Assembling the mitochondrial ATP synthase

Mitochondria are known as the powerhouses of the cell. The F₁F_o-ATP synthase of the mitochondrial inner membrane produces the bulk of cellular ATP. The respiratory chain complexes pump protons across the inner membrane into the intermembrane space and thereby generate a proton-motive force that drives the ATP synthase. In a fascinating molecular mechanism, the ATP synthase couples the synthesis of ATP to the transport of protons into the matrix (1–3). Formation of the ATP synthase depends on the association of 17 different structural subunits of dual genetic origin. Whereas a number of assembly factors and steps have been identified in the model organism baker’s yeast, little has been known about the assembly of the human ATP synthase. In PNAS, He et al. (4) report the identification of key assembly intermediates in the formation of the ATP synthase in human mitochondria. In a tour de force, they define the molecular composition of distinct vestigial ATP synthase complexes in cell lines lacking individual subunits of the enzyme and develop a model of how the membrane-bound domain of the ATP synthase is formed (Fig. 1).

Fig. 1.

Assembly pathways of the mitochondrial ATP synthase. (Upper) Assembly of human ATP synthase. The free F₁ domain or an F₁–c-ring intermediate binds to the peripheral stalk. The supernumerary subunits e, g, and f associate and promote the insertion of ATP6 and ATP8. Addition of 6.8PL and DAPIT stabilizes the inserted ATP6/ATP8, leading to formation of the proton-conducting channel between ATP6 and the c-ring. The inhibitory protein IF₁ that blocks ATP hydrolysis of uncoupled ATP synthase is released. (Lower) Assembly of yeast ATP synthase. The chaperones Atp11 and Atp12 promote formation of the F₁ domain. The INAC binds to the c-ring, as well as to an assembly intermediate that contains Atp6, Atp8, the peripheral stalk, the F₁ domain, and the maturation factors Atp10 and Atp23. The INAC thus supports proper association of the c-ring with Atp6 to allow formation of the proton-conducting channel. Subsequently, the supernumerary subunits are added to promote dimerization of the enzyme. (Inset) Structural subunits of human and yeast ATP synthase. Asterisks mark mitochondrially encoded subunits. IM, inner mitochondrial membrane; IMS, intermembrane space; OSCP, oligomycin sensitivity-conferring protein.

The ATP synthase consists of the inner membrane-bound F_o region and the matrix-exposed F₁ region. The catalytic head of the F₁ region and the membrane-integrated rotor module of the F_o region are linked by two stalks, the F₁ central stalk and the F_o peripheral stalk (Fig. 1). ATP6, together with the c-ring, forms the proton-conducting channel of the F_o region. The central c-ring is composed of eight or 10 identical subunits in human and yeast mitochondria, respectively. Transport of protons across the F_o rotor module results in rotation of the c-ring. The torque of the c-ring is transmitted to the catalytic head of the F₁ region via the γ-subunit of the central stalk. Three heterodimers of subunits α and β form the reactive centers that undergo conformational changes upon rotation of subunit γ, leading to synthesis of ATP. The peripheral stalk prevents rotation of the catalytic head (1–3). Five so-called supernumerary subunits are part of the F_o region and contribute to the formation of dimers and oligomers of the ATP synthase (5–7). A recent high-resolution cryoelectron microscopic structure of the dimeric F_o region of yeast ATP synthase revealed that the subunits Atp6 and i/j form the contact sites between two ATP synthase monomers, supported by interaction between subunits e and k (8). In mitochondria, rows of ATP synthase dimers localize to the rims of the cristae membranes, which are invaginations of the inner membrane. The ATP synthase dimers bend the inner membrane and are crucial for forming the typical cristae shape (9, 10). The supernumerary subunits e and g, together with the N-terminal portion of the peripheral stalk subunit b, affect the curvature of the inner membrane (8, 9, 11). Thus, the ATP synthase not only synthesizes ATP but is also crucial for the architecture of the mitochondrial inner membrane.

The assembly of the mitochondrial ATP synthase is a complicated process that involves the coordinated association of mitochondrially and nuclear encoded subunits (12). In humans, the genes encoding the F_o subunits ATP6 and ATP8 are located in the mitochondrial genome. The other subunits are encoded in the nuclear genome, synthesized as precursor proteins on cytosolic ribosomes and imported into mitochondria. During assembly of the ATP synthase, the F₁ domain seems to form independently of the membrane-integrated domain. In yeast, Atp11 and Atp12 bind to the newly imported subunits β and α, respectively. Loss of these assembly factors leads to aggregation of subunits α and β, and blocks formation of the ATP synthase (13). According to the current view, association of the central stalk leads to the release of Atp11 and Atp12 and the formation of the F₁ domain (12).

