Abstract
We scrutinized the evolution of rotary membrane ATPases/ATP synthases by combining phylogenomic and comparative structural analyses. Regarding F-type ATPases, we found that organisms with more subunits in the membrane-embedded "proteolipid" c-rings tend to have shorter or absent δ-subunits. Organisms with such F-ATPases branch earlier in the phylogenetic tree of catalytic β-subunits. Additionally, we discovered that archaeal-type A-ATPases form at least three distinct branches, including both archaeal and bacterial A-ATPases. This finding suggests that A-ATPase genes have undergone several lateral transfers from archaea to bacteria. We posit that the common ancestor of rotary membrane ATPases possessed six catalytic subunits and over 20 subunits in their Na⁺-transporting c-rings. This ancestor likely had two ion-translocating peripheral stalks and was optimized for low membrane potentials. The loss of half the catalytic subunits, which occurred separately in bacterial and archaeal lineages, enabled the number of c-subunits to be halved without losing efficiency. As cell membranes gained the ability to hold increasingly high membrane voltages, the loss of one of the ion-translocating peripheral stalks, separately in A- and F-ATPases, doubled their efficiency. The mechanism of this loss differed between the two types of ATPases; in F-ATPases, the gain in efficiency came at the cost of mechanistic stability. Thus, the large-scale lateral gene transfer of A-ATPases to bacteria may have been caused by the lower mechanical stability of prokaryotic F-ATPases. The concomitant acquisition of novel subunits resulted in the now-typical Na⁺- and H⁺-transporting A- and F-ATPases geared toward high membrane potentials. Overall, rotary ATP synthases evolved to produce more net ATP.