Natural Resistance to Inhibitors of the Ubiquinol Cytochrome c Oxidoreductase of Rubrivivax gelatinosus: Sequence and Functional Analysis of the Cytochrome bc₁ Complex

Abstract

Biochemical analyses of Rubrivivax gelatinosus membranes have revealed that the cytochrome bc₁ complex is highly resistant to classical inhibitors including myxothiazol, stigmatellin, and antimycin. This is the first report of a strain exhibiting resistance to inhibitors of both catalytic Q₀ and Q_i sites. Because the resistance to cytochrome bc₁ inhibitors is primarily related to the cytochrome b primary structure, the petABC operon encoding the subunits of the cytochrome bc₁ complex of Rubrivivax gelatinosus was sequenced. In addition to homologies to the corresponding proteins from other organisms, the deduced amino acid sequence of the cytochrome b polypeptide shows (i) an E303V substitution in the highly conserved PEWY loop involved in quinol/stigmatellin binding, (ii) other substitutions that could be involved in resistance to cytochrome bc₁ inhibitors, and (iii) 14 residues instead of 13 between the histidines in helix IV that likely serve as the second axial ligand to the b_H and b_L hemes, respectively. These characteristics imply different functional properties of the cytochrome bc₁ complex of this bacterium. The consequences of these structural features for the resistance to inhibitors and for the properties of R. gelatinosus cytochrome bc₁ are discussed with reference to the structure and function of the cytochrome bc₁ complexes from other organisms.

Ubihydroquinone cytochrome c oxidoreductase (cyt bc₁ complex) is an integral membrane complex involved in energy transduction in a wide range of organisms. Anoxygenic photosynthetic bacteria contain a cyt bc₁ complex, which serves for both light-driven electron transfer and dark respiration. For Rhodobacter sphaeroides and Rhodobacter capsulatus this enzyme catalyzes electron transfer from ubiquinol to soluble ferricytochrome c coupled to proton translocation across the membrane, thus generating an electrochemical gradient used for ATP synthesis. Oxygenic photosynthetic organisms contain a similar complex, i.e., cyt b_6f, which functions as a plastoquinol plastocyanin oxidoreductase (12). Structural and functional similarities have been reported for both types of cyt bc (3, 9, 41). The cyt bc complex was found to be made up of at least three subunits: cyt b (or b₆), containing the two b-type hemes of low (b_L) and high (b_H) redox potentials, cyt c₁ (or f) containing a c-type heme, and the Rieske protein (ISP) containing the [2Fe-2S] cluster. In addition to subunits containing the prosthetic groups and cofactors, the mitochondrial cyt bc₁ complex contains other subunits (4, 29, 30, 46, 47).

Photosynthesis and aerobic respiration are affected by inhibitors. Some of these molecules are specific to the cyt bc complex. In fact, in many species the ubiquinol-cyt c oxidoreductase activity was found to be sensitive to quinone analogs such as myxothiazol, stigmatellin, antimycin, hydroxyquinoline N-oxide (HQNO), and diuron. Extensive kinetic and genetic studies (7, 13, 16, 17, 19, 28) as well as recent structural data available from X-ray crystallography of the mitochondrial cyt bc₁ complex (4, 29, 30, 47) in the presence of these inhibitors have allowed better knowledge of the properties of the two sites of quinone binding, the Q₀ and the Q_i sites (3). The Q₀ pocket, which is the site of oxidation of quinol, is located toward the periplasmic side of the membrane and is essentially composed of cyt b residues close to the heme b_L and of the extrinsic domain of the Rieske protein. It is the site of two concerted electron transfers, one from quinol through ISP and cyt c₁ to the external electron acceptor and the other through hemes b_L and b_H to the reducing site of quinone Q_i, located near the cytoplasmic side of the membrane close to heme b_H. Much experimental data support the idea that this bifurcated electron flow through cyt bc₁ is controlled by the movement of the periplasmic domain of the Rieske protein containing the [Fe-S] prosthetic group (4, 17, 18, 47). This Q cycle results in the oxidation of two molecules of ubiquinol with the release of four H⁺s at the Q₀ site and the reduction of one molecule of quinone with uptake of two H⁺s at the Q_i site. Myxothiazol and stigmatellin bind to the Q₀ pocket, displacing ubiquinol, and thus block the electron pathway to ferricytochrome c and to heme b_L. Antimycin, HQNO, and diuron block the reduction of quinone at the Q_i site.

