The K^C Channel in the cbb₃-Type Respiratory Oxygen Reductase from Rhodobacter capsulatus Is Required for Both Chemical and Pumped Protons

Abstract

The heme-copper superfamily of proton-pumping respiratory oxygen reductases are classified into three families (A, B, and C families) based on structural and phylogenetic analyses. Most studies have focused on the A family, which includes the eukaryotic mitochondrial cytochrome c oxidase as well as many bacterial homologues. Members of the C family, also called the cbb₃-type oxygen reductases, are found only in prokaryotes and are of particular interest because of their presence in a number of human pathogens. All of the heme-copper oxygen reductases require proton-conducting channels to convey chemical protons to the active site for water formation and to convey pumped protons across the membrane. Previous work indicated that there is only one proton-conducting input channel (the K^C channel) present in the cbb₃-type oxygen reductases, which, if correct, must be utilized by both chemical protons and pumped protons. In this work, the effects of mutations in the K^C channel of the cbb₃-type oxygen reductase from Rhodobacter capsulatus were investigated by expressing the mutants in a strain lacking other respiratory oxygen reductases. Proton pumping was evaluated by using intact cells, and catalytic oxygen reductase activity was measured in isolated membranes. Two mutations, N346M and Y374F, severely reduced catalytic activity, presumably by blocking the chemical protons required at the active site. One mutation, T272A, resulted in a substantially lower proton-pumping stoichiometry but did not inhibit oxygen reductase activity. These are the first experimental data in support of the postulate that pumped protons are taken up from the bacterial cytoplasm through the K^C channel.

INTRODUCTION

The heme-copper superfamily of oxidoreductases includes both oxygen reductases and nitric oxide (NO) reductases (1–5). The oxygen reductases include the proton-pumping respiratory enzymes that terminate the aerobic respiratory chains in the mitochondrial inner membranes of eukaryotes as well as in the cytoplasmic membranes of aerobic bacteria and archaea. The heme-copper oxygen reductases all catalyze the 4-electron reduction of O₂ to two H₂O molecules, using the free energy available from this reaction to generate a transmembrane voltage and proton motive force (6–10). The respiratory oxygen reductases are classified into three major families (A, B, and C families) (3, 4) based on phylogenetic as well as structural analyses using the one subunit (i.e., subunit I) that they all have in common. Subunit I contains 12 transmembrane helices (at least), two histidine axial ligands for a low-spin heme Fe, one histidine axial ligand for a high-spin heme Fe, and three histidine ligands for a copper (Cu_B) atom. One of the histidine ligands to Cu_B is cross-linked to a tyrosine, forming a unique cofactor that is diagnostic of all heme-copper oxygen reductases (11–16).

Essential structural features of the heme-copper oxygen reductases are proton-conducting channels, which are needed to convey protons to the enzyme active site, where they are consumed to form water (“chemical” protons), as well as to convey protons that are pumped across the membrane (“pumped” protons). Both the chemical and pumped protons are taken from the bulk aqueous phase on the electrically negative side (N side) of the membrane, corresponding to the bacterial cytoplasm or mitochondrial matrix (8, 17–19). For each electron delivered to the enzyme active site, one chemical proton must also be transferred to this site to maintain charge neutrality, and the proton channels ensure that all these protons originate from the N side.

Within the prokaryotic members of the A family enzymes, there are two proton channels, named the D channel and the K channel (19, 20). Conserved polar residues define each of these proton pathways within the A family enzymes, and the functional importance of these residues has been confirmed by numerous site-directed mutagenesis studies (20). The K channel is required to deliver the chemical protons to the active site concomitant with each of the first two electrons added to the fully oxidized heme-copper center. Once the heme-copper pair is doubly reduced, it rapidly reacts with O₂, resulting in the formation of a high-valence iron-oxo form of the active-site heme as well as a neutral tyrosyl radical. The last two electrons return the enzyme to the fully oxidized state, completing the catalytic cycle, and the chemical protons delivered to the active site concomitant with these two electrons come through the D channel. All of the pumped protons are taken up through the D channel. The strongest argument in support of pumped protons exclusively using the D channel is the existence of a set of mutations within the D channel which selectively eliminate proton pumping while, remarkably, having little or no effect on the delivery of the chemical protons, some of which use the same D channel (20–24). These mutations are referred to as “decoupling” proton pumping from oxidase activity, and there is no consensus on the mechanism by which this is accomplished. No decoupling mutant within the K channel of any A family enzyme has been reported.

