Quinones operate as proton-collecting antennas in energy-transducing membranes

Significance

Quinones are central carriers of redox energy in biological membranes. Here, we show that ubiquinone Q₁₀, the main lipophilic redox carrier in respiratory chains, strongly influences the protonation reactions at the membrane surface. Our combined findings suggest that quinones interact with protons by noncovalent interactions between the Q headgroup and protonated water species at the membrane surface. The interaction kinetically favors proton exchange along the membrane plane, over the membrane-bulk exchange, and results in a higher concentration of protons at the membrane surface. We suggest that in addition to their function as redox carriers, quinones support the role of the membrane as a proton-collecting antenna and mediate the local proton currents between bioenergetic complexes.

Abstract

The bioenergetic complexes of energy-transducing membranes generate a proton current that powers the synthesis of adenosine triphosphate. Yet, since the early days of the chemiosmotic theory, it has remained elusive and much debated whether the proton motive force (PMF) delocalizes into the bulk solvent surrounding the energy-transducing membrane or if the thermodynamic force is exerted as a localized proton current along the membrane surface. To elucidate the molecular principles underlying protonation dynamics at biological membranes, we combine here proteoliposome experiments with fluorescence correlation spectroscopy and multiscale molecular simulations. We show that ubiquinone (Q₁₀), which is an essential electron carrier of inner mitochondrial membranes, interacts with protons at the membrane, and alters the rate of the protonation reactions along the surface. We find that physiological Q₁₀ concentrations increase the integrity of the liposome membranes to sustain a PMF and enhance the rate of surface protonation reactions of lipid-conjugated pH-sensitive fluorophores, occurring on a microsecond timescale. Our multiscale simulations reveal that the quinone headgroup localizes at the membrane surface and stabilizes protonated water species by cation–π and hydrogen-bonded interactions amplifying the proton exchange on the surface relative to the bulk solvent. We suggest that in addition to the well-established role of quinones as redox mediators in energy-transducing membranes, Q₁₀ also promotes the proton-collecting antenna effect, mediating proton exchange along the membrane and supporting a local proton circuit model. Our combined findings provide molecular insight into propagation of proton currents along biological membranes and reveal key principles underlying the energy conversion mechanisms in biology.

Biological energy conversion is powered by a membrane-bound protein machinery that transfers electrons via redox cofactors and converts the energy into an electrochemical proton gradient across biological membranes (1, 2). This proton-coupled electron transfer (PCET) process creates a proton motive force (PMF) that drives the synthesis of adenosine triphosphate (ATP) and powers active transport (1, 3). The chemiosmotic energy transduction mechanism is conserved across all domains of life, from bacteria and archaea to mitochondria and chloroplasts of the eukaryotic cell. Despite the thermodynamic insight provided by the chemiosmotic theory (1), the exact molecular principles by which the PMF drives the energy conversion remain unclear and much debated. In this regard, models favoring a delocalized PMF driving ATP synthesis have been proposed (1), but several findings also support that the PMF could operate as a local proton circuit along the membrane (4–10) (Fig. 1A).

Fig. 1.

Chemiosmotic coupling principles and structure of ubiquinone Q₁₀. (A) In biological energy conversion, redox-driven proton pumps generate a proton motive force (PMF) that power ATP synthesis. Two models are schematically presented, where the PMF is mediated via localized PMF transported along the lipid-membrane (cf. refs. 4 and 5) or via a delocalized PMF that extends to the bulk solvent (1). (B) The structure of ubiquinone (Q₁₀), with the headgroup highlighted in yellow.

Proton transfer reactions take place by a Grotthuss-type proton hopping process (11, 12), where the charge rather than the proton itself is transferred along continuous chains of hydrogen-bonded water molecules, followed by reorganization of the hydrogen-bonded network. The Grotthuss-mediated proton transfer reactions were recognized early on (4, 5, 13, 14), and account for the significantly faster proton conduction relative to other ions (15). Proton wires, which enable such proton transfer reactions, have been observed in a wide range of bioenergetic membrane complexes that drive the proton transfer reactions by redox-driven electric field effects (16–22, cf. also refs. 23–27).

While electrons can tunnel across large (>10 Å) molecular distances between redox-cofactors within the bioenergetic protein complexes (28, 29), the electron transfer between the protein complexes require mobile electron carriers, such as the soluble redox proteins cytochrome c, ferredoxin, or plastocyanin (2). In this context, quinone molecules serve as the main membrane-bound electron carriers in biological membranes that shuttle electrons in respiratory and photosynthetic energy conversion chains. This charge-carrier function is enabled by the quinone (Q) headgroup, comprising a redox active benzoquinone moiety (Fig. 1B) that undergoes a two-electron proton-coupled oxidoreduction (2 e⁻ + 2 H⁺) from its oxidized quinone (Q) form to the reduced quinol (QH₂) form (E_m ∼+90 mV vs. NHE) (30), catalyzed by the bioenergetic enzyme complexes. Quinones also form stable one-electron reduced semiquinone radicals (Q^•/− or QH^•), which are of central mechanistic importance in several enzymes, but can lead to the generation of harmful reactive oxygen species (ROS) upon reactions with molecular oxygen (31). The quinone headgroup is further attached to a long isoprenoid tail, comprising up to 10 isoprenoid units in ubiquinone (Q₁₀) (Fig. 1B).

In the mitochondrial electron transport chain (ETC) and oxidative phosphorylation (OXPHOS), Q₁₀ mediates the electron transfer between complexes I and III (2). Complex I transfers electrons from nicotinamide adenine dinucleotide (NADH) to Q, which is bound in the hydrophilic domain of Complex I, with the formation of quinol (QH₂) driving proton pumping across the membrane (16, 32). Complex II oxidizes succinate to fumarate and also contributes to the formation of the quinol pool, but does not generate a PMF. The quinol then carries the electrons onward to Complex III that reoxidizes QH₂ to Q, and releases the protons across the membrane by a Mitchell-Lundegårdh redox-loop process (1, 33, 34). The electrons are shuttled onward to Complex IV, which reduces oxygen to water and pumps protons across the membrane (34). The PMF generated by CI, CIII, and CIV subsequently powers the synthesis of ATP from ADP and P_i via the proton-driven rotary motion of ATP synthase (3). Cryoelectron tomography (cryo-ET) studies have revealed that the OXPHOS complexes associate into supercomplexes in the flat regions of the inner mitochondrial membranes (IMMs) (35), whereas ATP synthesis predominantly takes place in the curved cristae (35), thus necessitating the effective transport of the protons between spatially distant regions of the mitochondria. Despite different molecular principles, the photosynthetic energy conversion machinery also relies on quinones that carry the redox energy from the chlorophyll-based photochemistry driving water oxidation to (plasto)quinone reduction, while the (plasto)quinol pool is reoxidized by the b₆f complex. Similar to mitochondria and respiring bacteria, the photosynthetic PMF drives the synthesis of ATP (36).

