ATP Synthase, an Energy-Generating Rotary Motor Engine
Discovery Institute recently released a stunning animation of the mechanics of ATP synthase, a biomechanical power generator found almost ubiquitously across life. Let's consider the engineering prowess of this amazing molecular machine.
There exists three main types of membrane-embedded ion-translocating ATPases: F-type, V-type and P-type. I will discuss here the F-type ATPases (also called ATP synthase). V-type ATPases facilitate the acidification of intracellular organelles, and use the energy from adenosine triphosphate (ATP) hydrolysis to pump protons into cells and organelles (Beyenbach and Wieczorek, 2006). P-type ATPases are involved in the pumping of cations, also using the energy of ATP hydrolysis (Bublitz et al., 2011; Kuhlbrandt, 2004). The F-type ATPase discussed here is unique inasmuch as it, rather than hydrolysing ATP, actively synthesizes it using the energy from the flow of protons down an electrochemical gradient. There are also A-type ATPases which are found in archaea and perform a similar function to F-type ATPases (Bickel-Sandkötter et al., 1998).
ATP synthase has been described as "a splendid molecular machine," and "one of the most beautiful" of "all enzymes" (Boyer, 1997). This "bona fide rotary dynamo machine" (Mulkidjanian et al., 2007) is the key enzyme involved in the biochemical process known as oxidative phosphorylation, which is very closely coupled to the electron transport chain since the electrochemical proton gradient that is produced by electron transport supplies the energy necessary for the production of the ATP from adenosine diphosphate (ADP) and a phosphate group. The ATP synthase machine is able to crank out approximately 100 ATP molecules per second. With its near-100% efficiency, far surpassing human technology, ATP synthase manifests clear evidence not merely of engineering but of brilliant engineering. A recent paper in Nature offered a "resolution map of the H+-driven ATP synthase from Thermus thermophilus obtained by electron cryomicroscopy of single particles in ice," (Lau and Rubinstein, 2012).
The electron transport chain and oxidative phosphorylation occur following three previous stages of aerobic respiration (glycolysis, pyruvate oxidation, and the tricarboxylic acid, or TCA, cycle), and differ between eukaryotes and bacteria. For one thing, in eukaryotes, the process takes place within the cell's mitochondria (or chloroplasts), whereas, in bacteria (which lack mitochondria and chloroplasts), the process takes place in the cell's plasma membrane. Other differences relate to the individual proteins involved (eukaryotes use five main protein complexes, whereas bacteria use many different enzymes). The purpose of this article is to unravel the molecular complexity of this nanotechnological marvel. I will dwell only very briefly on the electron transport chain, offering an explanation merely by way of background.
The Electron Transport Chain
Briefly, the electron transport chain involves the flow of electrons through a respiratory chain. Electrons pass through three protein complexes that are embedded in the inner mitochondrial membrane: NADH-Q oxidoreductase (Complex I); Q-cytochrome c oxidoreductase (Complex III); and cytochrome c oxidase (Complex IV). Complex I, a large multi-subunit protein, is the enzyme that catalyzes the transfer of electrons from the reducing agent (electron donor) NADH to coenzyme Q. The electrons are relayed to cytochrome c at Complex III, and Complex IV transfers the electrons to O2, which is thus reduced to H2O. There are some substrates, such as succinate which has more positive redox potentials than NAD+/NADH, which involve succinate-Q reductase (Complex II) instead of NADH-Q oxidoreductase (Complex I).
The transmembrane electrochemical gradient across the inner mitochondrial membrane (the matrix side now has a net negative charge) creates a proton motive force that drives the process of ATP synthesis. Complexes I, III and IV serve as proton pumps, transporting protons from the matrix into the intermembrane space. The complexes utilize the energy given up by the flow of electrons. The inner mitochondrial membrane is impermeable to protons, leading to their accumulation in the intermembrane space. Like water behind a dam, this build-up of protons stores potential energy. The principle of ATP synthase is to facilitate the flow of protons down their concentration gradient from the inner membrane space to the matrix, using the energy released in the process to create ATP.