Surprisingly, analyses of the conserved polar residues of the B and C family oxygen reductases predict that the enzymes in each of these families lack the D channel (8, 17, 18). The subsequently determined X-ray structure of the C family oxygen reductase from Pseudomonas stutzeri shows that hydrophobic residues effectively block the space equivalent to that occupied by the D channel in the A family enzymes (13). Both the B and C family heme-copper oxygen reductases have putative proton channels in the same region of the protein as the K channel in the A family enzymes. The sets of conserved polar residues in the B and C families are different from each other and also differ from the conserved residues in the A family, so these putative proton channels are referred to as the K^B and K^C channels, respectively. In all three families of enzymes, the K/K^B/K^C channels have an entrance defined by a glutamic acid residue within a subunit that is different from subunit I (subunit II for the A and B family enzymes and subunit CcoP for the C family enzymes) and terminate at the cross-linked tyrosine at the active site.

If the K^C channel is the only proton input channel in the C family enzymes, it must be critical for delivering both chemical and pumped protons. Site-directed mutagenesis has confirmed the critical role of the K^C channel in oxygen reductase activity in the enzymes from Vibrio cholerae (18) and Rhodobacter sphaeroides (25). However, the influence of mutations in the K^C channel on proton pumping has not been reported. The current work was designed to test whether any mutation(s) within the K^C channel that eliminates or strongly decreases the stoichiometry of proton pumping could be found. One mutation was found to have these properties, consistent with the K^C channel being used to convey pumped protons as well as chemical protons during the catalytic cycle in the C family heme-copper oxygen reductases.

MATERIALS AND METHODS

Microorganisms and growth conditions.

The properties of the strains and plasmids used in this work are listed in Table 1. All Escherichia coli and Rhodobacter capsulatus strains were grown as described previously (26, 27).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Genotype or description	Phenotype	Source or reference(s)
Strains
E. coli
XL1-Blue	recA endA1 gyrA986 thi-1 hsdr17 supE44 relA1 lac		Stratagene
XL1-Blue/pMOZIT272A		Ampr	This work
XL1-Blue/pMOZIY280/281F		Ampr	This work
XL1-Blue/pMOZIN346V		Ampr	This work
XL1-Blue/pMOZIY374F		Ampr	This work
HB101	F− proA2 hsdS20 (rB− mB−) recA13 ara14 lacY1		41
HB101/pMOZI		Ampr
HB101/pACYC177	F− ara-14 leu fhuA2 Δ(gpt-proA)62 lacY1 glnV44 galK2 rspL20 xyl-5 mtl-1 Δ(mcrC-mrr)HB101		New England Biolabs
HB101/pMOZII		Ampr	29, 42
HB101/pMOZIIT272A		Ampr	This work
HB101/pMOZIIY280/Y281F		Ampr	This work
HB101/pMOZIIN346V		Ampr	This work
HB101/pMOZIIY374F		Ampr	This work
HB101/pOX15	15.3 kb with pOX15 DNA	Tetr	43
HB101/pOX15T272A		Tetr	This work
HB101/pOX15Y280/Y281F		Tetr	This work
HB101/pOX15N346V		Tetr	This work
HB101/pOX15Y374F		Tetr	This work
R. capsulatus
MT1131	crtD121 Rifr	Wild type	26
GK32	Δ(ccoNO::Kan)	Kanr	43
KZ101	Δ(ccoNO::Kan) Δ(cydA::Spect)	Tetr Kanr Spcr	Fevzi Daldal
GK32/pOX15		Tetr Kanr	43
GK32/pOX15T272A		Tetr Kanr	This work
GK32/pOX15Y280/Y281F		Tetr Kanr	This work
GK32/pOX15N346V		Tetr Kanr	This work
GK32/pOX15Y374F		Tetr Kanr	This work
KZ101/pOX15		Tetr Kanr Spcr	This work
KZ101/pRK415		Tetr Kanr Spcr	This work
KZ101/pOX15T272A		Tetr Kanr Spcr	This work
Plasmids
pBluescript	SKII+	Ampr	Fermentas
pMOZI	pBluescript SKII with 2.8-kb NO′ XhoI inserta	Ampr	29, 42
pMOZIT272A	Substitution of threonine to alanine in pMOZI	Ampr	This work
pMOZIY280F/Y281F	Substitution of tyrosine to phenylalanine in pMOZI	Ampr	This work
pMOZIN346V	Substitution of asparagine to valine in pMOZI	Ampr	This work
pMOZIY374F	Substitution of tyrosine to phenylalanine in pMOZI	Ampr	This work
pACYC177		Ampr Kanr	Fermentas
pMOZII	pACYC177 with 2.8-kb NO′ XhoI insert	Ampr Kanr	29, 42
pMOZIIT272A	The 0.6-kb BsaI fragment of pMOZII was replaced with the same fragment from pMOZIT272A	Ampr	This work
pMOZIIY280F/Y281F	The 0.6-kb BsaI fragment of pMOZII was replaced with the same fragment from pMOZIY280F/Y281F	Ampr	This work
pMOZIIN346V	The 0.6-kb BsaI fragment of pMOZII was replaced with the same fragment from pMOZITN346V	Ampr	This work
pMOZIIY374F	The 0.6-kb BsaI fragment of pMOZII was replaced with the same fragment from pMOZIY347F	Ampr	This work
pOX15	pRK404 with 4.7-kb ccoNOQP operon	Tetr	42
pOX15T272A	The 2.8-kb XhoI fragment of pOX15 was replaced with the same fragment from pMOZIIT272A	Tetr	This work
pOX15Y280F/Y281F	The 2.8-kb XhoI fragment of pOX15 was replaced with the same fragment from pMOZIIT280/281F	Tetr	This work
pOX15N346V	The 2.8-kb XhoI fragment of pOX15 was replaced with the same fragment from pMOZIIN346V	Tetr	This work
pOX15Y374F	The 2.8-kb XhoI fragment of pOX15 was replaced with the same fragment from pMOZIIY374F	Tetr	This work
pOX12	pOX15 lacking the 2.8-kb NO′ fragment (12.5 kb)	Tetr	29, 42
pRK2013	Self-transmissible plasmid; tra+ (RK2)	Kanr helper	30

