Membrane protein complexes can support both the generation and utilisation of a transmembrane electrochemical proton potential (‘proton‐motive force’), either by transmembrane electron transfer coupled to protolytic reactions on opposite sides of the membrane or by transmembrane proton transfer. Here we provide the first evidence that both of these mechanisms are combined in the case of a specific respiratory membrane protein complex, the dihaem‐containing quinol:fumarate reductase (QFR) of Wolinella succinogenes, so as to facilitate transmembrane electron transfer by transmembrane proton transfer. We also demonstrate the non‐functionality of this novel transmembrane proton transfer pathway (‘E‐pathway’) in a variant QFR where a key glutamate residue has been replaced. The ‘E‐pathway’, discussed on the basis of the 1.78‐Å‐resolution crystal structure of QFR, can be concluded to be essential also for the viability of pathogenic ε‐proteobacteria such as Helicobacter pylori and is possibly relevant to proton transfer in other dihaem‐containing membrane proteins, performing very different physiological functions.
According to chemiosmotic theory (see Mitchell, 1979, for a review), the energy released upon the oxidation of electron donor substrates, in both aerobic and anaerobic respiration, is transiently stored in the form of a proton potential, Δp, across the energy‐transducing membranes, which can then be used by the ATP synthase for ADP phosphorylation with inorganic phosphate. Fundamentally, there are two mechanisms (Richardson and Sawers, 2002) by which integral membrane proteins can act as catalysts in this coupling of electron transfer reactions to Δp generation. Firstly, the redox loop mechanism essentially involves transmembrane electron transfer coupled to protolytic reactions on opposite sides of the membrane. In a simple form, this mechanism is represented by the formate dehydrogenase (Jormakka et al, 2002) and membrane‐bound nitrate reductase (Bertero et al, 2003) enzymes of anaerobic respiration and in a more complicated form by the Q‐cycle of the cytochrome bc1 complex, complex III of the aerobic respiratory chain (Mitchell, 1976). Secondly, the proton pump mechanism involves transmembrane proton translocation by a Grotthuss‐type mechanism (von Grotthuss, 1806; Agmon, 1995), driven by the transfer of electrons between sites of catalysis via the prosthetic groups of the membrane protein complex. Such a mechanism is thought to be operative, for example, in cytochrome c oxidase, complex IV of the respiratory chain (Faxén et al, 2005).
In a specific case of a single respiratory membrane protein complex, the experimental results support (Haas et al, 2005; Lancaster et al, 2005; Mileni et al, 2005) both of these mechanisms being harnessed together, so as to ensure the counterbalancing of their effects for energetic reasons. The membrane protein in question is the quinol:fumarate reductase (QFR) from the anaerobic ε‐proteobacterium Wolinella succinogenes. QFR contains two haem b groups oriented along the membrane normal and bound by the transmembrane subunit C (see Figure 1A), which are termed the ‘proximal haem’, bP, and the ‘distal haem’, bD, according to their relative proximity to the hydrophilic subunits A and B. It is the terminal enzyme of fumarate respiration, a form of anaerobic respiration that allows anaerobic bacteria to use fumarate instead of dioxygen as the terminal electron acceptor (Kröger, 1978; Lancaster, 2004). QFR couples the two‐electron reduction of fumarate to succinate to the two‐electron oxidation of quinol to quinone. This reaction is part of an electron transfer chain that enables the bacterium to grow with various electron donor substrates such as formate or hydrogen (Supplementary Figure S1).