Rubrivivax gelatinosus, a facultative phototrophic nonsulfur bacterium belonging to the β subclass of purple bacteria, can grow either aerobically, deriving its energy from aerobic respiration, or photosynthetically by using the cyclic electron transport chain. In this paper we report the study of natural resistance to electron transport inhibitors myxothiazol, stigmatellin, and antimycin in this purple bacterium. The level of the resistance of the R. gelatinosus cyt bc₁ complex to these inhibitors was determined by in vitro enzymatic activity and compared to those observed for Rhodobacter sphaeroides, Blastochloris viridis, and Allochromatium vinosum. To explain the observed phenotypes, the sequence of the petABC operon was determined and compared to known pet sequences and to a compilation of substitutions located in the cyt b subunit of the bacterial and mitochondrial cyt bc₁ complexes associated with sensitivity or resistance to inhibitors (7, 13, 21, 22, 26). The primary sequence comparison between different photosynthetic bacteria revealed the presence of substitutions in R. gelatinosus and in A. vinosum petB (cyt b) and suggested that the resistance to inhibitors of both species may be attributed to substitution of these residues in the quinone binding sites. The identified substitutions in the cyt b sequence may account for the unusual properties of both Q₀ and Q_i sites in the R. gelatinosus cyt bc₁ complex.

MATERIALS AND METHODS

Chemicals.

Decylubiquinone, horse heart cyt c, and antimycin were purchased from Sigma, myxothiazol was from Boehringer, and stigmatellin was from Fluka.

DBH₂ preparation.

The reduction of the 2,3-dimethoxy-5-methyl-6-decyl-1,4-benzoquinone (decylubiquinol; DBH₂) was done as described previously (38). The concentration of DBH₂ in ethanol was determined by measuring the absorbance using an extinction coefficient (ɛ) at 290 nm of 4 mM⁻¹ cm⁻¹.

Bacterial strains and growth media.

R. gelatinosus wild-type strain S1 (45), Rhodobacter sphaeroides, and B. viridis were grown anaerobically in light in a malate growth medium (1). R. gelatinosus mutant strain SΔC2 (36) was grown semiaerobically in the dark in the same medium supplemented with kanamycin (50 μg/ml). A. vinosum strains D and DSM180T (German collection) were grown under reducing photoheterotrophic conditions as described previously (33).

Membrane preparation.

Membranes were prepared by cell disruption with a French press in the presence of 0.1 M sodium phosphate buffer, pH 7.4, containing 1 mM phenylmethylsulfonyl fluoride, followed by differential ultracentrifugation as previously described (1). Subsequently, the Rhodobacter sphaeroides membranes were resuspended in the same buffer and R. gelatinosus, B. viridis, and A. vinosum membranes were resuspended in 0.1 M Tris-HCl buffer, pH 8. Protein concentration was determined by using the Pierce BCA protein assay reagent method with bovine serum albumin as the reference standard.

Isolation of soluble electron carriers of R. gelatinosus.

It has been shown that the main periplasmic redox proteins in R. gelatinosus are cyt c₈ and the high-potential iron protein (HiPIP) and that, in cells grown under semiaerobic dark conditions, the amount of cyt c₈ is much larger than that in cells grown under photosynthetic conditions (34). Therefore, we purified this cytochrome from the R. gelatinosus SΔC2 strain grown in the dark. The cyt c₈ was recovered from the supernatant of cell extracts after ultracentrifugation and purified by precipitation with ammonium sulfate (45% saturation). The precipitate was dialyzed against 10 mM MES (morpholineethanesulfonic acid)-NaOH buffer, pH 6, and subsequently adsorbed onto a carboxymethyl (CM)-Sepharose ion exchange resin equilibrated with the same buffer. The cytochrome was eluted at 25 mM NaCl. The HiPIP was prepared by a method resembling that of Osyczka et al. (35). It was recovered from the supernatant of cell extracts and purified from the fraction precipitating at 100% saturation of ammonium sulfate, which was then dialyzed against 10 mM MES-NaOH buffer, pH 6. Subsequently the dialyzed fraction was adsorbed on CM-Sepharose equilibrated with the same buffer. The HiPIP was eluted at 50 mM NaCl, and the fractions with absorbance ratios (A₂₈₀/A₃₈₈) of 3 to 4 were pooled and concentrated. The sample was further purified by gel filtration on a Sephadex G-50 column equilibrated with 50 mM Tris-HCl, pH 8, and fractions with A₂₈₀/A₃₈₈ values of 2 to 2.2 were pooled.

