Oxygen as Acceptor

VITALIY B. BORISOV¹ AND MICHAEL I. VERKHOVSKY²*

[SECTION EDITOR: VALLEY STEWART]

Posted August 26, 2009

Department of Molecular Energetics of Microorganisms, Belozersky Institute of Physico-Chemical Biology, Lomonosov Moscow State University, Moscow 119991, Russia,¹ and

Helsinki Bioenergetics Group, Institute of Biotechnology, University of Helsinki, P.O.Box 65 (Viikinkaari 1), 00014, Helsinki, Finland²

*Corresponding author. Mailing address: Helsinki Bioenergetics Group, Institute of Biotechnology, P.O. Box 65 (Viikinkaari 1), 00014 University of Helsinki, Helsinki, Finland. Phone: 358-9-191 58005, Fax: 358-9-191 59920, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

TWO TYPES OF METABOLISM, TWO TYPES OF OXIDASES

Anaerobiosis versus Aerobiosis in Escherichia coli

Like most bacteria, Escherichia coli has a flexible and branched respiratory chain that enables the prokaryote to live under a variety of environmental conditions, from highly aerobic to completely anaerobic. E. coli induces the expression of those respiratory components that are best suited to a particular environment. In general, the bacterial respiratory chain is composed of dehydrogenases, a quinone pool, and reductases. Substrate-specific dehydrogenases transfer reducing equivalents from various donor substrates (NADH, succinate, α-glycerophoshate, formate, hydrogen, pyruvate, and lactate) to a quinone pool (menaquinone, ubiquinone, and demethylmenoquinone). Then electrons from reduced quinones (quinols) are transferred by terminal reductases to different electron acceptors. Under aerobic growth conditions, the terminal electron acceptor is molecular oxygen. A transfer of electrons from quinol to O₂ is served by two major oxidoreductases (oxidases), cytochrome bo₃ and cytochrome bd. When oxygen is not available (under anaerobic conditions), alternative terminal electron acceptors, including nitrate, nitrite, dimethyl sulfoxide, trimethylamine N-oxide, and fumarate, can be used, and the reaction is catalyzed by nitrate reductases, nitrite reductase, dimethyl sulfoxide reductases, trimethylamine N-oxide reductase, and fumarate reductase, respectively (reviewed in references 67 and 221).

Two Quinol Oxidases Only

Terminal oxidases of aerobic respiratory chains of bacteria, which use O₂ as the final electron acceptor, can oxidize one of two alternative electron donors, either cytochrome c or quinol. Oxidases utilizing cytochrome c are called cytochrome c oxidases, whereas oxidases oxidizing quinol are referred to as quinol oxidases (66). Cytochrome c oxidases cannot directly accept reducing equivalents from quinol. For this purpose, there is an additional, middle component of the respiratory chain between dehydrogenase and oxidase, the cytochrome bc₁ complex, which enables the transfer of electrons from quinol to cytochrome c. Respiratory chains of many aerobic bacteria, such as Paracoccus denitrificans and Azotobacter vinelandii, contain the bc₁ complex and both types of terminal oxidase, cytochrome c and quinol oxidases. For instance, P. denitrificans has one quinol oxidase (ba₃) and two cytochrome c oxidases (aa₃ and cbb₃) (55, 183). A. vinelandii has two quinol oxidases (bo₃ and bd) and one cytochrome c oxidase (cbb₃) (reviewed in references 24 and 164). As shown in Fig. 1, the aerobic respiratory chain of E. coli is much simpler than those of P. denitrificans and A. vinelandii. It lacks the cytochrome bc₁ complex and any cytochrome c oxidase but contains instead two major quinol oxidases, bo₃ and bd (reviewed in references 4, 142, and 164).

Fig. 1Simplified view of the E. coli respiratory chain under aerobic and microaerobic conditions. The two NADH-quinone oxidoreductases called NDH-I and NDH-II and succinate-quinone oxidoreductase (SQR) transfer reducing equivalents to ubiquinone-8 (UQ-8) to yield reduced UQ-8, ubiquinol-8. Two quinol-oxygen oxidoreductases, cytochrome bo₃ (CyoABCD) and cytochrome bd (CydAB), oxidize ubiquinol-8 and reduce O₂ to H₂O. CydAB can also oxidize menaquinol-8. NDH-I, CyoABCD, and CydAB are coupled (Δμ_H⁺ generators), NDH-II and SQR are uncoupled (no Δμ_H⁺ generation). The contributions of the third putative quinol-oxygen oxidoreductase, cytochrome bd-II (CyxAB), to O₂ reduction and Δμ_H⁺ generation are unknown. The energetic efficiency of each enzyme is indicated as the number of protons delivered to the periplasmic side of the membrane per electron (H⁺/e⁻ ratio).

Physiological Functions

Cytochrome bo₃ predominates under high aeration, whereas cytochrome bd is expressed under low oxygen tension (51, 65, 182) (Table 1). It is of interest that the cytochrome bo₃ level increases about 150-fold during aerobic growth but the cytochrome bd level falls only 3-fold, i.e., the change in the cytochrome bd level in response to oxygen is much smaller than that of the cytochrome bo₃ level (51). Both oxidases catalyze the oxidation of ubiquinol-8 to allow cellular respiration with oxygen as the terminal electron acceptor. Cytochrome bd can also oxidize menaquinol-8 (239, 244), which replaces ubiquinol-8 upon a change of growth conditions from aerobic to anaerobic (221).

Both cytochrome bo₃ and cytochrome bd are primary generators of a transmembrane gradient of electrochemical H⁺ potentials (Δμ_H⁺), because the reaction arising from the transfer of reducing equivalents from quinol to O₂ is coupled directly to transmembrane charge separation (21, 22, 98, 109, 134, 175). The energy conserved in the form of Δμ_H⁺ can be used by the E. coli cell for ATP synthesis, the transport of nutrients, and other useful work. Thus, the main function of both oxidases is energy conservation. The two enzymes, however, are different in their bioenergetic efficiencies (transmembrane proton translocation ratios, or the number of protons delivered to the periplasmic side of the membrane per electron [H⁺/e⁻ ratios]), with an H⁺/e⁻ ratio of 2 for cytochrome bo₃ and an H⁺/e⁻ ratio of 1 for cytochrome bd (175) (Table 1). This difference is because cytochrome bo₃ is a true proton pump whereas cytochrome bd is not capable of proton pumping (175). As sources of oxidizing power, cytochrome bo₃ and cytochrome bd can support disulfide bond formation upon protein folding catalyzed by the DsbA-DsbB system (14).

Apart from energy conservation, cytochrome bd endows E. coli with a number of specific physiological functions. Cytochrome bd can serve as an oxygen scavenger and inhibit the degradation of O₂-sensitive enzymes present under anaerobic and microaerophilic conditions (85). In a recent systematic mutational analysis to elucidate the contribution of the respiratory pathways to the abilities of commensal and pathogenic (enterohemorrhagic) E. coli strains to colonize a streptomycin-treated mouse intestine, mutants lacking cytochrome bd failed to colonize whereas cytochrome bo₃ was found not to be necessary for colonization (101).

The cytochrome bd contents increase not only at low oxygen concentrations but also under some unfavorable conditions, such as alkaline pH (9), high temperature (56, 230), the presence of uncouplers-protonophores (9, 26, 27), and low concentrations of cyanide (6) in growth media. Mutants defective in cytochrome bd are sensitive to H₂O₂ (56) and a self-produced extracellular factor that inhibits their growth (50, 128). Mutants that cannot synthesize cytochrome bd are also unable to exit from the stationary phase and resume aerobic growth at 37°C (201, 202). The expression of cytochrome bd, instead of cytochrome bo₃, may enhance bacterial tolerance to nitrosative stress (34, 35, 36).

Inhibitors

Table 2 compares the effects of different inhibitors on the respiratory activities of cytochrome bo₃ and cytochrome bd in E. coli. Inhibitors of the quinol oxidases can be divided into two groups: quinol-like compounds acting at a quinol-binding site(s) and heme ligands (e.g., cyanide, azide, and NO) acting at the oxygen-binding/reducing site. A specific feature of cytochrome bd is that it is much less sensitive to cyanide and azide than cytochrome bo₃ (109) (Table 2). The lower sensitivity of cytochrome bd to anionic heme ligands may be a result of an elevated electron density on the central ion of iron due to the breaking of the circle conjugate π-electron structure in the d-type porphyrin ring and/or may point to a more hydrophobic environment for the O₂-reducing site of cytochrome bd than for that of cytochrome bo₃. It is of interest that the quinolone-type compound aurachin D and its derivatives in submicromolar concentrations specifically inhibit cytochrome bd but virtually do not affect cytochrome bo₃ (133). These outcomes may indicate some differences in the specific structures of quinol-binding sites in cytochrome bo₃ and cytochrome bd. It has been shown very recently that cytochrome bd in E. coli is a bacterial membrane target for a cationic cyclic decapeptide, gramicidin S (50% inhibitory concentration, ~5.3 μM) (Table 2), although it has been generally accepted that the main target of gramicidin S is the membrane lipid bilayer rather than the protein components (143). This finding can provide new insight for the molecular design and development of novel gramicidin S-based antibiotics. The effect of gramicidin S on cytochrome bd and some other membrane-bound proteins may be the alteration of the protein structure through binding to the hydrophobic protein surface (143).

GENETICS

Oxidase Encoding

Cytochrome bo₃.

