The Terminal Oxidase Cytochrome bd Promotes Sulfide-resistant Bacterial Respiration and Growth

Hydrogen sulfide (H2S) impairs mitochondrial respiration by potently inhibiting the heme-copper cytochrome c oxidase. Since many prokaryotes, including Escherichia (E.) coli, generate H2S and encounter high H2S levels particularly in the human gut, herein we tested whether bacteria can sustain sulfide-resistant O2-dependent respiration. E. coli has three respiratory oxidases, the cyanide-sensitive heme-copper bo3 enzyme and two bd oxidases much less sensitive to cyanide. Working on the isolated enzymes, we found that, whereas the bo3 oxidase is inhibited by sulfide with half-maximal inhibitory concentration IC50 = 1.1 ± 0.1 μM, under identical experimental conditions both bd oxidases are insensitive to sulfide up to 58 μM. In E. coli respiratory mutants, both O2-consumption and aerobic growth proved to be severely impaired by sulfide when respiration was sustained by the bo3 oxidase alone, but unaffected by ≤200 μM sulfide when either bd enzyme acted as the only terminal oxidase. Accordingly, wild-type E. coli showed sulfide-insensitive respiration and growth under conditions favouring the expression of bd oxidases. In all tested conditions, cyanide mimicked the functional effect of sulfide on bacterial respiration. We conclude that bd oxidases promote sulfide-resistant O2-consumption and growth in E. coli and possibly other bacteria. The impact of this discovery is discussed.

reported to be ca. 40-60 μM, as estimated by direct measurement of the gas in the rat cecum 8,9 and analysis of human faecal samples 10 .
E. coli is a ubiquitous member of the human gut microbiota, with more than one strain commonly colonizing the large intestine at the same time. Since E. coli, like the other microorganisms inhabiting the gut, lives in a particularly H 2 S-enriched microaerobic niche, the question arises as to whether this microorganism can accomplish O 2 -dependent respiration without being inhibited by H 2 S. The E. coli respiratory chain possesses three terminal oxygen reductases, utilizing quinols as reducing substrates: the cyanide-sensitive cytochrome bo 3 enzyme and the bd-I and bd-II oxidases, much less sensitive to cyanide 11,12 . Cytochrome bo 3 belongs to the superfamily of heme-copper oxygen reductases that includes mtCcOX. The enzyme contains three redox-active metal centres: the low-spin heme b involved in quinol oxidation and a binuclear site composed of heme o 3 and Cu B , where O 2 reduction to water takes place. On the contrary, bd-I and bd-II are cytochrome bd-type O 2 -reductases phylogenetically unrelated to heme-copper oxidases 12 . They have no copper, but contain three hemes: the low-spin heme b 558 (the primary electron acceptor from the quinol), and the two high-spin hemes b 595 and d (possibly forming a di-heme site for O 2 reduction, see 12 and references therein). Cytochrome bo 3 predominates in E. coli under high aeration, whereas O 2 -limiting conditions such as those found in the human gut stimulate the expression of the cytochromes bd-I and bd-II [13][14][15] . The three E. coli terminal oxidases all generate a proton motive force, but cytochrome bo 3 is the only one able to pump protons, thus being twice as effective as bd-type cytochromes in terms of energy transduction 16 . Besides its role in bacterial energy metabolism, cytochrome bd-I was suggested to serve other physiological functions, being implicated in the bacterial response to oxidative and nitrosative stress [17][18][19][20] .
In this work, we examined the effect of sulfide on the O 2 reductase activity of the three terminal oxidases of E. coli and tested the ability of these enzymes to sustain bacterial growth and O 2 consumption in the presence of sulfide.