Charged and polar amino acids at the entrance of the proton channels in the bioenergetic protein complexes together with physiologically relevant lipids, such as cardiolipin (37–39) are known to mediate the proton uptake from the bulk, and exceeding diffusion-limited reaction rates (40–43). These protonation sites function as proton-collecting antennas (PCAs) that kinetically favor the protonation reactions. However, protons can also directly associate with the lipid membrane, with several experiments supporting a 2D-diffusion of protons along the membrane surface. Heberle et al. (6) observed that the PMF generated by integral membrane proteins, such as bacteriorhodopsin, lead to a faster proton transfer along the membrane surface as compared to the proton exchange with the bulk phase, and suggested that protons could be transported between source and sink complexes and contribute to the chemiosmotic coupling mechanism. This was further corroborated by fluorescence correlation spectroscopy (FCS) measurements monitoring the local proton exchange dynamics of individual pH-sensitive fluorophores, and showing increased protonation rates of membrane-bound fluorophores compared to fluorophores freely diffusing in the same aqueous solution (8). The observations demonstrated that certain lipid headgroups (e.g., DOPC and DOPG) in membranes of small unilamellar vesicles (SUVs) can collectively act as a PCA. Sandén et al. (9) further observed that this PCA effect depends on the ionic strength of the bulk solution, charge of the lipid headgroups, as well as pH, with stronger PCA effects observed at higher pH, while low pH conditions favored the proton uptake directly from the bulk solvent. Thus, the surface properties strongly affect the protonation of a membrane-associated proton acceptor (e.g. a fluorophore), from a membrane-mediated protonation at higher pH, to a direct exchange with the bulk solution at lower pH, with the balance tuned by the relative buffer strength between the bulk as compared to the membrane (44). FCS measurements with pH-sensitive fluorophores further showed that the proton diffusion along membranes is time- and length-scale dependent (44), with two orders of magnitude lower diffusion coefficients for protons measured locally around individual fluorophores by FCS, as compared to the proton diffusions over hundreds of micrometers as probed in proton pulse experiments (7, 45, 46).

The nature of the lipid headgroup strongly influences the proton diffusion along the membrane (10, cf. also refs. 47–49), further emphasizing the importance of the microenvironment in the biological membranes (cf. also refs. 6 and 7). Notably, prior theoretical models (50–52), reactive molecular dynamics simulations (53), and semiempirical ab initio dynamics simulations (DFTB3) (54) also support that protons associate with biological membranes and lead to the PCA effects, e.g., by association of the protons to the phosphate headgroups.

The composition of bioenergetic membranes responsible for oxidative phosphorylation and photosynthesis is highly complex. For example, in addition to POPC, POPE, and cardiolipin (in a 2:2:1 ratio), the IMM comprises 1 to 3% ubiquinone (Q₁₀) (55). Nevertheless, it currently remains unknown how Q₁₀, which introduce both hydrophilic, lipophilic, as well as aromatic groups into the membrane, affect the protonation reactions at the membrane surface.

Here we combine proteoliposome and single-molecule FCS experiments with multiscale molecular simulations to address the effects of Q₁₀ on the proton exchange reactions at biological membranes. Our combined data reveal that the quinone, which is an essential component of all energy-transducing membranes, also controls protonation reactions at the membrane surfaces, thus contributing to proton currents in energy transduction.

Results

Q₁₀ Alters the Protonation Reactions in Proteoliposomes.

To probe the influence of quinone on protonation reactions in membranes, we first created liposomes using E. coli polar lipids (ECPL) and studied how incorporation of 0 to 3% Q₁₀ affects an external proton gradient (ΔpH) across the liposome membrane (Fig. 2 A and B). The pH of the liposome interior was monitored by the fluorescence emission of pyranine (8-hydroxypyrene-1,3,6-trisulfonic acid, HPTS) centered at 510 nm upon excitation at 404 nm and 454 nm (56), with the pH changes obtained based on calibration experiments (SI Appendix, Fig. S1A). The pH-gradient was generated by addition of potassium acetate (K⁺CH₃COO⁻), which rapidly diffuses across the liposome membrane in its neutral form (CH₃COOH), and establishes a 0.3 pH-unit acidification of the liposome interior, as determined by the liposome pH = 7.2 and the pK_a of acetate (pK_a = 4.8). The established ΔpH was dissipated after 9 min by addition of the ionophore nigericin (Fig. 2 A and B) (22). The electrical component of the PMF, ΔΨ, was eliminated by addition of valinomycin prior to the measurements, which equilibrates K⁺ ions across the membrane.

Fig. 2.

Effect of Q₁₀ on protonation dynamics in liposomes. (A) Experimental setup used for the pH-conduction assay. (B) Proton conduction experiments with increasing amount of Q₁₀ in liposomes. The pH gradient was created by addition of potassium acetate and monitored by the fluorescence changes of pyranine (HPTS) in presence of valinomycin. The PMF is dissipated at 9 min by addition of nigericin. (C) ΔpH burst as a function of Q₁₀ concentration and (D) Initial proton conduction rates. (E) Liposome size as a function of the Q₁₀ concentration. (F) Experimental setup used the ATPase-driven PMF generation and antiporter-mediated proton conduction. (G) ATP hydrolysis drives generation of a PMF, monitored by the fluorescence quenching of ACMA. (H) ACMA quenching rates driven by ATP synthase proton pumping. (I) Initial proton transfer by ATP synthase and antiporter module (Nqo12) monitored by fluorescence quenching of ACMA. (J) Q₁₀ affects the rate of proton transfer between ATP synthase and antiporter module (Nqo12).

