ATP synthase

The ATP synthase, also known as FoF1 ATPase or F-ATPase, mediates the energy interconversion between the proton motive force (pmf) across membranes and the free energy of ATP hydrolysis via a rotary catalysis mechanism (Abrahams et al. 1994; Yoshida et al. 2001; Noji et al. 2017). The ATP synthase is composed of two rotary motors, F1 and Fo (Fig. 1) (Junge et al. 1997). F1 is the catalytic core domain responsible for ATP synthesis, showing an active ATPase activity when isolated (Yasuda et al. 2001; Spetzler et al. 2006; Bilyard et al. 2012; McMillan et al. 2016). Upon ATP hydrolysis, F1 rotates the inner subunit (γε) against the catalytic stator ring (α3β3). Fo is the membrane-embedded domain and conducts proton translocation across the membrane. Upon proton translocation, Fo rotates the oligomeric ring formed by the proton-carrying c-subunits against the stator complex (ab2). In the whole ATP synthase complex, the rotor parts of F1 and Fo are bound together, forming the common rotary shaft (Junge et al. 1997; Oster and Wang 2000; Yasuda et al. 2001). The stator parts of F1 and Fo are connected via the peripheral stalk to transmit the torque without slippage. When pmf is sufficient, Fo generates a larger torque than F1, reversing the rotation of the rotor shaft in F1 to induce ATP synthesis. In contrast, when pmf is low, F1 reverses the rotation of the rotor ring in Fo, forcing Fo to actively pump protons and generate pmf.

Fig. 1
figure 1

The two rotary motors of ATP synthase, F1 and Fo. The subunit composition of F1 and Fo in bacterial types is α3β3γδε and ab2cn, respectively, where n varies among species. F1 rotates the rotary shaft, composed of the γ and ε subunits (red) against the α3β3 stator ring (blue). Fo rotates the oligomer ring of the c-subunits (red) against the ab2 stator complex (blue) during proton translocation across the membrane. In the whole ATP synthase complex, the rotor complexes F1 and Fo form the common rotary shaft (red) and stator complexes (blue), which are connected via the peripheral stalk formed by the b2 and δ subunits

Full size image

Stepping rotation of F1

The minimum F1 complex as a rotary motor is the α3β3γ subcomplex, which rotates the rod-shaped γ subunit against the α3β3 stator ring in a counterclockwise direction when viewed from the Fo side. The catalytic reaction centers for ATP hydrolysis reside at the three pairs of α−β, with the main catalytic residues harbored in each of the β subunits (Weber and Senior 1997). The three β subunits conduct the catalytic reaction in a highly sequential manner, resulting in a sequential power-stroking conformational change that rotates the γ subunit unidirectionally.

As expected from the pseudo threefold symmetry of F1, the unitary rotational step is 120° rotation, coupled with a single turnover of ATP hydrolysis (Yasuda et al. 1998). The rotation dynamics of the γ subunit in F1 from thermophilic Bacillus PS3 (TF1) has been intensively characterized to establish a standard reaction scheme for bacterial F1 domains. TF1 makes 80° and 40° sub-steps in a single 120° rotation, which means that TF1 makes rotational steps intervened with 6 pauses per turn (Yasuda et al. 2001; Shimabukuro et al. 2003; Nishizaka et al. 2004; Adachi et al. 2007). Other F1 domains from bacteria and mitochondrial F1 from yeast (yMF1) were reported to make 6 pauses per turn (Steel et al. 2015). Thus, a 6-step rotation is widely conserved across microorganism species.

On the other hand, rotation assays in mammalian F1 domains have found an additional pause in 120° rotation, which translates into 9 steps per turn (e.g., three step rotations at 65°, 25°, and 30°) in human mitochondrial F1 (hMF1) (Suzuki et al. 2014). Similarly, bovine mitochondrial F1 (bMF1), the gold standard model for structural analysis of F1, was studied in the rotation assay and found to have an additional pause in 120° rotation (Kobayashi et al. 2020). However, the position in bMF1 is different from hMF1, making three step rotations of 10–20°, 60–70°, and 40°. These observations suggest that a 9-step rotation is conserved in mammalian mitochondrial F1 domains.

We should be able to progressively detect smaller sub-steps by improving the spatiotemporal imaging resolution and the data analysis methods. In fact, we analyzed the data of rotation trajectories with elaborated mathematical methods and found that TF1 makes an additional small step of 10° between the 80° and 40° sub-steps (Li et al. 2015). In this review, we aimed at a coarse-grained classification of the rotation behavior of F1. Therefore, we only considered the experimentally distinctive steps: the step size must be over 10°, and/or the intervening pause must be long enough to set the pace of the overall rotation rate under a certain condition, typically in the range of sub- or milliseconds.

Stoichiometry of H+ per turn of Fo

Fo is a membrane-embedded motor with the minimum subunit composition of a1b2cn. The stoichiometry (n) of the c-subunits varies from 8 to 15 among species (Meier et al. 2005; Pogoryelov et al. 2009; Watt et al. 2010; Saroussi et al. 2012; Preiss et al. 2014, 2015; Morales-Rios et al. 2015; Guo et al. 2019). The c-subunits form an oligomer ring that is rotated against the ab2 stator complex upon proton translocation across the membrane. According to the two half-channel model (Vik and Antonio 1994; Junge et al. 1997), which is well supported by the recent cryoEM studies, the a-subunit has two half-channels, one exposed on each side of the membrane (Allegretti et al. 2015). Each proton enters through one of the half-channels and is transferred to one of the c-subunits. After one turn of the c-ring against the ab2 stator, the proton is transferred to the other half-channel of the a-subunit facing the opposite side of the membrane. Thus, a proton is translocated by a c-subunit, and therefore, the total number of protons translocated per turn is determined by n, the number of c-subunits in the oligomer c-ring.