^aNO′ refers to the entire ccoN gene and part of the ccoO gene.

Mutagenesis of the cbb₃-type oxygen reductase from R. capsulatus.

Molecular biology techniques were performed according to methods described previously (28). The oligonucleotides used for mutagenesis are listed in Table 2 and were synthesized by Thermo Fisher Scientific, Germany. Stratagene QuikChange kits were used to perform site-directed mutagenesis. Restriction enzymes were obtained from Fermentas. The pMOZI vector containing the 2.8-kb XhoI fragment from the ccoNOQP operon from R. capsulatus, encoding the subunit CcoN, was used as a template for site-directed mutagenesis. Sequence verification of the mutagenesis reactions were performed at the DNA sequencing facilities at Iontek (Turkey) and the University of Pennsylvania. The strains containing the mutated plasmids were named XL1-Blue/pMOZIT272A, XL1-Blue/pMOZIY280/281F, XL1-Blue/pMOZIN346V, and XL1-Blue/pMOZIY374F. After sequence verification, the 0.6-kb BsaAI (Ppu21I) fragment containing the desired mutation from pMOZI was inserted into pMOZII from which the 0.6-kb BsaAI fragment had been removed. Plasmids containing the inserted fragment in the correct orientation were detected by SalI digestion. Strains were named HB101/pMOZIIT272A, HB101/pMOZIIY280/281F, HB101/pMOZIIN346V, HB101/pMOZIIT219M, and HB101/pMOZIIY374F. The 2.8-kb XhoI fragments of pMOZIIT272A, pMOZIIY280/281F, pMOZIIN346V, and pMOZIIY374F containing the mutations were used to replace the equivalent region of the ccoN gene in pOX15 and transferred into E. coli HB101. The correct orientation of the inserted DNA fragments was checked by using HindIII, and the strains were named pOX15T272A, pOX15Y280/281F, pOX15N346V, and pOX15Y374F. Finally, each of these plasmids was transferred into strain GK32 of R. capsulatus, from which the ccoNO genes had been deleted (29). These strains were used for characterization of cytochrome c oxidase activity and of assembly of the mutant enzymes. Selected clones were transferred by triparental matings in the presence of HB101/pRK2013 (30) to R. capsulatus strain KZ101, in which both the ccoNOQP and cydAB operons, encoding the native respiratory oxygen reductases, were disabled. These strains were used to test for aerobic growth supported by the mutant enzymes and for measurements of proton pumping.