Although experiments performed with inverted vesicles and proteoliposomes containing QFR demonstrated that the reaction catalysed by the dihaem‐containing QFR from W. succinogenes is not associated directly with Δp generation (Geisler et al, 1994; Biel et al, 2002; Kröger et al, 2002), the three‐dimensional structure of QFR, initially solved at 2.2 Å resolution (Lancaster et al, 1999), revealed locations of the active sites of fumarate reduction (Lancaster et al, 2001) and of menaquinol oxidation (Lancaster et al, 2000) that are oriented towards opposite sides of the membrane (Figure 1A), thus indicating that menaquinol oxidation by fumarate, as catalysed by W. succinogenes QFR, should be associated directly with the establishment of a Δp. To reconcile these apparently conflicting experimental observations, the so‐called ‘E‐pathway hypothesis’ (Lancaster, 2002) was proposed (Figure 1B). According to this hypothesis, the transfer of two electrons via the two QFR haem groups is strictly coupled to a compensatory, parallel transfer of two protons across the membrane via an unprecedented proton transfer pathway, which is transiently open during the reduction of the two haem groups and closed in the oxidised state of the enzyme. The latter is a requirement for the strict coupling of transmembrane proton transfer to electron transfer and for preventing the QFR from operating as an uncoupler of fumarate respiration (Supplementary Figure S1). The two most prominent constituents of the proposed pathway were suggested to be the haem bD ring C propionate and the amino‐acid residue Glu C180, which is conserved in dihaem‐containing QFR from ε‐proteobacteria (Lancaster, 2002; Supplementary Figure S2) and after which the ‘E‐pathway’ was named.
Since the proposal of this hypothesis, a number of theoretical (Haas and Lancaster, 2004) and experimental results (Haas et al, 2005; Lancaster et al, 2005; Mileni et al, 2005) have been obtained that supported it, but did not prove it. Here we provide, for the first time, evidence for the presence of such a compensatory ‘E‐pathway’ in the wild‐type enzyme and the non‐functionality of the ‘E‐pathway’ in a variant enzyme lacking Glu C180.
Results and discussion
In its detergent‐solubilised state, the enzyme variant E180Q, obtained by replacing Glu C180 with Gln, exhibited only 1/10 of the specific activity of the wild‐type enzyme for 2,3‐dimethyl,1,4‐naphthoquinol (DMNH2) oxidation by fumarate, whereas the Michaelis constant KM for DMNH2 was unchanged (Lancaster et al, 2005). However, compared to the wild‐type enzyme, a significant increase of approximately 50 mV in the oxidation‐reduction midpoint potential of haem bP was found for the E180Q variant (EM7=+39 mV; Lancaster et al, 2005). Considering that the [3Fe‐4S] cluster (Supplementary Figure S1) has been reported to exhibit a redox midpoint potential of −61 mV (Mileni et al, 2006), such an increase in the midpoint potential makes electron transfer from bP to the [3Fe‐4S] cluster two times more energetically unfavourable than the corresponding process in the wild‐type enzyme. In addition, the redox midpoint potential of haem bP in the E180Q variant is higher than that of the fumarate/succinate couple (EM7=25 mV; Ohnishi et al, 2000). Both of these features serve as explanations for the reduced activity of the detergent‐solubilised E180Q variant. However, when reconstituted into proteoliposomes and therefore bound by sealed membrane vesicles, the E180Q variant exhibited even less and ostensibly negligible enzymatic activity of DMNH2 oxidation by fumarate (left half of Figure 1C). Closer inspection (Figure 1D) indicated that this lack of activity was only obvious after approximately 20 s. Prior to that, the membrane‐bound variant enzyme was weakly catalytically active, which, according to the scheme in Figure 1A, should lead to the generation of a Δp, against which the reaction could no longer be catalysed (see Supplementary discussion for a more quantitative analysis). We attributed this rapid loss in activity to be caused by the small difference in oxidation‐reduction midpoint potential between the 2,3‐dimethyl,1,4‐naphthoquinone (DMN)/DMNH2 couple routinely used (EM7.3=−35 mV; see Table I) and the fumarate/succinate couple, rendering the overall reaction only mildly exergonic (ΔG≈−12 kJ/mol) under standard conditions (at pH 7), and apparently not exergonic enough to support sustained Δp generation (Figure 1A). These conclusions are supported by our finding that addition of the protonophore carbonyl cyanide m‐chlorophenylhydrazone (CCCP) to ostensibly inactive proteoliposomal E180Q‐QFR enabled the catalysis of DMNH2 oxidation (right half of Figure 1C).