The proton conduction assays show a rapid acidification burst of the liposome interior upon addition of the acetate (Fig. 2 B and C), followed by a slower alkalinization phase, arising from proton leakage across the ECPL membranes. Interestingly, higher Q₁₀ concentrations result in a larger initial ΔpH burst and slower (0.15 pH min⁻¹) alkalinization phase in the conduction assays (Fig. 2 C and D). In the liposomes without Q₁₀, the PMF completely dissipates during the 9-min experiment, with nigericin addition leading to a negligible pH change. In stark contrast, liposomes with 3% Q₁₀, retained more than 1/3 of the ΔpH during the experiments, with a gradual increase of the effect with the Q₁₀ concentration. The incorporated Q molecules establish tighter liposomes (Fig. 2E), leading to a ca. 25 nm reduction of the liposome size (from 125 nm to 100 nm), as shown by dynamic light scattering (DLS) experiments. However, the initial ΔpH burst and proton conduction kinetics are independent of the liposome size in this range (SI Appendix, Figs. S2 and S3), showing that the enhanced proton conduction properties do not arise from the liposome size. Similarly, while addition of cholesterol results in an increase in liposome size, addition of 0.5 to 1% Q₁₀ consistently increases the initial ΔpH burst as well as the membrane integrity in all conditions (SI Appendix, Fig. S3). Taken together, these findings suggest that the Q₁₀ affects the integrity of the lipid membrane (cf. also refs. 57 and 58, and Discussion), allowing the liposome to maintain a high PMF. However, the increasing acidification burst at higher Q concentrations, indicates that Q₁₀ also stimulate the uptake of acetate, e.g., by acting as surface catalyst that kinetically favors the rapid proton removal during the CH₃COOH → CH₃COO⁻ + H⁺ reaction inside the liposome.

Q₁₀ Enhances Proton Transport between Source and Sink, Resulting in an Enhanced Chemiosmotic Coupling.

To further probe how Q₁₀ affect the proton conduction along the membrane surface, and how this contributes to the chemiosmotic coupling between bioenergetic protein complexes, we co-reconstituted ATP synthase with an engineered variant of the proton-conducting antiporter-module (Nqo12^ΔTH, cf. ref. 22) of Complex I (Fig. 2F). To this end, a pH-gradient was created by the hydrolysis of ATP to ADP and P_i by ATP synthase that couples to proton pumping across the liposome membrane and establishes a proton source, while the proton conduction via the antiporter module was introduced as a proton sink, dissipating the generated PMF. The pH gradient was monitored by fluorescence quenching of 9-amino-6-chloro-2-methoxyacridine (ACMA), which works over a broader pH range than pyranine and is often used together with the kinetically efficient ATP synthase (59), pumping around 2,400 H⁺ s⁻¹ during ATP hydrolysis (60). The ATP hydrolysis leads to an 80% quenching of the ACMA fluorescence (Fig. 2 G and H), with the dissipation of the PMF restoring the base fluorescence level. Remarkably, addition of Q₁₀ in the proteoliposomes lead to a 25% increase in the steady-state quenching rate (Fig. 2 G and H), while coreconstitution of Nqo12^ΔTH resulted in a significantly faster dissipation of the ATPase-driven ΔpH, suggesting that Q₁₀ enhances the conduction rate of the protons between ATP synthase and the antiporter complex (Fig. 2 I and J). Importantly, control experiments show that the activity of ATP synthase is not affected by the presence of Q₁₀ in the proteoliposomes (SI Appendix, Fig. S13). For the combined system (Fig. 2J), the maximal quenching effect is obtained at around 1% Q₁₀, while for ATP synthase alone, the maximum effect arises at around 2% Q₁₀ (Fig. 2H) and could reflect differences in the local proton currents between the source and the sink and the relative proton transport rate of the individual proteins (Discussion). However, due to the nonlinear dependence of these effects on the Q₁₀ concentration (Fig. 2 H and J), we expect that there could be a competition between a proton-collecting antenna effect, leading to an increase of the rate of proton conduction, while the Q₁₀–lipid interactions result in tighter liposome that prevent the proton leak across the membrane. Taken together, these findings suggest that Q₁₀ enhances the transport of protons between the bioenergetic protein complexes and affects the bioenergetic coupling in membranes.

Q₁₀ Alters the Surface Protonation Dynamics Monitored at the Single Molecule Level.