ATP synthase consists of two protein complexes, the F1 domain and the F0 domain, each of which is comprised of several different subunits. The F0 complex is a proton channel and is embedded within the mitochondrial membrane. The F1 complex is the site of ATP synthesis and is located in the mitochondrial matrix. The F1 domain is made up of alpha, beta, gamma, delta and epsilon subunits. The alpha and beta subunits, present in three copies each, make up the catalytic core of the F1 domain, while the gamma subunit comprise the rotating central stalk, which connects the F1 domain to the F0 domain. The epsilon subunit is an ATPase inhibitor, and can take two conformations: extended and contracted. The conformation taken depends on the rotational direction of the gamma subunit and the presence of ADP. When it switches from the contracted to the extended conformation (triggered by the presence of ADP), where the c-terminus extends towards the F1 domain, the epsilon subunit is believed to inhibit ATP hydrolysis (but not synthesis), thereby operating as a safety lock to limit the wasteful hydrolysis of ATP (Feniouk and Junge, 2005).
The alpha and beta subunits are attached to the central stalk and form a hexameric cylindrical structure that surrounds it: Each also possesses a substrate-binding site (the ones on the alpha subunits are regulatory, while the ones on the beta subunits are catalytic). There is also a peripheral stalk, which extends from the membrane to the top of the F1 domain and is comprised of b and d subunits. It is attached to the α3β3 hexamer by the delta subunit. The purpose of the peripheral stalk is to anchor the alpha and beta subunits of the F1 domain, preventing them from rotating as the central stalk rotates.
The principle subunits in the F0 domain are a, b and c. Ten to fifteen (depending on the species) c-subunits form a transmembrane ring known as the "c ring", which is the rotor of the F0 domain (Meier et al., 2005). The F1 domain has its own rotor, comprised of the central gamma subunit inside the cylinder formed by the alpha and beta subunits. The rotors move in opposite directions. When the c ring rotor dominates, it uses the energy of the proton motive force to drive the reverse rotation of the gamma subunit rotor (clockwise), facilitating the production of ATP (Okuno et al., 2011; Yasuda et al., 2001). When the gamma subunit rotor dominates, it uses the energy from the hydrolysis of ATP to drive the reverse rotation of the c ring rotor (counter-clockwise), and protons are pumped against their electrochemical gradient.
Proton flow through the F0 domain causes the complex to rotate, driving ATP synthesis in the F1 domain. Some bacterial species, such as the obligate anaerobe Propionigenium modestum, use sodium ions instead (Dimroth et al., 1999; Dimroth, 1992; Laubinger and Dimroth, 1989). As the axis rotates, the conformation of the alpha and beta subunits in the F1 complex's active site is altered such that it switches from an "open" state (where ADP and phosphate can enter the active site) to a "closed" state (where ADP and phosphate are bound loosely) to a "tight" state (where the ADP and phosphate molecules are forced together, covalently bonding to form ATP) (Boyer, 1997). The active site then undergoes a further conformational change, resulting in the breaking of the hydrogen bonds that were stabilizing the ATP in the active site (releasing the newly-formed molecule), and reverting back to the original open state, ready for another reaction cycle.
How does the flow of protons cause rotation of the F0 domain? Site-directed mutagenesis studies in Escherichia coli have determined that two amino acid residues are crucial to the function of the F0 motor (Fillingame et al., 2002; Valiyaveetil and Fillingame, 1997). These are: "aArg-210 in one of the transmembrane α-helices (TMH), TMH4, of the a-subunit and cAsp-61 in the outer TMH of each c-subunit, the primary proton binding sites located in the middle of the membrane hydrophobic layer," (Aksimentiev et al., 2004).
Garrett and Grisham (2008) explain the remarkable process of motor rotation:
The a-subunit contains two half-channels, a proton inlet channel that opens to the intermembrane space and a proton outlet channel that opens to the matrix. The c-subunits are proton carriers that transfer protons from the inlet channel to the outlet channel only by rotation of the c-ring. Each c-subunit contains a protonatable residue, Asp61. Protons flowing from the intermembrane space through the inlet half-channel protonate the Asp61 of a passing c-subunit and ride the rotor around the ring until they reach the outlet channel and flow out into the matrix... Each c-subunit in the c-ring has an inner helix and an outer helix. Asp61 is located midway along the outer α-helix. When protonated, the Asp carboxyl faces into the adjacent subunit. Rotation of the entire outer α-helix expses Asp61 to the outside when it is protonated. Arg210, located midway on a transmembrane helix of the a-subunit, forms hydrogen bonds with Asp61 residues on two adjacent c-subunits. The half-channels of the a-subunit extend up and down from Arg210. The inlet channel terminates in Asn214, whereas the outer channel terminates at Ser206.