Two components contribute to Δp. These are the concentration difference of protons across the membrane, ΔpH, and the difference in electrical potential between the two membrane‐separated aqueous phases, ΔΨ. In principle, according to Figure 1A, we should have measured the generation of a ΔpH based on the acidification of the interior of proteoliposomes containing E180Q‐QFR and the generation of a ΔΨ by monitoring the uptake of tetraphenylborate (TPB−) with a TPB−‐selective electrode. However, due to the small energy difference of the redox couples involved, the resulting low rates (prior to the addition of uncoupler) and the short duration of turnover, any detection of ΔpH or ΔΨ for this reaction was not reliably possible with our experimental setup.
These developments prompted us to design and synthesise a substrate analogue with suitable substituents to lower the redox midpoint potential of the respective naphthoquinone/naphthoquinol couple, thus increasing the driving force of the catalysed oxidation by fumarate. Based on textbook literature (Fieser and Fieser, 1965), we expected substituents such as amino or methylamino groups (Supplementary Figure S4), when replacing the 2‐methyl group of DMN to lower the redox midpoint potential. As expected, both 2‐amino‐3‐methyl‐1,4‐naphthoquinone (AMN) and 2‐methyl‐3‐methylamino‐1,4‐naphthoquinone (MMAN) exhibited lower redox midpoint potentials (see Supplementary data). However, the Michaelis constant KM (see Table I) for AMNH2 as a QFR substrate was an order of magnitude higher than the others, resulting in a correspondingly lower specific enzymatic activity of quinol oxidation (see Table I), so that further characterisation focussed on DMNH2 and MMANH2. The in situ redox midpoint potential for the protein‐bound MMAN/MMANH2 couple was found to be approximately 90 mV lower than that of the DMN/DMNH2 couple (Table I), rendering the overall reaction of MMANH2 an estimated two‐and‐half times more exergonic (ΔG≈−30 kJ/mol) under standard conditions at pH 7 and possibly exergonic enough to support measurable Δp generation.
We determined ΔpH by monitoring the acidification of the interior of QFR‐containing proteoliposomes in the presence of valinomycin and 100 mM KCl. We found that, despite being catalytically active (Figure 2A), wild‐type QFR did not contribute to the generation of a ΔpH (Figure 2B), irrespective of whether DMNH2 or MMANH2 was provided as substrate (see Table I). When wild‐type QFR was replaced by the E180Q variant, the membrane‐bound enzyme did not significantly support the oxidation of DMNH2 (left half of Figure 1C). It did, however, catalyse the oxidation of MMANH2 by fumarate (Figures 2A and B), establishing a ΔpH of 1.6 pH units (see Supplementary data). The H+/e− ratios associated with the quinol oxidation by fumarate were determined from the ratios of the rate constant of acidification of the proteoliposomal interior and that of quinol oxidation in the same experiment. We found that the oxidation of 1 mol MMANH2 by E180Q‐QFR resulted in an increase of 2.0 mol H+ (Supplementary Table SII), corresponding to an H+/e− ratio of 1.0 (Table I). Specific inhibition at the quinol oxidation site upon addition of the competitive inhibitor 2‐decyl‐3‐hydroxy‐1,4‐naphthoquinone (DHN) with an IC50 of 8 μM (see Supplementary data) abolished the effects observed for E180Q‐QFR (Figures 1C, 2A and B). The apparent slowing and eventual reversal of the MMANH2 oxidation process after approximately 30 s (Figure 2A) can be attributed to the protonation of the product MMAN to MMANH+. The latter has a much lower extinction coefficient Δε280−320 versus MMANH2 than MMAN (Supplementary Figure S5A). This is supported by the observation that the rate of acidification of the proteoliposomal interior (Figure 2B) is still significant, even after 50 s, but is eventually only one fifth of the initial rate (see also Supplementary Figure S8). We conclude that, after significant acidification of the proteoliposomal interior, one of the two protons liberated upon oxidation of MMANH2 to MMAN is no longer released to the interior of the proteoliposomes but instead protonates the product MMAN to MMANH+.