Before the cyt bc₁ activities were measured, horse heart cyt c, cyt c₈, and the HiPIP were oxidized with potassium ferricyanide. The excess oxidant was then removed on a Sephadex PD-10 column.

Spectrophotometric measurements.

The absolute and differential absorption spectra of different samples were recorded with a Cary 2300 absorption spectrometer. Ubiquinol cyt c, cyt c₈, and HiPIP oxidoreductase activities were measured with a DW-Aminco Chance spectrometer in the dual-wavelength mode. The membranes were suspended in a 3-ml stirred cell containing 4 mM KCN (to prevent cyt c oxidase activity), and oxidized cytochromes or HiPIPs were then added. The mixture in the presence or absence of inhibitors was stirred for 5 min in the dark, and the activity measurements were initiated by the addition of DBH₂.

The kinetics of the enzymatic reduction of oxidized cyt c (cyt c_ox) and cyt c_8ox were monitored as an increase in the value of A₅₅₀ − A₅₄₀ (isobestic point) in the dark. From the initial rate of absorption increase (V_i), the amount of reduced cyt c per minute was quantified by using a value for the differential extinction coefficient (Δɛ) between 550 and 540 nm of 19 mM⁻¹ cm⁻¹ (23)_. Specific activities were determined as micromoles of cyt c reduced per nanomole of cyt bc₁ per hour as described previously (25). When the HiPIP was used as the electron acceptor, its reduction was monitored as the decrease of A₄₈₃ − A₆₀₀ with a Δɛ of 7 mM⁻¹ cm⁻¹. Under our experimental conditions the nonenzymatic reduction of ferricytochrome c or the HiPIP by DBH₂ was not detected.

Estimation of the cyt bc₁ concentration in the membrane fraction.

The amount of cyt bc₁ complex was estimated by determining the cyt b concentration at 562 nm from the absorption difference spectra for the membrane suspension (dithionite minus ascorbate); Δɛ [562 to 575 nm] = 28 mM⁻¹ cm⁻¹) (11). Alternatively, the amount of cyt b was estimated by the pyridine hemochrome method (5) for B. viridis, R. gelatinosus, and A. vinosum.

Molecular biology techniques.

Standard methods were performed, if not otherwise indicated, as described by Sambrook et al. (39). Plasmids were purified with the Bio-Rad Quantum Prep plasmid kit. DNA was treated with restriction enzymes and other nucleic acid-modifying enzymes according to the manufacturers' specifications. DNA fragments were analyzed on agarose gels, and restriction fragments were purified with the Geneclean kit (Bio 101, Inc.). PCRs were carried out by using genomic or plasmid templates in a 50-μl reaction mixture containing the DNA, the PCR buffer, 200 μM (each) deoxynucleoside triphosphate, 1 μM (each) primer, 5% dimethyl sulfoxide, and 2.5 U of Taq DNA polymerase. Twenty cycles were performed in a Hybaid thermal cycler; each cycle consisted of 30 s at 92°C, 40 s at 60°C, and 40 s at 72°C. The primers used to clone the DNA fragment of R. gelatinosus containing the pet genes were 5′-CGGTCGAGCACATCATGC-3′ and 5′-CACCGCGCGCAGCATCGA-3′. The primers used to clone the A. vinosum 0.4-kb fragment of the petB gene (5′-CGCTGGAGCGACAGGCTCCG-3′ and 5′-CGGTCGCCCACTTGAAGATC-3′) were based on the previously published sequence (10). DNA sequencing was performed with an ABI 373 automatic DNA sequencer.

Nucleotide sequence accession number.

The sequences reported in this paper were deposited in GenBank under accession no. AF380164.

RESULTS

Effects of inhibitors on decyl quinol oxidoreductase activity of R. gelatinosus membranes.

Most of the cyt bc₁ reductase activity measurements for R. gelatinosus were carried out with horse heart cyt c as the electron acceptor to allow a comparison with the cyt bc₁ activities from Rhodobacter sphaeroides, B. viridis, and A. vinosum. Nevertheless, in several experiments physiological carriers cyt c₈ and HiPIP were used as substrates.