The structure of the c-subunit complex is exquisitely suited for proton transport. When a proton enters the a-subunit inlet channel and is transferred from a-subunit Asn214 to c-subunit Asp61, the α-helix of that c-subunit rotates clockwise to bury the Asp carboxyl group. Each Asp61 remains protonated once it leaves the a-subunit interface, because the hydrophobic environment of the membrane interior makes deprotonation (and charge formation) highly unfavorable. However, when a protonated Asp residue approaches the a-subunit outlet channel, the proton is transferred to Ser206 and exits through the outlet channel. The a-subunit Arg210 side chain orients adjacent Asp61 groups and promotes transfers of entering protons from a-subunit Asn214 to Asp61 and transfers of existing protons from Asp61 to a-subunit Ser206. Arg210, because it is pronated, also prevents direct proton transfer from Asn214 to Ser206, which would circumvent ring rotation and motor function.
For further discussion of the rotary mechanism of ATP synthase, see Arechaga and Jones (2001).
An interesting review article reported on the incredible engineering found in the F0 motor (von Ballmoos et al., 2009):
The rotational mechanism of the ATP synthase demands ingeniously designed interfaces between rotor and stator subunits, particularly between the rotating c ring and the laterally abutted subunit a, because rotation speeds up to 500 Hz must be tolerated in the absence of a stabilizing rotor axis. This proteinous interface also acts as the critical scaffold for torque generation and ion translocation across the membrane. To prohibit charge translocation without rotation, ion leakage at the interface must be efficiently prevented.
For non-specialists it can often be difficult to visualize, from a written description alone, exactly what is going on at the molecular level. This is why computer animations can prove invaluable in understanding the operation of these systems. To get a better handle on how rotation of the F0 component drives the conformational changes of the active site in the F1 component, readers are referred to this animation. For a much closer inspection of the conformational changes that take place in the F1 complex at the molecular level, readers are directed to this animation. For an animated overview of the whole process from electron transport to oxidative phosphorylation, see this video.
The Evolution of ATP Synthase
ATP synthase is found almost ubiquitously across life, and is even found in fermentative organisms that lack the electron transport chain and do not undergo oxidative phosphorylation, such as Clostridium pasteurianum (Das and Ljungdahl, 2003). This is because many cellular processes use energy from the proton motive force rather than ATP. The motor is reversible, and hydrolysis of ATP by the α3β3-catalytic hexamer can supply the torque for the rotation in the opposite direction, catalyzing the active transport of protons from inside to outside the cell. Hydrolysis of ATP sometimes has to be halted in the cell (e.g. during hornworm moulting or during glucose deprivation). V-type ATPases have a nifty trick in this regard. In such cases, the V1 domain has been observed to reversibly detach from the V0 domain (Drory and Nelson, 2006; Beyenbach and Wieczorek, 2006; Iwata et al., 2004), owing to the "socket like function" of subunit C "in attaching the central-stalk subunits of the V1 domain," (Iwata et al., 2004).
ATP synthase bears similarity in some respects to other systems. Ion channels are very common in biology -- for example, bacterial flagella are also driven by a proton motive force across the membrane (although some bacteria, such as Vibrio species, use a flow of sodium ions instead). FliI serves as an inessential ATPase involved in the flagellar export apparatus, and it exhibits significant degrees of homology to the alpha and beta subunits of ATP synthase (Imada et al., 2007; Vogler et al., 1991). FliI also has a homologue in Yersinia pestis called YscN, which serves to energize the type III secretion system (Woestyn et al., 1994). It is hypothesized by some that an F or V-type ATPase was recruited by these systems (Pallen et al., 2006; Blocker, 2003; Aizawa, 2001). What is interesting, however, is that the ATPase of these systems lacks the characteristic central stalk found in F and V-type ATPases (Mulkidjanian et al., 2007).
The hexameric structure formed from the three alpha and three beta subunits also resembles hexameric DNA helicases which, like the ATP synthase, form a ring with a central channel, and three-fold rotational symmetry. DNA helicases also possess ATPase activity, and use the energy from ATP hydrolysis to move directionally along the phosphodiester backbone of DNA and separate the two nucleic acid strands.
