Intrinsic Disorder in the Structure of IF₁.

Bovine IF₁ is a homodimer in which the C-terminal regions from residues 44–84 form an antiparallel α-helical coiled-coil (4, 7). In this dimeric state the N-terminal inhibitory regions extend in approximately opposite directions, allowing the dimer to bind to two F₁-ATPase complexes simultaneously in vitro (8). However, there is no convincing evidence that an equivalent structure forms in the membrane-bound F₁F_o-ATPase, which is dimerized by an independent mechanism involving protein–protein interactions between specific subunits in the F_o membrane sector of the enzyme (9). In crystals of the dimeric IF₁ (4), where the inhibitors form fibers involving extensive intradimer interactions, the α-helical C-terminal region extends in an N-terminal direction as a continuous α-helix into the inhibitory region, so that the structure of each IF₁ molecule consists of a single α-helix from residues 18–80, and in solution a fragment consisting of residues 44–84 forms a dimeric antiparallel α-helical coiled-coil. However, NMR spectra of a fragment of bovine IF₁ consisting of residues 10–48 contain sharp resonances at or near random coil chemical shifts, showing that this fragment is largely or completely unstructured in solution (7). In contrast, in the structure of the complex of bovine F₁-ATPase inhibited by a monomeric version of IF₁ consisting of residues 1–60, residues 8–50 are resolved with residues 21–50 forming a single α-helix (5). These observations indicate that the inhibitory region of free bovine IF₁ is intrinsically disordered and that it becomes folded either by binding to F₁-ATPase or by self-association into the fibers as observed in crystals of the inhibitor (4). The resistance to heat denaturation of bovine IF₁ (1), the presence in residues 1–50 of a high number of charged residues (five lysines, five arginines, seven glutamates, and two aspartates), and the scarcity of bulky amino acids (two each of leucines, valines, and phenylalanines, one tyrosine, and no isoleucines or tryptophans) are characteristic features of intrinsically unfolded proteins (10). Moreover, programs that predict the presence of disorder in proteins support the presence of an intrinsically disordered N-terminal region of bovine IF₁. Disorder is predicted in residues 1–45 by FoldIndex (11), and in all but residues 12–17 by PONDR (12), and disorder or a loop conformation with high flexibility in residues 1–30 is predicted by DisEMBL (13). Thus, it seems that the N-terminal inhibitory region of bovine IF₁ is intrinsically disordered in free solution and that only the C-terminal region from around residue 44 to approximately residue 84 has a secondary structure in the form of an α-helix.

Initial Interaction of Bovine IF₁ to F₁-ATPase.

The structures of F₁(I1-60His)₂ and F₁(I1-60His-K39A)₃ can be interpreted as providing insights into changes in the secondary structure of the inhibitory region of bovine IF₁ that occur when the inhibitor binds to F₁-ATPase, and this interpretation allows a model of the pathway of binding and folding of the inhibitory region to be constructed. This pathway leads from the unstructured N-terminal inhibitory region of free IF₁, which is also present in monomeric I1-60His, to the largely folded structure observed previously in F₁-I1-60His, and also of I1-60_DP in both F₁(I1-60His)₂ and F₁(I1-60His-K39A)₃. In Fig. 4 the binding of IF₁ to F₁-ATPase is depicted as a four-state process whereby the disordered protein passes through two intermediate states (as observed in the I1-60_E and I1-60_TP inhibitors) before reaching the final folded state observed in I1-60_DP. This process is accompanied by an increase in enthalpy from the formation of interactions between residues in IF₁ and other residues in F₁-ATPase (Fig. 4A) and an increase in entropy from the burial of hydrophobic residues as IF₁ passes via two intermediates from the unbound unfolded state to the final inhibited state (Fig. 4B) (14).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

The folding of the intrinsically disordered IF₁ upon interaction with F₁-ATPase. In solution, the inhibitor is disordered and becomes progressively folded after initial interaction at the α_Eβ_E-catalytic interface (E) of F₁-ATPase, and subsequently the catalytic interface changing to a α_TPβ_TP interface (TP) and finally a α_DPβ_DP interface (DP). (A) The inhibitor residues that are involved in charged interactions with F₁-ATPase. These interactions increase during the binding of the inhibitor leading to an increase in enthalpy that facilitates folding of the IF₁. (B) The inhibitor residues that are in hydrophobic interactions with F₁-ATPase. These interactions increase during the binding of the inhibitor, leading to an increase in entropy of water molecules that were surrounding the unfolded inhibitor. Residues colored yellow, blue, or orange interact with the neighboring β-subunit, γ-subunit, or the closest nonneighboring β-subunit, respectively.