Cytochrome bo₃ is composed of four different subunits (70, 130, 131) encoded by the cyoABCDE operon (43, 48) (Table 1). The cyoABCDE operon, located at 10.2 min on the E. coli genetic map (12, 43), has been cloned and sequenced (8). Subunits I (75 kDa), II (33.5 kDa), and III (20.5 kDa) of cytochrome bo₃ appeared to be homologous to the counterparts of the eukaryotic and prokaryotic aa₃-type cytochrome c oxidases (48) and were referred to as cyoB, cyoA, and cyoC gene products, respectively, as determined by protein sequencing (138) and other approaches (8, 149, 150). Thus, cytochrome bo₃ is a member of the heme-copper terminal oxidase superfamily (48, 192). The cyoD gene encodes subunit IV (12 kDa) (47, 149, 188), which is homologous to the counterpart in cytochrome c oxidases from gram-positive bacteria and terminal quinol oxidases but unrelated to eukaryotic cytochrome c oxidases (44, 192). The cyoE gene, located at the 3′ end of the cyo operon, encodes no subunit of cytochrome bo₃ but does encode the enzyme which catalyzes the transformation of heme B (protoporphyrin IX, or protoheme) to heme O (uppercase letters in heme designations highlight the chemical nature of hemes, as opposed to protein-bound hemes) by attaching a long farnesyl side chain to the former (141, 186, 187). Heme O is specifically required for the binuclear oxygen-reducing site of cytochrome bo₃. Subunit I (CyoB) carries all three metal redox cofactors: low-spin heme b, high-spin heme o₃, and a copper ion (Cu_B) (66, 119, 137, 150). Heme o₃ and Cu_B form the heme-copper binuclear center for dioxygen reduction. Unlike aa₃-type cytochrome oxidase, subunit II (CyoA) does not have any metal redox cofactors. Subunits III (CyoC) and IV (CyoD) can be removed from cytochrome bo₃ without any loss of catalytic activity (110) but seem to be required for the assembly of the metal redox cofactors in subunit I (142, 189).

Cytochrome bd.

Cytochrome bd is a two-subunit oxidase (136) encoded by the cydAB operon (43, 74, 116). The aforementioned cydAB operon, located at 16.6 min on the E. coli genetic map (12, 43), has been cloned (75) and sequenced (74). The molecular masses of subunit I (CydA) and subunit II (CydB) determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, 57 and 43 kDa (135), respectively, are consistent with those determined based on the corresponding DNA sequences, 58 and 42.5 kDa (74). Neither of the two subunits of cytochrome bd has homology to any subunit of other known respiratory chain oxidases, such as cytochrome bo₃ and cytochrome c oxidases (74, 163). Cytochrome bd contains no copper atoms and does not pump protons (109, 134, 135, 175). Thus, cytochrome bd is not a member of the heme-copper terminal oxidase superfamily. Cytochrome bd subunits carry three hemes: b₅₅₈, b₅₉₅, and d (124, 136). Heme b₅₅₈ is located on subunit I (CydA), whereas hemes b₅₉₅ and d are likely to be in the area of the subunit contact (151). CydA can be expressed and purified without CydB by using mutant strains defective in cydB (76). The purified CydA retains heme b₅₅₈ but lacks hemes b₅₉₅ and d (76). In addition to the cydAB operon, two other genes, cydC and cydD of the cydCD operon, located at 19 min on the E. coli genetic map (71, 166, 170), are essential for the assembly of cytochrome bd (16, 71, 165, 166). CydC and CydD, however, are not subunits of cytochrome bd. It was shown previously that cydCD encodes a heterodimeric ATP-binding cassette-type transporter that is a glutathione transport system (160).

Regulation of Gene Expression

Under high oxygen tension, E. coli expresses cytochrome bo₃ (encoded by cyoABCDE) whereas cytochrome bd (encoded by cydAB) is moderately repressed (51, 65, 182). The expression of the cyoABCDE and cydAB operons is controlled by the two global transcriptional regulators Arc and Fnr (51, 52, 53, 73, 77, 96, 197, 214, 221). Arc is a two-component regulatory system that includes ArcA, a cytosolic response regulator, and ArcB, a transmembrane histidine kinase sensor. ArcA controls several hundred genes (126) and responds to the oxidation state of the quinone pool, which is sensed by ArcB (69). ArcB is activated in the course of the transition from aerobic to microaerobic growth and remains active during anaerobic growth. Upon stimulation, ArcB autophosphorylates and then transphosphorylates ArcA (68, 69). Under microaerobic conditions (i.e., oxygen tension of 2 to 15% of air saturation), the increased level of phosphorylated ArcA represses the cyoABCDE operon and activates the cydAB operon (3).

Another global regulator, Fnr (an oxygen-labile transcription factor regulating hundreds of genes), controls the induction of anaerobic processes in E. coli (107, 157). The Fnr protein has an Fe-S cluster which serves as a redox sensor. The levels of the Fnr protein are similar under both aerobic and anaerobic conditions (17, 77), but the protein is active only during anaerobic growth. The active Fnr protein represses both cyoABCDE and cydAB operons during a transition to anaerobic conditions (i.e., oxygen tension of less than 2% of air saturation) (17, 53, 73).

PROSTHETIC GROUPS

Quinones

Quinones are lipophilic molecules dissolved within the lipid bilayer of the cytoplasmic membrane. The E. coli membrane contains three types of quinones which all have an octaprenyl side chain (C₄₀). These are a benzoquinone, ubiquinone, and two naphthoquinones, menaquinone and dimethylmenaquinone. Both cytochrome bo₃ and cytochrome bd catalyze the two-electron oxidation of ubiquinol-8 to ubiquinone-8 (Fig. 2) (apparent midpoint redox potential [E_m] = +110 mV [221]), coupled to the four-electron reduction of O₂ to H₂O (E_m = +820 mV). Cytochrome bd can also oxidize menaquinol-8 to menaquinone-8 (Fig. 2) (E_m = −80 mV [221]) (239, 244). Ubiquinone-8 is predominant during aerobic growth but is replaced by menaquinone-8 upon the transition from aerobiosis to anaerobiosis (49, 198, 221). It was shown previously that cytochrome bo₃ contains a tightly bound ubiquinone (196). The presence or absence of bound quinone in solubilized cytochrome bd depends on the purification protocol. In some preparations of purified cytochrome bd, there is no quinone (98, 109, 135, 140), whereas others clearly contain bound quinone (18, 22).

Fig. 2Structures of ubiquinone-8, the reduced ubiquinone-8 (ubiquinol-8), menaquinone-8, and the reduced menaquinone-8 (menaquinol-8).

Cytochrome bo3 Prosthetic Groups

Cytochrome bo₃ contains three redox-active metal groups: a low-spin heme b, which is involved in quinol oxidation, heme o₃, and Cu_B; the latter two groups compose a binuclear center, which is the site of the binding of O₂ and its reduction to water. The chemical cofactors are heme b, corresponding to protoporphyrin IX (protoheme, or heme B); heme o₃, corresponding to the protoheme with a long farnesyl side chain attached (heme O); and the Cu_B center, represented by the Cu atom ligated by three histidine residues. In the three-dimensional (3D) structure of cytochrome bo₃, all prosthetic groups were found to be within subunit I (1). In addition to the metal centers, there is also one tightly bound ubiquinone cofactor. Although the X-ray structure did not show any bound quinone in the crystallized enzyme, the site-directed mutagenesis studies identified residues that modulate the properties of the bound quinone (81, 82, 127).

Heme b.

Subunit I of the bo₃ oxidase contains 15 transmembrane helices. Heme b is located between helices II and X at a depth of about one-third of the membrane thickness from the P (positive) side and oriented perpendicular to the membrane plane, such that its propionates are exposed to the P side of the membrane. In both reduced (S [spin quantum number] = 0) and oxidized (S = ¹/₂) states (10, 190), the low-spin heme iron is bound to the four nitrogen atoms of the porphyrin ring and to the two conserved histidines of subunit I (H106_I and H421_I, subscript “I” identifies the subunit where the residue is in). Heme b is a direct electron donor for the catalytic binuclear site. It transfers electrons obtained from the bound ubiquinone. The E_m of this heme is about +280 mV (without redox interaction) (191). The optical spectrum of this heme is rather characteristic for low-spin six-coordinate heme b, with the α-band of the reduced heme at 562 nm in the absolute and the difference (reduced-minus-oxidized) absorption spectra. The maxima of the β- and Soret bands of the reduced-minus-oxidized heme b spectrum are at 530 and 430 nm, respectively. At a high level of cytochrome bo₃ expression, heme B in the low-spin heme site can be replaced by heme O (177). Such heme alteration does not decrease the enzyme activity but lowers the functional K_m of the enzyme for oxygen (228).

Heme o₃.

Heme o₃ is the oxygen-binding heme, and the subscript 3 has been used historically to indicate this feature by analogy to the other heme-copper oxidases. Heme o₃ is a high-spin heme in both the fully reduced ferrous state (S = 2) (11) and the oxidized ferric state (S = ⁵/₂) (46). Depending on the conditions, heme o₃ may be penta- or hexacoordinate (46): the permanent bonds of the heme iron include four bonds with nitrogen atoms of the porphyrin ring and one extra bond with the conserved H419_I from helix X at the same depth in the membrane as heme b. Fivefold coordination of the heme iron leaves one side of the heme empty and available for the binding of ligands such as O₂. This free coordination points towards the Cu_B, together with which heme o₃ forms the bimetallic catalytic site where the reduction of oxygen to water takes place. The spectrum of heme o₃ is characteristic of high-spin hemes. It has a broad α-band centered at 560 nm in the reduced-minus-oxidized spectrum with small extinction and an intense Soret band with the maximum at 430 nm (177). The redox properties of the high-spin heme o₃ are very similar to those of heme b (191). Both of these hemes have a redox potential of about 280 mV when the neighboring heme is oxidized, but the reduction of the neighbor results in a 100-mV lowering of the heme potential (redox interaction).