Results
Effect of NaHS on isolated E. coli terminal oxidases. The effect of sulfide on the O 2 reductase activity of the E. coli respiratory oxidases, cytochromes bo 3 , bd-I and bd-II, was initially investigated testing the ability of each purified oxidase to consume O 2 before and after addition of the sulfide donor NaHS. In these assays, O 2 consumption was measured in the presence of dithiothreitol (DTT) and 2,3-dimethoxy-5-methyl-6-(3methyl-2-butenyl)-1,4-benzoquinone (Q 1 ) as the reducing system. As shown in Fig. 1A, NaHS (~7 μM) rapidly and effectively inhibits the O 2 reductase activity of the isolated cytochrome bo 3 . The enzyme is inhibited with an apparent half-maximal inhibitory concentration IC 50 = 1.1 ± 0.1 μM (Fig. 2). The inhibition of cytochrome bo 3 is fully reversible. A rapid and complete recovery of the O 2 reductase activity of the isolated enzyme was observed, when sulfide was quickly removed from solution by addition of an excess of O-acetyl-L-serine (OAS) and catalytic amounts of the sulfide-consuming O-acetylserine sulfhydrylase enzyme from Entamoeba histolytica (EhOASS, Fig. 1A). Sulfide consumption by EhOASS in the presence of OAS was assessed independently using a H 2 S-selective electrode ( Figure S1). Notably, while being an effective inhibitor of E. coli cytochrome bo 3 , NaHS proved to be unable to inhibit the two E. coli bd-type oxidases. Addition of NaHS, even at high concentration (58 μM), did not alter the O 2 consumption catalyzed by the bd-I or bd-II enzyme in the presence of DTT and Q 1 (Fig. 1A). No O 2 consumption stimulation by the OAS/EhOASS sulfide-scavenging system was observed in control oxygraphic experiments carried out in the absence or presence of the isolated oxidases (not shown).
Effect of NaHS on E. coli respiration. The striking results obtained with the isolated enzymes prompted us to explore the effect of sulfide on E. coli cell respiration. To this end, we investigated aerobic cultures of E. coli (see Methods for details) and tested the effect of NaHS on cell respiration along cell growth, i.e., at increasing cell density. We initially assayed three mutant strains each expressing a single terminal oxidase (bo 3 , bd-I or bd-II). The results were remarkably similar to those obtained with the isolated enzymes. O 2 consumption by E. coli cells expressing solely cytochrome bo 3 was quickly and fully inhibited upon addition of 50 μM NaHS (Fig. 1B). As observed with the isolated bo 3 enzyme, the inhibition was promptly and fully restored upon sulfide depletion by the EhOASS/OAS system (Fig. 1B). In contrast, no inhibition was observed following the addition of 50 μM NaHS to E. coli cells expressing either bd-I or bd-II as the only terminal oxidase (Fig. 1B). The results on the three mutant strains proved to be independent of the density at which cells were collected and assayed (Fig. 3, top panel). Similarly to NaHS, cyanide (50 μM) almost completely abolished O 2 -consumption in E. coli cells expressing only the bo 3 oxidase, whereas it was essentially ineffective when respiration was sustained by either bd oxidase (Fig. 3, bottom panel).
The effect of NaHS on respiration of the wild-type strain was assessed in the same way. Namely, we investigated aerobic cultures in which a change in oxidase expression from cytochrome bo 3 to the cytochromes of the bd-type is expected to take place along cell growth, following a progressive reduction in O 2 availability in the medium 21,22 . Accordingly, when cells were assayed in an early phase of the culture (OD 600 < 0.7), most of respiration (65-70%) proved to be sensitive to NaHS or cyanide (both at 50 μM, Fig. 3). In contrast, with cell growth bacterial O 2 -consumption became progressively less sensitive to sulfide inhibition and, in a late phase of the culture (OD 600 > 2.5), NaHS or cyanide caused only marginal effects on respiration (Fig. 3).
Altogether these results show that, unlike the heme-copper bo 3 oxidase, E. coli bd oxidases enable O 2 -dependent respiration in the presence of sulfide.
Effect of NaHS on E. coli cell growth. The lack of sulfide inhibition of cytochromes bd-I and bd-II, as opposed to the high sensitivity displayed by the bo 3 oxidase, prompted us to test whether the bd-type oxidases, besides enabling respiration, promote E. coli cell growth in the presence of sulfide. We investigated the effect of sulfide on the growth of both the wild-type and the three respiratory mutant strains. Following the addition of Scientific RepoRts | 6:23788 | DOI: 10.1038/srep23788 200 μM NaHS, the wild-type strain showed a delayed growth (Fig. 4A), while the growth of the bo 3 -expressing strain was severely impaired (Fig. 4B). Lacking bd oxidases, the latter strain proved to be highly sensitive to sulfide, with 6 μM NaHS causing ~25% reduced cell growth, as evaluated at 2 hours after NaHS addition (inset Fig. 4B). In contrast, no or very little effect on cell growth was observed over the same time window after addition of 200 μM NaHS to the strains expressing either bd-I or bd-II as the only terminal oxidase (Fig. 4C,D). Altogether, these data show that, unlike the bo 3 oxidase, the cytochromes bd-I and bd-II sustain E. coli growth in the presence of sulfide.