TABLE 2.

Oligonucleotides used for site-directed mutagenesis^a

^aF and R indicate forward and reverse primers, respectively. The bases in boldface type correspond to the genetic codes for amino acids to be mutated.

Isolation of chromatophore membranes, heme staining, and Western blotting.

Chromatophore membrane proteins of R. capsulatus were isolated as described previously (31) for the following strains (Table 1): MT1131 (wild type), GK32 (cbb₃ minus), GK32/pOX15, GK32/pOX15T272A, GK32/pOX15280/281F, GK32/pOX15N346V, and GK32/pOX15Y374F. Membranes were analyzed by using SDS-PAGE as described previously (32, 33). The subunits CcoO and CcoP, which contain covalently attached heme, were identified by heme staining (34), and the CcoN subunit was detected by using specific polyclonal antibodies as described previously (29).

Enzyme assays.

Visual detection of cytochrome c activity in colonies on agar plates was done by using the NADI staining test (35) (where NADI refers to alpha-naphthol + N′,N-dimethyl-p-phenylenediamine + O₂ → indophenol blue + H₂O reaction). Colonies of strains expressing wild-type cytochrome cbb₃ turn deep blue after 30 s, and this phenotype is designated NADI positive. Quantitative measurements of cytochrome c oxidase activity were carried out spectrophotometrically by monitoring the oxidation of reduced horse heart cytochrome c (Sigma, St. Louis, MO) at 550 nm at 25°C using a U3210 UV-visible spectrophotometer (Hitachi) as described previously (36).

Whole-cell proton pumping assay.

Cytochrome cbb₃ is located in both the cytoplasmic membrane and chromatophores of R. capsulatus. For convenience, the proton translocation characteristics of the wild type and the T272A mutant of cytochrome cbb₃ were determined by a whole-cell pumping assay with some modifications, as described previously (8). Proton translocation was determined by the pH change, monitored with a pH electrode. Cells were grown under photosynthetic conditions in anaerobic jars (Oxoid) at 35°C for 3 days. Whole-cell pumping experiments were done by using starved intact cells placed into a buffer containing 150 mM KCl, 100 mM potassium thiocyanate (KSCN) (Merck), and 0.5 mM HEPES (pH 7.4) (Sigma). Electron transfer from the cytochrome bc₁ complex to cytochrome cbb₃ was blocked by including 10 μM myxothiazol (Sigma) in the cell suspension. N,N,N′,N′-tetramethyl-p-phenylenediamide (TMPD) (Sigma) was used as the electron donor (1 mM final concentration). After the pH of the solution reached a constant value, whole cells were pulsed with 10 μl of air-saturated water at 25°C, and pH changes were monitored. Calibration was done by adding 10 μl of 1 mM HCl. To demonstrate that the pH changes thus observed required an active cytochrome cbb₃, the same experiments were performed in the presence of 100 μM potassium cyanide (KCN) (Sigma).

RESULTS

The genome of wild-type R. capsulatus (e.g., strain MT1131) encodes two different respiratory oxygen reductases, the cbb₃-type cytochrome c oxidase (ccoNOQP) and the bb′-type quinol oxidase (cydAB). Either of the two respiratory oxygen reductases can support aerobic growth under laboratory growth conditions, allowing either one to be genetically eliminated. In the absence of oxygen, or if both oxygen reductases are nonfunctional, cells cannot grow aerobically, but they can be grown photosynthetically. This physiological flexibility was exploited in the current work to characterize mutants of the R. capsulatus cbb₃-type oxygen reductase. Strain GK32 contains a chromosomal deletion of the ccoNO genes from the ccoNOQP operon and lacks functional cytochrome cbb₃. Operons encoding site-directed mutants of cytochrome cbb₃ were each expressed in GK32 by using a plasmid carrying the mutated ccoNOQP operon under the control of its native promoter. The following four mutants were engineered in subunit I (CcoN): Y374F, N346V, Y280F/Y281F, and T272A (Fig. 1). In each case, hydrophobic side chains replaced polar side chains. These R. capsulatus residues are equivalent to the following residues in P. stutzeri cytochrome cbb₃ (13): Y317, N289, Y227/Y228, and T215.

FIG 1.