Cu_B.

The second partner in the binuclear catalytic site is Cu_B. The oxidized Cu_B is a tetragonal center (155, 181); it has three permanent axial histidine imidazole ligands and one mobile oxygen ligand with an exchangeable proton(s) (155, 181). The imidazole ligands originate from H284_I in helix VI and from H333_I and H334_I, both located in a loop fragment between helices VII and VIII. This redox center is often called the invisible Cu site because it does not show any changes in optical spectra upon enzyme reduction and oxidation. Usually, the electron paramagnetic resonance (EPR) signal from the oxidized Cu can be easily detected, but the close proximity of the iron atom of heme o₃ results in strong magnetic interaction and the absence of any detectable spectrum of Cu_B. Such strong magnetic interaction, however, can help to define the E_m of the tetragonal center. The reduction of the Cu ion brakes magnetic interaction with the high-spin heme, and the appearance of a high-spin EPR signal can give information on the reduction level of Cu_B. With such an approach, an E_m of +370 mV for Cu_B was obtained (191).

Bound Ubiquinone.

It has been proposed that the bo₃ oxidase can have two ubiquinone-binding sites with different affinities. The bound ubiquinone in the site with high affinity for ubiquinone (the Q_H site) can be considered to be the enzyme cofactor (195, 196). At the same time, in the low-affinity (Q_L) site, fast exchange of the ubiquinone molecules occurs, and it was proposed that electrons are transferred from the Q_L site to the next electron acceptor (heme b) via the Q_H site (196). A functional study of mutants obtained by site-directed mutagenesis was used to create a model for the possible Q_H binding site, which is located in subunit I, close to heme b (1). According to this model, the Q_H site is predicted to form up to four hydrogen bonds with D75 and R71 to the 1-carbonyl oxygen and with H98 and Q101 to the 4-carbonyl oxygen. EPR spectroscopy has demonstrated that the Q_H site stabilizes a semiquinone anion radical of bound ubiquinone (94, 194).

Cytochrome bd Prosthetic Groups

Cytochrome bd is composed of two subunits, which are typical integral membrane proteins. The subunits carry three metal-containing redox centers, such as two protoheme IX groups (hemes b₅₅₈ and b₅₉₅) and a chlorin molecule (heme d) (Fig. 3), which are proposed to be in 1:1:1 stoichiometry per the enzyme complex, but no copper ion (83, 105, 132, 184, 185). Heme b₅₅₈ appears to be located within subunit I. Both subunits are required for the assembly of heme b₅₉₅ and heme d, suggesting that these two hemes may reside at the subunit interface (151). Both heme b₅₅₈ and heme d are presumed to be oriented with their heme planes perpendicular to the membrane plane. Heme b₅₉₅ is possibly oriented with its heme plane at ~55° with respect to the plane of the membrane (95).

Fig. 3Structures of heme B (protoheme IX), heme O, and heme D (chlorin), which are redox cofactors of cytochrome bo₃ and/or cytochrome bd from E. coli.

Heme b₅₅₈.

Heme b₅₅₈ was shown to be located on subunit I. Although two subunits of the isolated cytochrome bd complex cannot be split apart without denaturing the enzyme, some genetic approaches have allowed subunit I to be synthesized in the absence of subunit II (76). Antibodies directed against subunit I (59, 117), as well as selective proteolysis of this subunit (60, 122), inhibit the ubiquinol oxidase activity of cytochrome bd. These findings suggest that heme b₅₅₈ is associated with subunit I and involved in quinol oxidation. The α-band of the reduced heme b₅₅₈ reveals a peak at 560 to 562 nm in the absolute and the difference (reduced-minus-oxidized) absorption spectra at room temperature. The maximum of the β-band of the reduced heme is at 531 to 532 nm (161). The γ-bands of the oxidized and the reduced forms of heme b₅₅₈ in the absolute absorption spectra are likely characterized by peaks at 414 and 428 to 430 nm, respectively (31). The α- and β-band positions were confirmed upon redox titration by separating the composite difference absorption spectra of the enzyme into the contributions of the individual heme components (112, 124). Low temperature (77 K) shifts all bands by 1 to 4 nm to the blue (161). Heme b₅₅₈ seems to be a low-spin hexacoordinate (29), and amino acid residues H186 and M393 of subunit I were identified as its axial ligands (62, 106, 204). The location of heme b₅₅₈ is predicted to be near the periplasmic surface (156, 242).

Heme b₅₉₅.

The band with a maximum at 595 nm in the difference (reduced-minus-“air-oxidized”) absorption spectrum of cytochrome bd was long ascribed to heme a₁ because of the relatively bathochromic position of the extremum (162). Subsequently, it was established that the prosthetic group of this component is not heme a but protoheme IX (124, 135), whereupon the component has been called heme b₅₉₅. Magnetic circular dichroism (MCD) studies confirmed such a conclusion (29). The difference absorption spectrum of heme b₅₉₅ in the visible region was resolved from a set of composite spectra of the isolated cytochrome bd recorded at different redox potentials (112, 124). It turned out that this spectrum is similar to that of catalases and peroxidases, containing pentacoordinate (high-spin) protoheme IX (124). Heme b₅₉₅ shows an α-band at 595 nm and a β-band near 560 nm in the difference absorption spectrum. A trough at 645 nm in the difference spectrum of heme b₅₉₅ is indicative of the disappearance of a charge transfer to the ligand band, characteristic of the oxidized high-spin heme b, as in the case of peroxidases. The γ-band of ferrous heme b₅₉₅ is characterized by a maximum at ~440 nm, as clearly revealed by femtosecond spectroscopy (229). Heme b₅₉₅ is a high-spin pentacoordinate (29) ligated by the histidine (H19) of subunit I (206) and located on the periplasmic surface (156, 242). The role of heme b₅₉₅ remains obscure. It is proposed that heme b₅₉₅ participates in the reduction of oxygen, forming, together with heme d, a diheme oxygen-reducing site somewhat similar to the heme-Cu oxygen-reducing site in the aa₃- and bo₃-type oxidases (22, 29, 37, 38, 39, 84, 216, 229). In favor of this hypothesis is the finding that the circular dichroism (CD) spectrum of the reduced wild-type cytochrome bd in the Soret band shows strong excitonic interaction between ferrous hemes d and b₅₉₅ (5). Modeling the excitonic interactions in absorption and CD spectra yields an estimate of the Fe-to-Fe distance between heme d and heme b₅₉₅ to be about 10 Å (5). In the opinion of other researchers, the function of heme b₅₉₅ is limited to the transfer of an electron from heme b₅₅₈ to heme d (80, 169). Some authors believe that heme b₅₉₅ can form a second site capable of reacting with oxygen (57, 185).

Heme d.

Heme d is a chlorin-type molecule (213). The α-band of the reduced heme d in the absolute absorption spectrum shows a peak at 628 to 630 nm. However, under usual conditions, heme d is in the stable oxygenated (oxygen-ligated ferrous) form, which is characterized by a band with a maximum at 647 to 650 nm in the absolute absorption spectrum. It is suggested that heme d makes a minor contribution to the γ-band of cytochrome bd (161). Heme d is predicted to be located near the periplasmic surface (156, 242). This heme is the place for capturing and reducing O₂ to H₂O. Being free of external ligands, heme d seems to be in the high-spin state. The protein ligand of heme d is not known. The data on the nature of the heme d axial ligand are controversial. Authors of resonance Raman and electron nuclear double resonance studies have claimed that it cannot be an ordinary histidine, cysteine, or tyrosinate but is either a weakly coordinating protein donor or a water molecule (86, 100, 206). In contrast, EPR studies have indicated that the heme d axial ligand is histidine in an anomalous condition or another nitrogenous amino acid residue (88). Finally, it has been reported recently that the highly conserved glutamate 99 of subunit I may be a candidate for such a role (140).

The millimolar extinction coefficients used commonly for the determination of the E. coli cytochrome bd concentration are listed in Table 3.

Redox Potentials of Hemes in Cytochrome bd

The E_m values of hemes b₅₅₈, b₅₉₅, and d in the E. coli cytochrome bd solubilized in n-dodecyl-β-d-maltoside at pH 7.0 are +176, +168, and +258 mV, respectively (19) (Table 1). Similar values were reported for cytochrome bd contained in bacterial membranes (132, 174, 184), reconstituted in liposomes (125), or solubilized in Tween 20 or Triton X-100, in which the enzyme is active (125). Notably, the E_m value of heme b₅₅₈ can depend on the detergent used for solubilization (125). In particular, octylglucoside and cholate cause a large decrease in the E_m value of heme b₅₅₈ that also correlates with the reversible inactivation of the enzyme (125). The E_m values of all three heme components of cytochrome bd are sensitive to pH values between 5.8 and 8.3, with dependence of −61 mV per pH unit for heme d and −40 mV per pH unit for hemes b₅₅₈ and b₅₉₅, indicating that the reduction of the bd complex is accompanied by enzyme protonation (125).

STRUCTURE OF bo3 AND bd OXIDASES

3D Structure of bo3 Oxidase

The 3D structure of the bo₃ oxidase from E. coli at 3.5-Å resolution was reported in 2000 (1). The structure confirms that the overall architecture of this complex is very similar to those of all other members of the heme-copper oxidase superfamily. The whole integral protein contains 25 transmembrane helices, of which subunit I has 15, subunit II has only 2, subunit III has 5, and subunit IV has 3 helices. All known enzyme cofactors were found in subunit I. The transmembrane helices of this subunit are not perpendicular to the membrane plane but are tilted about 20 to 35° against it. When viewed from the top (P side), they are arranged in an anticlockwise direction and form three semicircular arcs, organized in a quasithreefold axis of symmetry. Three pores are formed in the center of the arcs. One pore houses the binuclear center (heme o₃ and Cu_B) of the oxidase and includes the proton-conductive K channel directed from the binuclear center towards the N (negative) side of the membrane. Another pore retains heme b; the last pore is devoid of cofactors but is used for the proton-conductive D channel.