Discussion
Together with NO and CO, H 2 S is presently considered a highly relevant signalling molecule in human (patho)physiology. It has long been recognized that many prokaryotes, including the model organism E. coli and numerous other members of the human gut microbiota, generate H 2 S (see 6 and references therein). Bacteria can accomplish H 2 S production by several pathways, including cysteine degradation by L-cysteine desulfhydrase, and dissimilatory sulfate reduction by SRB (see 6 and references therein). In a recent study, it was reported that orthologs of the mammalian H 2 S-synthesizyng enzymes CBS, CSE and 3-MST are widespread in the bacterial world and contribute to H 2 S generation, as demonstrated for several bacteria by genetic manipulation 23 . As an example, E. coli was shown to harbour an ortholog of 3-MST significantly contributing to bacterial H 2 S synthesis. Notably, in the same study H 2 S production was shown to enhance antibiotic resistance in all tested bacteria, thereby providing an adaptive advantage. The presence of numerous H 2 S-producing bacteria in the human gut makes this compartment particularly enriched in H 2 S compared to other tissues, with the free gas reaching in the intestinal lumen concentrations as high as 40-60 μM [8][9][10] . Relevant to human (patho)physiology, bacteria-derived H 2 S is emerging as a key regulator of several physiological functions not only in the gastrointestinal system, but also throughout the human body 1 . Moreover, it has been recently suggested that the differential susceptibility of mutualistic microbes to sulfide toxicity may contribute to shape the human gut microbiota 6 , a recognized factor contributing to human health and disease. In turn, the host H 2 S systemic bioavailability and metabolism have been found to be profoundly affected by the microbiota in studies on germ-free mice 24 . Altogether these observations provide evidence for interplay between H 2 S and the human microbiota, with important consequences on human health.
Though currently considered a key signalling molecule, H 2 S has long been known as a mere poison. Toxicity has been related to the ability of H 2 S to bind heme proteins and inhibit cellular respiration targeting mtCcOX (see 3 and references therein). Indeed, H 2 S is a potent (K i = 0.2-0.45 μM 3,4 ), non-competitive inhibitor of this respiratory enzyme, the inhibition being reversible, independent of oxygen concentration 25 , but dependent on pH 26 . Sulfide inhibition of isolated mtCcOX in turnover with ascorbate and cytochrome c is relatively fast, occurring at an initial rate constant of 2.2 × 10 4 M −1 s −1 , as measured at pH 7.4 3 . The inhibited enzyme exhibits sulfide bound to ferric heme a 3 27,28 , with Cu B in the cuprous state possibly bound to a second H 2 S molecule, as revealed by electron paramagnetic resonance (EPR) spectroscopy 29 . The mechanism of inhibition of mtCcOX is only partly understood, yet the reaction was suggested to involve the binding of H 2 S to the enzyme in turnover at cupric or cuprous Cu B , followed by intramolecular transfer of H 2 S to ferric heme a 3 , eventually blocking the reaction with O 2 3 . The well-known toxicity of H 2 S on mitochondrial respiration prompted us to address whether bacterial O 2 -dependent respiration can be accomplished in a H 2 S-enriched environment such as the human gut, thereby providing an adaptive advantage in terms of bacterial growth. This issue was addressed in the present study working on the model organism E. coli, a ubiquitous member of the human gut microbiota. Namely, we investigated the effect of sulfide on the O 2 reductase activity of each of the three terminal respiratory oxidases of this bacterium (cytochromes bo 3 , bd-I and bd-II), and tested the ability of these enzymes to sustain O 2 consumption and bacterial cell growth in the presence of sulfide. Using NaHS as a H 2 S donor, we carried out experiments on the isolated enzymes, as well as on the wild-type and three respiratory mutant E. coli strains each expressing only a single terminal oxidase. NaHS is commonly used as a donor of the cell permeant H 2 S, because in aqueous solution HS − equilibrates with H 2 S and S 2− , according to the pK a1 ~7.0 (H 2 S/HS − ) and pK a2 ~19 (HS − /S 2− ) measured at 25 °C. At pH = 7.0-7.4, ~30-50% of HS − is thus expected to be protonated to H 2 S, with S 2− being present in negligible amounts.