To address how Q₁₀ affects the rate of surface protonation reactions, we next incorporated a lipid-conjugated pH-sensitive fluorophore (fluorescein, Flu) into the membrane of the liposomes (Materials and Methods), and studied the proton exchange kinetics of Flu by FCS (Fig. 3A). In these experiments, we monitored the fluorescence intensity fluctuations of single Flu fluorophores at the membrane surface, as the liposomes diffuse through an open confocal detection volume. This allowed us to monitor the steady-state proton exchange kinetics of individual fluorophores at the membrane surface (cf. refs. 9, 61, and 62). Apart from a relaxation related to translational diffusion of the liposomes, the FCS curves reveal three major relaxation processes, well-separated on different timescales (Fig. 3A). These relaxation processes can be attributed to singlet-triplet transitions (with a relaxation occuring on ~µs timescales), proton exchange reactions (on ~10 to 50 µs timescales), and redox state transitions observed on longer timescales (~150 to 500 µs) in the Flu fluorophores (9, 44, 61, 63). Since these reactions are well separated on unique timescales, the equilibration of each process is independently established and assigned within our kinetic model (SI Appendix, Fig. S4). The experimental FCS curves can be accurately fitted to this model, including the translational diffusion and three exponential decay terms (Eq. 1, see Materials and Methods). Interestingly, with increased Q₁₀ concentrations in the liposomes, we observed a significant decrease in the protonation relaxation time, ${\tau }_P$, of the fluorophore in the liposome membrane (Fig. 3B), reflecting a higher exchange rate of protons to and from the fluorophore. At the same time, the amplitude of the protonation relaxation (reflecting the fraction of protonated fluorophores) in the FCS curves, P, increases significantly. At concentrations of Q₁₀ reaching 1% in the liposome membranes, the decrease in ${\tau }_P$ and increase in P were found to saturate at around 20 μs and at a 30% amplitude, respectively (from ca. 50 μs and a 10% amplitude, respectively for liposomes without Q₁₀, compared at the same bulk pH of 10, see Fig. 3C). We find that with a higher buffer concentration in the bulk solvent (20 mM instead of 2 mM glycine), P is lower and higher Q₁₀ concentrations are required to reach the same saturation in P and ${\tau }_P$ (SI Appendix, Fig. S4 C–F). This is consistent with the protonation model of Fig. 3D and results from the delicate balance between bulk-mediated and membrane surface-mediated proton exchange kinetics (44), where direct proton exchange between the bulk and the fluorophore at the membrane (III) is balanced against proton exchange with the fluorophore along the membrane (II), preceded by proton exchange between the membrane surface itself and the bulk (I). Here, we find that higher Q₁₀ concentrations favor the latter (II), while higher bulk buffer concentrations favor the former (III). To explore if the altered protonation kinetics can be attributed to mechanical properties of the membranes or to the liposome size, we performed FCS experiments on liposomes of different sizes with no Q₁₀ added (diameters from 30 to 400 nm), and with different concentrations of cholesterol (0 to 1%). Importantly, we observe no significant change in the protonation kinetics, as observed upon addition of Q₁₀, neither due to liposome size changes, nor to altered mechanical properties in the studied regime (SI Appendix, Fig. S5).

Fig. 3.

Fluorescence correlation spectroscopy (FCS) experiments on fluorescein-labeled liposomes. (A) FCS curve recorded from fluorescein-labeled liposomes (2 mM glycine buffer, pH 8.6), without Q₁₀. Fitting residuals plotted below. Four major relaxations are observed in the FCS curve, attributed to singlet-triplet state transitions (T, at a ~1 µs time-range), protonation kinetics (P, ~10 µs), redox state transitions ($\dot{R^{+}}$, ~100 µs), and translational diffusion (D, ~ms). (B) Normalized FCS curves recorded from fluorescein-labeled liposomes at pH = 10 and 2 mM glycine buffer with different Q₁₀ concentrations. Fitting residuals plotted below. (C) Fitted protonation relaxation rates (τ_P) and corresponding fitted amplitudes (P, see Eq. 1 and Materials and Methods) from the FCS curves in B. See SI Appendix, Fig. S3 C–F for data on other conditions. (D) Illustration of the major proton exchange pathways and proposed mechanism for the observed membrane protonation dynamics dependence on the Q₁₀ concentration, pH, and bulk buffer concentration I: direct proton exchange between the membrane surface itself and the bulk. II: proton exchange with the fluorophore at the membrane. III: direct proton exchange between the bulk and the fluorophore.

The FCS data thus show that Q₁₀ has an active role in catalyzing proton exchange at the membrane surface, strongly modulating the rate of the surface protonation reactions. However, due to its high pK_a, the oxidized quinone molecule cannot undergo protonation changes within the pH range covered in the FCS measurements. Moreover, while formation of radical states (e.g., Q^•/−) could result in a Q-induced protonation reaction, we note that any concomitant redox transitions in Flu takes place on much longer timescales (~150 to 500 µs), and therefore cannot account for the protonation phase (occurring at ~15 µs under saturating Q₁₀ concentrations, Fig. 3C and SI Appendix, Fig. S4). The observed effect from Q₁₀ on the protonation reactions is thus rather analogous to lipid molecules with protonatable headgroups that act as 2D buffers in membranes and enhance the proton exchange (8). The proton exchange rate (to and from) the fluorophore, as monitored by FCS, thus shows that Q₁₀ induces faster protonation relaxation rates. This suggests that the Q₁₀-induced effective protonation reaction mimics the effect of regular buffering molecules at the membrane surface, which is not attributed to mechanical effects, but seems to arise from a physically distinct interaction mechanism. Our QM/MM free energy calculations (SI Appendix, Fig. S8H) and estimation of diffusion coefficients (SI Appendix, Fig. S11) give further support for such mechanism and provide a basis for the local 2D buffering effect (see below).

Molecular Mechanism Underlying the Proton-Collecting Antenna Effect of Q₁₀.

To probe the mechanistic principles underlying how the Q₁₀ affects the protonation dynamics at the membrane, we performed atomistic molecular dynamics simulations of both pure (POPC) or mixed (POPC/PE/CDL) lipid membranes incorporated with Q₁₀ (SI Appendix, Fig. S6). During the microsecond MD simulations, the Q headgroup localizes at the membrane–water interface, but shows a rapid flip-flop motion between the bilayer leaflets on the ~150 ns timescales (Fig. 4 A–C, cf. also refs. 64 and 65). Overall, the Q₁₀ also affects the membrane properties, leading to a somewhat thicker membrane (40 Å without Q, 40.5 Å with Q), while also influencing the membrane dielectric (SI Appendix, Fig. S7). The 10-isoprenoid unit of the Q₁₀ tail is longer relative to the C16-C18 tails of POPC/POPE and cardiolipin tails, and leads to a change in the lipid ordering (SI Appendix, Fig. S7). Taken together, these findings could rationalize how the Q-saturated liposomes maintain the PMF, and account for the overall more compact liposomes (Fig. 2E).

Fig. 4.

Molecular simulations probing the effect of Q₁₀-mediated PCA effect. (A) Q₁₀ localizes to the surface of the lipid membrane in atomistic MD simulations. (B) Position of the Q headgroup along the Z-axis of the membrane plane from MD simulations shows the flip-flop motion of the headgroup. (C) Kinetics of the flip-flop motion extracted by computing the time-correlation of the flip-flop motion. (D and E) QM/MM-MD simulations of the proton at the membrane surface, shows interaction of the Q headgroup with a protonated water species. (F) Ab initio quantum chemical calculations (SCS-MP2/def2-TZVP) of H₃O⁺-Q interaction. (G) Thermodynamic model suggests that the Q-proton interactions lead to an effective lowering of the membrane surface pH.