The K^C channel with conserved residues on the structure of the cbb₃-type oxidase from P. stutzeri (PDB accession number 3MK7) (13), with Rhodobacter capsulatus numbering. Heme groups, iron ions, copper ions, and calcium ions are labeled. Targeted amino acids are circled. This figure was generated by using VMD software, version 1.9 (44).

The GK32 strains expressing cytochrome cbb₃ variants (Table 1) were grown under chemoheterotrophic conditions in which cytochrome cbb₃ is expressed. In the absence of a functional cbb₃-type oxygen reductase, the cells and membranes isolated from the cells are devoid of cytochrome c oxidase activity. This was evaluated qualitatively in colonies grown on agar plates by the NADI reaction, which monitors the oxidation of N,N-dimethyl-p-phenylenediamine via cytochrome c and cytochrome c oxidase. Representative NADI-stained plates are shown in Fig. 2.

FIG 2.

NADI-stained plates of positive (GK32/pOX15) (panels 1) and negative (GK32) (panels 2) controls with targeted mutants. (a3) GK32/pOX15T272A; (b4) GK32/pOX15Y280F/Y281F; (c5) GK32/pOX15N346V; (d6) GK32/pOX15Y374F.

Table 3 shows that wild-type strain MT1131 was NADI positive, as was strain GK32 carrying a plasmid expressing wild-type cytochrome cbb₃. In these strains, the deep-blue color developed in <30 s. The strains expressing the N346V and Y374F mutants were NADI negative, indicating little or no cytochrome c oxidase activity. Strains with the T272A or Y280F/Y281F mutation were slow to develop the NADI color, suggesting that they contained some functional cytochrome cbb₃.

TABLE 3.

NADI staining phenotypes of mutants with controls and quantitative analysis of cytochrome c oxidase activity in R. capsulatus membranes using a spectrophotometric assay^a

Strain	NADI phenotype	% of wt activity
GK32/pOX15	Positive	100
GK32	Negative	0
MT1131	Positive	47
GK32/pOX15T272A	Slow	169
GK32/pOX15Y280F/Y281F	Slow	79
GK32/pOX15N346V	Negative	9
GK32/pOX15Y374F	Negative	0

^aFor cytochrome c oxidation, 100% activity corresponds to the activity of strain GK32/pOX15, 1,602 nmol of cytochrome c/mg of membrane protein per min (cells grown under semiaerobic conditions). Each value shown is the mean of three independent measurements. wt, wild type.

These qualitative evaluations were confirmed by isolating chromatophore membranes from each strain and assaying spectrophotometrically the cytochrome c oxidase activity by monitoring the oxidation of reduced horse heart cytochrome c. The activity values (per mg of membrane protein) are also shown in Table 3. These values are dependent on both the total amount of cytochrome cbb₃ protein present in the membranes (i.e., the expression level) as well as the specific activity (per mg of pure enzyme) of the variant. The data show that the total cytochrome c oxidase activity present when the native enzyme was carried by a plasmid was about twice that of the wild-type strain, reflecting the increased copy number of the plasmid-borne operon compared to the chromosomal copy. The N346V and Y374F mutants had 9% and 0% of cytochrome c oxidase activity of membranes of the control strain, in agreement with their NADI-negative phenotype.

The NADI-slow phenotype on plates for both the Y280F/Y281F and T272A mutants correctly indicated that functional cytochrome cbb₃ was present, but this assay is not quantitative. Indeed, the total cytochrome c oxidase activity in the isolated membranes was high in each case, with the Y280F/Y281F mutant having 79% and the T272A mutant having 169% of the activity of membranes from the control strain. Clearly, these mutant enzymes are active, but their specific activities cannot be determined without quantifying the amount of enzyme present in the membranes.

The amount of enzyme present in the membranes was estimated by heme staining of the covalent heme c in the CcoO and CcoP subunits on an SDS-PAGE gel (Fig. 3) and by Western blotting of the subunit CcoN by using polyclonal antibodies against this subunit (Fig. 4).

FIG 3.

Cytochrome c profiles of wild-type (MT1131), negative-control (GK32), complemented positive-control (GK32/pOX15), and mutant (GK32/pOX15T272A, GK32/pOX15Y280F/Y281F, GK32/pOX15N346V, and GK32/pOX15Y374F) strains. Approximately 50 μg of protein was loaded into each lane of a 16.5% SDS-polyacrylamide gel. c-type cytochromes were detected by 3,3′,5,5′-tetramethylbenzidine (TMBZ) staining (34). Lanes: 1, GK32/pOX15Y374F; 2, GK32/pOX15Y280F/Y281F; 3, GK32/pOX15N346V; 4, GK32/pOX15T272A; 5, GK32/pOX15; 6, GK32; 7, wild-type strain MT1131. Cytochromes c_p and c_o are CcoP and CcoO of cytochrome cbb₃ oxidase, and cytochromes c₁ and c_y correspond to the cytochrome c₁ subunit of the bc₁ complex and the membrane-attached electron carrier c_y, respectively.