As a new finding we report that, whereas the heme-copper bo 3 oxidase is highly sensitive to sulfide inhibition (IC 50 = 1.1 ± 0.1 μM , Figs 1 and 2), the two bd oxidases (bd-I and bd-II) are remarkably insensitive to sulfide (Fig. 1), as confirmed by measuring the effect of NaHS on O 2 consumption by the purified terminal oxidases (Fig. 1A) or by whole cells (Figs 1B and 3). In agreement with these finding, cell growth proved to be severely impaired by sulfide in an E. coli mutant strain expressing only the bo 3 oxidase (Fig. 4B), but unaffected in mutant strains expressing either bd-I or bd-II as the only terminal oxidase (Fig. 4, panel C,D). Consistently, in the wild-type strain, H 2 S affected cell growth and respiration only in the early phase of the culture, when O 2 availability is expected to be still sufficiently high to favour the expression of the bo 3 oxidase, but it caused no effect in a late phase of the culture, when O 2 limitation is expected to stimulate the expression of bd oxidases (Fig. 4A).
Altogether, these observations led us to conclude that, at variance with the heme-copper bo 3 oxidase that is potently and reversibly inhibited by sulfide, both E. coli bd oxidases are sulfide-insensitive and thus able to sustain cell respiration and growth in the presence of considerably high levels of sulfide. Although the molecular basis for the remarkable sulfide insensitivity of the E. coli bd oxidases remains to be elucidated, it may originate from the lack of Cu B , which was indeed suggested to be implicated in sulfide inhibition of mtCcOX 3 . In this regard, still possibly due to the lack of Cu B , it is noteworthy that bd oxidases are not only more resistant to NO inhibition than heme-copper oxidases [30][31][32] , but also poorly sensitive to other commonly used oxidase inhibitors, such as cyanide 12 and azide 33 . On this basis, cyanide and sulfide are expected to exert similar inhibitory effects on E. coli respiration, as observed in the present study (Fig. 3).
As shown here for E. coli, it is likely that bd oxidases confer sulfide resistance also to other microorganisms. The bd oxidases are indeed widespread in the prokaryotic world and have been identified in numerous enterobacteria 34 , where expression of these oxidases is likely stimulated in the microaerobic conditions found in the human colon. In view of the novel results presented here, it will be important to test whether bd oxidases, by conferring sulfide resistance, play a role in shaping the human gut microbiota, thereby impacting human (patho)physiology. Furthermore, based on these data, bd oxidases may represent very attractive targets for the development of next-generation antimicrobials against pathogenic enterobacteria 18,20,35 . Finally, the finding that bd oxidases enhance bacterial resistance to sulfide, if representing a hallmark of this protein family, may pave the way to biotechnological applications aimed at increasing bacterial sulfide resistance. Comparison of the effect of cyanide and sulfide on cell respiration: respiratory activity measured after the addition of 50 μM NaHS or 50 μM NaCN to wild-type and mutant E. coli cells. Data (mean ± standard deviation) refer to the control activity measured before the addition of inhibitors (taken as 100%).
Scientific RepoRts | 6:23788 | DOI: 10.1038/srep23788 Methods Materials, bacterial strains and growth conditions. All chemicals were purchased from Sigma unless otherwise indicated. NaHS stock solutions were prepared by dissolving NaHS in degassed water or phosphate buffer saline, and the overall concentration of sulfide species (H 2 S/HS − /S 2− ) in solution was determined spectrophotometrically according to 36 . All E. coli strains used were K-12 derivatives; MG1655 (RKP5416) was the wild type 37 from which the respiratory mutants, TBE025 (MG1655 ΔcydB nuoB appB::kan), TBE026 (MG1655 ΔcydB nuoB cyoB::kan) and TBE037 (MG1655 ΔappB nuoB cyoB::kan) were derived, respectively expressing cytochrome bo 3 , bd-II and bd-I as the only terminal oxidase (mutants kindly given by Alex Ter Beek and Joost Teixeira de Mattos, University of Amsterdam). These strains carry the same mutant alleles as described by Bekker et al. 38 . E. coli cells were grown in 50 mL-Falcon tubes, in 5 mL Luria Bertani (LB) medium supplemented with 30 μg/ mL kanamycin, at 37 °C and 200 rpm. For growth studies, cells were grown as described above in the absence or presence of NaHS (6-200 μM) added to cells at an OD 600 of about 0.05. 3 were isolated from the E. coli strains GO105/pTK1, MB37 and GO105/pJRhisA, respectively, as previously described [39][40][41] . The concentration of the cytochromes bd-I and bd-II was determined from the difference absorption spectrum using Δε 628-607 = 10.8 mM −1 cm −1 for the dithionite-reduced minus 'as prepared' proteins. Cytochrome bo 3 concentration was estimated from the Soret absorption band of the oxidized enzyme using ε 407 = 183 mM −1 cm −1 . UV-visible absorption spectra were acquired in an Agilent Cary 60 spectrophotometer.

