0/e = 38 s−1), 2.5 (green; v0/e = 32 s−1), and 5 mM azide (red; v0/e = 26 s−1). The temperature was 5 °C, and final NADH concentrations after mixing were 0.15 and 0.18 mM in panels A and B, respectively. The final O2 concentration was 0.4 mM in both experiments.
Figure 3. Enzyme-monitored turnover of C44S SpyNox at 5 °C. (A) C44S SpyNox [7.9 μM after mixing, in 0.1 M potassium phosphate (pH 7.0), containing 0.5 mM EDTA] was mixed in the stopped-flow spectrophotometer at 5 °C with NADH buffer. Both enzyme and substrate solutions were equilibrated with ambient O2, on ice, prior to the experiment. Final NADH and O2 concentrations after mixing were 0.15 and 0.4 mM, respectively, and spectral data were acquired with the photodiode-array detector interfaced with the Applied Photophysics stopped-flow instrument. Spectra shown, in order of increasing absorbance at 438 nm, correspond to the reaction mixture at 4.431 (red), 12.53 (orange), 12.97 (green), 13.42 (lime), 13.90 (blue), and 14.89 s (purple), as well as the final oxidized enzyme (19.64 s; black). (B) In a parallel experiment, the C44S enzyme was mixed at 5 °C with 0.15 mM NADH in the presence of 0.4 mM O2 and different concentrations of azide (0, 0.2, 0.5, 1.25, and 2.5 mM), and the reactions were followed at 438 nm. End point A438 values at 20 s have been normalized to match the control (no azide) value; the maximal offset introduced is ∼10%.
with wild-type SpyNox (O2 → 2H2O). Initial rates [ΔA340 (Figure 4B)] in the absence of either anion were linear for ∼300 ms with the wild-type enzyme, allowing direct determination of v 0 /e as a function of N 3 − or Cl − concentration. At 5 mM N3−, the wild-type enzyme is inhibited by 43%; 50 mM Cl−, in contrast, gives only 32% inhibition in the parallel experiment. Possible mechanisms for monovalent anion inhibition of wild-type SpyNox in turnover include decreased rates of E-FADHOOH → E-FADHOH and/or EFADHOH → E-FAD conversion,3,8,9 processes that do not occur with C44S Nox. The basis for C44S Nox activation will be considered in the Discussion. Structure Solution and Quality of the Model. Given the structures available for Ef Npx and LsNox,12,21,31 BLASTP analysis demonstrated that both sequences were 37% identical to SpyNox. In addition, the LsNox structure had itself been determined by molecular replacement with the Ef Npx coordinates. Our approach therefore was to determine the SpyNox structure by molecular replacement, using Ef Npx (PDB entry 1JOA) as the search model. Well-diffracting crystals of native His-tagged SpyNox were grown, and a complete data set was collected to 2.0 Å. It is important to note that crystal preparation, as optimized, occurred over 4−9 days at pH 5.6, under an ambient atmosphere at 20 °C. Molecular
The linear initial rates of NADH oxidation, measured simultaneously at 340 nm (Figure 4A), support this observation. In addition, estimates for Km(NADH) from these experiments with C44S SpyNox, in the absence of azide, give values of ∼1 μM. This is consistent with the essentially linear ΔA340 traces, maintained well beyond the standard initial velocity condition ([NADH]t ≥ 0.95[NADH]0). Direct determination of enzyme turnover gives a v0/e of 1.1 s−1 at 5 °C. Consistent with the observations in Figure 3B, v0/e increases from 1.1 to 3.1 s−1 in a concentrationdependent manner, as the N3− concentration is increased to 2.5 mM. The parallel analysis over the range of 0−25 mM Cl− increased v0/e from 1.4 to 5.5 s−1. Both monovalent anions are activators with C44S SpyNox (O2 → H2O2). Direct observation of wild-type SpyNox in turnover reveals a qualitatively similar effect of N3− on the approach to steady state, but the return to oxidized enzyme is now slowest at 5 mM N3−. As expected, and consistent with earlier published reports for wild-type Ef Nox,11 both N3− and Cl− are inhibitors 6820
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry
Figure 5. Structural analysis of the redox state of SpyNox Cys44. (A) Final 2Fo − Fc map, in stereo, for Cys44-SO2H, His11, and other active-site residues of chain A, together with the refined model. Density is not shown for FAD, a portion of which has been omitted for the sake of clarity. Carbon atoms are color-coded as follows: protein residues, sand; FAD, yellow. All other color-coding is by atom type. The depicted contour level is 1.5σ. Also shown is the positive difference electron density (green, contoured at 4σ) obtained from OMIT Fo − Fc maps, superimposed on the Cys44-SO2H residue. Hydrogen bond interactions are indicated by black dashes. (B) Active-site overlay, in stereo, for SpyNox, Ef Npx, and LsNox. This superposition was performed with each respective chain A. Color-coding for SpyNox is as in panel A, with residues labeled in black. Colorcoding for the Cys-SOH residues of Ef Npx and LsNox (only one Cys-SOH conformer shown) is also by atom type. All other atoms and labels for Ef Npx are colored cyan, and those for LsNox are colored slate blue.