Figure 1 shows the time course for reduction of cyt c by DBH₂ catalyzed by R. gelatinosus membranes in the presence and absence of specific inhibitors of the cyt bc₁ complex acting either on the Q₀ site (myxothiazol and stigmatellin) or on the Q_i site (antimycin). The reduction of cyt c by cyt bc₁ displays monoexponential decay kinetics with an average rate constant (k) of 0.046 s⁻¹; values of k in the presence and absence of inhibitors were similar. Furthermore, the activities were measured with different cyt bc₁ concentrations in the presence of different amounts of each inhibitor (Table 1). From inspection of Fig. 1 and Table 1 it is obvious that the R. gelatinosus cyt bc₁ complex is resistant to the three inhibitors. A weak decrease in the activity in the presence of a very high concentration (0.15 mM) of myxothiazol was observed, namely, a maximum of 50% inhibition of cyt bc₁ activity was reached (not shown).

FIG. 1.

Decyl ubiquinol horse heart cyt c oxidoreductase activity of R. gelatinosus membranes in the presence and absence of specific inhibitors of the cyt bc₁ complex. Membranes were suspended at 0.66 mg of protein/ml in 0.1 M Tris-HCl buffer, pH 8, containing 4 mM KCN and 20 μM horse heart cyt c. The reaction was started by the addition of 30 μM DBH₂. The time course of activity follows a monoexponential decay with time constants as follows: without inhibitor (•), 0.046 s⁻¹; with 25 μM antimycin (○), 0.046 s⁻¹, with 25 μM myxothiazol (∗), 0.04 s⁻¹; with 10 μM stigmatellin (▴), 0.04 s⁻¹. (Inset) Initial linear rates of absorption increase (derived from the larger graph) from which specific activities (A) were determined as described in Materials and Methods. •, A = 36; ○, A = 32; ∗, A = 30; ▴, A = 30.

TABLE 1.

Measurements of cyt bc₁ activity in the membranes isolated from different species^a

Species	Protein concn (mg/ml)	Amt (nmol) of cyt bc1	Inhibitor, concn (μM)	Vi	cyt bc1 activity
R. gelatinosus	0.16	7	0	12	110
			Ay, 10	13.6	125
			Mx, 10	10	87.4
			St, 10	15.5	133
	0.66	28	0	17	36
			Ay, 25	15	32
			Mx, 25	14	30
			St, 12	14	30
	1.1	46	0	8	10.4
			Ay, 20	12.6	16.5
			Mx, 25	8.2	10.7
			St, 10	11	14.3
Rhodobacter sphaeroides	0.24	45	0	10.3	14
	Ay, 20	0.2	0.3
			Mx, 25	0.3	0.4
			St, 12	0.4	0.5
B. viridis	0.66	7	0	29	248
			Ay, 25	3	27
			Mx, 25	1.5	13
			St, 10	4	34
	1.5	15	0	18	72
			Ay, 20	3.8	15
			Mx, 20	3.4	13.4
			St, 10	4	17
A. vinosum	0.6	11	0	7	38
			Ay, 20	7.5	41
			Mx, 25	7	38
			St, 10	12	65
	3	58	0	7	7.3
			Ay, 25	8	8.3
			Mx, 25	6	6.2
			St, 10	10	10.3

^aThe cyt bc₁ activity in micromoles of cyt c reduced per nanomole of cyt bc₁ per hour is derived from the initial rate, V_i (micromoles of cyt c reduced per minute). The enzymatic activity was initiated by addition of 20 to 30 μM decyl ubiquinol to different concentrations of the membrane suspensions containing 15 to 20 μM oxidized horse heart cyt c. Activity was tested in the absence (0) or presence of inhibitors antimycin (Ay), myxothiazol (Mx), and stigmatellin (St).

When R. gelatinosus cyt bc₁ activity in the presence of its physiological electron acceptors cyt c₈ and HiPIP was determined, a higher turnover of the enzyme was obtained. For cyt c₈, the activity increased by a factor of 2 to 3 relative to that for equine cyt c, whereas, with the very reactive HiPIP, the activity was 30 times higher than that for cyt c reductase measured under the same experimental conditions (not shown). As with cyt c, cyt c₈ and the HiPIP reductase were not affected by the inhibitors (data not shown). These results demonstrate that the cyt bc₁ complex from R. gelatinosus is resistant to both Q₀ and Q_i site inhibitors, i.e., stigmatellin, myxothiazol, and antimycin, irrespective of its external electron acceptor.