According to this interpretation, the unfolded inhibitory region enters the enzyme at the α_Eβ_E-catalytic interface of F₁-ATPase, the most open of the three catalytic interfaces, which provides an accessible surface in the β_E-subunit where the unfolded inhibitory region of IF₁ can make its initial contacts. This initial mode of interaction has been proposed previously (5), and the present structures strongly favor it over an alternative proposal that the inhibitor enters the enzyme via the partially closed α_TPβ_TP-catalytic interface (15). The interactions that form at this initial stage help to stabilize a 19-residue α-helical segment from residue 31–49 (Fig. 3A).

Completion of the folding of IF₁ cannot take place at this initial phase of binding because a number of key interactions required to stabilize the maximally folded state are unable to form. For example, residue E30 of I1-60_E cannot make a salt bridge with R408 of the β_E-subunit, which can form in both I1-60_TP and I1-60_DP. Superimposition of maximally folded IF₁ on the partially folded I1-60_E in the α_Eβ_E-catalytic interface demonstrated that inhibitor residue E30 and β_E-R408 were 4 Å apart, and that β_E-R408 and β_E-E454 formed a salt bridge, and β_E-R408 clashed with IF₁-E31. The importance of E30 in IF₁ is demonstrated by the mutation E30A, which abolishes the binding of IF₁ to F₁-ATPase (6). It was also apparent from this docking experiment that IF₁ residues V15 and F22 cannot interact with the γ- and β_E-subunits, respectively, at the initial stage of inhibitor binding. In contrast, in the maximally folded IF₁ bound to the α_DPβ_DP-catalytic interface, these residues are positioned appropriately to allow the interactions to form and stabilize the structure of the N-terminal region of IF₁.

Progression of Binding of Bovine IF₁ to F₁-ATPase.

After the initial binding of IF₁ at the α_Eβ_E-catalytic interface, in the next stages of the inhibitory process the γ-subunit rotates through two 120° steps in an anticlockwise direction (as viewed from the membrane domain of intact F₁F_o-ATPase), using energy released by the hydrolysis of two ATP molecules at the α_DPβ_DP-catalytic interface. In this way the enzyme becomes inhibited by IF₁ bound ultimately at the α_DPβ_DP-catalytic interface.

In the conversion of the α_Eβ_E-catalytic interface to a α_TPβ_TP-catalytic interface, the extent of the α-helical inhibitory region of IF₁ increases toward its N terminus by residues 23–30 becoming folded, so that now the α-helical segment of IF₁ extends from residues 23–49, with residues 1–21 remaining evidently disordered and probably accommodated in the central aqueous cavity of F₁-ATPase. In this step, the number of interactions between IF₁ and the enzyme increases (Fig. 3B). The salt bridge between E30 of the inhibitor and R408 of the β_TP-subunit (discussed above) has now formed, R25 of IF₁ has made a salt bridge with γ-E241, and IF₁ residues R37 and E40 interact in an interinhibitor salt bridge. This increase in the extent of folding is accompanied by a rise in buried surface area of hydrophobic residues in IF₁ relative to I1-60_E bound at the α_Eβ_E-catalytic interface. This rise is a consequence of an enhanced interaction between F34 of IF₁ and the adjacent residues, V404, R408, S405, and E454, in the β_TP-subunit. However, the maximally folded inhibitor cannot be accommodated in the α_TPβ_TP-catalytic interface because residues 21 and 22 of IF₁ would clash with the γ-subunit. Thus, the extent of formation of secondary structure of IF₁ toward its N-terminal end is restricted by the conformation of the γ-subunit in the α_TPβ_TP interface.

