Rotational Mechanism of FO Motor in the F-Type ATP Synthase Driven by the Proton Motive Force
where i = aE223 or aE162; each MC phase begins with the probabilistic selection of the protonation states of aE223 and aE162 with these values.
Next, we realized probabilistic proton transfer between the c-subunit and a-subunit glutamates. For proton transfer between pairs, the following conditions must be satisfied: (1) only one glutamate molecule in each pair is protonated; (2) the two glutamates are close to each other; (3) they are located at an appropriate angle; and (4) they are not located across aR176, which is a highly conserved arginine in the a-subunit that separates the two half channels. The kinetic weight factor for the three last criteria can be given as follows:
where f(r) is the decreasing function of the distance r between two glutamates; g(θ) is the function of angle θ between the two vectors that emulate the side chain conformation of the donor and acceptor; g(θ) assumes a Gaussian form centered around the corresponding angle in the cryo-EM structure; and haR176 represents the role of aR176 in preventing proton leakage between two half channels. We assumed that haR176 was 0 when the proton donor and acceptor were located across aR176; otherwise, haR176 = 1.
Using the kinetic weight factor wj←i, the probability of proton transfer from the ith site to the jth site can be given as follows:
where E is the energy change upon proton transfer; k is the overall rate constant; and τ is the time step between MC trials. We set τ = 105 MD steps. As the value of the overall rate constant of proton transfer k is unknown, we simply set kτ = 1. H+(before) and H+(after) represent the protonation states before and after the proton transfer trial, respectively. Accordingly, energy change upon proton transfer can be given as follows:
where the second and third terms in Vtotal are relevant to proton transfer, and the last term VpKa(H+) represents the intrinsic propensity for protonation at each site (i.e., 2 half-channels and 10 c-subunits). In a previous study (Kubo et al., 2020), we set pKaaE223 = 6.0, pKaaE162 = 9.0, and pKacE59 = 8.0, based on discussion by Srivastava et al. (2018). The reliability of these values was also examined using the pKa calculation tool PROPKA (Li et al., 2005). Note, the all parameter used in our previous study are listed in Table 1.
Earlier theoretical models assume that after a proton is released from the c-subunit into the matrix, a single-step rotation of the c-ring is followed by the receipt of a proton from the IMS site to the same or the nearest neighbor c-subunit (Elston et al., 1998) (left panel in Figure 1B). In other words, during the cycle, the number of deprotonated proton-carrier residues in the c-subunit is between zero and two. Subsequently, high-resolution cryo-EM structures revealed the accurate position of the highly tilted α-subunit, suggesting that the deprotonated c-subunit is be immediately protonated in the next step (right panel in Figure 1B). In fact, our hybrid MC/MD simulations indicated that proton-carrier residues in the two c-subunits are simultaneously deprotonated in the ground state (Figure 4B). In addition, a recent study using all-atom MD simulation supports the presence of doubly deprotonated c-subunits (Marciniak et al., 2022).
Starting from the doubly deprotonated c-ring configuration, which occurs next: further deprotonation to the triple-deprotonated state (deprotonation-first pathway) or protonation to the single-protonated state (protonation-first pathway)? How is this coupled to c-ring rotation? Based on free energy profiles obtained from our simulations along the two scenarios (Figures 3A,B), the deprotonation-first pathway has a lower free energy cost, whereas the protonation-first pathway requires high free energy to achieve the single-protonated state. Interestingly, the deprotonation-first pathway was dominant in both ATP synthesis and hydrolysis. However, the proposed pathway does not fully consider the dynamical aspects of c-ring rotation coupled to proton transfer. Thus, through further dynamic hybrid MC/MD simulations, we obtained the most prominent pathway in the synthetic direction (Figure 4C) (Kubo et al., 2020), as follows:
In this pathway, mean c-ring rotational position in the double-deprotonated state DEP(c1, c2) is at −18, with marked fluctuations (here, the reference of the angle is arbitrary; therefore, the change in angle rather than its absolute value is important). When the value fluctuates to −2, the third c-subunit c3 releases a proton, resulting in the triple-deprotonated state DEP(c1, c2, c3), with relatively limited fluctuations. After a while, the c-subunit c1receives a proton, followed by c-ring rotation to an angle of 18.