Our MD simulations suggest that the surface exposed Q₁₀ headgroup could interact with water molecules, ions, as well as protons at the membrane (Fig. 4 A–C). To test the molecular principles of such putative interactions, we performed density functional theory (DFT)-based quantum/classical molecular dynamics (QM/MM-MD) simulations. The QM/MM simulations suggest that the proton hopping near the membrane surface involve transitions among Zundel (H₅O₂⁺) and Eigen (H₉O₄⁺) species, coupled to a rotation of a surrounding water molecule (SI Appendix, Fig. S8), consistent with multistate empirical valence bond (MS-EVB2) simulations of proton transfer reactions in bulk water on longer timescales (66, 67). Interestingly, during our QM/MM-trajectories, the solvated water species forms transient π-cation interactions with the surface exposed Q headgroup, leading to a Q-Zundel ion (H₅O₂⁺) contact at the membrane surface, and stabilized by transient hydrogen-bonded contacts with the Q₁₀ carbonyl headgroup (Fig. 4 D and E, SI Appendix, Fig. S8, and Movie S1), with the noncovalent interactions also supported at the correlated ab initio theory level (Fig. 4F). Despite the neutral headgroup, we find that the localization of the Q₁₀ to the membrane surface affect the electrostatic potential of the membrane surface (SI Appendix, Fig. S9) that could direct the proton toward the membrane surface (SI Appendix, Fig. S8). Based on computation of the mean squared displacement (MSD) of the excess charge from our unbiased QM/MM-MD simulations, we obtain a 2D proton diffusion coefficient of (3.91 ± 0.01) × 10⁻⁵ cm² s⁻¹ (SI Appendix, Fig. S11), which compare rather well to diffusion coefficients derived from spectroscopic studies of local proton diffusion near lipid membranes, over length-scales of 10 µm or longer (3.5 × 10⁻⁵ cm² s⁻¹ in 2D) (10), but is somewhat faster relative to previous DFTB3 calculations (1.5 to 3.4 × 10⁻⁵ cm² s⁻¹) (54). Moreover, our QM/MM free energy calculations (SI Appendix, Fig. S8H) further support the favorable Q-proton interaction, suggesting that a protonated water species, transitioning between Zundel and Eigen forms, interacts with the quinone headgroup both via hydrogen-bonding and π-cation interactions (SI Appendix, Fig. S8H). In this regard, we do not observe any significant interactions between Q₁₀ and Na⁺ ions in the MD simulations (SI Appendix, Fig. S12 C–F) or in the FCS experiments (SI Appendix, Fig. S12 A and B), despite that these species could, in principle, also form a similar π-cation interaction. However, we speculate that Na⁺ ions may interact with other types of quinone species (e.g., menaquinone), in biological membranes that rely on a sodium motive force (cf. ref. 68), although a systematic analysis of these potential effects are still missing.

To further probe the Q–proton interaction, we performed QM/MM calculations of optical absorption spectra of Q₁₀ at the time-dependent density functional theory (TDDFT) as well as corrected ab initio (LT-SOS-RVS-ADC(2)) level. Our QM/MM calculations suggest that the Q–proton interaction result in a small, but statistically significant redshift that can be traced to the interaction of the cation with the π* orbital of the quinone (SI Appendix, Fig. S10 A–F and Table S4). Although we note that experimentally trapping such transient state is highly challenging, we nevertheless observe a small, but reproducible redshift of the Q absorbance at 276 nm upon addition of acid to a solution with Q₁ (SI Appendix, Fig. S10G). The predicted absorption shift could thus open up a basis for the future spectroscopic characterization of the transient Q-proton interaction.

To understand how the Q interaction affects the local membrane surface pH, we further developed a thermodynamic solvation model (SI Appendix, Extended Methods), where we computed the proton solvation free energy based on its relative Born desolvation energy, and considered the effect of a tunable membrane interaction, as well as the Q-proton interaction as observed in our QM/MM calculations. Interestingly, for an interaction energy between Q₁₀ and the proton of ca. 2 kcal mol⁻¹ (based on our quantum chemical calculations), we obtain an effective lowering of the surface pH by ca. 2 relative to the bulk solvent (Fig. 4G, cf. also refs. 8 and 9), and conceptually supporting the PCA effect.

Discussion

Previous experiments suggest that Q₁₀ reduces the membrane permeability of small hydrophilic solutes and causes mechanical deformation (57, 58) that could contribute to maintaining the PMF, as observed in our current experiments. However, we also observed enhanced proton exchange kinetics at the membrane surface, and an increased uptake of acetate in the liposome, which points toward a “catalytic effect” of Q₁₀, e.g., by kinetically favoring the net proton transport at the surface. Yet, due to the significantly higher pK_a of the benzoquinone headgroup, the Q is unlikely to undergo a protonation reaction itself. Although the mechanical properties of the liposomes are modified by addition of Q₁₀, the enhanced proton conduction and exchange kinetics are not likely to be attributed to such effects. For our measurements of cholesterol-enriched liposomes and liposomes with a wide range of different sizes, we only observe minor, if any, changes in the proton conduction and exchange kinetics, and thus significantly smaller effects as compared to those observed upon Q₁₀ addition (SI Appendix, Figs. S3 and S5).

On a general level, this effect implies that Q₁₀ could enhance the proton exchange between a source and a sink in bioenergetic energy transduction chains, e.g., between the respiratory complexes I, III, IV, and ATP synthase. The average distance between the OXPHOS complexes is around 7 nm in the IMM (69, 70), while ATP synthase is located on the mitochondrial cristae (35), and thus further away from the proton source. Our data suggest that Q₁₀ enhance the effective rate of 2D proton diffusion in the membranes (τ = <r²>/4D) to exceed the 3D diffusion (τ = <r²>/6D), even if the diffusion distance could be longer in the curved membranes. In this regard, Xu et al. (44) observed that the radius for the PCA effect is around 10 nm in DOPC vesicles (without added buffering molecules). Thus, with Q₁₀ present in the membrane, the resulting 2D diffusion radius is likely to exceed the average distance between the OXPHOS complexes in the IMM.