Comparison of decyl quinol cyt c oxidoreductase activities of Rhodobacter sphaeroides and B. viridis.

For reference, we measured the decyl quinol cyt c oxidoreductase activity in Rhodobacter sphaeroides and B. viridis membranes under the same conditions. A high sensitivity to the inhibitors in both species has already been demonstrated (15, 25, 44). B. viridis and R. gelatinosus membranes contain much less cyt bc₁ than Rhodobacter sphaeroides membranes (Table 1). However, the small amount of cyt bc₁ in these species was compensated for by its high enzymatic turnover. In the presence of the inhibitors, enzymatic activity was completely abolished in Rhodobacter sphaeroides (Fig. 2) and was strongly reduced in B. viridis to only 10 to 20% residual activity (Fig. 3). Table 1 summarizes the cyt c reductase activities measured with different cyt bc₁ concentrations in the presence of different amounts of each inhibitor.

FIG. 2.

cyt bc₁ activity of Rhodobacter sphaeroides chromatophores in the presence and absence of the inhibitors. The chromatophores were suspended at 0.24 mg of protein/ml in 0.1 M sodium phosphate buffer, pH 7.5, containing 4 mM KCN and 20 μM horse heart cyt c. The reaction was started with 20 μM DBH₂. Additions and conditions were as in Fig. 1. •, specific activity (A) = 14; ○, A = 0.28; ∗, A = 0.54; ▴, A = 0.42.

FIG. 3.

cyt bc₁ activity of B. viridis chromatophores in the presence and absence of inhibitors. The chromatophores were suspended at 0.66 mg/ml in 0.1 M Tris buffer, pH 8, containing 20 μM cyt c; 30 μM DBH₂ was added to start the reaction. •, specific activity (A) = 248; ○, A = 27; ▴, A = 34; ∗, A = 13.7.

We conclude that Rhodobacter sphaeroides and B. viridis are sensitive to the inhibitors as already reported (15, 25, 44) and confirm that the R. gelatinosus cyt bc₁ complex is resistant to myxothiazol, stigmatellin, and antimycin.

Because the resistance or sensitivity to the cyt bc₁ inhibitors is primarily related to the cyt b primary structure, we then analyzed the cyt b sequence from R. gelatinosus and compared it to the sequences of those from other species.

Sequence of the petABC operon from R. gelatinosus.

A genomic R. gelatinosus DNA library (37) was screened by PCR for the presence of the petB gene encoding cyt b. A plasmid containing an approximately 7-kb DNA fragment was isolated, and a 2.3-kb EcoRI fragment containing three open reading frames (ORFs) was subcloned and sequenced. BLASTX (2) sequence similarity searches against the nonredundant protein database suggested that the ORFs are homologues of the petA, petB, and petC genes, which encode the Rieske Fe-S protein (ISP), cyt b, and cyt c₁, respectively. The DNA fragment contains the entire sequence of petA and petB and the 5′ end of petC. The three genes have the same transcriptional orientation, with no putative promoter sequences in the intergenic region, suggesting that they might constitute an operon.

cyt b amino acid sequence analyses.

