A Conserved Steroid Binding Site in Cytochrome c Oxidase^†,‡

Cytochrome c oxidase (CcO)¹ catalyzes a reaction vital to energy production, reducing oxygen to water while translocating protons across the membrane to form the electrochemical gradient used for ATP synthesis. Because of its importance in energy metabolism, a variety of regulatory mechanisms are expected and observed, to achieve control of activity and efficiency. For the mammalian mitochondrial enzyme, besides being controlled on the transcriptional (1), translational (2), and molecular assembly levels (3), its activity is also regulated by secondary messengers such as cAMP (4, 5), Ca²⁺ (5), and nitric oxide (6), with consequences that are not yet well defined.

The detailed mechanism of how CcO couples electron transfer to proton pumping, and various modes of regulation, are not fully understood. Some mechanistic insight has been gained by studying the bacterial homologues of the mammalian enzyme, such as CcO from Rhodobacter sphaeroides (Rs), by mutational, kinetic, and crystallographic methods. In the bacterial system, it has been established that during each catalytic cycle protons are taken up from the inside of the membrane via two pathways, labeled D and K (7). The D pathway transports protons destined for translocation across the membrane, plus protons used for generation of water at the active site, while the K pathway transports only protons involved in reduction of oxygen to water.

Mutations of a key residue at the entrance of the K path, E101_IIA (Rs numbering, with the subscript representing the subunit number), leads to severe inhibition of overall activity (<5% of that of the wild type), but the remaining activity is still coupled to proton pumping. We previously reported that certain carboxylic acids, including arachidonic acid and the bile acid, cholate, can rescue the activity of the E101_IIA mutant, from a few percent to more than 50% of the wild-type activity (8). We hypothesize that the rescuing effect is due to the ability of the exogenous lipophilic carboxylate to bind to a position near the entrance of the proton pathway, partially compensating for the loss of the carboxylate group. Arachidonic acid has a similar effect on a D path mutant, D132_IA (9), but cholate activates only the K path (8), suggesting a more specific interaction.

Here we report the potent rescuing effect of deoxycholate on the K path mutant of RsCcO (Figure 1), and the X-ray crystal structure of wild-type RsCcO with deoxycholate bound in close proximity to the K pathway, with its carboxyl group positioned at the entrance, as predicted from the rescuing effect (Figure 2).

Figure 1 shows the comparison of the activating effects of cholate and deoxycholate on the E101_IIA mutant. The final activation level is similar, giving approximately 50% of wild-type activity (700 vs 1200 e⁻/s), but half-maximal stimulation was achieved at 40 μM for deoxycholate and 350 μM for cholate. This suggests that deoxycholate may bind to RsCcO more tightly and in a more specific manner, which makes it a superior candidate for cocrystallization in determining the liganded structure.

The structure of wild-type CcO with deoxycholate bound was determined at 2.15 Å resolution (Figure 2). A new and more extensive packing interface was present in the deoxycholate-bound versus unbound CcO structure. A deoxycholate molecule (dochl) is clearly resolved, as shown in Figure 2A,B. There are two molecules of CcO in the asymmetric unit, but only one of them has the deoxycholate molecule bound.

Detailed examination of the binding interaction, shown in Figure 3A, reveals that the sterol ring of the deoxycholate is associated with a hydrophobic portion of subunit I that is normally within the lipid bilayer, via multiple van der Waals interactions involving P358_I, I361_I, P315_I, T105_II, and I102_II. The 3-OH group is hydrogen bonded to the carbonyl oxygen of Y318_I and to the side chain of A322_I, while the carboxylate group is situated at the K path entrance at the inner surface of the membrane. The binding of the deoxycholate is further strengthened by the ligation of its carboxyl group to a cadmium ion that is also ligated to E101 and H96 of subunit II. Cadmium is a known inhibitor of wild-type RsCcO, as well as an important ingredient for successful crystallization (10). The cadmium ion bound to E101_II and H96_II is also observed in RsCcO structures without deoxycholate (8, 10).

In the vicinity of the deoxycholate, a detergent molecule, decyl maltoside (DM), is also clearly resolved (Figures 2B and 3A). This decyl maltoside molecule associates with one side of the deoxycholate, with its alkyl tail forming multiple van der Waals contacts with the hydrophobic edge of the sterol group of deoxycholate, while the polar sugar headgroup forms interactions with the carboxyl group of the deoxycholate. It has been previously shown that certain detergents can mimic the binding of membrane lipids to membrane proteins and occupy their specific binding sites (10). In fact, a lipid in this position, also associated with the hydrophobic surface of cholate, is resolved in the bovine CcO structure (Figure 3B). Therefore, it is conceivable that in the native membrane, a lipid is present at this site in RsCcO as well and could contribute to the stabilization of steroid binding.

It has long been known that cholate, while being widely used as a solubilizing agent in preparing bovine mitochondrial CcO, promotes dimerization of the enzyme (11) and is also a strong reversible inhibitor (12). The mechanism for such inhibition is unclear, but electron transfer between heme a and heme a₃ is observed to be markedly slowed in the presence of cholate (13). In the structure of bovine CcO (PDB entry 2DYR, 1.8Å), which is a homodimer, a total of four cholate molecules are resolved per monomer, two of them at the dimer interface (14). These cholate molecules bind to residues in both monomers of the dimer, but for the site that matches the RsCcO binding site (Figure 3B), the interaction with a nuclear subunit, VIa, was considered to be of primary interest (15). Careful examination of this bovine CcO VIa binding site shows that the cholate is also bound to subunits I and II of the other monomer, near the entrance of the K pathway. The position is similar to and involves some of the same residues as the deoxycholate bound in RsCcO (Figure 3B). The binding of cholate at this position in bovine CcO is different in detail from the deoxycholate binding in Rs, which could account for the observed inhibitory effect in bovine, not seen in wild-type RsCcO (8). In bovine CcO, the carboxyl oxygen of E62_II at the K pathway entrance (the equivalent of E101_II in Rs) is tightly hydrogen bonded to the 7-OH group of cholate, possibly affecting proton transport.

It was previously proposed that the binding sites of cholate in the bovine CcO structure indicated potential nucleotide binding sites, since cholate is similar to ADP in size and shape (16). The two proposed nucleotide binding sites are at the interface of the dimer: one between subunits I and III close to a conserved Y304 (Y347 in RsCcO) of subunit I which is phosphorylated in bovine CcO (4) and the other in the same location as the RsCcO deoxycholate site and also close to a known site of phosphorylation in bovine CcO, T11 of subunit VIa (14). There is considerable evidence for cholate and adenine nucleotide regulation of the mammalian CcO (15).

Given the previous observations of regulation of CcO activity by a variety of ligands, including bile salts (12) and glucocorticoids (17), and our current evidence of a conserved site that impacts an essential proton uptake path, it is reasonable to propose that the conserved site identified in this study could represent a physiological binding site for steroids or other effector molecules. This possibility is being explored.