A crucial step during assembly of the ATP synthase is the formation of the proton-conducting channel between ATP6 and the c-ring within the F_o rotor module. A proton-conducting channel that is not properly inserted into an ATP synthase would lead to a dissipation of the proton gradient across the inner membrane. He et al. (4) generated human cell lines with disruption of individual genes for the supernumerary subunits e, f, g, 6.8PL (6.8-kDa mitochondrial proteolipid), and diabetes-associated protein in insulin-sensitive tissues (DAPIT). They found that lack of individual supernumerary subunits leads to respiratory defects and loss of ATP synthase dimers or oligomers. They also determined the molecular composition of vestigial ATP synthases in various knockout cell lines that lacked single supernumerary subunits, peripheral stalk subunits, or the c-ring, as well as in ρ⁰ cells devoid of mitochondrial DNA (4, 14, 15). Based on their findings, they propose an elegant model of how the membrane domain of human ATP synthase is built (4) (Fig. 1, Upper). In one branch, an F₁–c-ring intermediate associates with the peripheral stalk and the supernumerary subunits e and g (15, 16). In the other branch, the F₁ domain first assembles with the peripheral stalk and supernumerary subunits e, g, and f. Both pathways merge in a key assembly intermediate that contains the F₁ domain, the c-ring, the peripheral stalk, and the supernumerary subunits e, g, and f. In all these vestigial ATP synthase complexes, the inhibitory protein IF₁ is enriched to prevent ATP hydrolysis by the uncoupled ATP synthase (4). The presence of the supernumerary subunits e, g, and f is crucial for the subsequent integration of the mitochondrially encoded subunits ATP6 and ATP8 that are stabilized by addition of 6.8PL. Thus, the proton-conducting channel between ATP6 and the c-ring is formed. At this stage, ATP synthesis is coupled to the proton-motive force and the inhibitory protein IF₁ is released. Finally, DAPIT is added to the assembly line to promote dimerization and oligomerization of the ATP synthase.

In addition, He et al. (4) assigned 6.8PL and DAPIT as the likely functional orthologs of the yeast subunits i/j and k, respectively. Thus, the 17 main structural subunits of human and yeast ATP synthase are now directly comparable (Fig. 1, Inset).

Interestingly, several maturation steps of yeast and human ATP synthase differ (Fig. 1). Yeast subunit c is mitochondrially encoded, whereas three nuclear genes code for identical c subunits in humans (only the precursors carry different mitochondrial targeting sequences). Yeast Atp6 and Atp8 associate first with a peripheral stalk-F₁ intermediate, and the c-ring is added subsequently (17). In human mitochondria, ATP6 and ATP8 insert into an intermediate form of the ATP synthase that contains the c-ring, F₁ domain, peripheral stalk, and supernumerary subunits (4). A number of assembly factors for formation of the membrane-bound F_o region have been identified in yeast. The two-domain protein Atp25 stimulates synthesis and assembly of the subunit c into the c-ring (18). The protease Atp23 processes membrane-inserted yeast Atp6, whereas human Atp6 is not proteolytically cleaved. In addition to its proteolytic activity, Atp23, along with the Atp10 chaperone, stabilizes and assembles yeast Atp6 (19, 20). The inner membrane assembly complex (INAC), composed of Ina17 and Ina22 (21), binds to two distinct assembly intermediates of the yeast ATP synthase: the newly assembled c-ring and an assembly intermediate composed of the F₁ domain, the peripheral stalk, Atp6/Atp8, and the assembly factors Atp10 and Atp23 (17) (Fig. 1, Lower). The INAC prevents a premature interaction of the intermediates and supports the proper assembly of the c-ring with Atp6 to form the proton-conducting channel. Thus, although the overall composition of human and yeast ATP synthase are similar, the assembly pathways of the central proton-conducting channel differ.

Taken together, the analysis of vestigial ATP synthase complexes by He et al. (4, 14, 15) leads to a stepwise assembly model of the membrane domain of human ATP synthase. Future studies will address the molecular mechanisms and regulation of the distinct assembly steps. Important questions are if and which assembly factors promote the assembly of the human ATP synthase. Although humans and yeast belong to the same supergroup of eukaryotes, the opisthokonts, they use different assembly pathways for the proton-conducting channel. Studies addressing the assembly of the mitochondrial ATP synthase in different eukaryotic supergroups will likely reveal novel mechanisms for building this central cellular machinery.