The petB gene consists of 1,272 bp and encodes a polypeptide of 423 amino acids with a molecular mass of 48 kDa. PetB has significant similarity to the cyt b polypeptides from Pseudomonas aeruginosa (65% identity and 76% similarity), Neisseria meningitidis (63% identity and 74% similarity), and PetB from A. vinosum (61% identity and 74% similarity). Lower matches to the PetB polypeptides from Rhodobacter sphaeroides (42% identity and 59% similarity), B. viridis (41% identity and 58% similarity), and Rhodobacter capsulatus (40% identity and 57% similarity) were seen. According to TMpred and SOSUI transmembrane region detection programs, PetB possesses eight hydrophobic transmembrane helices, in agreement with the published structure of the mitochondrial cyt bc₁ complex (46). In addition, according to the helical wheel analysis (SOSUI), an extra amphipathic helix could be exposed on the periplasmic side of the membrane linking helices III and IV. The alignment of the deduced amino acid sequence of cyt b from R. gelatinosus with homologues from bacteria and Saccharomyces cerevisiae (not shown) revealed three interesting features. (i) Whereas the spacing (13 residues) between the histidines supposed to serve as axial ligands to the hemes is conserved in all cyt b sequences in helix II, there are 14 residues between the heme binding histidines in helix IV in R. gelatinosus (Fig. 4A), as already observed in cyt b₆ (41) and in the cyt b subunits of the β, γ, and ɛ subdivisions of proteobacteria, with the exception of A. vinosum (41). (ii) The highly conserved PEWY motif found in hundreds of cyt b sequences from different organisms is changed to PVWY in R. gelatinosus and A. vinosum (Fig. 4B). The glutamic acid in the PEWY loop is involved in inhibitor binding at the Q₀ site, as shown by the structure of the chicken cyt bc₁ cocrystallized with stigmatellin (47). The replacement of glutamic acid 303 with the hydrophobic valine might then prevent the binding of stigmatellin to cyt b (see Discussion). (iii) Several substitutions of residues localized in the regions presumed to be involved in the Q₀ binding site were identified in PetB sequences from R. gelatinosus, i.e., V151L, T163Q, T166I(V), E303V, L307T, T145E, and F331G. In A. vinosum only the first five substitutions are conserved (Fig. 4B). The possible consequences of these changes for the properties of the Q₀ pocket are discussed below. Mutations that confer resistance to antimycin at the Q_i site were also reported. A comparison to the Rhodobacter sequence identified substitutions that might be responsible for resistance to antimycin in R. gelatinosus in the putative Q_i site. In particular, we focus on the I49F and the A55M substitutions (Fig. 4C); both amino acids are directly involved in antimycin resistance in Rhodobacter (7). Also, the I254T substitution is located in a critical region involved in antimycin resistance, as shown by the different mutations mapped in this region (7, 14).

FIG. 4.

Partial amino acid sequence comparison of the deduced petB gene products (cyt b) of R. gelatinosus (R. gel), A. vinosum (A. vin), and Rhodobacter sphaeroides (R. sph). ∗, identical residues; :, similar residues (from the Clustal W program). (A) Comparison of the helix IV sequences showing the two histidines that bind the hemes (shaded) and the extra residue (boldface and underlined) in the R. gelatinosus sequence. (B) Amino acid sequence comparison of the putative Q₀ pockets. ▾, most-relevant residues, shown to be involved in sensitivity or resistance to cyt bc₁ inhibitors stigmatellin and myxothiazol in other organisms (7) and likely to be responsible for the resistance to Q₀ pocket inhibitors in R. gelatinosus and A. vinosum. (C) Amino acid sequence comparison of the putative Q_i pockets. ▿, substitutions which might be responsible for the resistance to the Q_i site inhibitor, antimycin.

Inhibitor effects on A. vinosum cyt bc₁ activity.

The sequence homologies between the cyt b primary sequences of R. gelatinosus and A. vinosum, especially in the area of the Q₀ site (E303V substitution), suggest that the cyt bc₁ from A. vinosum would be resistant to at least stigmatellin. To test this hypothesis, we measured the cyt bc₁ reductase activity of A. vinosum membranes in the presence of the three inhibitors (Table 1). As for R. gelatinosus, the A. vinosum cyt bc₁ activity was not inhibited by the addition of antimycin, myxothiazol, and stigmatellin. This resistance of A. vinosum strain D to inhibitors was confirmed with the A. vinosum DSM180T collection strain.

Concerning the resistance to myxothiazol and antimycin, our results are different from the previous ones obtained with A. vinosum membranes (6, 11, 44). In the work of Tan et al. (44) cyt c reductase activity was measurable only after a treatment with dodecyl maltoside detergent. This implies that the quinone binding site could be masked in situ. In our system, cyt c reductase activity on intact chromatophores was measured, showing that quinone has access to the cyt bc₁ complex. However, to check whether the inhibitors have access to the binding sites, we preincubated chromatophores with the detergent as described previously (44). Detergent treatment did not allow the complex to regain sensitivity, suggesting that resistance is an intrinsic property of the complex.

In the cyt b sequence of A. vinosum strain D the E303V substitution in the PEWY motif was reported by Chen et al. (10). To check if the PVWY motif was present in the A. vinosum strain D used in our work, a 0.4-kb DNA fragment of the region flanking this motif in the petB gene was cloned. The DNA sequence confirmed the presence of the GTC valine codon.

DISCUSSION

In the present work we have studied the cyt bc₁ complex from R. gelatinosus by measuring the decylquinol oxidoreductase activity of membranes with an artificial electron acceptor, the horse heart ferricytochrome c, and with its physiological electron carriers, namely, cyt c₈ and HiPIP, in the presence of three different inhibitors.