replacement using AMoRe (CCP4 suite)22,23 and the native SpyNox data set, with chain A of 1JOA as a search model, led to successful phasing. The final model, refined at 2.0 Å resolution, has an Rcryst of 19.5% (Rfree = 23.4%) with good geometry, as analyzed with WHATCHECK (Table 1).32 Ramachandran analysis of the structure shows that 96−97% of all residues fall within the core regions, as defined by Kleywegt and Jones.33 The crystals have two SpyNox molecules in the asymmetric unit, and the final model includes residues Ser2−Asp456 for both chains A and B of the biological dimer (with residues from chain B being designated by a prime symbol), two FAD cofactors, two Cys-SO2H, and 586 ordered solvent molecules. The 2Fo − Fc map in the active-site region [chain A (Figure 5A)] illustrates the quality of electron density for this important, well-ordered substructure, and an OMIT Fo − Fc map documents the density peaks for the two oxygen atoms of Cys44-SO2H. The Met1Gly replacement introduced by design during cloning was also confirmed in the final model, and electron density is also present for several residues from the His tag, including Trp0. For chain B, there is no density for the Tyr199′ side chain, and only weak density was observed for a flexible surface loop (Ile125′−Thr135′). Overall Structure. Earlier analytical ultracentrifugation and gel filtration analyses with Ef Nox demonstrated that the homodimeric quaternary structure is predominant, particularly at enzyme concentrations of >15 μM.16 The two SpyNox molecules per asymmetric unit are related by an approximate noncrystallographic 2-fold symmetry axis, and there are slight differences when the two monomers are superimposed (Cα rmsd = 0.57 Å). Formation of this tightly associated dimer decreases the accessible surface area by 3650 Å2 per monomer, or ∼17% of each monomer surface. We conclude that this represents the biologically relevant dimer. SpyNox is organized
into three compact domains, based on structural alignment with Ef Npx.31 These are FAD-binding [residues 1−112 (FAD-1 segment) and residues 258−334 (FAD-2 segment)], NADHbinding (residues 113−257), and interface (residues 334−456) domains; this fold is common to both peroxidase-oxidasereductase and disulfide reductase subgroups of flavoprotein disulfide reductases.1,2 As such, the refined structure is very similar to those of LsNox (PDB entry 2CDU),12 Ef Npx (PDB entries 1JOA and 1NPX),21,31 and, in particular, the reduced forms of Bacillus anthracis coenzyme A-disulfide reductase (BaCoADR; PDB entries 3CGE and 3CGD).13 A search using the DALI server,34 with wild-type SpyNox (PDB entry 2BC0; chain A) as a query, ranks the three top scores (distinct proteins) as LsNox (Z = 48.0; rmsd = 1.5 Å; 36% sequence identity), the reduced BaCoADR·NAD(P)H complex (Z = 47.2; rmsd = 1.6 Å; 27% sequence identity), and Ef Npx (Z = 46.8; rmsd = 1.6 Å; 37% sequence identity). Because the LsNox structure has been described in detail, here we highlight the features of SpyNox that are key to the catalytic mechanism. The first area of focus is the Cys redox center (Figure 5A). In both monomers of the final wild-type SpyNox model, all atoms of Cyo44, including both oxygen atoms of the sulfinic acid, are refined at an occupancy of 1.0. At a resolution of 2.0 Å, we would anticipate negative difference electron density for one or both Cys-SOH conformations. Model bias can also be excluded, because positive difference density matches the positions of the two sulfinate oxygen atoms. With LsNox,12 the electron density for the redox-active Cys42 residue was interpreted as a Cys-SOH with two conformations, each with an occupancy of 0.5. Refinement models, including Cys42-SH, Cys42-SO2H, and Cys42-SO3H, were analyzed, but all were reported to produce positive difference density. Preservation of the Cys42-SOH redox state in the final refined LsNox model is 6821
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry still somewhat surprising, given that 20 mM DTT was present during crystal growth. The original 2.16 Å structure of Ef Npx included the hyperoxidized Cys42-SO3H in the final model, and mass spectrometric and 13C NMR analyses of the H2O2inactivated enzyme35 gave results consistent with Cys42-SO2H formation. Ef Nox, in an early study, was demonstrated to be inactivated upon being treated with increasing concentrations of H2O2.16 We therefore interpret the electron density for the redox-active Cys44 residue in SpyNox as Cys-SO2H. The final Cys44-SO2H model for SpyNox preserves the His11 and FADO2′ interactions with the two sulfinate oxygens. In both chains A and B, Cyo-OD1 and -OD2 are 2.6−2.7 and 2.6−2.8 Å, respectively, from the corresponding His11-NE2 and FAD-O2′ partners. The active-site structures for SpyNox(Cys44-SO2H), native Ef Npx(Cys42-SOH), and LsNox are compared in Figure 5B. The His10···Arg303 hydrogen bond identified in the Ef Npx structure is thought to maintain His10 in the unprotonated state throughout the catalytic cycle.4,31 Despite the loss of the Arg303 interaction, the crystal structure of the R303M Npx mutant indicated almost no change in His10 conformation.17 The interactions of the conserved His (LsNox His10 and SpyNox His11) with elements of the respective Cys42-SOH and Cys44-SO2H centers have been described. More notable is the substitution of Val for Ef Npx Arg303 in both Nox enzymes (Val304 and Val314, respectively). Figure 6 shows that all five Nox enzymes represented have either Val or Leu at this position, within the α7 helix. van der Waals volumes for Val, Leu, and Met are 105, 124, and 124 Å3, respectively,36 so the Ef Npx R303M mutant does mimic the Nox consensus for a hydrophobic residue at this position. Additional electron density for LsNox was interpreted as tightly bound ADP, which was reported to copurify with the enzyme, occupying the NAD(P)H-binding site.12 The only ADP structure present in SpyNox is that component of the tightly bound FAD cofactor, and the NADH-binding site is unoccupied except for ordered solvent molecules. LsNox is unusual in its kinetic properties, as was reported, with similar kcat/Km values for NADH and NADPH of 2.7 and 0.9 × 107 M−1 s−1, respectively (presumed to be at 25 °C). Earlier work with Ef Nox37 and with the enzymes from S. mutans38 and Lactococcus lactis12 demonstrated exclusive NADH specificity. Our earlier comparison of the LsNox·ADP and reduced BaCoADR·NADPH structures13 led to a possible mechanism for LsNox dual NAD(P)H specificity, implicating His181, Lys187, and Lys213. The alignment of known functional Nox enzymes given in Figure 6 identifies LsNox His181 and Lys187 within a segment immediately following the βαβ supersecondary structural element of the NAD(P)H-binding domain. The conserved GxGxxG motif and Asp179 here are expected, but for strict NADH specificity.39 The three basic LsNox side chains are either unique to LsNox (His181 and Lys213) or conserved in only the NADH-specific Ef Nox (Lys187). Without offering any support for the LsNox kinetic analysis, the His-Lys-Lys triad is unique to that enzyme and provides at least a viable mechanism for dual NAD(P)H specificity. The well-documented dual NAD(P)H specificity of BaCoADR involves the Glu180−Thr187 loop, structurally equivalent to LsNox Asp179−Lys187, which is poorly ordered in the oxidized BaCoADR structure. The equivalent segment in SpyNox (Asp190−Gly198) is well-ordered, in contrast, consistent with its strict NADH specificity. In the oxidized
Figure 6. Sequence alignment for SpyNox and known functional Nox enzymes from S. pneumoniae (SpnNox), S. mutans (SmuNox), Ef Nox, and LsNox. The residue numbering corresponds to that of native wildtype SpyNox. Red and yellow blocks are conserved residues and conservative substitutions, respectively. Secondary structure assignments [color-coded by type; η (green), 310 helix] correspond to SpyNox (top, PDB entry 2BC0) and LsNox (bottom, PDB entry 2CDU). The interdomain junction connecting FAD-1 and NADHbinding domains is indicated with a black box (SpyNox residues 108− 115); the LsNox residues implicated in the NAD(P)H dual-specificity mechanism (His181, Lys187, and Lys213) and their equivalents are indicated in blue boxes (SpyNox residues 192, 198, and 224), and SpyNox Val314 and its equivalents are indicated in a green box.
