Cytochrome bd Displays Significant Quinol Peroxidase Activity

Cytochrome bd is a prokaryotic terminal oxidase that catalyses the electrogenic reduction of oxygen to water using ubiquinol as electron donor. Cytochrome bd is a tri-haem integral membrane enzyme carrying a low-spin haem b558, and two high-spin haems: b595 and d. Here we show that besides its oxidase activity, cytochrome bd from Escherichia coli is a genuine quinol peroxidase (QPO) that reduces hydrogen peroxide to water. The highly active and pure enzyme preparation used in this study did not display the catalase activity recently reported for E. coli cytochrome bd. To our knowledge, cytochrome bd is the first membrane-bound quinol peroxidase detected in E. coli. The observation that cytochrome bd is a quinol peroxidase, can provide a biochemical basis for its role in detoxification of hydrogen peroxide and may explain the frequent findings reported in the literature that indicate increased sensitivity to hydrogen peroxide and decreased virulence in mutants that lack the enzyme.

The bioenergetic efficiency of cytochrome bd is half that of the oxygen-reducing cytochrome oxidases, which in addition to consuming chemical protons also pump protons across the membrane (reviewed in 20 ).
Cytochrome bd is a tri-haem protein carrying haem b 558 which is ligated by His186 and Met393 (Escherichia coli numbering), haem b 595 ligated by His19 and Glu99, and haem d ligated by Glu445 21 . Haems b 595 and d are proposed to constitute a functional binuclear site, similar to the binuclear haem-Cu B site in haem-copper oxidases where the oxygen chemistry takes place 20,[22][23][24][25][26][27] (but see 21 ). An important mechanistic feature in both classes of enzymes is that the reduction of oxygen occurs in a concerted 4-electron redox reaction preventing the formation of reactive oxygen species (ROS): superoxide (O 2 − ), hydrogen peroxide (H 2 O 2 ) and the hydroxyl radical 28 . ROS are produced endogenously when molecular oxygen is partially reduced to superoxide and H 2 O 2 by redox enzymes, especially flavoenzymes, including the respiratory chain Complex I [29][30][31][32][33][34] . Two superoxide anions dismutate to H 2 O 2 and O 2 in the cell either spontaneously or catalysed enzymatically by superoxide dismutase (SOD). When H 2 O 2 is reduced by cellular Fe 2+ through Fenton chemistry, hydroxyl radicals are produced leading to a wide spectrum of damage to biological molecules 35,36 . In addition to lipids and DNA, protein targets of ROS, which lead to enzyme inactivation, include solvent-exposed Fe-S clusters of dehydratases among which aconitases and fumarases and the Isc system responsible for Fe-S cluster synthesis 37,38 . Cells not only have to cope with endogenous ROS. Microorganisms must also detoxify ROS produced extracellularly by competing microorganisms and in the case of pathogenic microorganisms by host immune systems [39][40][41] .
In order to protect themselves from oxidative damage, prokaryotes express different ROS scavenging enzymes and employ low-molecular weight agents such as ascorbate and glutathione 36,42,43 . In addition to SOD, E. coli synthesizes a number of specific cytoplasmic H 2 O 2 -scavenging enzymes: the catalases KatG and KatE 36 ; NADH-dependent alkyl hydroperoxide reductase (Ahp) 44 , glutathione peroxidase (GPX) 45,46 and thiol peroxidase 47,48 . In aerobically growing E. coli cells, the main H 2 O 2 scavengers are KatG, KatE and Ahp 42 .
Collectively, these studies indicate that cytochrome bd can play a role in scavenging exogenous H 2 O 2 produced e.g. during infection in a manner similar to periplasmic catalases/peroxidases or SOD, which have been proposed as virulence factors in highly pathogenic bacterial strains among which E. coli O157:H7 and several other species [64][65][66][67][68] .
Two recent studies have suggested that cytochrome bd from E. coli is endowed with very low guaiacol peroxidase activity 69 and a significant catalase activity 70 , proposed to explain the protective phenotype of the enzyme in vivo.
In the present study, we aimed to investigate the in vitro activity of a highly purified preparation of cytochrome bd towards hydrogen peroxide. Mass spectrometry showed the presence of a third subunit, CydX. We further show that cytochrome bd has quinol peroxidase (QPO) activity and lacks catalase activity. We discuss how the newly discovered QPO activity of cytochrome bd can contribute to detoxification of exogenous hydrogen peroxide, therefore potentially contributing to the virulence of pathogenic microorganisms.