Heme b is located at a depth of about one-third of the membrane thickness from the P side and oriented perpendicular to the membrane plane, such that its propionates are pointing towards the P side of the membrane (Fig. 4). The heme o₃ is located at the same depth as heme b, the plane of heme o₃ is also perpendicular to the membrane, and the propionates point towards the P side in a manner similar to heme a. Heme b and heme o₃ are facing each other at an angle of 104 to 108^o. At an ~5-Å distance from the iron of heme o₃, a copper atom designated Cu_B is located. In addition to this Cu atom that is coordinated by the three histidines, there is another important structure identified in all heme-copper oxidases and represented by the covalently bound H284_I and Y288. The three histidines and the tyrosine form a conjugated π-electron system around the Cu_B center. Cytochrome bo₃ has been proposed to have two ubiquinone-binding sites, one with high (Q_H) and one with low (Q_L) affinity for ubiquinone (196). It has also been postulated that electrons are transferred from the Q_L site to heme b via the ubiquinone bound at the Q_H site. Site-directed mutagenesis studies (81, 82) have identified residues that modulate the properties of the Q_H site. The model of a Q_H quinone-binding site including R71, D75, H98, and Q101 residues is also supported by the results of X-ray crystallography (1).

Fig. 4Structure of cytochrome bo₃ from E. coli. Only two main subunits are shown: subunit I in gray and subunit II in yellow. Hemes are shown in red (heme o₃ on the right and heme b on the left). The cyan sphere near heme o₃ represents the Cu_B center. The amino acid residues of possible proton-conducting pathways, the D (red-tag) and K (blue-tag) channels, in subunit I are shown. The most likely position of the membrane is depicted by the gray background.

The protein medium itself can not facilitate proton delivery towards the binuclear center or across the membrane, and in order to overcome this limitation, the oxidase has special proton-conductive structures. It is proposed that these structures are based on chains of hydrogen bonds between hydrogen-bonding protein side groups (polar and/or protonatable) and water molecules, whereby the proton is transferred via a Grotthuss type mechanism. At least two proton-conductive channels have been identified primarily by site-directed mutagenesis (89, 211, 212), and these findings were later confirmed by X-ray crystallography (1). One of them is the so-called K pathway, named after the highly conserved lysine 362 (89, 211), which is situated approximately halfway through the channel (Fig. 4). This pathway starts with S315 and S299 and continues through conserved residues K362_I and T359_I towards the hydroxyethyl farnesyl side chain of heme o₃ and Y288_I in the proximity of the binuclear center. The other channel is named D after the highly conserved D135_I (64, 212), which is situated near the surface of the enzyme on the N side. D135_I, together with T211_I, forms a mouth that leads via polar residues N124_I, N142_I, S145_I, T149_I, T201_I, and T204_I to E286_I (Fig. 4), which is an important residue for the proton-pumping mechanism.

Proposed Structure of Cytochrome bd

To date, the X-ray structure of cytochrome bd is not available; however, the findings from conventional studies of the protein topology in the membrane suggest that all three hemes are located near the periplasmic side of the membrane (156, 242). Figure 5 shows topological models of subunits I (CydA) and II (CydB) of cytochrome bd from E. coli (239). Both subunits are integral membrane proteins showing no sequence homology to the subunits of the heme-copper oxidase superfamily (e.g., cytochrome bo₃). Subunit I consists of nine transmembrane helices, with the N terminus in the periplasm and the C terminus in the cytoplasm (156). Subunit II is composed of eight transmembrane helices, with both N and C termini in the cytoplasm (156). Subunit I contains a large hydrophilic domain, which is called the Q-loop, connecting transmembrane helices VI and VII. As shown by many experimental approaches, the Q-loop is involved in quinol binding and oxidation (59, 60, 117, 122, 129, 139). Thus, the quinol-oxidizing site in cytochrome bd is located on the periplasmic side of the membrane.

Fig. 5Proposed topology of the CydA and CydB subunits of cytochrome bd from E. coli. The axial ligands of heme b₅₉₅ (H19) and heme b₅₅₈ (H186 and M393) in the CydA subunit are shown in purple and red, respectively. The protein sequence data have been taken from information available at http://genolist.pasteur.fr/Colibri/. The alignment has been made using the TOPO2 program available at http://www.sacs.ucsf.edu/TOPO-run/wtopo.pl. The model is very similar to that reported in reference 239.

At this point in time, there are about 815 sequences of the cytochrome bd gene available, which allows one to search for highly conserved residues in the corresponding protein (239). It was shown previously that subunit II evolved significantly faster than subunit I, with the result that subunit II exhibits greater sequence diversity (79). A number of residues in subunit I are totally (>99%) conserved in the 815 sequences (239). These residues include H19 (the heme b₅₉₅ axial ligand [206]), H186 and M393 (the heme b₅₅₈ axial ligands [62, 106, 204]), K252 and E257 (involved in quinol binding [139]), R448 (having an unknown function), and E99, E107, and S140 (proposed to be components of a proton channel [33, 156] and important for binding in the heme d-heme b₅₉₅ diheme site [140, 239]). Slightly less conserved (95 to 99%) are E445 (required for charge compensation at the heme b₅₉₅-heme d oxygen-reducing site upon the full reduction of oxygen by two electrons [22]), N148 (a plausible component of a proton channel), and R9 (having an unknown function) (239). Somewhat less conserved (~85%) are R391 (which stabilizes the reduced form of heme b₅₅₈ [243]) and D239 (having an unknown function); however, these residues are totally conserved within the group of cytochromes bd with a long Q-loop, to which the E. coli enzyme belongs (239). Other conserved residues are glycines, prolines, phenylalanines, and tryptophans, which may play structural roles. There is only one totally (>99%) conserved residue (W57) in subunit II (239). Residues R100, D29, and D120 of subunit II are totally conserved within the family of long-Q-loop cytochromes bd, and the residue at position 58 in subunit II (according to the numbering for the E. coli enzyme) is either an aspartate or a glutamate. The N-terminal portion of subunit II is thought to be involved in the binding of heme d and heme b₅₉₅ (152, 239).

Since the active site of O₂ reduction is located near the periplasmic surface and protons for H₂O production are taken from the bacterial cytoplasm, there must be at least one transmembrane proton-conducting pathway to convey protons from the cytoplasm to the heme b₅₉₅-heme d site (22, 33, 98, 156) (Fig. 6).

Fig. 6Schematic model of electron and proton transfer pathways in cytochrome bd from E. coli. There are two protonatable groups, X_P and X_N, redox coupled to the heme b₅₉₅-heme d active site. A highly conserved residue, E445, was proposed to be either the X_P group or the gateway in a channel that connects X_P with the cytoplasm or the periplasm (22). A strictly conserved E107 residue is a part of the channel mediating proton transfer to X_N from the cytoplasm (33). X_b, a group at the periplasmic side of the membrane that picks up and releases a proton as heme b₅₅₈ is reduced and oxidized.

COMPARISON OF AFFINITIES OF THE TWO OXIDASES FOR OXYGEN

In E. coli, the affinity of cytochrome bd for oxygen {dissociation constant for O₂ [K_D(O₂)], 0.28 μM (20)} is about 1,000-fold higher than that of cytochrome bo₃ [K_D(O₂), >0.3 mM (208)], which allows us to consider the bd and bo₃ enzymes as the high- and low-affinity oxidases, respectively. Such a striking difference in the K_D(O₂) values correlates with the following facts.

1. Cytochrome bo₃ predominates under high aeration, whereas cytochrome bd is expressed maximally under low aeration (51, 65, 182).

2. A peculiar feature of cytochrome bd is that it is purified mainly as a stable oxygenated (oxy)complex (heme b₅₅₈³⁺-heme b₅₉₅³⁺-heme d²⁺-O₂) characterized by an absorption peak at 645 to 650 nm (40, 104, 123, 167). The fact that a stable oxycomplex can be generated by the reversible binding of oxygen to the one-electron-reduced cytochrome bd can be used for direct measuring of the K_D(O₂) of cytochrome bd (20). This is not the case for cytochrome bo₃ or any other heme-copper oxidase, which in any redox state under normal conditions does not form a stable oxycomplex; therefore, the K_D(O₂) for cytochrome bo₃ can be determined only indirectly (208). For cytochrome bo₃, the K_D(O₂) is more than 100-fold higher than the apparent K_m for O₂, which allows us to conclude that the bo₃ oxidase is designed to trap O₂ kinetically by reducing it to an oxoferryl species (228). Due to its ability to function efficiently under microaerobic conditions, cytochrome bd is required for commensal and pathogenic E. coli strains to colonize mouse intestine (101). It turns out that E. coli mutants lacking cytochrome bd, with high affinity for O₂, are eliminated by competition with wild-type strains competent in respiration and that cytochrome bo₃, with low affinity for O₂, is not necessary for colonization (101). The colonization defects of the cytochrome bd mutants challenge the traditional view that the intestine is anaerobic (13) but support the hypothesis that a microaerobic niche is critical for both establishing and maintaining E. coli in the intestine (101).