LsNox−ADP complex, however, the Lys187 side chain, which is implicated in the dual NAD(P)H specificity mechanism, is disordered in both subunits, and Arg183′ also exhibits alternate conformations. High B factors approaching 50 Å2 are also observed within the LsNox NAD(P)H-binding motif. The reduced BaCoADR·NADH structure confirms the conformational change required for this loop and the novel mode of NADPH binding. Also as shown in Figure 6, the segment corresponding to the flexible surface loop in SpyNox (Ile125′− Thr135′), for which only weak density is observed for chain B, 6822
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry
Figure 7. Active-site structures of C44S SpyNox and the azide complex. (A) Active-site overlay, in stereo, for C44S SpyNox chains A and B. Carbon atoms for chain A of the C44S mutant are colored pale green; FAD carbon atoms are colored yellow (a portion of the FAD has been omitted for the sake of clarity), and all other color-coding is by atom type. Carbon atoms for chain B are colored slate blue, as is the FAD. (B) Active-site overlay, in stereo, for SpyNox(Cys44-SO2H) and the C44Sin conformer. Color-coding for wild-type SpyNox is as in Figure 5. Carbon atoms for the C44Sin conformer are colored pale green, as is the FAD. Hydrogen bond interactions with the conserved active-site WAT are given as black dashes. A new WAT, also colored red, is recruited into the active site of the mutant, where it interacts via hydrogen bonds with Ser44 and His11. (C) Active-site overlay, in stereo, for the C44Sin conformer and the azide complex. Color-coding for the mutant is as in panel A. Carbon atoms for the azide complex are colored magenta, as is FAD. Bound azide is colored orange, and its hydrogen-bonding interactions are given as black dashes. The α2 helices of the C44Sin conformer and the azide complex are represented as pale green and magenta transparencies. In particular, the contraction of this helix in the azide complex, and the concomitant shift for Ser44, is seen to accommodate the bound azide.
Ef Npx,42 glutathione reductase,43 and other disulfide reductase and peroxidase-oxidase-reductase enzymes. However, we have noted that the putative bound ADP in the LsNox structure may in fact be NAD+, because partial ordering of bound NAD(P)/H is very common among these structures. In unliganded human glutathione reductase, the closest approach between Tyr197OH and FAD is with FAD-N10, at 3.2 Å. In the clearly unliganded SpyNox structure, Tyr170 adopts the same conformation as glutathione reductase Tyr197, giving a Tyr170-OH−FAD-N10 distance of 3.1 Å. As noted by Stehle et al.31 in the original Ef Npx structure report, the conserved active-site Tyr159-OH interacts with FAD-O4α (3.4 Å). The
is present in the three streptococcal Nox enzymes but is absent in both Ef Nox and LsNox. The Ile108−Pro115 segment preceding this is conserved among the disulfide reductase and peroxidase-oxidase-reductase subgroup enzymes;1,40 in SpyNox, this spans the junction between strands β7 and β8 that connect the FAD-1- and NADH-binding domains. In Es. coli glutathione reductase, this conserved element was also noted to contribute to β-sheet structures in both domains.41 The SpyNox flexible surface loop closely follows this segment, near the beginning of the NADH domain. In the LsNox active-site structure description, with bound ADP, Tyr159 was noted to assume the “out” conformation commonly associated with the NAD(P)/H complexes of 6823
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry properties of SpyNox mutants lacking His11 and Tyr170 will be addressed in a separate communication. Structural Analysis of the C44S SpyNox·Azide Complex. C44S SpyNox crystallized in the same space group as the wild-type enzyme, and the structure was refined using the overoxidized wild-type (Cys44-SO2H) structure as the starting model. As summarized in Table 1, the final model has an Rcryst of 19.8% (Rfree = 24.1%) as refined at 2.15 Å resolution; chain A displays the lower B factors for both protein and FAD. As shown in Figure 7A, Ser44 is modeled in two equal conformations in chain A, resembling the two equal conformations observed for Cys42-SOH in the LsNox structure. The two Ser44 conformers result from an approximately 90° rotation about the Ser44 CA−CB bond. For the internal (“in”) conformer, Ser44 is angled in toward the flavin on the si face, with a Ser44-OG−FAD-N5 distance of 3.1 Å. Comparison of SpyNox(Cys44-SO2H) and C44S active-site structures (Figure 7B; chain A in each case) indicates that Ser44-OGin occupies essentially the same position taken by Cyo44-SG in the wild-type structure. A new ordered solvent molecule is recruited to compensate for the loss of the sulfinate oxygens. This water forms a network of active-site hydrogenbonding interactions with Ser44-OGin (2.2 Å), Ser44-N (2.9 Å), and His11-NE2 (3.2 Å). The latter interaction mimics that between wild-type Cyo44-OD1 and His11. Importantly, in consideration of Ser44 as an isosteric replacement for Cys44SH in the reduced wild-type enzyme, the internal Ser44 conformer is expected to represent a productive charge-transfer orientation, relative to FAD-C(4a). While Ser44-OGout retains the hydrogen bond with the ordered solvent molecule, it is angled away from the flavin, it does not interact with FAD-N5, and it is not expected to support a charge-transfer orientation for Cys44-SH. To an extent, it resembles the external (“out”) conformation observed for Cys42-SH in the structure of the reduced BaCoADR·NADH complex.13 In SpyNox chain B, Ser44 adopts a single well-ordered conformation similar to that for Ser44in in chain A. Our remaining discussions will focus on the internal Ser44in conformer. A second ordered solvent, similarly recruited into the mutant active site, is positioned just above the isoalloxazine system and is hydrogen-bonded to Ser44-O. In C42S Ef Npx, a wellordered solvent molecule (WAT453) is recruited to bridge Ser42-N and -OG (2.6 and 2.7 Å, respectively) with His10-NE2 (2.9 Å).4 Replacement of the oxidized Cys with Ser in Ef Npx led to the formation of a cavity between Ser42 and the isoalloxazine but had no impact on the isoalloxazine conformation or position. We also reported that the flavin redox potential