Protein preparation and activity assays. Expression and purification of the wild type cytochrome bd was performed using a cydABX pACYC177 overexpression vector as described earlier 28 . For production of the His 6 -tagged protein, the vector was modified by addition of six histidine triplets (CACCATCACCACCATCAC) at the 3′ -end of cydA (C-terminal His 6 -tag). Overexpression of the His 6 -tagged protein and membrane isolation were done as in 28 . The protein was purified over a HisTrap Nickel column (GE Healthcare) eluting at ~ 0.3 M imidazole (0.02-0.5 M imidazole gradient). The protein was further purified using a Superdex 200 gelfiltration column (GE Healthcare). The haem d content was determined spectrophotometrically from the dithionite-reduced minus as isolated difference spectrum using ε 630-650nm = 24 mM −1 cm −172 . The protein content was determined with the BCA assay (Uptima, Interchim). The purity of the protein was assessed based on the haem d/protein ratio we found (9.26 μ mol haem d/g protein) which corresponds to ~97% using a molar weight of 105.5 kDa for His 6 -CydABX. Polarographic oxidase activity measurements and lack of catalase activity were performed and confirmed in two groups either using a home-built setup with a Clark-type oxygen electrode 73 (Group Simon de Vries) or an Oxygraph+ Clark-type oxygen electrode from Hansatech (Group Thorsten Friedrich). Determination of quinol peroxidation rates was conducted inside a Coy anaerobic chamber equipped with an Avantes DH-2000 spectrophotometer. Due to the high 278-nm absorbance at high quinol/quinone concentrations, the reaction progress was monitored at 260 nm rather than 278 nm. The extinction coefficient ε 260nm = 6.23 mM −1 cm −1 was determined from the UV spectrum of decylubiquinone based on ε 278nm = 12.7 mM −1 cm −1 . The reactions were performed in the standard buffer: 50 mM MOPS, 100 mM NaCl, 0.1% LM, pH 7 unless otherwise noted. Aliquots of nitric oxide (NO) were added from a NO-saturated (2 mM) aqueous solution.
Igor Pro version 6.1 (Wavemetrics) was used for data analysis and creating graphs.
Analysis of the steady-state kinetics. Initial QPO rates were determined for varying H 2 O 2 concentrations while keeping the decylubiquinol (dQH 2 ) concentration constant and vice versa. The rates were simulated using the model for a Ping-Pong Bi Bi reaction mechanism according to:  The dQH 2 concentration was determined in the same experiment from the absorbance change at 260 nm as described above.
Determination of catalase activity in membranes. The catalase activity in membranes was determined by following the oxygen production (see above) in standard buffer without detergent at different H 2 O 2 concentration. To test whether the catalase activity was membrane-associated, the membranes of E. coli overexpressing cytochrome bd were washed by first diluting the membrane suspension (1:13) in standard buffer without detergent. The diluted suspension was sonicated (10 min in a Biorupter sonicator from Diagenode at maximum intensity) to disrupt possible membrane vesicles containing cytosolic proteins. The sonicated membrane suspension was centrifuged for 1 h at 300,000 g for membrane recovery. The membrane pellet was resuspended in buffer prior to the polarographic assay. The dilution/sonication procedure was repeated (second wash) using the product from the first step and the polarographic assay was performed again.

Tandem MS analysis and identification of CydX.