We determined ΔΨ by monitoring the uptake of TPB− with a TPB−‐selective electrode (Figure 2C). Catalysis of MMANH2 oxidation by fumarate by E180Q‐QFR established a ΔΨ of 86 mV (positive inside, Table II). This was inhibited by the addition of DHN (see Figure 2C and Supplementary discussion). No ΔΨ was detected during analogous catalysis by wild‐type QFR. In the case of both the wild type and the E180Q enzyme, no generation of a ΔΨ by means of TPB− uptake could be detected after replacement of MMANH2 with DMNH2.
Considering the measurements of both ΔpH and ΔΨ, Δp generation can be demonstrated only for MMANH2 oxidation by fumarate as catalysed by E180Q‐QFR. Comparison to catalysis by the wild‐type enzyme clearly proves the presence of the compensatory ‘E‐pathway’ in the wild‐type enzyme and its non‐functionality in E180Q‐QFR. Comparison of these results to the inability of membrane‐bound E180Q‐QFR to support sustained DMNH2 oxidation in the absence of CCCP demonstrates that this latter observation is due to the insufficient driving force of DMNH2 oxidation by fumarate in the absence of a compensatory E‐pathway. In contrast, catalysis of the same reaction by the membrane‐bound wild‐type enzyme demonstrates the facilitation of transmembrane electron transfer by transmembrane proton transfer.
The X‐ray crystal structure, refined at 1.78‐Å resolution (see Supplementary Table S1), of the oxidised W. succinogenes QFR at pH 6 most importantly reveals the side chain of Glu C180 to be found in two approximately equally populated alternate conformations, one a so‐called ‘intermediate’ orientation, oriented between the two haem groups, and a second conformer in a so‐called ‘distal’ conformation, oriented more towards the distal haem group (Figure 3A and B). These conformers are analogous to those predicted by multiconformation continuum electrostatic calculations (Haas and Lancaster, 2004), where their relative occupancy and degree of protonation are predicted to depend on the redox state of the haem groups, and therefore be functionally relevant within the context of the E‐pathway hypothesis. These theoretical results are supported by experimental results from redox‐induced Fourier‐transform infrared difference spectroscopy (Haas et al, 2005), which indicate that the side chain of Glu C180 undergoes a combination of a change in protonation and its environment upon reduction of the enzyme.
Although all available QFR crystal structures are those of the oxidised enzyme and the E‐pathway is required to be non‐functional in the oxidised state, inspection of the crystal structure provides clues as to which further groups could contribute to the ‘E‐pathway’ (Figure 3A and B). Large parts of transmembrane helix VI and a significant part of transmembrane helix V are amphipathic (Figure 3C), and their hydrophilic sides are oriented towards one another, thus providing, in addition to the previously proposed and supported roles of the transmembrane helix V residue Glu C180 and the haem bD ring C propionate, further possible constituents (Figure 3D). The role of these additional residues as possible constituents of the ‘E‐pathway’ is currently being investigated by site‐directed mutagenesis. However, the results presented here, in agreement with previous theoretical and experimental results, strongly support the view that Glu C180 is the central ‘switch’ in the coupling of transmembrane electron and proton transfer.
What we have now proved to be the ‘E‐pathway’ has been demonstrated to be essential for the growth of W. succinogenes with fumarate as the terminal electron acceptor (Lancaster et al, 2005). In turn, QFR has been shown to be essential for the viability of the pathogenic ε‐proteobacterium Helicobacter pylori in the murine stomach (Ge et al, 2000). It is therefore reasonable to conclude that the ‘E‐pathway’ is also essential for the viability of H. pylori in its host.