for C42S Ef Npx, −219 mV, is ∼100 mV higher than that for the Ef Npx EH2 form. The Cys42-S− → FAD charge-transfer interaction (and its absence in the mutant) was considered a major factor in this difference. For the structural analysis of the C44S SpyNox·azide complex, crystals were prepared in a three-step soaking protocol as described in Experimental Procedures. During refinement of the resulting structure at 2.10 Å, additional density in the vicinity of the isoalloxazine si face was identified and modeled as a single bound azide (N3−). Figure 7C gives an active-site view for the C44S SpyNox·N3− model as fully refined (Rcryst = 19.1%; Rfree = 23.6%). Ser44 is well-ordered in both subunits of the dimeric complex, and there is one azide in each active site. Comparison of the active sites for the complex and unliganded C44S enzyme (Figure 7C) reveals the displacement of one ordered solvent by azide, a different conformation for
the His11 side chain, a contraction of the helical element (Gly43−Ile50) that includes Ser44 and helix α2, and a change in Ser44 side chain conformation. Azide lies within the cavity created between the C44S mutation and the si face of the isoalloxazine, with the N1−N2−N3 axis roughly parallel to the FADN10−N5 axis. Azide-N3 is hydrogen bonded to both Ser44-N (2.8 Å) and -OG (2.7 Å) and lies 3.7 Å above FADN5. Azide-N1 is 3.6 Å from His11-NE2, in its altered position, and forms a hydrogen bond with FAD-O2′, at 2.7 Å. In the azide complex, all elements of the Ser44 side chain are displaced by >2 Å from their positions in unliganded C44S. This is primarily due to the contraction of the active-site helical element and has the net effect of shifting the side chain away from the si face of the flavin. Modeling the C44S SpyNox C(4a)-Peroxyflavin Intermediate. In earlier studies, the crystallographic coordinates for a C(4a)-N(5)-epoxyethanolumiflavin derivative were used to provide a structural model for the PHBH-FADHOOH intermediate.44 In subsequent work, Schreuder et al.45 surveyed all possible positions of the distal oxygen of the hydroperoxide, by rotating this oxygen about the single bond between FADC(4a) and the proximal oxygen. The position of the distal oxygen, following this rotation, was found to be almost ideal for nucleophilic attack from C3 of the bound PHB substrate. This description differentiates the two analytical stages of (1) providing a structural model for the enzyme-bound hydroperoxide and (2) evaluating how the protein controls peroxyflavin chemistry. In earlier studies reported by our laboratory, 11 we took the structure of the E. faecalis Npx(Cys42-SO3H)-FAD·NADH complex as a model for the Ef Nox-FADH2·NAD+ complex. The reduced C44S SpyNoxFADH2·NAD+ complex is clearly implicated in turnover (Figure 3). With NADH bound at the flavin re face in the Ef Npx complex, it was necessary to use the coordinates for the complementary enantiomer of the peroxyflavin model. As such, upon introduction into the Ef Npx active site, this model of the Ef Nox-FADHOOH·NAD+ complex supported si face reactivity of the distal oxygen with a reduced active-site Cys42-SH. However, contact violations were noted for the proximal oxygen in this model with elements of the Cys42 side chain. Because these issues could not be resolved by simple rotation about the FAD-C(4a)−O bond, we concluded that some change in the active-site orientation of Cys42 in the reactive Ef Nox-FADHOOH·NAD+ intermediate, relative to this model, might alleviate these violations. On the basis of this earlier work, we had the appropriate FADHOOH model. Experimentally, we lacked a Nox structure that, when used in combination with this FADHOOH structure, would provide an acceptable model for the reactive Nox-FADHOOH·NAD+ intermediate. One clear explanation for the elimination of Cys44-S− → FAD charge transfer upon binding of azide to wild-type SpyNox EH2 (Figure 2A) could be a conformational shift for the thiolate charge-transfer donor. “Out” conformations correlated with the absence of charge transfer have similarly been described for the active-site Cys42 and Cys47 residues in the reduced BaCoADR·NADH complex13 and the fluorescent EH2 species I observed with Es. coli lipoamide dehydrogenase, respectively.46 With the two active-site conformations described in the structures of C44S SpyNox with and without (Ser44in) azide, and taking the Ser44 side chain as an uncharged isosteric replacement for Cys44thiolate, we first modeled the peroxyflavin analogue, as described above for the Ef Npx(Cys42-SO3H)-FAD·NADH 6824
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry
Figure 8. Conformational shift induced upon binding of azide to C44S SpyNox impacts the C(4a)-peroxyflavin model. Active-site view, in stereo, for the C(4a)-peroxyflavin model with C44S SpyNox. The model was generated by omitting azide from the refined structure for the C44S·N3− complex (PDB entry 2BCP) and manually superimposing the peroxyflavin analogue as described in the text. The same operation using the structure of the azide-free C44S mutant led to contact violations between Ser44 and the proximal oxygen of the peroxyflavin analogue. Carbon atoms for the C44S mutant are colored magenta, with those for FAD colored yellow. The α2 and α5 helical elements are also colored magenta. Carbon atoms for the C(4a)-peroxyflavin model are colored white. A rotation of the peroxyflavin C(4a)−O bond provides an optimal structural view of the reaction between the terminal peroxide oxygen and Cys44-SH, where Ser44 is taken as an isosteric model for the reduced Cys44 in the native SpyNox.
model, we cannot be certain whether this network is present in the Nox hydroperoxide. With respect to the peroxyflavin model, there are no protein interactions with either FAD-N5 or FAD-O4α. Tyr170-OH is in the vicinity of FAD-N5, but its distance (3.9 Å) and geometry do not support any significant interaction. The position of FAD-O4α is shifted in the peroxyflavin model, consistent with the introduction of tetrahedral geometry at FAD-C4a. The oxygen atom does approach Pro438 and Phe436, and Ser44-OG is now 3.0 Å away; however, no polar interactions with any of these residues exist, and the geometry required for a hydrogen bond with Ser44 is not observed.