Purified His-tagged cytochrome bd was loaded on a Native-PAGE, the protein band of interest was excised from the gel and subjected to in gel proteolytic digestion using either trypsin, chymotrypsin or GluC (enzyme: protein ratios ~1:15-1:20 (w/w) in 25 mM ammonium bicarbonate, pH 8.1) overnight at 37 °C. Prior to digestion, cysteines were reduced with dithiothreitol (DTT) in ammonium bicarbonate for 30 min, followed by alkylation with iodoacetamide in ammonium bicarbonate in the dark for 45 min. In-gel digests were separated and analyzed on EASY-nLC 1000 system directly coupled to a Q Exactive Plus mass spectrometer (Thermo, Bremen, Germany). Peptides were separated on a reversed-phase column (Acclaim PepMap, 50 μ m × 150 mm, 2 μ m, 100 Å, Thermo, Bremen, Germany). The gradient was from 100% Solvent A (0.1% formic acid in water) to 60% solvent B (acetonitrile) in 25 min. at a flow rate of 500 nl/min. The column effluent was directly electrosprayed in the ESI source of the mass spectrometer using a nano-ESI emitter (Nano-bore emitter, Thermo, Bremen, Germany). MS data was acquired in the positive ion mode using a data-dependent top10 analysis method. Full scan spectra were acquired in the m/z range 400-1200 at a resolution of 70.000, a target value of 3e6 and a maximum injection time of 100 msec. HCD fragmentation events were dynamically triggered at an underfill ratio of 5%. Isolation of precursor ions was done with a window of 2,5 amu, a target value of 2e5 and maximum injection time of 50 msec. Normalized collision energy of 27 eV was used and fragment ions were acquired at a resolution of 17.500 with m/z 100 as first mass. The raw data was processed with Proteome Discoverer 1.4 (Thermo, Bremen, Germany) and spectra were matched against the Uniprot protein database using mascot. Search parameters used were; 5 ppm for precursor mass, 0.02 Da for fragment ions, taxonomy restrain E. coli, carbamidomethylcysteine as fixed modification and oxidized methionine as variable modification and no cleavage enzyme was specified. CydX from E. coli consists of 37 amino acid residues (1-MWYFAWILGTLLACSFGVITALALEHVESGKAGQEDI-37).

Analytical chromatography.
To verify the monodispersity of the pure cytochrome bd, 500 μ g of the enzyme was loaded onto a gel filtration column (Superose 6 10/300 gl, GE Healtcare) equilibrated with the standard buffer (VE2001 GPC solvent/sample module, Viscotek). The UV absorbance at 280 nm (UV Detector 2600, Viscotek) as well as the refractive index and the right angle light scattering were monitored during the run (TDA 305 triple detector array, Viscotek).

Results
The cytochrome bd preparation is highly pure and contains the CydX subunit. Using decylubiquinol (dQH 2 ) as electron donor, the purified cytochrome bd displayed a turnover number of 185 ± 15 dQH 2 s −1 consistent with the value for the wild-type enzyme 72 indicating that the His-tag did not interfere with the activity of the enzyme. Cytochrome bd has long been considered a hetero-dimer throughout literature [1][2][3]20 . However, recent mutational studies in E. coli, Brucella abortus and Shewanella oneidensis suggested that the small protein, CydX (37, 64 and 38 amino acids in E. coli, B. abortus and S. oneidensis, resp.) is important for assembly, stability and activity of cytochrome bd in vivo and in vitro 63,[74][75][76] . The presence of CydX has also been confirmed in purified cytochrome bd 75 . To confirm the presence of CydX in our preparation we performed a mass-spectrometric analysis. Using trypsin, chymotrypsin and Glu-C to cleave the protein we detected the peptides 23-ALEHVESGKAGQEDI-37, 29-SGKAGQEDI-37 and 23-ALEHVESGKAGQEDI-37, respectively, unequivocally confirming the presence of CydX in our preparation. To verify the monodispersity of the purified cytochrome bd, the enzyme was subjected to analytical chromatography (Fig. 1). The UV absorption showed a single peak, corresponding to the mass of the cytochrome bd tetramer including the LM micelle (approx. 480 kDa). Refractive index and right angle light scattering exhibited a second peak (approx. 70 kDa) corresponding to the average size of an empty LM micelle. Based on the haem d/protein ratio (see Materials and Methods) the protein purity was approximated as ~97%. These results show that the isolated cytochrome bd is pure, active and complete.
Cytochrome bd is a quinol peroxidase. We tested whether cytochrome bd could function as a peroxidase with its natural oxidase substrate, ubiquinol. Reduced decylubiquinone was used as replacement for the natural  Experiments were carried-out strictly anaerobically to prevent interference between the oxidase and peroxidase reactions. Cytochrome bd was found to catalyse the oxidation of dQH 2 in the presence of H 2 O 2 ( Fig. 2A). To confirm that the oxidation of dQH 2 is due to dQH 2 :H 2 O 2 oxidoreduction, i.e. QPO activity, we measured the amounts of both H 2 O 2 and dQH 2 consumed in time in order to determine the reaction stoichiometry. Figure 2B shows the calculated ratios of dQH 2 /H 2 O 2 , which average to 1.05 ± 0.19. This is consistent with the 1:1 stoichiometry predicted for a genuine QPO reaction (Eq. 3).