BINDING OF LIGANDS (OTHER THAN OXYGEN)

Ligand Binding to the Cytochrome bo3 Binuclear Center

Ligands that bind to the binuclear center of cytochrome bo₃ can be divided into two classes: uncharged molecules like CO and NO, which preferably bind to the binuclear center in the reduced form and induce the transition of the high-spin heme to the low-spin state, making it six coordinate, and ionizable molecules, like HCN and NaN₃, or formate, which preferably bind to the oxidized heme. The binding of the second group of ligands can result in a different spin state for heme o₃. Some of the ligands can bind either to heme o₃ only or to a site between the two metals forming the binuclear center. The dynamics of ligand exchange can be used to characterize the possible dynamics of the binding of the substrate and the partial products of the reaction generated during the catalytic cycle.

CO and NO binding.

The molecules of CO and NO mimic the oxygen molecule and can be bound to the catalytic site of cytochrome bo₃. This binding occurs with the high-spin heme of the reduced binuclear center. Carbon monoxide reacts with the reduced enzyme with a stoichiometry of 1:1, and the K_D for this reaction was determined to be 1.7 × 10⁻⁶ M (46). The value of a second-order rate constant for association (k_on) is 6.1 × 10⁴ M⁻¹·s⁻¹.

The CO ligation of heme o₃ results in a blue shift of the heme absorption bands. The characteristic CO-binding spectrum has two small peaks in the visible part of the spectrum at 530 and 570 nm and a pronounced spectral shift in the Soret region, with the maximum λ of ~415 nm and the minimum λ of ~430 nm.

The photolysis of the reduced CO-bound enzyme at low temperatures results in the dislocation of the iron-bound CO to Cu_B, where it can be recognized by the specific C≡O stretch at ~2,065 cm⁻¹ due to CO bound to copper (2). At room temperature, the photodissociation of CO from the heme iron and its subsequent binding to Cu_B is an extremely fast reaction, and then CO remains bound to Cu_B for only a short time (two-component dissociation with the time constants of ~14 and 140 μs [15]). After carbon monoxide dissociation from the binuclear site, the rebinding to the heme iron via Cu_B occurs at a much lower rate (238). Yoshikawa et al. (241) have reported the crystallographic structures of the reduced bovine enzyme in the presence and absence of CO. While this X-ray study did not find any significant changes in the structure of Cu_B relative to the fully oxidized bovine enzyme, the extended X-ray absorption fine-structure (181) investigation of the first shell of Cu_B showed a dramatic change upon the addition of CO, which involves the dissociation of one of the Cu_B histidine ligands and its possible replacement by a chloride ion.

Fully reduced cytochrome bo₃ also binds NO with a stoichiometry of 1:1 and very high affinity (K_D < 10⁻⁸ M) (42), forming the ferrous nitrosyl (Fe²⁺-NO) species, as determined by EPR spectroscopy (46). The heme o₃-NO complex yields well-resolved EPR signals from the ¹⁴N atoms of both NO itself and the proximal histidine ligand of heme o₃, showing nuclear hyperfine coupling (46).

Reaction of Cytochrome bo₃ with Cyanide.

Cyanide reacts almost exclusively with oxidized cytochrome bo₃. The reaction is manifested by a red shift of the Soret band from approximately 411 to 415 nm (93). A characteristic feature of the ligand-binding reaction in the α-band region is the loss of the broad charge transfer band at 630 nm. This band is attributable to the fully oxidized binuclear center in which heme o₃ is in a high-spin state. The results of EPR studies (93) seem to indicate that these absorption changes occur because the ferric high-spin heme o₃ becomes a ferric low-spin ligand complex. Due to the magnetic coupling with Cu_B²⁺, the EPR spectrum (g_z [constant of magnetization] = 3.3) is observed only when the copper of the binuclear center is reduced. This signal is characteristic of the conversion of the ferric high-spin heme into the low-spin state by the binding of strong-field ligands such as cyanide (222). Cytochrome bo₃ binds a single equivalent of cyanide (K_D, 1 × 10⁻⁶ to 2 × 10⁻⁶ M) with monophasic kinetics (232) and k_on of 37 M⁻¹·s⁻¹ (at pH 6.0) (144). This rate constant is slightly pH dependent and increases about 1.8-fold over the pH range between 6.0 and 8.5. (144). Room-temperature MCD spectra in the near-infrared region (45), results from infrared and EPR studies (219), and also resonance Raman detection (159) of the product of the reaction of the oxidized cytochrome bo₃ with cyanide have led us to propose the bridging structure of the cyanide complex to be as follows: Fe_o3³⁺—C=N—Cu_B²⁺, where Fe_o3 is the iron of heme o₃. It was also shown that the reduction of the enzyme results in the release of the Cu_B ligation and the formation of an Fe_o3²⁺—C=N moiety.

Reaction of Cytochrome bo₃ with Azide.

Azide binds to cytochrome bo₃ with a stoichiometry of 1:1; the K_D for this reaction is about 2 × 10⁻⁵ M (121). Contrary to the addition of cyanide, the addition of azide to the oxidized isolated enzyme causes a relatively rapid but small blue shift in the Soret band from approximately 411 to 409 nm. Simultaneously, the broad charge transfer band at 630 nm becomes more pronounced and shifts its maximum to 640 nm. These small changes induced in the electronic absorption spectrum are consistent with heme o's becoming hexacoordinate but remaining high spin, which can also be seen by MCD spectroscopy in the range of 350 to 1,100 nm (45, 121). The kinetics of azide binding is an order of magnitude faster than that observed for the binding of cyanide. The calculated k_on for the binding of azide to cytochrome bo₃ is about 800 M⁻¹·s⁻¹ at pH 7.5. The k_on shows a marked increase upon acidification, indicating that the active species is electroneutral hydrazoic acid. Analyses of EPR, electronic, and MCD spectra (232) were used to prove that, unlike cyanide, azide does not bind to heme o₃ but rather to the Cu_B site, whereas the resolved 3D structure of the bovine cytochrome c oxidase in the presence of azide revealed the bridging structure of the complex (Fe_a3³⁺—N=N=N—Cu_B²⁺) (241). The Fourier transform infrared (FTIR) study of the bovine enzyme complex (215) showed a major infrared band at 2,051 cm⁻¹. Subsequently, an FTIR spectroscopic study of the E. coli bo₃ oxidase (217) showed an infrared band at 2,041 cm⁻¹, which was assigned to the bridging structure. Discrepancy in the ligation geometry found by the different methods was resolved by detailed analysis of the FTIR spectroscopy of the complex of cytochrome bo₃ with asymmetrically ¹⁵N-labeled azide (218). The experiments showed time-dependent evolution of the geometry of azide binding. In the air-oxidized form, a major infrared azide antisymmetric stretching band corresponds to the bridging geometry. An additional band developing at 2,062.5 cm⁻¹ during longer incubation reflects the appearance of the Cu_B²⁺—N=N=N structure. In addition, the partial reduction of the oxidase with β-NADH caused the appearance of new infrared bands indicating the emergence of the Fe_o³⁺—N=N=N configuration (218).

Ligand Binding to Cytochrome bd

Since hemes d and b₅₉₅ in cytochrome bd are in the high-spin pentacoordinate state, these redox centers can potentially bind ligands. It has to be expected that the reduced form of the bd enzyme can bind mainly electroneutral molecules like O₂, CO, and NO and that the oxidized cytochrome bd binds ligands in the anionic form, such as cyanide and azide. It appears that heme d binds ligands readily but that the ligand reactivity of heme b₅₉₅ is minor (29, 32, 39). It was reported previously that heme b₅₅₈, although a low-spin hexacoordinate, may also bind ligands (e.g., CO and cyanide) to some extent (29, 32). Such marginal reactivity is due possibly to the weakening of the bond of the methionine axial ligand (M393) with the heme b₅₅₈ iron caused by the isolation procedure and/or protein denaturation (32).

CO Binding.

CO brings about a red shift of the heme d band, with a maximum at 643 to 644 nm, a minimum near 624 nm, and a peak at 540 nm. In the Soret band, CO binding to fully reduced cytochrome bd results in an absorption decrease and minima at 430 and 442 to 445 nm. Absorption perturbations in the Soret band and at 540 nm occur in parallel with the changes at 630 nm and reach saturation at 3 to 5 μM CO. The peak at 540 nm is probably either the β-band of the heme d-CO complex or part of its split α-band (32). A peculiar W-shaped curve in the Soret region of the difference spectrum can be caused by a small band shift for unligated heme b₅₉₅ induced by CO interaction with the nearby heme d (38, 229). Only a small fraction (<5%) of heme b₅₉₅ in cytochrome bd binds CO at room temperature or low temperatures (from −70 to −100°С) (29, 39). CO binding with about 15% of heme b₅₉₅ in the membrane-bound cytochrome bd at a cryogenic temperature (4 K) was observed with the aid of FTIR spectroscopy (84). The apparent K_D for the CO complex with fully reduced cytochrome bd appeared to be ~80 nM (32). This value is markedly higher than that for cytochrome bo₃ (1.7 μM) (46). The fully reduced cytochrome bd can form a photosensitive heme d-CO complex, and following flash photolysis, CO recombines with ferrous heme d proportionally to the CO concentration, with k_on of 8 × 10⁷ M⁻¹·s⁻¹ (83) (Table 4). This value is much higher than that for cytochrome bo₃ (6.1 × 10⁴ M⁻¹·s⁻¹) (46).

CO can also react with one-electron-reduced oxygen-free cytochrome bd of E. coli, forming a mixed-valence (MV) CO compound in which both b-type hemes remain oxidized (heme b₅₅₈³⁺-heme b₅₉₅³⁺-heme d²⁺-CO) (36, 38, 229). The heme d α-band is also positioned at 635 to 636 nm in the absolute absorption spectrum.