complex, with the unliganded structure (Ser44in; PDB entry 2BC1). The analogue was superimposed directly with the experimentally determined C44S Nox isoalloxazine, based on all non-hydrogen atoms of the planar dimethylbenzene moiety. As with the early Ef Npx-based model, contact violations exist between the proximal peroxide oxygen and Ser44-CB and -OG. When the azide complex (PDB entry 2BCP) is used, the only positional overlaps involve the two peroxide oxygens and bound azide. Via simple omission of azide from the structure of the complex, a satisfactory fit of the C(4a)-peroxyflavin model is observed (Figure 8). This result is attributed to the contraction of the helical element containing Ser44, away from the isoalloxazine si face, eliminating the unfavorable contacts seen with unliganded C44S SpyNox. Binding of azide to the wild-type EH2 form in the same location, relative to Cys44-SH, His11, and the si face of the flavin, would be expected to dissipate the charge-transfer interaction via the movement of Cys44-SH, commensurate with contraction of the helical element. The extensive description of the model of the Nox hydroperoxide, as the plausible reaction intermediate, requires a survey of the possible positions of the distal oxygen of the peroxyflavin.45 When Ser44-CB in the C44S Nox·azide complex (and in the Nox hydroperoxide model) is taken as the positional equivalent of Cys44-SG in the reduced, wild-type enzyme hydroperoxide [Nox(FADHOOH, Cys44-SH)], the distal oxygen atom of the Nox peroxyflavin model (Figure 8) is pointed away from Ser44-CB, and at a distance of 5.7 Å. This orientation and distance do not support chemistry between the Nox hydroperoxide and Cys44-SH, which is why the further survey of distal oxygen positions will be required before an extensive description of peroxyflavin chemistry can be given. With the model presented here, we have analyzed the potential for interactions of protein with FAD-N5 and FADO4α. As shown in Figures 5A and 7, an active-site water is networked through hydrogen-bonding interactions with Glu174 and with FAD-N5 and FAD-O4α, in both wild-type and C44S SpyNox structures. This water, and the hydrogenbonding network, are preserved in the C44S Nox·azide complex. Given the change in bond order for FAD-N5, the tetrahedral geometry centered on FAD-C4a, and the concomitant shift observed for FAD-O4α in the peroxyflavin
■
DISCUSSION Distinctions within the Peroxidase-Oxidase-Reductase Subgroup Provide Keys to Microbial Physiology. The peroxidase-oxidase-reductase subgroup is represented by the structures of Ef Npx,21,31 LsNox,12 and SpyNox (this work), and Staphylococcus aureus and B. anthracis coenzyme A-disulfide reductases.13,47 The major distinction relative to all disulfide reductase and alkyl hydroperoxide reductase subgroup enzymes is the presence of a single active-site Cys in each peroxidaseoxidase-reductase enzyme. In Npx/Nox, this Cys is oxidized to the Cys-sulfenic acid (Cys-SOH) redox center, but in CoADR, the Cys-SSCoA mixed disulfide is present. Two structural variants of CoADR, both of which include a C-terminal rhodanese homology domain, have been identified; the structures and proposed mechanisms for two of these CoADR-RHD proteins have been reported.48,49 Pfam analysis of the Npx, Nox, and CoADR sequences leads to three domain annotations. The Pfam Pyr_redox_2 (PF07992, SpyNox residues 3−126 and 262−298) segment corresponds to the FAD-binding domain. Pfam Pyr_redox (PF00070, SpyNox residues 162−238) is the NAD(P)H-binding domain, and Pyr_redox_dim (PF02852, SpyNox residues 346−441) is the interface domain. The absence of an interface domain in alkyl hydroperoxide reductase subgroup proteins distinguishes these enzymes. The peroxidase-oxidase-reductase, disulfide reductase, and alkyl hydroperoxide reductase subgroups of the twodinucleotide-binding domain flavoproteins superfamily, together with the thioredoxin fold proteins,50 account for a major portion of the protein thiolomes in pathogenic bacteria. 6825
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry
classified with the flavoprotein disulfide reductases but exhibits oxygen reactivity with similarities to the flavoprotein monooxygenases like PHBH. In the absence of PHB, no oxygenated flavin intermediates are observed in stopped-flow analyses of the oxidative half-reaction for PHBH, and H2O2 is the product. When reduced PHBH (wild type or Asn300Asp mutant) is mixed with O2 and 0.1 M (final) azide, 60% of the enzyme is observed transiently as the C(4a)-peroxyflavin intermediate.8,9 Binding of azide to this site(s), which selectively inhibits the peroxyflavin decay rate, is rapid; there is no difference regardless of whether azide is included with the enzyme or introduced in the O2 syringe. The azide site with reduced PHBH is considered rapidly accessible to solvent, likely at the protein surface. From steady-state analyses, azide exhibits mixed-type inhibition with respect to NADPH.51 While there are no published structures for a PHBH·azide complex, Gatti et al.52 published two structures for ternary PHBH complexes with bromide ion, which is a competitive inhibitor with respect to NADPH.53 Br− occupancy at either of two distinct sites, respectively, was proposed to alter the dynamic steady state between the two isoalloxazine conformations that correspond to coupled (hydroxylation) and uncoupled (H2O2 formation) PHBH catalysis. The resolution for the C44S SpyNox·N3− complex (2.0 Å) is very similar to that for the two PHBH·Br− structures (2.0 and 2.3 Å). Several important distinctions are notable. (1) N3− binds at the si face, not the re face. At the same time, the respective active-site models with the stable C(4a)-peroxyflavin analogue use opposite enantiomers. As a result, the C(4a)hydroperoxide of the model projects into space above the si face with SpyNox but above the re face with PHBH, giving a distance of 2.9 Å between the terminal oxygen of the hydroperoxide and the 3-OH of the bound 3,4-dihydroxybenzoate product, in the latter case. (2) There is only one N3− site in C44S SpyNox (resting, oxidized enzyme), not two as with Br− and PHBH. (3) The isoalloxazine conformation in C44S SpyNox is essentially identical in the absence and presence of N3−, but there is a protein conformational shift relating to the active-site helical element that includes α2 and Ser44. The side chain conformations for Ser44 and His11 are altered, as well. The isoalloxazine conformations for the PHBH complexes with PHB and 2,4-dihydroxybenzoate are dramatically different in the presence of Br−, and Br− binding is proposed to perturb the conformational dynamics of the flavin in and out states in PHBH. (4) N3− interacts directly with Ser44-OG, Ser44-N, and FAD-O2′ in C44S SpyNox. In the re face Br1 position of PHBH, Br− interacts with the backbone amides of Gly297 and Lys298, and with a water molecule. As reviewed by Ghisla and Massey in 1989,54 for all flavoprotein monooxygenases (e.g., PHBH), an initial oneelectron transfer from FADH2 to oxygen leads to the paramagnetic complex of superoxide (O2•−) and flavin semiquinone. Spin inversion gives, first, the biradical complex and, second, the peroxyflavin. Azide (N3−) is an inhibitor of Es. coli Fe(III) superoxide dismutase.55 It is a commonly used substrate (superoxide) analogue for superoxide dismutase. It possesses the same charge as superoxide and similar frontier orbitals.56 It is believed to bind Fe(III) dismutases in the same way as superoxide. While azide stabilizes the PHBH peroxyflavin in reactions of the free reduced enzyme with O2,8,9 it does not stabilize the Nox peroxyflavin in the same context. Stopped-flow measurements at 390 and 438 nm3 demonstrate that the second-order rate constant for the O2
Nox as a Peroxidase-Oxidase-Reductase Subgroup Enzyme. On a mechanistic level, the precise structural basis separating Nox and Npx function remains to be elucidated. The one major active-site side chain substitution is SpyNox Val314, which replaces Ef Npx Arg303. In Ef Npx, the interaction of Arg303 with His10 has been considered to maintain the latter active-site side chain in its unprotonated, neutral state during the catalytic cycle.4,17,29 In addition, Stehle et al. first described a “domain zipper” of alternating electrostatic interactions between elements of the FAD-1 and -2 segments of the Npx FAD-binding domain.31 The network proceeds from Arg303 (active site) to Glu14 to Arg307 to WAT515 to Glu18 (surface). In addition to the Arg303 → Val314 substitution in SpyNox, we observe the pattern of replacements given below, for those four charged side chains in Npx:
The domain zipper of salt bridges and other polar interactions that connects helices α1 and α7 within the Ef Npx FAD-binding domain is systematically eliminated in each of the functional Nox sequences from Figure 6. The presence of hydrophilic/polar side chains in place of Npx Glu18 reflects the location of these residues on the protein surface. The active-site His (Ef Npx His10 or SpyNox His11) is conserved at the N-terminus of the α1 helix in all six proteins. The respective helix dipoles are expected to influence the His side chain pKa, and the presence or absence of the domain zipper could modulate the role of this His toward different catalytic ends in Nox versus Npx. Given the similarities in physicochemical properties for Val (van der Waals volume of 105 Å3, hydrophobicity of −1.5 kcal/ mol) and Met (van der Waals volume of 124 Å3, hydrophobicity of −1.3 kcal/mol),36 our earlier study of the Ef Npx R303M mutant is instructive.17 Substitution of Arg303 was expected to eliminate the side chain hydrogen bonding with and electrostatic effect on His10 while also disrupting part of the domain zipper. The crystal structure of the mutant confirmed the absence of any large structural perturbations and also showed that WAT131 now links His10 and Glu14, adopting one role of Arg303. The rest of the domain zipper is unaffected by the Arg substitution. The R303M mutant mimics Nox in two ways. (1) In direct contrast to wild-type Npx, and more like Nox, the mutant EH2 form can be directly reduced to a Nox-like E(FADH2, Cys42-SH)·NAD+ species, and (2) linked to this, the flavin redox potential of the mutant EH2 form is 51 mV more positive than in wild-type Npx EH2. However, NADH oxidase activity in the absence of H2O2 was unchanged relative to that of wild-type Npx. There is no evidence to support the premise that Npx can form the C(4a)peroxyflavin intermediate essential to Nox catalysis. Azide as a Probe of C(4a)-Peroxyflavin Mechanisms: Flavin versus Protein Conformational Dynamics. This work with SpyNox is focused on a “bifunctional” enzyme that is 6826
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry
may reflect some hysteretic transition59 during reduction to the steady state, as induced by azide, remains to be seen.