To investigate the steady-state kinetics of the QPO reaction, the initial peroxidation rates were measured at different enzyme, H 2 O 2 and dQH 2 concentrations. The plot of initial rate versus the amount of enzyme shows a linear relationship (Fig. 3A). The K M values for H 2 O 2 and dQH 2 were determined at 6.6 ± 1.1 mM and 72 ± 20 μ M, respectively, with the latter being similar to the K M (dQH 2 ) of 85 ± 5 μ M 72 of the oxidase reaction (Fig. 3B). The maximal QPO k cat calculated according to Eq. 2 was 101 ± 10 H 2 O 2 s −1 yielding a specificity constant k cat /K M (H 2 O 2 ) = 1.5 10 4 M −1 s −1 . The QPO pH-dependence profile (inset Fig. 3B) is similar to that of the oxidase reaction 77 , with the highest activity at around pH 7. However, the oxidase reaction is completely inhibited at pH values lower than 5.5 77 whereas at this pH the QPO reaction retains ~1/3 of its maximal value at pH 7.0.

The QPO reaction is inhibited by oxidase inhibitors. NO which mainly binds haem d 27 , is a reversible
inhibitor of the oxidase reaction 78 . Interestingly, our data show that also the QPO reaction is inhibited by NO as well (Fig. 4). Upon addition of 6 μ M NO, dQH 2 oxidation was drastically decreased. The inhibition was reversible as the activity slowly restored (Fig. 4), likely due to slow reaction between NO and dQH 2 which was observed in a separate experiment (22 nM s −1 NO at 100 μ M NO and 100 μ M dQH 2 , data not shown). We did not detect a reaction between NO and H 2 O 2 or any quinol:NO reductase activity, in agreement with others 78 . We also found that titration of the QPO and oxidase activities with HQNO shows that 50% inhibition is obtained at ~10-15 μ M for both reactions.
Cytochrome bd does not show catalase activity. Catalases produce oxygen and water from hydrogen peroxide (Eq. 4) allowing the detection of their activity polarographically using a Clark-type oxygen sensor.
Recently it was reported that cytochrome bd from E. coli had catalase activity 70 . We tested our pure cytochrome bd preparation polarographically in standard buffer (Fig. 5A) and in the buffer (50 mM KP i + 0.1 mM EDTA + 0.05% N-lauroylsarcosine, pH 7.0) used in ref. 70 (Fig. 5B). Even at enzyme concentrations as high as 1 μ M cytochrome bd (Fig. 5B), catalase activity was absent. In Fig. 5C we show our attempt to reproduce the mid-turnover catalase activity measurement shown in ref. 70. Albeit we noticed a decrease of oxidase activity of 9 ± 2% upon addition of 1 mM H 2 O 2 to the assay during turnover (Fig. 5C), we were not able to detect catalase activity under neither of these assay conditions. The lack of post-turnover catalase activity shown in Fig. 3 in ref. 70 was confirmed using similar reaction parameters (Fig. 6A). Interestingly, we observed that the quantity of released oxygen upon catalase addition is dependent of the incubation time of the enzyme with H 2 O 2 (Fig. 6B). This supports the conclusion that H 2 O 2 is consumed during the incubation process, but no oxygen is released, i.e. due to the QPO activity (Eq. 3).
We hypothesized that the catalase activity detected by the authors in ref. 70 might be due to impurities in their enzyme preparation. Therefore, isolated membranes from E. coli that overexpress cytochrome bd were assayed A theoretical activity profile (solid bars) is shown representing the expected remaining catalase activity for a soluble entity (7.7% and 0.60% remaining activity after the first and second washing steps, respectively). The results are presented as means ± SD of duplicates (n = 2).
Scientific RepoRts | 6:27631 | DOI: 10.1038/srep27631 for catalase activity. Interestingly, the membranes did show a weak catalase activity (Fig. 7). It is notable that the relation between activity and H 2 O 2 concentration (Fig. 7A) is very similar to that presented in the inset of Fig. 1 in ref. 70 [79][80][81] . These data suggest that the catalase activity reported in ref. 70 is due to an impurity in the cytochrome bd preparation, although we cannot rule out the possibility that the catalase activity may be dependent on the experimental conditions chosen for protein expression and purification.
To determine if the catalase activity we found in the membrane suspension is membrane-associated and not a cytosolic entity, the activity was measured after washing and sonicating the membranes in buffer containing no detergent (see Materials and Methods). The weak catalase activity decreased slightly after each washing step (Fig. 7B, striped bars), which is prescribed to inactivation and loss of material during the washing procedure. The results show that this catalase activity is resistant to washing suggesting it is membrane-associated. The catalases (KatG and KatE) in E. coli are soluble proteins and to our knowledge, no membrane-bound catalases have been reported in E. coli. Our results show the presence of an unknown membrane-associated catalase activity in E. coli. This hitherto unidentified catalase activity was not further investigated in this study.