Succinate oxidation by menaquinone, an endergonic reaction under standard conditions, is catalysed by dihaem‐containing succinate:menaquinone reductases in several Gram‐positive bacteria, for example, B. subtilis. These enzymes are similar to W. succinogenes QFR also with respect to the location of the four histidine residues ligating the haem groups (Hägerhäll and Hederstedt, 1996), the conservation of an acidic residue at the position of Glu C66, demonstrated to be selectively essential for menaquinol oxidation (Lancaster et al, 2000), the location of the site of menaquinone reduction close to the haem bD (Matsson et al, 2000), but different in that they lack an acidic residue at the position of Glu C180 (Supplementary Figure S2). Experimental results on whole cells and crude membranes indicated that succinate oxidation by menaquinone results in charge imbalance across the membrane, that is, is electrogenic, in B. subtilis and is driven by Δp (Schirawski and Unden, 1998), and that catalysing the reaction in the opposite direction generates a Δp (Schnorpfeil et al, 2001). In this context, the results presented here on E180Q‐QFR also present the first experimental evidence on electrogenic catalysis by an isolated and reconstituted succinate:menaquinone reductase.
The electric compensation of transmembrane electron transfer against membrane potential may be a more general phenomenon. Under conditions where the free energy gap of electron transfer is small, it appears that even in the cytochrome bc1 complex, transmembrane electron transfer can be electroneutral (Mulkidjanian et al, 1991). However, this electroneutrality need not involve proton transfer and other principles of electric compensation (Lancaster, 2003) may be applicable in this case (Mulkidjanian, 2005). On the other hand, transmembrane proton transfer pathways similar to the ‘E‐pathway’ may be relevant also in other, hitherto less well‐studied membrane proteins, containing two haem groups, which are oriented along the membrane normal so as to support transmembrane electron transfer. These membrane proteins can have very different physiological roles, such as NADPH oxidases, which, through the generation of superoxide and its derivatives, are thought to be involved in a variety of important cellular functions in phagocytic and non‐phagocytic cells and whose involvement in transmembrane proton translocation is currently highly disputed (DeCoursey et al, 2002; Henderson and Meech, 2002; Maturana et al, 2002).
Materials and methods
QFR production, crystallisation, data collection and processing, and crystallographic refinement of wild‐type QFR
Wild‐type and E180Q QFR were produced and purified as described earlier (Lancaster et al, 1999; Lancaster et al, 2005). Wild‐type QFR was crystallised also as described earlier (Lancaster et al, 1999). X‐ray diffraction data were collected at the European Synchrotron Radiation facility (ESRF, Grenoble, France). The crystal structure of QFR was determined to 1.78 Å resolution and refined to R and Rfree values of 0.229 and 0.237, respectively (Supplementary Table S1). The atomic coordinates of the oxidised QFR wild‐type enzyme have been deposited in the Protein Data Bank with the accession number 2BS2.
Synthesis of naphthoquinones
DMN was synthesised as described earlier (Lancaster et al, 2005). The synthesis of AMN, MMAN and 2‐decyl‐3‐hydroxy‐1,4‐naphthoquinone is described in Supplementary methods. The oxidation‐reduction midpoint potentials of DMN, AMN and MMAN in solution were determined (see Supplementary methods) to be 20, 50 and 80 mV lower, respectively, than that of menaquinone‐4 (vitamin K2).
The oxidation of quinols by fumarate was recorded as the absorbance difference. For DMNH2, the difference in absorption at 270 nm minus that at 290 nm (Δε270−290=15.2 mM−1 cm−1) (Lancaster et al, 2000), for MMANH2 the difference in absorption at 280 nm minus that at 320 nm (Δε280−320=27.6 mM−1 cm−1), and for AMNH2 the difference in absorption at 270 nm minus that at 310 nm (Δε270−310=24.6 mM−1 cm−1) were used (Supplementary Figure S5).