reaction in the presence of azide is decreased 20-fold, relative to that of reoxidation in the absence of azide; this corresponds to an unfavorable ΔΔG⧧ value of 1.7 kcal/mol. The possibility that azide, as a structural and electronic mimic of superoxide, could interfere with the C42S Ef Nox O2 reaction leads to the prediction that neutral semiquinone would be detected. This would require favorable rates of semiquinone formation and decay, as measured at 565 nm, the optimal wavelength for neutral semiquinone detection. Anaerobic dithionite titration of wild-type SpyNox, in the presence of DTT (Figure 1A), does yield approximately 50% neutral semiquinone intermediate, and similar behavior is seen with the C44S mutant. Azide as a Nonessential Activator of C44S SpyNox. Our original steady-state analysis of azide inhibition with wildtype Ef Nox led to a linear mixed-type inhibition scheme, in which only the reductive half-reaction was considered.11 From the pre-steady-state analyses of the azide effect with C44S SpyNox presented here, it is clear that the rate of reduction to the steady state is much slower with azide present. However, for a simple flavoenzyme mechanism, the rate of reduction to the steady state equals the sum of the first-order rate constants for reduction and oxidation, or kred + kox.57 The ratio of reduced to oxidized enzyme molecules in the steady state is kred/kox. From the approach to steady state, N3− clearly has an inhibitory effect on kred + kox, but at steady state, the kred/kox ratio is not dramatically different. Diode-array analyses of the enzyme at steady state, in each azide experiment, confirm full formation of the E-FADH2·NAD+ intermediate. The earlier detailed kinetic study with C42S Ef Nox3 led to the conclusion that kcat (2.2 s−1, similar to the value of 1.1 s−1 determined with C44S SpyNox) is limited by a chemical or conformational change. The two active-site E-FADHOOH intermediates per homodimer release H2O2 at different rates (29 and 1.8 s−1). The conformational change in the “slow” active site is followed by rapid O2 reactivity, and the symmetric oxidized dimer initiates the next catalytic cycle. In this work, we have demonstrated that azide binds to both C44S and wild-type EH2 forms of SpyNox. Azide binding is mutually exclusive with respect to Cys44-S− → FAD charge transfer in the latter case. This result is consistent with the structure of the C44S Nox·N3− complex, in which both the Ser44 side chain and the α2 helix in which it resides are shifted. Only this active-site conformer of C44S Nox accommodates the C(4a)-peroxyflavin model, provided the bound azide is excluded. The general kinetic scheme for nonessential activation of an enzyme58 is very similar to that applied in our early, and limited, analysis of azide inhibition with Ef Nox. When it is applied together with a kinetic analysis of C44S SpyNox, however, we observe discrepancies between some kinetic parameters and their static equivalents. For example, the Kd for azide binding of 0.33 mM as determined by spectral titration clashes with the kinetic equivalent, KA (4.36 mM), determined in the scheme for nonessential activation. On the basis of these considerations, we conclude that the nonessential activation of turnover observed with azide and C44S Nox is linked to the rate-limiting conformational shift described in the C42S Ef Nox oxidative half-reaction. The activation, though small (from 1.1 to 3.1 s−1 at 2.5 mM azide), is consistent with the active-site perturbation observed in the structure of the oxidized complex. Whether this results from a direct ligand-induced change in the active site or whether it
■
ASSOCIATED CONTENT
Accession Codes
Coordinates have been deposited with the Protein Data Bank as entries 2BC0, 2BC1, and 2BCP.
■
AUTHOR INFORMATION
Corresponding Author
*Tel.: (336) 716-3914, Fax: (336) 713-1283. E-mail: alc@csb. wfu.edu. Present Addresses ∥
T.C.M.: Rigaku Americas Corp., The Woodlands, TX 77381. T.O.: Faculty of Science, Yamagata University, Kojirakawamachi 1-4-12, Yamagata 990-8560, Japan.
⊥
Author Contributions
J.R.W. and T.C.M. contributed equally to this work. Funding
This work was supported by National Institutes of Health Grant GM-035394 (A.C.), National Science Foundation Grant INT-9803674 (A.C.), and by grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology (Monbusho), Japan, and the Japan Society for the Promotion of Science (JSPS, to T.T.). T.C.M. was the recipient of International Fellowship P-99934 from JSPS, and A.C. was the recipient of Short-Term Invitation Fellowship RC20137002 from JSPS. Data for this study (Proposal 99G119) were collected at the Photon Factory, KEK, Tsukuba, Japan. Notes
The authors declare no competing financial interest.
■
ABBREVIATIONS Nox, H2O-forming NADH oxidase; Ef Nox, E. faecalis NADH oxidase; Cys-SOH, cysteine-sulfenic acid; Ef Npx, E. faecalis NADH peroxidase; PHBH, p-hydroxybenzoate hydroxylase; LsNox, L. sanf ranciscensis NADH oxidase; SpyNox, S. pyogenes NADH oxidase; Cys-SO2H and Cyo, cysteine-sulfenic acid; EH2, two-electron reduced enzyme; DTT, dithiothreitol; FADHOOH, C(4a)-peroxyflavin; FADHOH, C(4a)-hydroxyflavin; BaCoADR, B. anthracis coenzyme A-disulfide reductase; Cys-SO3H, Cys-sulfonic acid; CoADR, coenzyme A-disulfide reductase; rmsd, root-mean-square deviation.