Currently, this pathway has not been directly confirmed by other studies. Although very challenging, direct observations of rotational motions together with proton transfer are warranted. However, a recent study demonstrating the coupling among neighboring c-subunits supports the involvement of a multiply deprotonated state, as described below (Mitome et al., 2022).
Proton transfer-coupled molecular simulations suggested that proton-carrier residues in two or three c-subunits are simultaneously deprotonated while the FO motor rotates. Recently, we examined the coupling of multiple c-subunits in FO through a collaborative study of biochemical assays and molecular simulations of mutant FO motors (Mitome et al., 2022). First, Mitome et al. (2004) built a genetically fused single-chain c-ring of Bacillus PS3 ATP synthase. In the single-chain c-ring, 10 c-subunits were labeled in the order “a” through “j” (Figure 5A). We then introduced a mutation of the proton-carrier cE56 to aspartic acid in a specific c-subunit of the single-chain c-ring. Due to its shorter side chain and lower pKa value, the cE56D mutation in the c-subunit “e” slowed down yet did not abolish ATP synthesis compared with the wildtype (c10 and “e” in Figure 5B). Furthermore, double mutants carrying cE56D mutations in two c-subunits exhibited further reduced ATP synthesis activity compared with the single mutants. Interestingly, the ATP synthesis activity in double mutants depended on the distance between the two mutated c-subunits. Specifically, among the five double mutants examined, the ones carrying mutations in the neighboring c-subunits (“ef”) exhibited the highest ATP synthesis activity. The activity tended to decrease with increase in distance between the two mutated subunits. Motivated by this finding, we performed molecular simulations on the corresponding mutant of mitochondrial FOF1 ATP synthase that we had used earlier and found that the FO rotation rate decreased as the distance between two mutation sites increased, which is consistent with the results of biochemical assays (Figure 5C). Although this experiment alone did not offer any direct mechanistic understanding, the corresponding simulations provided some clues.
Briefly, four phases were defined for each c-subunit: resting time, proton release duration, deprotonated rotation, and proton uptake duration (Figure 5D). The time series of the 10 c-subunit protonation states was plotted based on the four phases defined. Figures 5E,F show the four phases of “ef”, in which the neighboring “e” and “f” are mutated, and “ej”, in which the farthest subunits “e” and “j” are mutated. The time for the most prolonged duration for proton uptake was markedly shared in “ef” (Figure 5E), but not in “ej” (Figure 5F). Thus, the farther apart the mutated c-subunits are, the less latency they share for protonation and the lower is their activity. Thus, the results of biochemical assays are consistent with the view that the proton-carrier residues in multiple c-subunits are simultaneously deprotonated, indicating coupling among the c-subunits.
As previously described, mutations of the c-subunit proton-carrier residues as well as of the a-subunit affect rotational activity of the FO motor. The effect of mutation of the highly conserved aArg (aR176 in mitochondrial ATP synthase) was significant. From structural information, aArg is important for separating the two water-accessible regions beyond the two half-channels. In addition, Mitome et al. (2010) found that aArg mutation decreased ATPase activity and induced proton leakage. Thus, we tested leak events through simulations mimicking the aR176A mutation. We modeled the effect of the arginine mutation by replacing the factor haR176 with unity. Consequently, both ATP synthesis and hydrolysis were markedly suppressed. In particular, in the ATP hydrolysis pathway, protons leaked in the opposite direction. Proton leakage occurred both in ATP synthesis and hydrolysis during the waiting time for the c-subunit to receive protons (Kubo et al., 2020).
Using all-atom MD simulations, Marciniak et al. (2022) found that the aR176A mutation connects the two half channels such that water molecules can directly pass through, regardless of the c-ring rotation. The authors also noted similar results in the aR176K mutant, which may be due to the higher flexibility of the lysine side chain than that of the arginine side chain.