The combined findings suggest that incorporation of Q₁₀ into the membranes lead to a similar effect as for lipids with protonatable headgroups (8). Q₁₀ thus functions as a “2D-buffer” by enhancing the PCA effect of the membrane, so that the membrane-mediated proton exchange is favored relative to the bulk-mediated proton transfer. However, as the pK_a of the oxidized benzoquinone moiety is far into the basic pH range, the process is highly unlikely to take place by a direct protonation reaction of the quinone, but instead via noncovalent interactions

between the solvated protons with the Q headgroup, as suggested by our molecular simulations. Previous studies (8–10) show that both the charge and pK_a of the lipid headgroups can also influence the surface protonation rates. However, such charge effects are overall small (8), and the buffering effect of the quinone cannot be attributed to its pK_a, which is far from the local pH at the membranes in our experiments. Moreover, our experiments with cholesterol, also comprising an aromatic ring structure, do not result in a similar rate enhancement, suggesting that the effect can be specifically attributed to the quinone headgroup. Thus, together with the phosphate lipid headgroups, these interactions provide Q₁₀ buffer-like properties, which can promote proton exchange at the membrane surface and lead to the effective lowering of the surface pH.

Despite the rapid proton transfer along the biological membranes, which is enhanced by Q₁₀, the proton transfer between the OXPHOS complexes and ATP synthase is not likely to limit the overall energy flux of biological proton circuits (Fig. 5). An analogy can be made to electric circuits, where electrons travel significantly slower (cm s⁻¹) relative to the propagation of the energy flux (power flow), defined by the electrical Poynting vector, S = 1/μ₀ (E x B), where E is the electric field, B is the magnetic field, and μ₀ is the vacuum permittivity. The proton translocation results in a similar rapid global electrical response across the membrane that creates the overall PMF (Fig. 5). The proton conduction along the membrane is thus not limited by the number of free protons, which is overall low at a neutral pH, but by the propagation of the electric field created by the proton charge. In this regard, the proton current creates an overall charge displacement along the membrane between the proton translocating source (CI, CIII, CIV in the IMM), and the energy conserving sink (ATP synthase), defining the overall energy flux (Fig. 5). Here, we propose that the Q-enriched membrane results in a lowered internal resistivity, and an overall lowered effective resistance in the membrane between the source and the sink (R_mem) that will thus pass a higher proton current along the surface relative to the bulk (with a unique R_bulk) that could reduce dissipation effects and maximize the overall power output of energy conversion.

Fig. 5.

Schematic representation of PMF transduction along energy-transducing membranes. The proton pumping created by the OXPHOS complex (in blue, the proton source) and the proton consumption by ATP synthase (in red, the sink) creates a proton circuit, where the PMF across the membrane generates an electrical field (E), and the potential difference result in the proton current, I_H+. Q₁₀, together with specific lipid molecules, enhance the local proton current (I_H+^mem) along the membrane surface by lowering the effective membrane resistance (R_mem) relative to the bulk resistance (R_bulk). The moving proton charge creates a local magnetic field (B), with the Poynting vector, S = μ₀⁻¹ (E × B), defining the direction of the energy flux from the OXPHOS complex to the ATP synthase.

Conclusion

In addition to its central role as a redox-carrier in biological energy conversion chains, we show here that ubiquinone Q₁₀ alters the protonation dynamics at and along a membrane surface. By combination of proteoliposome experiments with single molecule fluorescence correlation spectroscopy, proton pumping assays, and multiscale simulations, we find that Q₁₀ enhances the proton-collecting antenna effect of the lipid membrane by stabilizing solvated protons via noncovalent interactions. In biological energy transduction, we suggest that the enhanced net transport of protons along the membrane reduces the overall internal resistivity of the proton translocation, in analogy to an electric circuit model. This may in turn reduce dissipation effects, and maximize the overall power output of the energy conversion process.

Materials and Methods

Preparation of Liposomes.

Liposomes were prepared based on a protocol described before (22). Briefly, E. coli polar lipids (ECPL) in chloroform (25 mg mL⁻¹, Avanti), were incubated together with ubiquinone Q₁₀ (Sigma-Aldrich) from 0 up to 3% molar ratio, and/or 0 to 5% cholesterol molar ratio, and dried under a N₂ (g) stream, followed by desiccation overnight. The liposome buffer A (2 mM MOPS/KOH pH 7.2, 50 mM KCl) was used for resuspension of the dried lipids at 5 mg mL⁻¹. Samples were prepared by eight freeze/thaw cycles before being extruded through a 100 nm pore membrane (Nuclepore membranes, Whatman Ltd). 5 mM 8-hydroxypyrene-1,3,6-trisulfonic acid (HPTS), a pH-sensitive dye, was incorporated inside of the formed liposomes by freeze/thawing three times. The excess of HPTS was removed using a PD-10 prepacked desalting column (Cytiva).

Proton Conduction Assays.

Proton conduction assays were based on our assay described before (22). In this regard, the internal pH of the PLs was measured by following fluorescence changes of HPTS, with excitations at λ_ex1 = 404 nm and λ_ex2 = 454, and emission at λ_em = 510 nm. The fluorescence measurements were performed using a Cary Eclipse Fluorescence Spectrophotometer (Agilent Technologies) at 37 °C for 10 min. 60 µL of liposomes in presence or absence of Q₁₀, were incubated together with 0.1 µM valinomycin (in absolute EtOH), and resuspension buffer (2 mM MOPS/KOH pH 7.2, 50 mM KCl) to a final volume of 2 mL. The proton conduction assay was mediated by addition of 10 mM potassium acetate, while the ΔpH generated was dissipated using 0.8 µM nigericin (in absolute EtOH) after a steady state was reached. Each sample was investigated by n = 6 independent measurements. The proton pumping in PLs experiments with ATP synthase was monitored by ACMA fluorescence, at 37 °C, using 20 µL of PLs, 4 nM valinomycin, and 4 µM ACMA added to a final volume of 1 mL (see below). The reaction was started by addition of 0.2 mM ATP, followed by dissipation of the ∆pH by nigericin. The ACMA fluorescence emission at 480 nm upon excitation at 410 nm was monitored using a Cary Eclipse Fluorescence Spectrophotometer (Agilent Technologies) in plastic cuvettes (d = 1 cm). Each sample was investigated by n = 3 to 6 independent measurements. The fluorescence response of pyranine or ACMA was not affected by addition of Q₁₀ (SI Appendix, Fig. S1). The ATPase activity was monitored by a coupled activity assay (cf. ref. 71), monitoring spectroscopically the consumption of NADH at 340 nm. The reaction was started by addition of 10 µL ATPase PLs into a 150 μL final volume, 50 mM phosphoenolpyruvate, 2 units mL⁻¹ pyruvate kinase, 5.5 units mL⁻¹ lactate dehydrogenase, and following the signal for 5 min.