It was proposed that the redox state of the b-type hemes in cytochrome bd, presumably that of heme b₅₉₅, controls the status of the pathway for ligand transfer between heme d and the bulk phase (36, 229). Two observations allowed us to draw such a conclusion.

1. Flash photolysis of the CO complex of the fully reduced cytochrome bd results in the complete photodissociation of the CO molecule into the bulk. If the experiment is repeated with the MV CO complex, a significant part of the CO flashed off heme d²⁺ (up to 70%) gets trapped inside the protein and undergoes geminate recombination with heme d²⁺ on a subnanosecond time scale (229).

2. The apparent off rate constant for the spontaneous dissociation (k_off) of CO from heme d²⁺ is markedly lower for the MV state of cytochrome bd than for the fully reduced state of the bd oxidase (36) (Table 4).

Interaction with Some Nitrogen-Containing Ligands.

A number of small nitrogen-containing molecules can react with fully reduced cytochrome bd from E. coli. NO₃⁻, NO₂⁻, N₂O₃²⁻ (trioxodinitrate), NH₂OH, and NO, when added to membranes containing cytochrome bd or the purified enzyme, give the decrease in amplitude and shift the 630 nm band of ferrous heme d to 641 to 645 nm (28, 29, 34, 36, 88, 91, 92, 102, 185). The common peak position was interpreted as all ligands’ yielding the same or very similar heme-nitrosyl compounds (102). A red shift of the α-band of heme d²⁺, observed upon the interaction of nitrite with the fully reduced membrane-bound cytochrome bd, was accompanied by the slower formation of a trough at 438 nm in the difference (nitrite-treated-minus-reduced) absorption spectrum. These changes in the Soret band were ascribed to the formation of the product of the interaction of heme b₅₉₅ with nitrite (heme b₅₉₅²⁺-NO) (185).

Cytochrome bd can also form a stable complex with NO in an MV state, in which ligand-bound heme d is reduced (to heme d²⁺) while the other two hemes (b₅₅₈ and b₅₉₅) are oxidized (36, 88). The rates of NO dissociation from heme d²⁺ of cytochrome bd in both the fully reduced and MV states were determined previously (36). In the fully reduced state, NO dissociates from heme d²⁺ at an unusually high rate, 0.133 s⁻¹ (36), which is ~30-fold higher than k_off measured for the ferrous heme a₃ of the mitochondrial cytochrome c oxidase, 0.004 s⁻¹ (193). These data are consistent with the proposal that in heme-copper oxidases, Cu_B acts as a gate, controlling ligand binding to the heme in the active site (120). Another remarkable feature of NO dissociation from cytochrome bd is that the k_off value for the MV state (0.036 s⁻¹), although still quite high, is significantly lower than that measured for the fully reduced enzyme (36) (Table 4). The same effect was observed with CO (see above) (36). These data show that the redox state of heme b₅₉₅ controls the kinetic barrier for ligand dissociation from the active site of cytochrome bd. The unusually high rate of NO dissociation from cytochrome bd may explain the observation (34) that the NO-poisoned cytochrome bd recovers respiratory function much more rapidly than a heme-copper oxidase. It is postulated that the expression of bd-type instead of heme-copper-type oxidases enhances bacterial tolerance to nitrosative stress, thus promoting the colonization of host intestine or other microaerobic environments (36).

Reaction with Cyanide.

An earlier spectrophotometric study of the reaction of cytochrome bd from E. coli with KCN was carried out with the membrane vesicles (173) and was confined to measurements in the α-band. That work revealed a decay of the absorbance at 650 nm induced by cyanide, and this finding was interpreted at that time to represent the disappearance of the free form of ferric heme d (173). This result has been reinterpreted later as the decay of the ferrous heme d oxycomplex (161). Cyanide, interacting with the oxygenated form of the isolated cytochrome bd, brings about significant absorption changes in the γ-region: a maximum at 434 to 437 nm and a minimum near 405 to 410 nm in the difference (KCN-treated-minus-oxygenated) spectrum. There is also considerable increase in the MCD signal in the Soret region. These data were interpreted to indicate the ligand-induced transition of high-spin heme b₅₉₅ to the low-spin state (118). Subsequently, a simple and fast method for the conversion of the oxygenated enzyme into the fully oxidized form with the use of lipophilic electron acceptors was developed (40). This approach enabled researchers to study the interaction of cyanide and other ligands with the homogenous oxidized preparation of cytochrome bd. It was found that the addition of KCN to the fully oxidized cytochrome bd brings about some absorption changes in the visible range of the difference absorption spectrum (the 624-nm peak is most pronounced) (29). These changes suggest the reaction of the ligand with heme d. A typical bathochromic shift of the γ-band is also observed. While the changes in the visible region are virtually saturated at 0.5 mM KCN, the Soret band effect continues to grow, indicating a second low-affinity ligand-binding site (29). The MCD spectrum of the fully oxidized cytochrome bd is dominated by an asymmetric signal in the Soret region. Submillimolar cyanide has no effect on the initial MCD spectrum. KCN at 50 mM induces minor changes of the MCD signal in the Soret band, which can be modeled as the transition of a part of the low-spin heme b₅₅₈ (15 to 20%) to its low-spin cyanocomplex (29). There is no evidence of the interaction of high-spin ferric heme b₅₉₅ with the ligand. The apparent discrepancy between data on the interactions of cyanide with oxygenated (118) and fully oxidized (29) forms of the bd enzyme may derive from the fact that in the former case there seemed to be partial reduction of heme b₅₉₅ associated with ligand-induced electron transfer from heme d rather than a change of the heme b₅₉₅ spin state. On the basis of EPR spectra, Tsubaki et al. (216) proposed that the treatment of air-oxidized cytochrome bd with cyanide results in a cyanide-bridging species with a heme d³⁺—C=N—heme b₅₉₅³⁺ structure. However, Tsubaki et al. (216) did not account for the electron released from heme d upon cyanide binding to as-prepared cytochrome bd.

Interaction with H₂O₂.

The addition of excess H₂O₂ to E. coli membranes containing cytochrome bd (168) and the purified enzyme in the as-prepared (37, 123) or the fully oxidized (30, 37, 98) state gives rise to an absorption band at ~680 nm. The reaction of H₂O₂ with fully oxidized cytochrome bd also induces a bathochromic shift of the γ-band (30, 37). H₂O₂ binds to ferric heme d with an apparent K_D of 30 μM, but it seems not to interact with heme b₅₉₅ (30, 37). The fully ferric cytochrome bd reacts with peroxide with k_on of 600 M⁻¹·s⁻¹. The decay of the H₂O₂-induced spectral changes upon the addition of catalase (at a rate of ~10⁻³ s⁻¹) is about 20-fold slower than expected for the dissociation of H₂O₂ from the complex, with heme d assuming a simple reversible binding of peroxide (k_off = K_D × k_on ~ 2 × 10⁻² s⁻¹) (30). This finding suggests that the interaction of H₂O₂ with cytochrome bd is essentially irreversible, giving rise to the oxoferryl state of heme d (30). The assignment of compound 680 to the oxoferryl state of heme d is confirmed by resonance Raman spectroscopy data (105). The results of resonance Raman spectroscopy studies suggest that heme d is in the high-spin pentacoordinate state when it is oxygenated (Fe²⁺-O₂) or exists as an oxoferryl species (Fe⁴⁺-O²⁻) or in a compound with cyanide (207).

PROPOSED MECHANISMS OF FUNCTIONING

Mechanism of Cytochrome bo3 Functioning

Cytochrome bo₃ catalyzes the final step of E. coli respiration—the oxidation of ubiquinol-8 and the reduction of molecular oxygen. The reduction of one dioxygen molecule to water requires four electrons, which are supplied by two molecules of ubiquinol on the P side, and four protons, taken up from the N side of the membrane. The reduction of oxygen to water is an exergonic process, coupled with the release of large amounts of energy. This energy is conserved in the form of Δμ_H⁺. The formation of Δμ_H⁺+ by cytochrome bo₃ is based on two principles: vectorial chemistry and proton pumping. Since the protons and electrons for oxygen reduction to water are taken from different sides of the membrane, the reduction results in the net transfer of four charges across the membrane. At the same time, the enzyme is able not only to catalyze the oxygen reduction but also to utilize the released energy for proton pumping. This process was discovered in 1977 (234) by the demonstration that the reduction of molecular oxygen to water by mitochondrial cytochrome c oxidase is linked to the pumping of additional four protons across the membrane dielectric (234). Later, such functional ability was shown for cytochrome bo₃ (176). Hence, the overall reaction catalyzed by cytochrome bo₃ can be described by the following equation:

2QH₂ + 8H_N⁺ + O₂ → 2Q +2H₂O + 8H_P⁺

where Q is ubiquinone, QH₂ is ubiquinol, H_N⁺ is proton taken from the N side of the membrane (the cytosol), and H_P⁺ is proton released to the P side of the membrane (the periplasmic space). The mechanism coupling electron transfer reactions with the transmembrane proton translocation is still under debate. Let's look through the main elements of this mechanism.

Electron Transfer Reactions in Cytochrome bo₃.

The general sequence of electron transfer in the bo₃ oxidase is well-established. The ubiquinone molecule occupying the Q_H binding site has a stable semiquinone form (240), which ensures the coupling of a one-electron redox reaction to a two-electron donor. A pulse radiolysis study showed that the quinone bound at the Q_H site is important for the rapid reduction of heme b but not for rapid electron transfer from heme b to the heme o₃-Cu_B binuclear center (111). The rate constant of electron transfer between semiquinone and heme b was found to be 1.5 × 10³ s⁻¹ (111).

