ATP synthase: Evolution, energetics, and membrane interactions

Abstract

The synthesis of ATP, life's "universal energy currency," is the most prevalent chemical reaction in biological systems and is responsible for fueling nearly all cellular processes, from nerve impulse propagation to DNA synthesis. ATP synthases, the family of enzymes that carry out this endless task, are nearly as ubiquitous as the energy-laden molecule they are responsible for making. The F-type ATP synthase (F-ATPase) is found in every domain of life and has facilitated the survival of organisms in a wide range of habitats, ranging from the deep-sea thermal vents to the human intestine. Accordingly, there has been a large amount of work dedicated toward understanding the structural and functional details of ATP synthases in a wide range of species. Less attention, however, has been paid toward integrating these advances in ATP synthase molecular biology within the context of its evolutionary history. In this review, we present an overview of several structural and functional features of the F-type ATPases that vary across taxa and are purported to be adaptive or otherwise evolutionarily significant: ion channel selectivity, rotor ring size and stoichiometry, ATPase dimeric structure and localization in the mitochondrial inner membrane, and interactions with membrane lipids. We emphasize the importance of studying these features within the context of the enzyme's particular lipid environment. Just as the interactions between an organism and its physical environment shape its evolutionary trajectory, ATPases are impacted by the membranes within which they reside. We argue that a comprehensive understanding of the structure, function, and evolution of membrane proteins-including ATP synthase-requires such an integrative approach.

Figure 1.

Comparative schematic and atomic models of F1Fo motors from bacteria, chloroplasts, and mitochondria. Map resolutions are as follows: 6.9 Å for E. coli ATP synthase (Sobti et al., 2016); 2.9 Å/3.4 Å for Spinacia oleracea (spinach) chloroplast F1/Fo, respectively (Hahn et al., 2018); and 7.0 Å for Pichia angusta mitochondrial ATP synthase (Vinothkumar et al., 2016). Subunits are color coded to show homologous domains across species: α (red); β (yellow), δ (gray in bacteria and chloroplasts, brown in mitochondria); OSCP (gray in mitochondria); b/b′ (dark purple); c-ring (light purple); F6 (pink); γ (blue); ε (green); and d (light blue).

Figure 2.

Three primary rotary states of intact F1Fo ATPases in bacteria (E. coli), chloroplasts (S. oleracea), and mitochondria (P. angusta), separated by an ∼120° rotation. States are listed from most to least populated; the most populated states for each organism (conformation 1) are as shown in Fig. 1.

Figure 3.

Illustration of one full revolution of chloroplast F1, Fo, and the peripheral stalk connection between the two motors. As shown in Fig. 2, image processing revealed three primary rotary states in intact chloroplast F-ATPase, separated by 112°, 103°, and 145° (Hahn et al., 2018). (a) A full revolution of F1 results in the generation of three ATP molecules, one at each of the three catalytic β-subunits via Paul Boyer’s proposed “binding change mechanism” (Boyer, 1993; Boyer, 1997; Boyer at al., 1977). The three β-subunits cycle between “open,” “tight,” and “loose” states. Tracking a single β-subunit through a full revolution: (1) ADP + Pi arrive and bind at the open site; (2) upon arrival of ADP and phosphate, the open subunit rotates into the loose conformation, followed by (3) a second rotary power stroke into the tight conformation, in which the catalysis into ATP takes place. (4) The final substep reverts the subunit back to its open conformation, releasing the newly synthesized ATP. (b) Rotation angles of the Fo c-ring rotor relative to subunit a. Primary states of the intact motor do not each correspond to an integral number of c-subunits (and consequently, neither to an integral number of ion translocations). (c) Two orthogonal views of superposed peripheral stalks, illustrating the flexible coupling between Fo and F1; this flexibility allows for the observed “symmetry mismatch” between the paired motors (see Fig. 4).

Figure 4.

Phylogenetic distribution of c-ring stoichiometries. Rooted phylogeny of organisms with experimentally determined c-ring structures obtained through the TimeTree knowledge base using the method outlined in Hedges et al., 2015. Leaves are marked with colored circles according to the number of subunits found in the ATP synthase c-ring. Shading corresponds to whether stoichiometries were determined for bacteria (blue), archaea (red), eukaryotic mitochondria (yellow), or chloroplasts (green). Stoichiometries and corresponding references are also provided in Table 1.

Figure 5.

Structure of mitochondrial ATP synthase dimers. (a) Formation of dimer rows relieves strain caused by dimerization-induced local membrane curvature. Perspective slices through simulated membrane patches illustrate how curvature profile changes when one, two, or four ATP synthase dimers assemble into a row (Davies et al., 2012). (b and c) Subtomogram images of ATP synthase dimers (b) and surface representations of cristae structure (c) from representative species of each dimerization class: type I, Saccharomyces cerevisiae (Davies et al., 2012); type II, Polytomella sp. (reproduced with permission from Blum et al. [2019]); type III, Paramecium tetraurelia (Mühleip et al., 2016); type IV, Euglena gracilis (Mühleip et al., 2017).

Figure 6.

Lipids associated with ATP synthase. (a) The evolution of membrane topology from the lightly curved inner membrane of gram-negative proteobacteria to the highly curved inner membrane of eukaryotic mitochondria. For both, the inner membrane contains asymmetric distribution of lipids that provide spontaneous membrane curvature. In mitochondria, asymmetry is more dramatic and likely acts alongside the dimerization of ATP synthase to support cristae structure. (b) Characteristic lipids of the IMM include CL and PE. CL is shown with four linoleic acid chains, as is typical in many mammalian tissues. These lipids are also enriched on the inner leaflet of the membrane, as they are gram-negative bacteria (Bogdanov et al., 2020). (c) Coarse-grained MD simulations of the bovine c₈-ring in a model inner membrane suggest transient interactions with CL, whose mean density distribution is shown in pink. It has been hypothesized that these interactions promote the rotation of the ring during enzyme cycles. Figure adapted from Duncan, et al., 2016.