Expression and Purification of ATP Synthase and the Nqo-Antiporter.

Nqo12 expression and membrane isolation were performed as described in ref. 22. Briefly, for the purification, solubilized membranes were diluted to 6 mg mL⁻¹ protein concentration in membrane resuspension buffer (25 mM HEPES, 150 mM NaCl) together with 2% Lauryl Maltose Neopentyl Glycol (LMNG). The resuspension was left stirring cold for 1 h before removing insolubilized membranes by 30 min of ultracentrifugation. The supernatant was then supplemented with 20 mM imidazole before binding to a 5 mL His-trap HP column (Cytiva) three times. The column was previously equilibrated with purification buffer A (25 mM HEPES pH 7.5, 300 mM NaCl, 20 mM imidazole, 0.005% LMNG). The purification was then performed using ÄKTA purifier and washing at 20 mM, 75 mM, 100 mM, 126 mM, and 150 mM imidazole, before eluting at 300 mM imidazole by mixing purification buffer A with purification buffer B (25 mM HEPES pH 7.5, 300 mM NaCl, 300 mM imidazole, 0.005% LMNG). Relevant fractions were pooled and concentrated using a concentrator with a 50 kDa cut-off. The concentrated sample was then loaded into a HiLoad 16/600 Superdex 200 pg column (Cytiva) for size exclusion chromatography using SEC buffer (25 mM HEPES pH 7.5, 300 mM NaCl, 0.005% LMNG). ATP synthase purification was performed as described in ref. 22.

Preparation of Proteoliposomes for ATPase-Driven PMF Generation.

Type II-S soybean lipids were resuspended in chloroform (25 mg mL⁻¹, Sigma-Aldrich), and incubated together with ubiquinone Q₁₀ (Sigma-Aldrich) from 0 up to 3% molar ratio, and dried under a N₂ (g) stream, followed by desiccation overnight. The liposome buffer B (2 mM MOPS/KOH pH 7.2, 50 mM KCl, 5 mM MgCl₂) was used for resuspension of the dried lipids at 5 mg mL⁻¹. Samples were prepared by eight freeze/thaw cycles before being extruded through a 100 nm pore membrane (Nuclepore membranes, Whatman Ltd). Reconstitution of the ATP synthase together with the antiporter-like subunit Nqo12 was performed using the LMNG-autoinsertion reintegration (LAiR) method (72). To this end, LMNG-purified ATP synthase and Nqo12 were added to 100 μL liposomes in a 1:1.3 molar ratio, samples were then diluted up to 200 μL final volume with buffer B. Reconstitution was performed at room temperature for 30 min before pelleting down the proteoliposomes using a benchtop ultracentrifuge. Pellet was resuspended in 100 µL buffer B.

FCS Sample Preparation.

Dioleoylphosphatidylcholine (DOPC) lipids, solubilized into chloroform (25 mg mL⁻¹, Avanti Polar Lipids), were mixed with DPPE-fluorescein lipids (10 mM, in CHCl₃) at 1:100,000 DPPE:DOPC ratio. Thereby, it was ascertained that the probability to have more than one Flu fluorophore per liposome was negligible. Ubiquinone-10 (Q₁₀, Sigma-Aldrich) and/or cholesterol were included up to 1% molar ratio each. Samples were dried under a N₂ (g) stream, while vortexing to create a thin lipid layer prior to desiccation overnight. The dried lipids were hydrated at 5 mg mL⁻¹ with the liposome buffer C (from 2 to 20 mM glycine/NaOH pH 8.6 or 10, with various NaCl concentration 0 to 200 mM) followed by vigorous vortexing to allow for incorporation of the quinone in the membrane. Unilamellar vesicles were formed by flash freezing and thawing the samples eight times, prior to 21 extrusions through a 100 nm pore membrane (Nuclepore membranes, Whatman Ltd). Liposomes were diluted four times in 200 µL final volume before measurements.

FCS Experiments and Fitting.

FCS measurements were performed on a setup similar to previous works (44), based on an inverted confocal laser scanning microscope (Olympus Fluoview 1000, Shinjuku, Tokyo, Japan) modified for FCS measurements. A linearly polarized diode laser (Picoquant GmbH, LDH-D-C-485, Berlin, Germany) ran in continuous wave (CW) mode with emission at 485 nm was reflected by a dichroic mirror (ZT405/488/635rpc-UF2, Chroma Technology, Rockingham, VT, USA) and then focused by a 60×, 1.2 NA water immersion objective (UPlanApo 60×, 1,2NA, Olympus, Japan) into the sample. The excitation power was kept constant (22 µW) throughout the experiments yielding an excitation intensity of ~10 kW cm² (with a laser beam radius in the focal plane of ~0.3 µm). The fluorescence emission was collected by the same objective, passed through the dichroic mirror, focused onto a pinhole of 50 mm in diameter, then split by a 50:50 beam splitter cube, passed through a pair of band-pass filters (HQ532/70; Chroma Technology, Rockingham, VT, USA) and finally detected by two avalanche photodiodes (APDs, SPCMAQR-14/16; Perkin-Elmer Optoelectronics, Wellesley, MA, USA). The FCS experiments were performed in triplicates, with a 2 min recording time. The APD signals were processed by the Symphotime software (Picoquant GmbH, Berlin) used to generate and analyze semi-log-scale correlation curves, based on the recorded fluorescence intensity fluctuations over the measurement time, as described in Eq. 1. The recorded correlation curves were analyzed using a Levenberg–Marquardt nonlinear least-square curve fitting algorithm written in Python (73). In the fitting, a model correlation function was used, assuming that the fluorescence intensity fluctuations are generated by i) diffusion (of free fluorophores, or of fluorophore-labeled SUVs), into and out of the confocal detection volume, and ii) transitions of the fluorophores back and forth into three different dark states (62),