In the next step, the low-spin heme delivers electrons to the binuclear site in a controlled fashion that is coupled to proton translocation across the membrane (23). The rate of this electron transfer without coupling to the reaction with protons has been studied extensively, in particular by the photodissociation of CO from the oxygen-binding heme in the so-called mixed valence form of the enzyme, where only the binuclear-site metals are initially reduced (99, 146). (Note that the term “mixed valence” means two-electron-reduced form for cytochrome bo₃ but one-electron-reduced form for cytochrome bd.) Primarily, it was found that this rate is about 3 μs (3 × 10⁵ s⁻¹), which is too low to correspond to the pure electron tunneling, especially since this rate is independent of pH and substitution with heavy water (78, 153) and therefore may not be presumed to be linked to proton transfer. The obtained value is, however, about three orders of magnitude lower than the value predicted by the empirical electron transfer theory (148, 158). The results of recent femtosecond experiments show that the electron transfer from CO-dissociated ferrous heme o₃ to the low-spin ferric heme b takes place at a rate of 8.3 × 10⁸ s⁻¹ (1.2 ns).

The final step of electron transfer reactions between heme o₃ and Cu_B has not been a subject for experimental determination, but from our modern understanding of electron transfer reactions, the rate of transfer between the two metal atoms in the binuclear catalytic site can be predicted to be in the order of picoseconds, taking into account the very small distance (~5 Å) between these atoms.

The suggested electron transfer sequence in cytochrome bo₃ can be described by the following equation:

Q_LH₂ ↔ Q_H ↔ b ↔ [heme o₃-Cu_B] → O₂

Proton Transfer Reactions in Cytochrome bo_3.

The proton transfer pathways in cytochrome bo₃ have been much less investigated than the electron transfer pathways. There are no well-identified places for proton localization during the transfer of protons across the membrane, except the aspartate residue at position 286 (210, 224, 233) in the middle of the membrane. The quality of the X-ray structure of the bo₃ oxidase was not sufficient to resolve individual water molecules in the proton-conducting channels. However, a very high level of structural homology to the other members of the heme-copper superfamily allows us to draw some conclusions about proton movements in the protein milieu including structural water arrays (180, 199) in two well-defined proton channels. Both of these channels serve as proton delivery pathways and cross about two-thirds of the membrane dielectric from the cytoplasmic side to the binuclear catalytic site. The functional separation of the channel is not yet completely clear. Originally, models of the proposed molecular mechanism of proton pumping predicted the existence of two proton-conductive structures with different functions. One channel was proposed to be responsible for the translocation of “pumped” protons and the other for the uptake of “chemical” protons for water formation (147, 235). The resolved structure revealed the existence of two such channels (named the D and K channels), and originally (97), it was proposed that the D channel was responsible for the translocation of pumped protons and that the K channel was used for the uptake of chemical protons for water formation. However, more recent results indicate that the D channel is involved in the uptake not only of all four pumped protons but also of two chemical protons used in the oxidative part of the catalytic cycle (115) and that the K channel is responsible for the uptake of another two chemical protons during the reductive part of the cycle (90, 236). The proton translocation mechanism also requires two proton exit channels—one for proton release upon the oxidation of bound ubiquinol and the other for the release of the pumped protons. The 3D structure did not show clear exit pathways. However, the exit pathways should be much shorter than the D and K channels, only one-third of the membrane dielectric. In addition, in the well-resolved structures of bovine (220) and Rhodobacter sphaeroides (180) aa₃ oxidases, areas rich in structural water molecules, which can serve as exit channels, were found above the heme propionates. The results of site-directed mutagenesis studies suggested that the exit for pumped protons may start at conserved residues R481 and R482 (179), which are hydrogen bonded to the Δ-propionates of the hemes, and then continue further through the chains of mobile water molecules.

Cytochrome bo₃ Catalytic Cycle.

The catalytic cycle for the reduction of oxygen to water is a rather complex process. One enzyme turnover includes the delivery of four electrons and four protons to the catalytic site, the binding of oxygen in this site, the translocation of four protons across the membrane, and the release of the product (two water molecules). This complexity is the reason why the real-time measurement of a single catalytic cycle shows a large number of intermediate states of the enzyme.

The reaction starts from the very fast (k_on = 1.6 × 10⁸ M⁻¹·s⁻¹ [208, 227]) formation of an oxygen adduct, so-called compound A, which is characterized by an oxygen molecule bound to a high-spin heme o₃, as in hemoglobin.

Unlike that of hemoglobin, the lifetime of this intermediate is very short. In 24 μs (227), the bound oxygen accepts electrons from Cu_B and heme b and a proton, which results in O—O bond splitting and the formation of a peroxy intermediate. This compound was named “peroxy” because it can be produced by an oxidase containing only two electrons per enzyme molecule, which formally corresponds to the reduction of O₂ to peroxide. But more recent examination by kinetic resonance Raman spectroscopy (172) and mass spectrometry (61) clearly demonstrated that the oxygen-oxygen bond is already broken and that heme o₃ is in the oxoferryl state (Fe_o3⁴⁺=O) with another oxygen atom being bound to Cu_B as a hydroxide ion. The “peroxy” state can also be generated directly in the reaction of the oxidized bo₃ with H₂O₂ (145), and the resulting spectrum has a characteristic peak at 582 nm and a shoulder at 550 nm.

Upon the arrival of the next electron and the proton, the “peroxy” form (208) is converted into a ferryl intermediate. The results of time-resolved resonance Raman studies showed the formation of a ferryl intermediate with a rate constant of about 2 × 10⁴ s⁻¹ (87, 231). A very similar rate was also detected by visible spectroscopy (154, 178, 208). The stable ferryl intermediate can be obtained as the end product in the reaction of the fully reduced enzyme with oxygen when no bound ubiquinone is in the Q_H site (178). Such an enzyme contains only three of the four electrons required for complete O₂ reduction, and so the catalytic cycle stalls at the ferryl state. In the visible spectroscopy, this state is characterized by a spectrum with peaks at 557 and about 420 nm.

The presence of bound ubiquinone at the Q_H site of cytochrome bo₃ (195) increases the electron capacitance of the enzyme and allows the reaction to proceed further to yield the oxidized form. The rate of formation of the oxidized state in the single-turnover experiments was estimated to be 0.3 × 10³ s⁻¹ by recording the electron and proton transfer reactions (209).

Mechanism of Δμ_H⁺ Formation by Cytochrome bo₃.

The oxidation of ubiquinol (E_m of redox couple Q/QH₂ ~ +0.1 V) by oxygen (E_m of redox couple O₂/H₂O ~ +0.8 V) is linked with free energy release in the order of ~0.7 V. The proton motive force created by the respiratory chain on the E. coli cellular membrane is about −0.2 V (108, 225). It is clear that excess free energy (~0.5 V) can be used for the translocation of more than one charge across the membrane. Indeed, measurement of the stoichiometry (175) of the proton transfer by the bo₃ oxidase on the E. coli cellular membrane showed that two protons are transferred per each electron used for the O₂ reduction. Of these two charges, the first is driven by the vectorial chemistry and the second is driven by the proton pump (176). The vectorial chemistry includes the oxidation of the ubiquinol molecule and the reduction of dioxygen, with the release and uptake, respectively, of one proton per electron. The charge separation in this case is achieved by the separation in the space of these two events.

The protons extracted from QH₂ due to its oxidation by heme b are released to the P side, and the protons for the O₂ reduction in the binuclear site are taken up from the N side of the membrane. Because heme b and the binuclear site are located at the same depth in the membrane, the release and uptake of the protons correspond to the translocation of one complete charge across the membrane. A second charge is transferred by the proton pump, the mechanism of which is still unclear. Several proposals for the proton translocation machinery of the heme-copper oxidases have been presented during the last 30 years since the discovery of this machinery in mitochondria (234) and bacteria (176, 223), but a definitive explanation of how it functions has not yet been presented. A high level of structural homology between different members of the heme-copper oxidase superfamily argues for similarity in the molecular mechanism of the transmembrane proton translocation. For cytochrome c oxidase, it was shown that, in the continuous-turnover regimen, the catalytic cycle consists of four sequential proton translocation steps (25, 226). During each step, the delivery of an electron to the binuclear site initiates a proton pump cycle, which is likely to occur by essentially the same mechanism every time when an electron arrives. There are a number of models explaining how the coupling between electron transfer and proton translocation occurs (63, 171, 200, 203, 237). Despite the different postulations regarding the drivers of these processes, the models are quite similar and employ electrostatic coupling between the movements of electrons and protons in the low-dielectric medium of the membrane protein. The delivery of an electron to the binuclear center drives proton pumping across the hydrophobic barrier. This translocation is followed by the uptake of the chemical proton to the active site, which leads to the release of the pumped proton out of the protein at the P side.

Mechanism of Cytochrome bd Functioning

Under physiological conditions, cytochrome bd can oxidize ubiquinol-8 and menaquinol-8. In vitro, the bd enzyme can also utilize shorter-chained ubiquinols, menadiol, duroquinol, and artificial electron donors such as N,N,N',N'-tetramethyl-p-phenylendiamine (TMPD) (in the presence of excess ascorbate). Of the in vitro substrates, ubiquinol-1 (plus excess dithiothreitol) shows the highest turnover numbers (103, 125). The apparent K_m values for some reductants are shown in Table 1. As noted above, the activity of the purified bd enzyme strongly depends on the nature of the detergent in which the enzyme is solubilized. Cytochrome bd is inactive in octylglucoside or cholate but shows high activity in Tween 20 or Triton X-100 (125) or N-lauroyl-sarcosine (34). The ubiquinol-1 oxidase activity of cytochrome bd has a broad optimum above pH 7.5 but decreases at more acidic pH values (125). Cytochrome bd possesses three distinct active sites—for quinol oxidation, TMPD oxidation, and oxygen reduction. All the three sites seem to be located at or close to the periplasmic surface of the membrane. Electrons donated from quinol transfer to heme b₅₅₈ and then to the b₅₉₅-d diheme site, whereas electrons donated from TMPD transfer directly to the b₅₉₅-d site, bypassing the quinol-binding site and heme b₅₅₈ (122).