In the conversion of the α_TPβ_TP- to the α_DPβ_DP-catalytic interface, the α-helical region of IF₁ increases further in an N-terminal direction, so that now it extends from residues 21–50. In comparison with the inhibitor bound at the α_TPβ_TP-catalytic interface, there are now fewer charged interactions between inhibitor and enzyme; residues R25 and E31 of IF₁ no longer form ionic interactions with the F₁-ATPase, and there is no interinhibitor salt bridge between residues R37 and E40. Residues R25, R37, and E40 are all essential for a functional inhibitor, although they are not in contact with the enzyme in the final inhibited complex represented by F₁-I1-60His; as such they were classified as “group 2” residues (6). The present structures demonstrate that they play important but transient roles in the formation of an intermediate state in the pathway of inhibition. There is as yet no explanation for the roles of two other group 2 inhibitor residues, Q27 and K39, and presumably they play a role in the formation of some other undefined intermediate in the folding pathway.

The reduction in interactions in the conversion of the α_TPβ_TP- to the α_DPβ_DP-catalytic interface is compensated by a 55% increase in the total buried surface area of hydrophobic inhibitor residues at the α_DPβ_DP-catalytic interface. This increase arises from greater interactions between the enzyme and IF₁ residues, L42 and L45 and the β_DP-subunit, and the formation of the new interactions between residues V15 and F22 of IF₁ with the γ- and β_TP-subunits, respectively, which can now form because the γ-subunit and the neighboring β_TP-subunit have adopted appropriate conformations. These conformations also allow a structured region to form from residues 11–20 of IF₁, including an α-helical turn from residues 15–18. This turn is immediately adjacent to the γ-subunit and prevents further rotation and hydrolysis of ATP.

One characteristic feature of IF₁ is that it is a unidirectional inhibitor of ATP hydrolysis without direct effect on ATP synthesis. Once a proton motive force in the appropriate sense is applied to the rotor of the inhibited complex, its direction of rotation will reverse, and bound IF₁ will be ejected from its inhibitory site at the α_DPβ_DP-catalytic interface. The pathway of ejection of IF₁ is presumed to be the reversal of the inhibitory pathway and would regenerate the active ATP synthesizing enzyme and the unfolded inhibitory region of IF₁.

Comparison of folding of IF₁ and other proteins.

The folding pathways of barnase and chymotrypsin inhibitor 2 (CI2) describe a progression from the denatured protein to a semistructured intermediate, to a near native transition state lacking a fully packed hydrophobic core, and finally to the fully folded native state (16, 17). A similar pathway has been described for the Engrailed homeodomain (En-HD) of Drosophilia melanogaster, where the structure of the immediate state in the protein folding pathway was determined by solution NMR (18). These pathways and that proposed for the inhibitory domain of IF₁ resemble each other insofar as an initial nucleus of secondary structure forms first, and then this nucleus develops into the final fully folded state. The structures of I1-60_E and I1-60_TP in the F₁(I1-60His)₂ and F₁(I1-60His-K39A)₃ complexes could be considered as being equivalent to the semistructured intermediate states in the folding pathways of barnase, CI2, and En-HD. However, the folding of IF₁ differs from barnase, CI2, and En-HD in that it is not a spontaneous process and it requires the participation of specific regions of F₁-ATPase.

Biological Roles of IF₁.

IF₁ is a unidirectional inhibitor of ATP hydrolysis that has no direct effect on ATP synthesis by F₁F_o-ATPase (2). The primary biological role suggested for IF₁ is to act as an emergency protein that intervenes to inhibit ATP hydrolysis by F₁F_o-ATPase when the proton motive force collapses, for example during anoxia. Experimental support for this role is that human cells where expression of IF₁ has been suppressed have reduced ATP levels relative to untreated cells (19). However, mice in which the gene for IF₁ had been disrupted were viable, suggesting that IF₁ is required only under restricted conditions (20). One proposed restricted role is that IF₁ is a mediator of apoptosis. Apoptosis is initiated by the release from mitochondria of proapoptotic molecules, such as cytochrome c, and release of cytochrome c from mitochondria increases when IF₁ expression is suppressed, and is reduced by the overexpression of IF₁ (21). Hence, it has been proposed that IF₁ influences whether or not a cell enters apoptosis, and in tumor cells where the levels of IF₁ are elevated it may be that IF₁ plays a role in suppression of cell death (22). One characteristic feature of intrinsically disordered proteins that fits with the multiple possible roles for IF₁ is that such proteins have the ability to recognize and interact with more than one partner protein (23, 24). It has been suggested, for example, that in addition to recognizing and binding to F₁-ATPase, IF₁ may interact with calmodulin (25), but the physiological relevance of such an interaction is unclear, and other possible interacting partners such as those that may mediate the influence of IF₁ on apoptosis remain to be identified.