How could a complex macromolecular machine like ATP synthase have evolved by natural selection? No other enzyme works in the same way. One hypothesis is that a proton motor came to be associated with a DNA or RNA helicase, and the ATPase of the helicase was driven forcibly in reverse by the proton motor (Mulkidjanian et al., 2007; Walker, 1998; Doering et al., 1995). It's certainly an elaborate story, but biology doesn't work like that. Molecular machines are not Lego bricks. They don't spontaneously combine to form new machines. Furthermore, explaining the F-type ATPases in terms of other ATPases only begs the question of the origin of ATPases in the first place. The ability of the F0 domain to cause such specific conformational changes in the active site of the F1 domain via proton-driven rotation requires foresight and planning, and ingeniously designed interaction.
It is interesting that even though "the stalk domains of the more closely related A1 and V1 are remarkably similar in shape and dimensions, and are different in these respects from the F1-ATPase", nonetheless, "A1A0 and F1F0 enzymes function as ATP synthases in cells whereas the V1V0 ATPase works as an ATP-driven ion pump," (Gruber et al., 2001). What is even more curious is that comparison of the F-type and V-type ATPases reveals non-homology between the subunits of the central stalk of the two systems (a component that is essential for rotation catalysis), even though there is homology between their membrane and catalytic subunits (Mulkidjanian et al., 2007; Gruber et al., 2001). One hypothesis that takes this into account is that "the conserved head structure, the membrane portion and the peripheral stalk (or stalks) together could have formed a translocase that coupled ATP hydrolysis to the transfer of RNA and/or proteins across the membrane, with the translocated polymer occupying the place of the central stalk," (Mulkidjanian et al., 2007). Such a hypothesis is interesting. After all, there are helicase/F1ATPase-like DNA translocases involved in bacterial chromosome segregation during cytokinesis (Graham et al., 2010) and in bacterial conjugation (Tato et al., 2005; Gomis-Rüth et al., 2001). Protein translocases are also common (e.g. Papanikou et al., 2007). Since the central stalks of the F and V-type ATPases are not homologous, such a scenario would entail that the nucleic acid or protein translocase was independently converted into an ion-translocating ATPase (unless the central stalk was later replaced following divergence from a single common ancestor, but this doesn't seem plausible). Moreover, it is hypothesized that the ATPase system underwent at least two, and possibly three, reversals in function; "from a progenitor proton-pumping ATPase to a proton-driven ATP synthase" and subsequently "transforming the synthase back into a proton-pumping ATPase" before perhaps transforming it "from an ATPase back to an ATP synthase" (Cross and Muller, 2004).
Such a scenario would suggest that a translocating protein somehow got stuck in a protein translocase, and subsequently evolved into the central stalk of the membrane ATPase. This hypothesis seems less than convincing to me, however. For one thing, the central stalk has to somehow be coupled to the c-ring (in F-type ATPases) or the A subunit (in V-type ATPases) in order for it to rotate. Are we to expect the broken protein translocase to stick around through multiple generations while we wait for it to evolve into the membrane ATPase system? In any case, protein (and nucleic acid) translocation is dependent upon ATP hydrolysis, so these systems already possess ATPase activity (which is left unaccounted for). ATP synthases also appear to be essential in nearly all life-forms, and ATPases more generally are absolutely essential in all known life. It should be noted, however, that the F-type ATPase described in this article appears to be non-essential in some life forms, since they are not found in any member of the bacterial phylum Deinococcus-Thermus (where V-type ATPases are the only category of membrane ATPase) (Lapierre et al., 2006). Moreover, a small subset of obligate parasitic bacteria, such as the BCc strain of Buchnera aphidicola (an endosymbiont of cedar aphid Cinara cedri), are able to get by without ATP synthase (Charles et al., 2011). This organism is an obligate parasite, however, meaning that it is unable to survive outside of its host organism. With the essential nature of the ATPases in mind, it seems highly unlikely that the ATPases could have evolved gradually by natural selection.
ATP synthase is truly a marvel of nanotechnology. With its ingenious design and remarkably high efficiency and speed, this amazing molecular energy turbine stands among the numerous examples of complex macromolecular machines that bear the unmistakable imprints of intelligence and foresight. As one recent review paper stated, the "unique energy transmission mechanism [found in ATP synthase] is not found in other biological systems. Although there are other similar man-made systems like hydroelectric generators, F0F1-ATP synthase operates on the nanometer scale and works with extremely high efficiency," (Okuno et al., 2011). If such a unique and brilliantly engineered nanomachine bears such a strong resemblance to the engineering of manmade hydroelectric generators, and yet so impressively outperforms the best human technology in terms of speed and efficiency, one is led unsurprisingly to the conclusion that such a machine itself is best explained by intelligent design.