It is conspicuous that the rate of reduction of the HiPIP is increased by a factor of 20 to 30 relative to the rate of cyt c reduction. The role of HiPIP as an efficient electron mediator between the cyt bc₁ complex and the reaction center (RC) has previously been suggested by the experiments of Schoepp et al. (40). The high reactivity of the cyt bc₁ complex with HiPIP could be explained by the presence of a specific docking site on the cyt c₁, as is present in the mitochondrial cyt c₁ subunit, shown to contain a binding site for cyt c (43, 47).

The specific inhibitors of the Q₀ and of the Q_i sites strongly reduce the enzymatic activity of the cyt bc₁ complex in both Rhodobacter sphaeroides and B. viridis, as has already been shown in the literature. In contrast, for R. gelatinosus membranes, this activity was insensitive to these inhibitors, irrespective of the nature of the final electron acceptor. Flash-induced electron transfer experiments carried out by Shoepp et al. (40) on R. gelatinosus whole cells have shown that the rereduction rate of the photo-oxidized RC was not affected by myxothiazol or antimycin. On the contrary, stigmatellin appreciably slowed the light-induced electron flow through the cyt bc₁ complex. It has been shown that stigmatellin is also an inhibitor of the photochemical RC Q_B site in B. viridis (32). Therefore it is very likely that, in these light-induced experiments, stigmatellin acts as an inhibitor of the RC by binding to the Q_B site. In this way it would prevent the regeneration of the ubiquinol pool necessary for cyt bc₁ turnover, decreasing the electron transfer rate.

A series of constructed mutations as well as spontaneous substitutions in the cyt b subunit of bacterial and mitochondrial cyt bc₁ are available (7, 13, 21, 22, 28). These mutations have been shown to confer either hypersensitivity or resistance to the cyt bc₁-specific inhibitors. Amino acids associated with resistance were mapped essentially in the quinone binding domains. To explain the natural resistance of R. gelatinosus to these inhibitors, we studied the petB gene sequence, encoding cyt b. The comparison of the primary structures of R. gelatinosus and A. vinosum cyt b polypeptides with those of Rhodobacter species and yeast allowed the identification of substitutions which may contribute to resistance. Among them, we identified a substitution which has been reported to induce resistance to both myxothiazol and stigmatellin, T163Q, and those inducing resistance only to stigmatellin, i.e., T166I(V) and E303V (7). The resistance-associated amino acids introduce different residue volumes, polarities, or charges into the Q₀ binding site. According to structural data (13), the accommodation of the inhibitor inside the Q₀ pocket is accompanied by a series of conformational changes, and we presume that mutations of some residues may prevent access of the inhibitor to, or its binding in, the pocket. Stigmatellin is bound in the distal domain of the Q₀ pocket by two H bonds, one from His161 (ISP) and the other from Glu303 (cyt b) of the highly conserved PEWY loop. In R. gelatinosus and A. vinosum the replacement of glutamic acid 303 with a hydrophobic valine might prevent stigmatellin H bonding to cyt b and hence the inhibitor might not be retained in the site. Nevertheless, Thiobacillus ferroxidans PetB contains a PPWY loop (instead of PEWY), and stigmatellin could bind in the proximity of heme b_L, as demonstrated by the red shift of its α band (24). Yet, like valine, the hydrophobic proline is unable to form an H bond with the inhibitor. Thus, an H bond provided by the cyt b subunit is not always required to bind stigmatellin, and the whole structure of the Q₀ site therefore should be taken into account when assessing binding.

In the R. gelatinosus Q₀ pocket, other substitutions were identified (V151L, T145E, L307T, and F331G) (Fig. 4B). Their proximity to the binding domains of the inhibitors may also explain the observed resistance.

Similarly, we propose that the resistance of the R. gelatinosus cyt bc₁ complex to myxothiazol may be explained by the fact that some of the substituted residues are involved in inhibitor binding at the Q₀ site. The weak sensitivity to myxothiazol detected suggests that its binding site in the Q₀ pocket differs from that of stigmatellin, as already shown by the crystallographic data for the mitochondrial cyt bc₁ complex cocrystallized with different inhibitors (13).