Mechanism of Δμ_H⁺ Formation by Cytochrome bd.

Cytochrome bd was shown to generate a transmembrane electric potential both in single turnover (21, 22, 98) and under multiple-turnover conditions (109, 113, 134) but without invoking a proton pump (H⁺/e⁻ ratio, ~1 [136, 175]; q/e⁻ ratio [the number of charges translocated across the membrane per electron], ~1 [235]). When reconstituted into liposomes, cytochrome bd generates an uncoupler-sensitive transmembrane voltage difference with a value of 160 to 180 mV (negative inside) (109, 134). The ubi(mena)quinol molecule generated by the dehydrogenases of the respiratory chain can diffuse laterally within the bilayer, finding its way into the quinol oxidase site located near the outer side of the membrane. Upon the oxidation of a quinol, two protons are released into the periplasmic space and two electrons are transferred through heme b₅₅₈ to the heme b₅₉₅-heme d oxygen-reducing site also located near the periplasmic surface of the membrane. The four protons used for O₂ reduction are taken up from the cytoplasm. Single-turnover electrometric experiments showed that membrane potential generation is associated with electron transfer from heme b₅₅₈ to the b₅₉₅-d active site (21, 22, 98). However, since all of the three hemes are likely to be located at the same depth of the membrane, close to the periplasmic side (156, 242), the electron transfer itself cannot be electrogenic (98). Rather, vectorial proton movement from the cytoplasm towards the active site on the opposite (periplasmic) side of the membrane, coupled with a redox reaction between hemes b and d, must occur (21, 22, 98). The latter means that there must be a proton-conducting channel connecting the cytoplasm to the b₅₉₅-d active site (22, 98) (Fig. 6). Thus, the generated potential must result primarily from protons moving from the cytoplasm to the O₂-reducing site on the opposite side of the membrane.

Reaction of Cytochrome bd with Oxygen: Sequence of Catalytic Intermediates.

For cytochrome bd, the following species were detected: fully ferrous (109, 135), fully ferric (40, 123), MV O₂ bound, (40, 104, 123, 167), fully reduced O₂ bound (A_R) (98), oxoferryl (105), and peroxy (21). A_R and the peroxy intermediate are transient, whereas the others can be generated in relatively stable forms.

The reaction of the fully ferrous cytochrome bd with oxygen was studied with the use of the flow-flash method (72) by means of spectroscopic and electrometric techniques (21, 22, 98). Recording absorption spectra and membrane potential development with 1-μs time resolution allowed us (i) to observe the sequence of the catalytic intermediates and (ii) to establish which catalytic steps are linked to electric potential generation (21).

The scheme describing the reaction is shown in Fig. 7. The initial complex of the fully ferrous cytochrome bd with CO is photolysed in the presence of oxygen. The unligated ferrous enzyme generated by the CO photolysis binds O₂ very rapidly, forming the ferrous heme d oxyspecies (A_R). The transition of the unligated fully ferrous enzyme to A_R is not electrogenic, and its rate is proportional to [O₂] (k_on = 1.9 × 10⁹ M⁻¹·s⁻¹ [21, 83]). The A_R formation is followed by electron transfer from heme b₅₉₅ to form the peroxy intermediate. The A_R→peroxy transition occurs at a rate of 2.2 × 10⁵ s⁻¹ (4.5 μs) and is also nonelectrogenic (21). Thus, electron transfer from heme b₅₉₅ to heme d is not coupled with membrane potential generation (21, 22). It is proposed that the peroxy intermediate is a peroxy complex of ferric heme d (21). If this is the case, the bound peroxide is likely not to be in the anionic form but at least singly protonated. The proton may come from one of the two protonatable groups linked to the b₅₉₅-d diheme site upon its oxidation (22). The peroxy intermediate is further converted into the ferryl intermediate at a rate of 2.1 × 10⁴ s⁻¹ (48 μs). This conversion is accompanied by the oxidation of heme b₅₅₈. The formation of the ferryl intermediate is coupled to the generation of a membrane potential (21, 22, 98). At the ferryl intermediate stage, the b-type hemes are in a ferric state and heme d is in an oxoferryl state. When cytochrome bd contains bound quinol, the reaction proceeds further to form the oxidized enzyme. The transition from the ferryl intermediate to the oxidized enzyme occurs at a rate of 0.9 × 10³ s⁻¹ (1.1 ms) and is electrogenic (21, 22).

Fig. 7Cytochrome bd reaction scheme. The three rhombuses represent hemes b₅₅₈, b₅₉₅, and d, respectively. The minus signs and red backgrounds in the rhombuses denote that the heme is in the ferrous state. R-CO, R, A_R, P, F, O are fully ferrous CO bound, fully ferrous unligated, fully ferrous O₂ bound, peroxy, oxoferryl, and fully ferric species, respectively.

Role of Heme b₅₉₅

Upon the addition of a ligand (e.g., CO or cyanide) to cytochrome bd, a majority of heme b₅₉₅ (>95%) does not bind the ligand (29, 32, 39). It is likely that heme b₅₉₅, although in the high-spin pentacoordinate state, is resistant to interaction with the classical ligands of the high-spin iron-porphyrin complexes. It cannot be excluded that despite the high-spin pentacoordinate state of the iron-porphyrin group, the specific features of the protein environment are such that this redox cofactor is protected from interaction with ligands. In such a case, the participation of heme b₅₉₅ in O₂ reduction in cooperation with heme d is unlikely and the role of heme b₅₉₅ is limited to the transfer of an electron to heme d. Another possible explanation that we favor is the following.

1. Both heme b₅₉₅ and heme d potentially can bind ligands.

2. The hemes are located very close to each other, forming a diheme active site.

3. The spatial proximity of hemes b₅₉₅ and d results in steric restrictions, allowing such a diheme site to bind only one ligand molecule.

4. Heme d has a higher affinity for ligands than heme b_595, in which case the final result observed upon the addition of a ligand will always be the binding of the ligand to heme d, whereas heme b₅₉₅ will remain mainly in the free state (29, 32, 37, 39).

The data on the redox coupling of the two hemes to the same ionizable groups (22) and the migration of CO within the protein from heme d to heme b₅₉₅ at cryogenic temperatures (84) are in agreement with this proposal. Modeling the excitonic interactions in absorption and CD spectra of cytochrome bd yields an estimate of the Fe_d-to-Fe_b595 distance of about 10 Å (5). This distance is markedly larger than that between the Fe-Cu_B pair in heme-copper oxidases (4 to 5 Å). If this is the case, heme b₅₉₅ cannot be functionally identical to Cu_B. A possible role of heme b₅₉₅, apart from electron delivery to heme d and/or to heme d-bound oxygen intermediates, would be as a binding site for hydroxide produced from heme d-bound oxygen upon the reductive cleavage of the O—O bond (5).

CYTOCHROME bd-II (CyxAB)

A putative second cytochrome bd in E. coli (cytochrome bd-II) encoded by cyxAB genes (previously called appBC or cbdAB) was identified (54). The cyxAB genes, located at 22 min on the E. coli genetic map, are upstream from a pH 2.5 acid phosphatase (appA) gene (54). The cyxAB and appA genes constitute the complex operon. The cyxA and cyxB genes encode 58.1- and 42.4-kDa integral membrane proteins, respectively. The deduced amino acid sequences of cyxA and cyxB gene products show homologies of 60 and 57%, respectively, to the sequences of subunit I (CydA) and subunit II (CydB) of the major cytochrome bd (encoded by cydAB) (54). A mutant lacking the cyo and cyd operons and a functional cyxB gene is unable to grow aerobically in rich medium, in contrast to a mutant lacking only the cyo and cyd operons, suggesting that cyxAB genes encode a third terminal oxidase in E. coli (54). The expression of the cyxAB-appA operon is induced by entry into the stationary phase and phosphate starvation (7). The cyxAB genes are also induced by anaerobic growth, and this induction is controlled by transcriptional regulators AppY and ArcA but is independent of Fnr, in contrast to the induction of the cyd operon (7, 41). Cytochrome bd-II is likely to function under even more oxygen-limiting conditions than cydAB-encoded cytochrome bd (41). Cytochrome bd-II has been extracted and partially purified (205). With the use of sodium dodecyl sulfate-polyacrylamide gel electrophoresis, two polypeptides, 43 kDa (subunit I) and 27 kDa (subunit II), were resolved from the preparation. These subunits show no cross-reactivity to subunit-specific polyclonal antibodies directed against the subunits of cydAB-encoded cytochrome bd (205). The spectral properties of cytochrome bd-II closely resemble those of cydAB-encoded cytochrome bd. Of the quinols tested, cytochrome bd-II utilizes menadiol as the preferred substrate (although ubiquinol-1, the most efficient in vitro substrate for cydAB-encoded cytochrome bd, was not tested). The TMPD oxidase activity of cytochrome bd-II is much more sensitive to cyanide than that of cydAB-encoded cytochrome bd (205). Cytochrome bd-II has been poorly studied, and so its physiological role is not known.

ACKNOWLEDGMENTS

Studies in our laboratories were supported by the Biocentrum Helsinki, the Sigrid Jusélius Foundation, a grant from the Academy of Finland (to M.I.V.), and the Russian Foundation for Basic Research grant 08-04-00093 (to V.B.B.).