FIG 4.

Western blot analysis with anti-CcoN (subunit I) antibodies. After SDS-PAGE (10 μg of membrane proteins per lane) and electrophoretic transfer onto an Immobilon-P membrane, CcoN was detected with horseradish peroxidase-conjugated anti-rabbit immunoglobulin G with NiCl₂-enhanced 3,3′-diaminobenzidine as the substrate. Lanes: 1, GK32/pOX15Y374F; 2, GK32/pOX15Y280F/Y281F; 3, GK32/pOX15N346V; 4, GK32/pOX15T272A; 5, GK32/pOX15; 6, GK32; 7, wild-type strain MT1131.

Heme staining following SDS-PAGE of membranes isolated from strains with the native cytochrome cbb₃ (MT1131) and the complemented positive control (GK32/pOX15) showed four membrane-bound c-type cytochromes with molecular masses of 32, 31, 29, and 28 kDa. These correspond to CcoP, cytochrome c₁ of the bc₁ complex, cytochrome c_y, and CcoO, respectively (Fig. 3, lanes 5 and 7). As expected, the negative control (GK32) (Fig. 3, lane 6) did not contain the CcoP or CcoO subunits of cytochrome cbb₃. Each of the four mutants showed the presence of both CcoO and CcoP, suggesting that both the Y280F/Y281F (Fig. 3, lane 2) and T272A (Fig. 3, lane 4) mutants may be present in the membranes at slightly higher concentrations than the positive control (Fig. 3, lane 5).

Immunodetection using antibodies against CcoN showed the presence of this subunit in the control strains (Fig. 4, lanes 5 and 7) and its absence in a strain (GK32) in which the chromosomal operon has been disrupted (Fig. 4, lane 6). Immunodetection analyses of membranes from strains expressing each of the four mutants indicated the presence of CcoN in all cases. Quantitation was not attempted, but the amount of CcoN appears to be comparable to that of the positive control for each of the mutants (Fig. 4, lane 5).

Proton-pumping activity of the cytochrome cbb₃ mutants could be assayed only with those that had active cytochrome c oxidases (i.e., the Y280F/Y281F and T272A mutants). Each of these mutant enzymes was expressed in strain KZ101, in which the chromosomal operons of both respiratory oxygen reductases (ccoNOQP and cydAB) have been interrupted. This oxidase-deficient strain grows photosynthetically but not aerobically. Strain KZ101 expressing the T272A mutant was able to grow aerobically, but the expression of the Y280F/Y281F mutant oxidase did not confer the ability to grow aerobically. As a result, proton pumping was measured only for the T272A mutant enzyme. The Y280F/Y281F double mutant is potentially very interesting but will require additional work to determine why this mutant could not be grown aerobically.

Data for representative proton pumping experiments with the T272A mutant are shown in Fig. 5. The results show that the strain expressing the native cytochrome cbb₃ (Fig. 5a) pumps 0.5 ± 0.068 H⁺/e⁻¹ (about 2 H⁺ per O₂) and that the T272A mutant (Fig. 5c) pumps 0.1 ± 0.041 H⁺/e⁻¹ (about 0.4 H⁺ per O₂). Proton pumping was not observed for the control with an empty plasmid (Fig. 5b) or in the presence of 100 μM KCN (Fig. 5d), which blocks the cbb₃-type oxygen reductase. These data demonstrate that the T272A mutation reduces the proton-pumping activity of cytochrome cbb₃ to a level of approximately 20% of that of the wild-type control.

FIG 5.

Change in pH upon addition of 10 μl of air-saturated water (about 2.5 nmol O₂) to suspensions of cells. (a) KZ101/pOX15; (b) KZ101/pRK415; (c) KZ101/pOX15T272A; (d) KZ101/pOX15T272A plus 100 μM KCN. The pH changes caused by oxygen addition were calibrated by adding a solution containing HCl.