Crystallographic and mutagenesis data have highlighted the structural and functional role of the glutamic acid in the PEWY loop in the cyt bc₁ and cyt b_6f complexes in the rate of ubiquinol (QH₂) oxidation (7, 16, 48). It was reported that this rate was appreciably decreased in mutants in which the glutamate was replaced by other residues (16, 48). Therefore, it has been inferred that the H bonding of quinol to glutamate stabilizes the binding of quinol to the site. On the other hand, it has been proposed that the carboxyl group of Glu303 is the acceptor of the second H⁺ released by QH₂ oxidation (16). Although the Q₀ pocket in the R. gelatinosus cyt bc₁ complex contains the PVWY motif, the ability of the enzyme to turn over demonstrates that Glu is not indispensable for binding QH₂. It is likely that the H bonding to ISP in addition to other interactions with the nearby residues is sufficient to stabilize the quinol in its active site.

The resistance to antimycin usually involves displacement of the quinone from the Q_i site. Several authors have shown that some substitutions of residues by site-directed mutagenesis in Rhodobacter capsulatus PetB conferred resistance to antimycin and HQNO (for reviews see references 7 and 14). Substitutions of residues involved in the resistance to antimycin at the Q_i sites of different organisms have been mapped (7, 14). Among the residues contributing to the folding of the R. gelatinosus Q_i pocket, relevant substitutions (I49F, A55M and I254T) that may affect antimycin access or binding to the pocket were found. However, due to the low conservation of the Q_i site sequences, it is difficult to ascertain the residues responsible for resistance to antimycin in the R. gelatinosus cyt bc₁ complex.

A. vinosum cyt bc₁ has properties that appear similar to those of R. gelatinosus in that it is resistant to inhibitors of both Q₀ and Q_i sites. The resistance of both species to stigmatellin could be correlated to the E303V substitution in the cyt b subunit. Nevertheless, part of these results are different from previously published results for A. vinosum chromatophores that showed inhibition of the cyt bc₁ complex by both antimycin and myxothiazol (6, 11, 44). However, in this paper, we have measured resistance to inhibitors and identified the sequence flanking the PVWY loop on the same strain to get reliable results. The resistance to myxothiazol and to antimycin, which is controversial and which disagrees with previous works, was found for A. vinosum strains D and DSM180T.

In the R. gelatinosus and A. vinosum PetA sequences (not shown), a substitution (Y165F) adjacent to the cluster ligand domain of the Rieske protein was identified. Site-directed mutagenesis experiments with yeast (20) and Rhodobacter sphaeroides (27) have shown that replacing the amino acids forming OH-S bonds to the cluster (S163 and Y165) by residues unable to provide H bonds decreases the midpoint redox potential of the protein as well as the catalytic activity of the cyt bc₁ complex. Therefore substitution Y165F in the cluster binding domain of the Rieske protein in R. gelatinosus could result in a midpoint potential lower than that for yeast or Rhodobacter sphaeroides. Mutations in the Rieske protein that confer resistance to stigmatellin in Rhodobacter capsulatus, in particular mutations that perturbed Q₀ site occupancy, have been reported (8). In the absence of experimental data, sequence alignment does not give any information on the possible contribution of the R. gelatinosus ISP subunit to the resistance to Q₀ inhibitors.

In addition to the PVWY motif and the Y165F substitution in the ISP subunits, the extra residue found between the conserved heme-binding histidines in helix IV could generate different redox potential properties of the R. gelatinosus cyt bc₁ complex. Indeed, it was shown by Kuras et al. (31) that addition of an extra residue in this spacing region by site-directed mutagenesis of petB from Rhodobacter sphaeroides increases the potential of the b_L heme, shifting the potential difference between the two b hemes from 140 mV in the wild-type to 55 mV in the mutant strain. Therefore, this arrangement in helix IV, as well as the substituted residues in the cyt b and the ISP subunits of the R. gelatinosus cyt bc₁ complex, could account for a modified structure of this enzyme at the level of the Q₀ and Q_i pockets, impeding the binding of the inhibitors.

Given the high level of resistance to quinone analogs in the cyt bc₁ complex from R. gelatinosus, it seems likely that the nature and the structure of the quinone binding sites in cyt bc₁ are significantly different from those of the corresponding sites in cyt bc₁ of Rhodobacter species. This study of a bacterial strain belonging to the β subclass of purple bacteria points to interesting features that help us to focus on important residues implicated in the functioning of the cyt bc₁ complex. These assumptions should be verified by site-directed mutagenesis, and purification of the complex is in progress. This will allow a deeper understanding of the functioning of this crucial enzyme.