■
REFERENCES
(1) Ojha, S., Meng, E. C., and Babbitt, P. C. (2007) Evolution of function in the ″two dinucleotide binding domains″ flavoproteins. PLoS Comput. Biol. 3, e121. (2) Argyrou, A., and Blanchard, J. S. (2004) Flavoprotein disulfide reductases: advances in chemistry and function. Prog. Nucleic Acid Res. Mol. Biol. 78, 89−142. (3) Mallett, T. C., and Claiborne, A. (1998) Oxygen reactivity of an NADH oxidase C42S mutant: evidence for a C(4a)-peroxyflavin intermediate and a rate-limiting conformational change. Biochemistry 37, 8790−8802. (4) Mande, S. S., Parsonage, D., Claiborne, A., and Hol, W. G. (1995) Crystallographic analyses of NADH peroxidase Cys42Ala and Cys42Ser mutants: active site structures, mechanistic implications, and an unusual environment of Arg 303. Biochemistry 34, 6985−6992. (5) Huijbers, M. M., Montersino, S., Westphal, A. H., Tischler, D., and van Berkel, W. J. (2014) Flavin dependent monooxygenases. Arch. Biochem. Biophys. 544, 2−17.
6827
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry (6) van Berkel, W. J., Kamerbeek, N. M., and Fraaije, M. W. (2006) Flavoprotein monooxygenases, a diverse class of oxidative biocatalysts. J. Biotechnol. 124, 670−689. (7) Entsch, B., Cole, L. J., and Ballou, D. P. (2005) Protein dynamics and electrostatics in the function of p-hydroxybenzoate hydroxylase. Arch. Biochem. Biophys. 433, 297−311. (8) Palfey, B. A., Entsch, B., Ballou, D. P., and Massey, V. (1994) Changes in the catalytic properties of p-hydroxybenzoate hydroxylase caused by the mutation Asn300Asp. Biochemistry 33, 1545−1554. (9) Entsch, B., Ballou, D. P., and Massey, V. (1976) Flavin-oxygen derivatives involved in hydroxylation by p-hydroxybenzoate hydroxylase. J. Biol. Chem. 251, 2550−2563. (10) Sheng, D., Ballou, D. P., and Massey, V. (2001) Mechanistic studies of cyclohexanone monooxygenase: chemical properties of intermediates involved in catalysis. Biochemistry 40, 11156−11167. (11) Ahmed, S. A., and Claiborne, A. (1992) Catalytic properties of streptococcal NADH oxidase containing artificial flavins. J. Biol. Chem. 267, 25822−25829. (12) Lountos, G. T., Jiang, R., Wellborn, W. B., Thaler, T. L., Bommarius, A. S., and Orville, A. M. (2006) The crystal structure of NAD(P)H oxidase from Lactobacillus sanf ranciscensis: insights into the conversion of O2 into two water molecules by the flavoenzyme. Biochemistry 45, 9648−9659. (13) Wallen, J. R., Paige, C., Mallett, T. C., Karplus, P. A., and Claiborne, A. (2008) Pyridine nucleotide complexes with Bacillus anthracis coenzyme A-disulfide reductase: a structural analysis of dual NAD(P)H specificity. Biochemistry 47, 5182−5193. (14) Mallett, T. C., Parsonage, D., and Claiborne, A. (1999) Equilibrium analyses of the active-site asymmetry in enterococcal NADH oxidase: role of the cysteine-sulfenic acid redox center. Biochemistry 38, 3000−3011. (15) Gibson, C. M., Mallett, T. C., Claiborne, A., and Caparon, M. G. (2000) Contribution of NADH oxidase to aerobic metabolism of Streptococcus pyogenes. J. Bacteriol. 182, 448−455. (16) Ahmed, S. A., and Claiborne, A. (1989) The streptococcal flavoprotein NADH oxidase. I. Evidence linking NADH oxidase and NADH peroxidase cysteinyl redox centers. J. Biol. Chem. 264, 19856− 19863. (17) Crane, E. J., 3rd, Yeh, J. I., Luba, J., and Claiborne, A. (2000) Analysis of the kinetic and redox properties of the NADH peroxidase R303M mutant: correlation with the crystal structure. Biochemistry 39, 10353−10364. (18) Larkin, M. A., Blackshields, G., Brown, N. P., Chenna, R., McGettigan, P. A., McWilliam, H., Valentin, F., Wallace, I. M., Wilm, A., Lopez, R., Thompson, J. D., Gibson, T. J., and Higgins, D. G. (2007) Clustal W and Clustal X version 2.0. Bioinformatics 23, 2947− 2948. (19) Robert, X., and Gouet, P. (2014) Deciphering key features in protein structures with the new ENDscript server. Nucleic Acids Res. 42, W320−324. (20) Otwinowski, Z., and Minor, W. (1997) Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307−326. (21) Yeh, J. I., Claiborne, A., and Hol, W. G. (1996) Structure of the native cysteine-sulfenic acid redox center of enterococcal NADH peroxidase refined at 2.8 Å resolution. Biochemistry 35, 9951−9957. (22) Navaza, J. (1994) AMORE - an automated package for molecular replacement. Acta Crystallogr., Sect. A: Found. Crystallogr. A50, 157−163. (23) Collaborative Computational Project Number 4 (1994) The CCP4 suite: programs for protein crystallography. Acta Crystallogr., Sect. D: Biol. Crystallogr. D50, 760−763. (24) Cowtan, K., and Main, P. (1998) Miscellaneous algorithms for density modification. Acta Crystallogr., Sect. D: Biol. Crystallogr. D54, 487−493. (25) Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M. (1991) Improved methods for building protein models in electrondensity maps and the location of errors in these models. Acta Crystallogr., Sect. A: Found. Crystallogr. A47, 110−119.