DISCUSSION

The current work is the first to examine the influence of a mutation on the proton-pumping stoichiometry of a C family respiratory oxygen reductase. For A and B family heme-copper oxygen reductases, proton pumping can be evaluated by using purified enzymes reconstituted in phospholipid vesicles. Efforts to demonstrate proton pumping by a C family enzyme in reconstituted vesicles have not been successful, with the exception of recent work in which the proton-pumping stoichiometry by the purified, reconstituted cytochrome cbb₃ from R. sphaeroides was demonstrated to be 1 H⁺/e⁻¹ (37). This work required that the enzyme be electrochemically reduced just prior to measurements of proton pumping. Alternatively, we measured proton pumping for several cbb₃-type oxygen reductases using intact cells in which cytochrome cbb₃ is the only respiratory oxygen reductase that is present (8). The stoichiometry obtained by this method is 0.5 H⁺/e⁻¹, which may reflect a lower proton-pumping efficiency in vivo or may be a low value due to an artifact of the technique, possibly uncoupling due to the use of TMPD as the electron donor (38). If there is an artifact due to the use of TMPD, it should be systematic and should not vary with the mutant being examined. For the purposes of the current work, the important points are that the native enzymes pump protons and that the assay is reliable, reproducible, and useful to evaluate mutant phenotypes.

The structure of cytochrome cbb₃ from P. stutzeri (13), the pattern of conserved polar residues (18), and the results of previous site-directed mutagenesis experiments (18) all suggest that the proton-conducting D channel does not exist within the C family enzymes. In the A family of heme-copper oxygen reductases, the D channel is required for proton pumping (20–24), but despite the absence of the D channel, enzymes of the C family pump protons (8, 39). If the K^C channel is the only proton input channel, it must be used as the conduit for all chemical as well as all pumped protons.

Previous site-directed mutagenesis studies of the C family enzymes from R. sphaeroides and V. cholerae identified mutations within the K^C channel which block catalytic turnover (18, 25), demonstrating that the channel is used by chemical protons to reach the active site. The Rhodobacter capsulatus N346V mutant (RcN346V) and RcY373F (Fig. 1, Table 3) have little or no cytochrome c oxidase activity, although the enzymes appear to be properly assembled and are present in the membrane at approximately the same level as that of the native enzyme. Mutations in the residues equivalent to RcN346 in the R. sphaeroides (RsN349G) and V. cholerae (VcN293D) enzymes have 0% and 26% cytochrome c oxidase specific activities, respectively (18), confirming the functional importance of this residue. Mutation of the residues equivalent to RcY373F in the R. sphaeroides (RsY377G) and V. cholerae (VcY321F) enzymes results in inactive enzymes, consistent with the blocking of chemical protons from reaching the active site via the K^C channel (18).

By computationally placing water within the structure of P. stutzeri cytochrome cbb₃, it was proposed (40) that the chemical and pumped protons share a pathway from the entrance of the K^C channel to P. stutzeri residue Y317 (PsY317) (RcY374), which is a bifurcation point. Beyond this residue, chemical protons follow a pathway involving PsT215 (RcT272), and pumped protons follow a different pathway toward PsE323 to the proposed proton loading site.

The most interesting mutant examined in this work is RcT272A (Fig. 1), which is approximately as active as the native enzyme in R. capsulatus. A previous molecular dynamics study (40) indicated that this threonine is hydrogen bonded to a chain of water molecules that leads to the active site of the enzyme. The equivalent residue was mutated in V. cholerae (VcT219V), and the purified enzyme had 48% of the wild-type cytochrome c oxidase activity (18).

The significant new data presented in this work (Fig. 5) show that the proton-pumping stoichiometry of the RcT272A mutant is only about 20% of that of the wild-type enzyme. As expected, no pumping was observed in the presence of 100 μM KCN, demonstrating that the observed pumping is due entirely to this enzyme. Hence, this mutation decouples the proton pump despite the relatively high catalytic turnover rate.

The mechanism by which RcT272 decouples the proton pump is not known. The mutation may alter a gating mechanism which would normally lower the rate of proton transfer to the active site relative to the rate of transfer to the proton loading site via the “pumping route” leading from the proposed bifurcation point RcY374 (40). In this way, it can be speculated that protons that would normally be directed through the pumping branch are instead consumed at the active site. Although these data do not definitively prove that pumped protons are taken up via the K^C channel, this is the most plausible explanation.