Sonicated liposomes were prepared with phosphatidylcholine (Avanti Polar Lipids) and di‐palmitoyl phosphatidate (Fluka, Buchs, Switzerland) as described (Biel et al, 2002), except that the liposome film was resuspended in 5 mM HEPES buffer (adjusted to pH 7.5 with NaOH) containing 100 mM KCl. For the acidification measurements, 10 mM pyranine (Molecular Probes, Leiden, NL) was added to the buffer. The suspension of the liposomes contained 10 g phospholipids/l. Proteoliposomes with reconstituted wild‐type or E180Q‐QFR from W. succinogenes were prepared as described earlier (Biel et al, 2002). Sonicated liposomes were treated with dodecyl‐β‐d‐maltoside (0.8 g/g phospholipid) in a suspension containing 5 mM HEPES buffer (adjusted to pH 7.5 with NaOH) and the naphthoquinone analogues for 3 h under constant stirring. The following addition of wild‐type QFR or E180Q‐QFR (0.16 g/g phospholipids) was carried out stepwise, stirring was continued for 1 h. The detergent was removed with Bio‐Beads SM‐2 (Bio‐Rad) (0.24 g ml−1) under continued stirring for 1 h. The suspension was sonicated for 20 s at 0°C before use. All steps were carried out using anaerobised buffers. The solution and the Bio‐Beads were separated by gentle centrifugation (30 s at 5000 g). The proteoliposomes were concentrated by harsh centrifugation (1 min at 16 100 g), and the supernatant was removed. The concentration of the proteoliposomes was then adjusted with fresh buffer to 10 g phospholipids/l.
Acidification measurements, H+/e− ratio
The proteoliposomes were added to 50 mM HEPES buffer (adjusted to pH 7.5 with NaOH) containing 100 mM KCl and 5 μM valinomycin per gram of phospholipid. MMAN was reduced with NaBH4. The reaction was started by addition of 40 μM fumarate. The MMANH2 oxidation by fumarate was recorded as the absorbance difference at 280 nm minus 320 nm (Supplementary methods). The ΔpH inside the proteoliposomes was recorded as the amplitude of the absorbance ratio at 450 and 415 nm calibrated with 119.3 mM HCl (Supplementary methods). The values expressing the ratio of the acidification of the proteoliposomal interior per reduced quinone (H+/e−) were calculated from the time constants monitored under steady‐state conditions. A diode‐array UV/VIS spectrophotometer (Agilent 4853) was used for recording different wavelengths simultaneously; therefore, no further scaling was needed.
Measurement of ΔΨ
The TPB− electrode was constructed as described (Karlovsky and Dadak, 1982). Proteoliposomes were suspended in 15 mM HEPES buffer (adjusted to pH 7.5 with NaOH) containing 100 mM LiCl. The buffers were anaerobised and flushed with N2 prior to use. The ΔΨ was calculated by means of the TPB− uptake. The concentration of TPB− within the proteoliposomes (Ti) and in the medium (Te) was calculated using the Nernst equation. Ti was calculated from the maximum amount of TPB− (Ts in mol/g phospholipid) taken up from the medium in the steady state of electron transport according to Equation (1) (Zaritsky et al, 1981; Geisler et al, 1994; Biel et al, 2002).
where Vi (3.5 ml g phospholipids−1) (Biel, 2002) represents the average internal volume of the proteoliposomes and KT is the binding constant for the reporter ion (3050 ml/g phospholipids) (Biel, 2002). The internal reporter ion concentration (Ti)n+1 was calculated by the iteration of an assumed value for (Ti)1 (Equation (1)) until (Ti)n+1 was equal to (Ti)n.
Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).
We thank O Schürmann and A Roth for technical assistance, T Prisner for access to the cyclic voltammetry equipment, B Trumpower for providing an initial sample of 2‐decyl‐3‐hydroxy‐1,4‐naphthoquinone, H Belrhali and S Monaco for being respective local contacts at ESRF beamlines ID14‐EH3 and ‐EH1, the Deutsche Forschungsgemeinschaft (SFB 472 ‘Molecular Bioenergetics’) and the Max Planck Society for funding.
- Copyright © 2006 European Molecular Biology Organization