Discussion
The purpose of this study was to investigate the in vitro activity of cytochrome bd with hydrogen peroxide to highlight its potential anti-ROS activity in vivo. We have demonstrated here that cytochrome bd from E. coli is a bifunctional enzyme equipped with quinol-linked oxygen and H 2 O 2 reduction activities. In addition, we have shown that the QPO reaction is inhibited by HQNO and NO similar to the oxidase reaction, which suggests a similar involvement of the haem centres and the quinol-binding site in both the oxidase and QPO reactions in respect to electron transfer and catalysis. Under the conditions employed in this study, the data showed that cytochrome bd does not function as a catalase. However, we did detect a membrane-associated catalase activity in isolated E. coli membranes not documented before that showed an unusual relation between activity and H 2 O 2 concentration.
Quinol peroxidation is quite rare in prokaryotes. Besides cytochrome bd, another QPO was found in the human pathogen Aggregatibacter actinomycetemcomitans. This enzyme is a tri-haem c membrane-bound protein with ~43% sequence identity with bacterial cytochrome c peroxidases but less than 13% with cytochrome bd 82,83 . Inhibition of AcQPO correlated to decreased pathogenicity of A. actinomycetemcomitans 83 , a phenotype typical for cytochrome bd mutants (see below). E. coli contains a homologue of AcQPO (YhjA 82 ) predicted to be a cytochrome c peroxidase 43 . YhjA was also tested for QPO activity and was found negative 82 . To our knowledge, cytochrome bd is the first quinol peroxidase characterized in E. coli.
The QPO activity of cytochrome bd demonstrated could provide direct biochemical underpinning for understanding some phenotypes displayed by organisms with non-functional cytochrome bd. For example E. coli with disrupted cytochrome bd accumulated temperature-sensitive growth defects, which could be reverted by exogenous addition of reducing agents as well as SOD and catalase suggesting that increased ROS concentrations (induced at higher temperatures) can be counteracted by the peroxidative cytochrome bd activity 84 . The localization of cytochrome bd in the membrane, suggests that the enzyme can reduce exogenous H 2 O 2 and is therefore functionally differentiated from Ahp, KatG and KatE that scavenge intracellular H 2 O 2 32, [85][86][87][88] .
As described in the Introduction 6,[10][11][12][13]51,52,58,62,63,89 , many examples indicate that pathogenic bacteria that lack cytochrome bd activity display compromised virulence and viability. A striking example is provided by the in vivo anti-ROS activity of cytochrome bd in the gram-negative pathogen B. abortus 6 . B. abortus devoid of a functional cytochrome bd had severely compromised survival in murine spleens, but in trans over-expression of SOD, catalase or cytochrome bd complemented this phenotype showing that H 2 O 2 accumulation is the main phenotype induced by lack of cytochrome bd activity 6 . Consistent with this finding, Staphylococcus aureus increases cytochrome bd expression 8-9 fold upon addition of 10 mM H 2 O 2 90 and M. tuberculosis with over-expressed cytochrome bd showed increased resistance to H 2 O 2 91 , whereas a cytochrome bd knockout in this strain resulted in decreased survival in the mammalian host 10 . The predicted localization of the cytochrome bd active site at the periplasmic side of the cytoplasmic membrane may testify to its protective function mainly against environmentally produced H 2 O 2 and against H 2 O 2 produced in the phagocyte oxidative burst experienced by pathogenic bacteria residing in human macrophages. The finding that the oxidase reaction is completely inhibited at pH values lower than 5.5 77 but the QPO reaction of cytochrome bd is not, may be relevant to its role in combatting the phagocyte oxidative burst in view of the low pH in the phagocyte 92 . It would be important to test cytochrome bd from pathogenic bacterial strains for QPO activity, and to evaluate the contribution of the QPO for survival in the host.
In summary, our finding that cytochrome bd exhibits QPO activity demonstrates that this respiratory complex can serve as a detoxifying enzyme.
In addition to indirectly decreasing the rate of intracellular ROS production via its oxidase reaction, cytochrome bd can also actively metabolize and detoxify hydrogen peroxide. As such, the very catalytic properties of cytochrome bd may explain how the enzyme can act as general virulence factor, which operates in concert with other virulence factors enhancing pathogenicity.