As high-resolution cryo-EM structure models of the FO motor, together with functional and theoretical information, offered a refined view of proton transfer-coupled FO rotation, the next aspect to be refined was how FO rotation is coupled to ATP synthesis. The fundamental mechanism of coupling has long been known; briefly, the FO motor rotates the central rotor, which is shared with the F1 motor. The rotation of the central rotor in F1, the γ-subunit, induces a series of conformational changes in the F1 stator, α3β3, which drives ATP synthesis at the interface of the α-β-subunits. An obvious question that remains to be addressed is the symmetry mismatch. The F1 motor exhibits 3-fold pseudo-symmetry in all species, whereas the FO motor exhibits diverse pseudo-symmetries, ranging from 7- to 13-fold. First, the units of rotational steps in FO and F1 are not consistent. Furthermore, the symmetries of FO and F1 are coprime in many species, rendering the three unit steps of F1 rotation unequal with respect to FO rotation. How does ATP synthase solve this symmetry mismatch issue, realizing high efficiency of energy transduction? Some earlier studies have suggested various possibilities: the elastic torsion of the central rotor may contribute (Cherepanov et al., 1999; Pänke and Rumberg, 1999; Okazaki and Hummer, 2015); the torsion of the peripheral stalk, b-subunit, may be of central importance (Zhou et al., 2015; Guo et al., 2019); or the δ-subunit may absorb the symmetry mismatch (Murphy et al., 2019). Despite various notions, however, no consensus has been reached as yet.
Recently, we sought to elucidate the means of resolving this symmetry through coarse-grained MD simulations for the holo-complex of Bacillus PS3 FOF1 ATP synthase (Kubo et al., 2021). In this study, instead of simulating proton transports in FO, we mechanically rotated the c-ring with 10 c-subunits at 36°-steps, examining changes in the other parts of the holo-complex structures. Overall, the mismatches between angular positions favored by the F1 and FO motors were absorbed by combination of a number of elements, depending on the degree of mismatch. Distortion and twist of the central rotor, γ-subunit, can absorb some mismatches. Distortion of the peripheral stalk, b-subunit, results in the rotation of the F1 stator, α3β3, relative to the FO stator, a-subunit. Furthermore, a comparative study of cryo-EM structure models among three species confirmed the results of MD simulations and elucidated significant deviation of the c-ring rotation angle from the one anticipated based on the symmetry (a multiple of 36°). Notably, in our simulations of the holo-complex, the FO c-ring was forced to rotate at steps of 36° without room for fluctuation; thus, the deviation of the c-ring rotation from its symmetric position could not be observed using this design. However, our previous simulation of proton transfer-coupled FO rotation revealed marked fluctuations in c-ring rotation (Figures 3A,C), which is consistent with the observation of c-ring rotation angle deviation from the ideal value. In the future, hybrid MC/MD simulations for proton transfer-coupled c-ring rotation should be performed on the holo-complex.
As described above, the side chain of glutamates, the proton-carrier residues of the c-ring, assumes two distinct conformations depending on the surrounding environment: open and closed. Changes in conformation induces a pKa shift, and their conformation is important for bidirectional rotation. Thanks to remarkable advances in cryo-EM technology, atomic details of both conformations have already been revealed. In addition, free energy surfaces that depend on both the conformation and surrounding environment have already been estimated. In our MC/MD simulations using a coarse-grained model, we did not treat the side chain explicitly but set the angular relationship to mimic the effect of side chain reorientation.
In the future, more detailed proton transfer pathways and rotational mechanisms of FO motors can be revealed through computational simulations more explicitly taking the orientation of side chains into account. Furthermore, it would be interesting to simulate the actual movement of protons; QM/MM simulations have been employed to explicitly mimic the actual motion of protons during photosynthesis. However, explicit investigations of proton transfer dynamics depending on the rotational motion and orientation of key glutamate side chains is expected to be challenging.
SK and ST conceived the project. SK made figures. Both authors were involved in the manuscript writing process. Both authors contributed to the article and approved the submitted version.
The study was supported partly by JSPS KAKENHI Grants JP21J00021 (SK), 20H05934 (ST) and 21H02441 (ST), and partly by JPMXP1020200101 as Program for Promoting Researches on the Supercomputer Fugaku (ST).
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.