$$\begin{array}{c}\begin{array}{cc}G\left(\tau \right)=\frac{\frac{1}{T_m}{{\displaystyle \int }}_0^{T_m}F\left(t\right)F\left(t+\tau \right)dt}{\frac{1}{T_m}{{\displaystyle \int }}_0^{T_m}{F}^2\left(t\right)dt}\hfill & =\left\{F\left(t\right)=\langle F\rangle +\delta F(t)\right\}=\frac{\langle \delta F\left(t\right)\delta F\left(t+\tau \right)\rangle }{{\langle F(t)\rangle }^2}+1\hfill \\ \hfill & =\frac{1}{N\left(1-P-T-R\right)}{\left(1-\frac{\tau }{{\tau }_D}\right)}^{-1}{\left(1-\frac{\tau }{{\beta }^2{\tau }_D}\right)}^{-1/2}\times \left(1-P-T-R+P{e}^{-\raisebox{1ex}{$t$}\!\left/ \!\raisebox{-1ex}{${\tau }_{prot}$}\right.}+T{e}^{-t/{\tau }_T}+{\mathrm{Re}}^{-t/{\tau }_R}\right)+1.\hfill \end{array}\hfill \end{array}$$ [1]

Here, F(t) is the detected fluorescence intensity at a time t, τ is the correlation time, T_m denotes the measurement time over which the fluorescence fluctuations are integrated, square brackets signify time average, τ_D is the average translational diffusion time of the fluorescent species through the confocal detection volume, N the mean number of fluorophores in the detection volume, and β the relationship between the axial and lateral extension of the detection volume. P signifies the fraction of protonated fluorophores, τ_prot = 1/k_prot is the proton relaxation time and k_prot the proton relaxation rate. The other dark states are attributed to triplet state formation (with its population T and relaxation time τ_T in Eq. 1) (63), and to redox state formation (74) (relative population R and relaxation time τ_R in Eq. 1) of the fluorescein dyes. In the fitting of the FCS curves, T and τ_T were fitted as global parameters, based on the assumption that these parameters are not changed in FCS curves recorded in the same session from liposomes with different Q₁₀ concentrations, in samples with the same bulk buffer concentration and pH and subject to the same excitation intensity.

Dynamic Light Scattering.

The liposome size in absence and presence of 1 to 3% Q₁₀ and/or 1 to 5% cholesterol, was measured using dynamic light scattering with a Zetasizer Lab instrument (Malvern PCS Instruments, UKl) at 25 °C. 100 µL freshly extruded liposomes were diluted up to 1 mL in the resuspension buffer (2 mM MOPS/KOH, 50 mM KCl) before the size determination.

Molecular Dynamics Simulations.

Classical atomistic MD simulations of POPC a mixed POPC/POPE/CDL (2:2:1) lipid membranes were performed with different Q₁₀ concentrations (0 to 5%) of either Q or QH₂ (SI Appendix, Table S1). The membrane system was built using the CharmmGUI membrane builder (75), followed by solvation of the membranes with TIP3P water molecules and neutralization with 150 mM NaCl. After minimization and heating, the equilibration was performed in an NPT ensemble with periodic boundary conditions at T = 310 K and P = 1 atm for 50 ns using a timestep of 2 fs, describing intermolecular interactions with the CHARMM36m force field parameters (76) in combination with in-house parameters for Q and QH₂. Long-range electrostatic interactions were treated using the Particle Mesh Ewald approach with a grid size of 1 Å. Under the same conditions, the systems were propagated for at least 400 ns in duplicates (SI Appendix, Table S1). The classical MD simulations were performed with NAMD2.14 (77, 78) and the trajectories were analyzed using Visual Molecular Dynamics (79) and MDAnalysis (80, see SI Appendix, Extended Methods).

QM/MM Simulations.

Hybrid quantum/classical (QM/MM) simulations of a protonated water at the membrane surface were studied using QM/MM-MD simulations. The QM region was described at the DFT level using B3LYP-D3/def2-SVP (81–84). The initial geometry was selected from a snapshot from the atomistic MD simulation by selecting all residues within 30 Å of the headgroup of a quinone molecule located at the membrane–water interface, and transforming one of the water molecules into a hydronium species, followed by structure optimization. The QM region comprised 164 atoms (with a total charge of +1), while the MM region comprised 27,471 atoms, which were modeled using the CHARMM36 force field, and coupled to the QM region by an additive electrostatic embedding approach. The position of the excess proton was determined by employing the modified center of excess charge reaction coordinate definition. (85) The QM/MM-MD simulations were performed with a 1 fs integration timestep and T = 310 K using FermiONs++ (86), and trajectories were analyzed using Visual Molecular Dynamics (VMD) (79) and MDAnalysis (80). To explore the free energy landscape of the proton interaction with the quinone, we applied a shared-bias well-tempered metadynamics-extended adaptive biasing force (MWE) (87, 88) method in combination with the modified center of excess charge (mCEC) (84) as a reaction coordinate. The MWE method enables the sampling of phase space by using a moving fictitious particle λ, known as the walker, coupled to a harmonic biasing potential B(ξ(x), λ), and by incorporating a time-dependent bias potential B_mwe(λ, t), adding Gaussian hills every 10 fs with heights and widths of 0.40 kJ mol⁻¹ and 0.25 Å, respectively. A single walker was used to sample a total of 25 ps using a timestep of 0.5 fs and at T = 310 K. The mCEC reaction coordinate was defined by the quinone oxygen atom (O₂, donor) and a restrained oxygen atom of a distant water molecule as the acceptor. The unbiased free energy profile was reconstructed using MBAR (89).

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Protonated water species interacting with Q₁₀ from QM/MM-MD simulations.

Data, Materials, and Software Availability

Study data are included in the article and/or supporting information.