Purification and H 2 S consumption by recombinant O-acetylserine sulfhydrylase from
Entamoeba histolytica. The O-acetylserine sulfhydrylase-encoding gene (EhOASS, Genbank XM_643199.1) was PCR-amplified from Entamoeba histolytica HM-1:IMSS genomic DNA using the forward primer 5′-CATATGATGGAACAAATAAGTATTAGC and the reverse primer 5′-AACGTTTTA TTCATTCAATAATGAATCAAG, containing the NdeI and HindIII restriction sites respectively. The PCR product was cloned into the Topo TA pCR2.1 vector, digested with the NdeI and HindIII restriction enzymes, and gel purified. The DNA insert was subcloned into the NdeI and HindIII restriction sites of the pET28b expression vector, yielding the pET-EhOASS construct encoding N-terminally 6xHis-tagged EhOASS. pET-EhOASS was used to transform E. coli BL21 (DE3). Cells were grown at 37 °C in LB broth supplemented indicates the ratio between the optical density measured at 600 nm in the presence of NaHS and the one recorded after the same period of time (2 hours) in the absence of NaHS. Data expressed as mean ± standard deviation.
with 25 mg/L kanamycin (Nzytech) until OD 600 reached 0.4-0.5. EhOASS expression was induced with 0.1 mM isopropyl-β -D-thiogalactoside addition and the cultures moved to 30 °C, 130 rpm for 4 h. Cells were harvested and the pellet resuspended in 10 mL/L culture of buffer A (50 mM potassium phosphate, 300 mM KCl, pH 7.5, 10% glycerol) containing 1 mg/mL lysozyme, 1 mM phenylmethylsulfonyl fluoride and deoxyribonuclease I. After 30-min incubation on ice, cells were disrupted by sonication, centrifuged at 8200 g (5 min, 4 °C) and imidazole was added to the supernatant to a final concentration of 10 mM. Protein purification steps were performed in an Åkta Prime (GE Healthcare) chromatography system. Affinity purification of the His-tagged protein was performed using a HisTrap FF crude 1-mL column previously equilibrated with buffer A containing 10 mM imidazole (buffer B). The cleared supernatant was loaded onto the column at 1 mL/min and the column was washed with 25 column volumes of buffer B followed by a linear gradient of 15 column volumes up to 500 mM imidazole. Pooled protein fractions were loaded onto a PD10 (GE Healthcare) desalting column for imidazole removal, equilibrated and washed with buffer A. EhOASS-containing fractions were concentrated with Amicon Ultra-15 centrifugal filter units (30 kDa cut-off) and loaded onto a size-exclusion 120-ml Superdex S-200 (GE Healthcare) column, equilibrated and eluted with buffer A at 0.7 mL/min. EhOASS fractions were pooled; protein purity was assessed by SDS-PAGE and protein concentration was determined by the Bradford assay. As previously reported 42 , pure EhOASS eluted as a dimer of ~38 kDa monomers ( Figure S1). H 2 S consumption by EhOASS was measured at 20 °C in 100 mM HEPES, 260 U/mL catalase, 100 μM EDTA pH 7.0, using an ISO-H2S-2 hydrogen sulfide sensor coupled to an Apollo 4000 Free Radical Analyzer (World Precision Instruments). In these assays the concentration of H 2 S in solution was obtained from the nominal concentration of the NaHS added, assuming 1:1 partition between HSand H 2 S at pH 7.0, according to the pK a of H 2 S. O 2 consumption measurements. Oxygraphic measurements were carried out at 25 °C in 100 mM Na/ phosphate pH 7.4, using a high-resolution respirometer (Oxygraph-2k, Oroboros Instruments) with a 1.5 mL chamber. The buffer was supplemented with 0.1 mM EDTA and either 0.05% N-lauroyl-sarcosine (cytochrome bd-I) or 0.02% dodecyl-β -D-maltoside (cytochrome bd-II and cytochrome bo 3 ) in the assays on isolated oxidases. The apparent IC 50 of NaHS for the O 2 -reductase activity of the isolated bo 3 oxidase was obtained by plotting the percentage inhibition of the enzyme as a function of NaHS concentration and fitting the data to the Hill equation 43 , assuming a Hill coefficient n = 1.