(26) Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr., Sect. D: Biol. Crystallogr. D54, 905−921. (27) Ahmed, S. A., and Claiborne, A. (1989) The streptococcal flavoprotein NADH oxidase. II. Interactions of pyridine nucleotides with reduced and oxidized enzyme forms. J. Biol. Chem. 264, 19863− 19870. (28) Müller, F., and Massey, V. (1969) Flavin-sulfite complexes and their structures. J. Biol. Chem. 244, 4007−4016. (29) Crane, E. J., 3rd, Parsonage, D., and Claiborne, A. (1996) The active-site histidine-10 of enterococcal NADH peroxidase is not essential for catalytic activity. Biochemistry 35, 2380−2387. (30) Miller, H., and Claiborne, A. (1991) Peroxide modification of monoalkylated glutathione reductase. Stabilization of an active-site cysteine-sulfenic acid. J. Biol. Chem. 266, 19342−19350. (31) Stehle, T., Ahmed, S. A., Claiborne, A., and Schulz, G. E. (1991) Structure of NADH peroxidase from Streptococcus faecalis 10C1 refined at 2.16 Å resolution. J. Mol. Biol. 221, 1325−1344. (32) Hooft, R. W., Vriend, G., Sander, C., and Abola, E. E. (1996) Errors in protein structures. Nature 381, 272. (33) Kleywegt, G. J., and Jones, T. A. (1996) Phi/psi-chology: Ramachandran revisited. Structure 4, 1395−1400. (34) Holm, L., and Rosenstrom, P. (2010) Dali server: conservation mapping in 3D. Nucleic Acids Res. 38, W545−549. (35) Crane, E. J., 3rd, Vervoort, J., and Claiborne, A. (1997) 13C NMR analysis of the cysteine-sulfenic acid redox center of enterococcal NADH peroxidase. Biochemistry 36, 8611−8618. (36) Creighton, T. E. (1993) Proteins, 2nd ed., W. H. Freeman and Co., New York. (37) Schmidt, H. L., Stöcklein, W., Danzer, J., Kirch, P., and Limbach, B. (1986) Isolation and properties of an H2O-forming NADH oxidase from Streptococcus faecalis. Eur. J. Biochem. 156, 149−155. (38) Higuchi, M., Shimada, M., Yamamoto, Y., Hayashi, T., Koga, T., and Kamio, Y. (1993) Identification of two distinct NADH oxidases corresponding to H2O2-forming oxidase and H2O-forming oxidase induced in Streptococcus mutans. J. Gen. Microbiol. 139, 2343−2351. (39) Scrutton, N. S., Berry, A., and Perham, R. N. (1990) Redesign of the coenzyme specificity of a dehydrogenase by protein engineering. Nature 343, 38−43. (40) Miller, H., Mande, S. S., Parsonage, D., Sarfaty, S. H., Hol, W. G., and Claiborne, A. (1995) An L40C mutation converts the cysteinesulfenic acid redox center in enterococcal NADH peroxidase to a disulfide. Biochemistry 34, 5180−5190. (41) Mittl, P. R., and Schulz, G. E. (1994) Structure of glutathione reductase from Escherichia coli at 1.86 Å resolution: comparison with the enzyme from human erythrocytes. Protein Sci. 3, 799−809. (42) Stehle, T., Claiborne, A., and Schulz, G. E. (1993) NADH binding site and catalysis of NADH peroxidase. Eur. J. Biochem. 211, 221−226. (43) Karplus, P. A., and Schulz, G. E. (1989) Substrate binding and catalysis by glutathione reductase as derived from refined enzyme: substrate crystal structures at 2 Å resolution. J. Mol. Biol. 210, 163− 180. (44) Schreuder, H. A., Hol, W. G., and Drenth, J. (1988) Molecular modeling reveals the possible importance of a carbonyl oxygen binding pocket for the catalytic mechanism of p-hydroxybenzoate hydroxylase. J. Biol. Chem. 263, 3131−3136. (45) Schreuder, H. A., Hol, W. G., and Drenth, J. (1990) Analysis of the active site of the flavoprotein p-hydroxybenzoate hydroxylase and some ideas with respect to its reaction mechanism. Biochemistry 29, 3101−3108. (46) Wilkinson, K. D., and Williams, C. H., Jr. (1979) Evidence for multiple electronic forms of two-electron-reduced lipoamide dehydrogenase from Escherichia coli. J. Biol. Chem. 254, 852−862. (47) Mallett, T. C., Wallen, J. R., Karplus, P. A., Sakai, H., Tsukihara, T., and Claiborne, A. (2006) Structure of coenzyme A-disulfide 6828
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829
Article
Biochemistry reductase from Staphylococcus aureus at 1.54 Å resolution. Biochemistry 45, 11278−11289. (48) Wallen, J. R., Mallett, T. C., Boles, W., Parsonage, D., Furdui, C. M., Karplus, P. A., and Claiborne, A. (2009) Crystal structure and catalytic properties of Bacillus anthracis CoADR-RHD: implications for flavin-linked sulfur trafficking. Biochemistry 48, 9650−9667. (49) Warner, M. D., Lukose, V., Lee, K. H., Lopez, K., Sazinsky, M. H., and Crane, E. J., 3rd (2011) Characterization of an NADHdependent persulfide reductase from Shewanella loihica PV-4: implications for the mechanism of sulfur respiration via FADdependent enzymes. Biochemistry 50, 194−206. (50) Atkinson, H. J., and Babbitt, P. C. (2009) An atlas of the thioredoxin fold class reveals the complexity of function-enabling adaptations. PLoS Comput. Biol. 5, e1000541. (51) Shoun, H., Arima, K., and Beppu, T. (1983) Inhibition of phydroxybenzoate hydroxylase by anions: possible existence of two anion-binding sites in the site for reduced nicotinamide adenine dinucleotide phosphate. J. Biochem. 93, 169−176. (52) Gatti, D. L., Palfey, B. A., Lah, M. S., Entsch, B., Massey, V., Ballou, D. P., and Ludwig, M. L. (1994) The mobile flavin of 4-OH benzoate hydroxylase. Science 266, 110−114. (53) Steennis, P. J., Cordes, M. M., Hilkens, J. H., and Müller, F. (1973) On the interaction of para-hydroxybenzoate hydroxylase from Pseudomonas f luorescens with halogen ions. FEBS Lett. 36, 177−180. (54) Ghisla, S., and Massey, V. (1989) Mechanisms of flavoproteincatalyzed reactions. Eur. J. Biochem. 181, 1−17. (55) Tierney, D. L., Fee, J. A., Ludwig, M. L., and Penner-Hahn, J. E. (1995) X-ray absorption spectroscopy of the iron site in Escherichia coli Fe(III) superoxide dismutase. Biochemistry 34, 1661−1668. (56) Gutman, C. T., Guzei, I. A., and Brunold, T. C. (2013) Structural, spectroscopic, and computational characterization of the azide adduct of Fe(III)(2,6-diacetylpyridinebis(semioxamazide)), a functional analogue of iron superoxide dismutase. Inorg. Chem. 52, 8909−8918. (57) Gutfreund, H. (1972) Enzymes: Physical Principles, John Wiley & Sons, New York. (58) Segel, I. H. (1975) Enzyme Kinetics, John Wiley & Sons, New York. (59) Frieden, C. (1979) Slow transitions and hysteretic behavior in enzymes. Annu. Rev. Biochem. 48, 471−489.
6829
DOI: 10.1021/acs.biochem.5b00676 Biochemistry 2015, 54, 6815−6829