Currently, there are not enough reports on the stepping rotation of Fo to discuss the experimental data in a comprehensive manner. Our working assumption is that the number of steps in Fo is determined by n.

Number of steps in F1 versus Fo

We analyzed the data on the following ATP synthases: thermophilic Bacillus PS3 TF, Escherichia coli EF, yeast yMF, and bovine bMF. We chose these ATP synthases because both single-molecule rotation assays on F1 (Watanabe et al. 2010; Bilyard et al. 2012; Steel et al. 2015; Kobayashi et al. 2020) and the structural data on the c-ring of Fo (Stock et al. 1999; Ballhausen et al. 2009; Watt et al. 2010; Guo et al. 2019) are available. Considering the evolutionary distance and the high-sequence homology of the c-subunits, it is highly likely that the ATP synthase from human mitochondria (hMF) also contains 8 c-subunits. Therefore, we added the data on hMF. The correlation between the number of steps in F1 and Fo is shown in Fig. 2. Clearly, a 6-step F1 is always paired with a 10-step Fo, whereas a 9-step F1 is paired with an 8-step Fo.

Fig. 2
figure 2

The number of steps in F1 versus the number of steps in Fo. TF represents data on ATP synthase from thermophilic Bacillus PS3, EF from Escherichia coli, yMF from yeast, bMF from bovine, hMF from human, EhV from Enterococcus hirae, and ThV from Thermus thermophilus. Structures of c8-ring of bMF (orange), c10-ring of TF (red), c12-ring of ThV (cyan), and c14-ring of Pisum sativum ATP synthase (brown) are shown

Full size image

To gain more data points, we added information gained from the studies on V-ATPases, which are evolutionarily highly related rotary ATPases. V-ATPases are also composed of two distinctive domains, V1 and Vo, corresponding to F1 and Fo, respectively. To date, there are only two well-characterized V-ATPases for which the number of rotational steps in V1 and number of proton-carrying units in Vo is known. One of them is the Enterococcus hirae V-ATPase (EhV), with a 6-step V1 (Iida et al. 2019) and a Vo with 10 proton-carrying units (Murata et al. 2005), providing support for the abovementioned correlation. The other one is the V-ATPase from Thermus thermophilus (ThV), which consists of a 3-step V1 (Furuike et al. 2011) and a Vo with 12 proton-carrying units (Toei et al. 2007). This data point from ThV appears to expand the correlation map to include 3-step ATPases paired with 12-step proton-conducting domains.

Implications and perspective

Figure 2 shows an obvious trend: ATPase motors with more steps have proton-conducting motors with less steps. Although the total number of steps varies from 15 to 17, this trend appears to be relevant in the design principle of rotary ATPases. One possibility is that rotary ATPases are designed to have potential minima around 16. It is highly likely that some angular pause positions in F1/V1 overlap with the pause positions in Fo/Vo. In that case, the above numbers should indicate the maximum numbers of rotary potential minima per turn in the ATPase complex.

In this letter, we only consider the data points of F/V-ATPases, of which the number of the proteolipid is 8, 10, or 12, due to the limited information. On the other hand, some ATPase’s have different numbers of proteolipids: 9, 11, 13, 14, or 15 (Meier et al. 2005; Pogoryelov et al. 2009; Saroussi et al. 2012; Preiss et al. 2014, 2015). Therefore, it is important to analyze other ATPases to investigate the universality and limitation of the found correlation between the step numbers of F1 and Fo. At least, the correlation line in Fig. 2 should be kinked or broken for Fo/Vo with proteolipids more than 12, because the number of rotational steps in F1/V1 should not be 2 or less, considering the conservation of the threefold symmetry of F1 without exception. A simple expectation is that when the number of proteolipids is 12 or more, F1/V1 is a 3-step motor.

In this regard, FoF1 from Caldalkalibacillus thermarum TA2.A1 (CtF) could be along this contention: CtFo has 13 proteolipids in the c-ring (Matthies et al. 2009), and the single-molecule rotation assay of CtF1 found only 3 distinctive pauses per turn. It should be mentioned that a few rotation trajectories of CtF1 seem to show a sign of the additional pauses in a turn. A more conclusive analysis is awaited. It would be also interesting to characterize the rotary catalysis of the F-ATPases from Paracoccus denitrificans (Morales-Rios et al. 2015), Pisum sativum (chloroplast) (Saroussi et al. 2012), and cyanobacteria bacteria species (Pogoryelov et al. 2007), in which the Fo contains 12, 14, and 13–15 c-subunits, respectively. It should be noted that the deviance from the found correlation may come from ATPases isolated from cyanobacteria species: the single-molecule rotation assay on a thermophilic cyanobacteria species shows the ADP-inhibition pause at a difference position from ATP-binding pause found in active rotation (Konno et al. 2006), suggesting cyanobacterial F1 make more than 3 steps per turn.

Summary

Recent progress in single-molecule rotation analysis and structural analysis on rotary ATPases has revealed a variety of functions and structures among species. This allows for comprehensive analyses. Here, we report a correlation between the number of steps in F1/V1 and that in Fo/Vo. There is a clear trend showing that ATPase motors with more steps have proton-conducting motors with less steps. In addition, ATPases with 6 steps are always paired with proton-conducting domains with 10 steps. To confirm the universality of these findings, we need more data on the rotation and structure of rotary ATPases. A theoretical approach is also needed to investigate the mechanism behind these rules.