How oxygen is activated and reduced in respiration

The advent of aerobic biology was heralded about two billion years ago, when primitive cyanobacteria evolved the ability to photooxidize water. Oxygen was released as a waste product, and atmospheric O₂ levels rose quickly. This rapid change to an oxygenic atmosphere introduced a devastating pollutant, but eventually organisms evolved that capitalized on the strong driving force for O₂ reduction. Enzyme active sites that were capable of binding and activating oxygen evolved, and new classes of biochemistry that use O₂ as a thermodynamic sink to drive otherwise unfavorable reactions became possible. The efficiency of food metabolism changed dramatically. The amount of ATP that could be produced by metabolizing glucose aerobically, for example, increased almost 20-fold. Eukaryotes appeared shortly after the oxygenic atmosphere and were eventually followed by the diverse array of multicellular organism that exist today. In our aerobic biochemistry, O₂ is used in a plethora of synthetic reactions that are fundamental to almost all aspects of cell growth, development, and reproduction.

Despite its biochemical versatility, however, >95% of the oxygen that we consume is used in respiration. High-energy electrons derived from food traverse the mitochondrial electron transport chain in a series of exergonic redox reactions. These energetically downhill electron transfers are used to develop the chemisosmotic proton gradient that ultimately produces ATP. Oxygen is the final electron acceptor in this respiratory cascade, and its reduction to water is used as a vehicle by which to clear the mitochondrial chain of low-energy, spent electrons. The enzyme that catalyzes this process, cytochrome oxidase, spans the mitochondrial membrane. It binds, activates, and reduces up to 250 molecules of O₂ per second and couples the energy released in this process to the translocation of protons that contribute to the chemiosmotic gradient. The mechanism by which cytochrome oxidase catalyzes this remarkable chemistry has been studied intensely. The results reported in this issue by Fabian, Wong, Gennis, and Palmer provide new insight into this process and support the growing notion that unifying concepts exist for the way in which oxygen-utilizing enzymes activate O₂ for O⩵O bond cleavage and reduction (1).

The reduction of O₂ in cytochrome oxidase occurs under severe constraints. The process takes place with little overpotential, the release of partially reduced, toxic oxygen intermediates from the active site is minimized, and the free energy available in O₂ reduction is coupled with high efficiency to proton translocation (2, 3). The enzyme operates under these constraints by using a heme Fe, called heme a₃, and a copper ion, termed Cu_B, in a binuclear center in which O₂ binds and is reduced (see Fig. 1). Electron input to this site occurs from cytochrome c by way of a second heme iron, heme a, and a second copper center, Cu_A. Recently, Yoshikawa’s group (4) and Michel’s group (5) independently and simultaneously provided crystal structures of the enzyme that have given deep insight into many aspects of the catalytic cycle, particularly how protons and oxygen are likely to move through the protein. The mechanism of O₂ reduction by oxidase has been pursued by a number of groups with a variety of spectroscopic techniques (for reviews, see refs. 6 and 7). From this work, a simplified reaction sequence that involves transient, but detectable, intermediates at the binuclear center can be written as follows (see also Fig. 2):

The P and F species, in particular, have attracted attention, as they have been implicated in the pumping mechanism that drives proton translocation (8). Recent work from Michel (9) and from Wikström and coworkers (10) has highlighted both the progress and the uncertainties in our understanding of the mechanism that couples exergonic electron transfers to oxygen with endergonic proton motion across the membrane.

Figure 1.

The binuclear center in cytochrome oxidase. Heme a₃ and Cu_B are shown along with the proximal ligand for the heme iron, H376, and the Cu_B ligand, H240, which is cross-linked to Y244 (24, 25). O₂ binding and reduction occurs in the region between the a₃ iron and Cu_B.

Figure 2.

A simplified scheme for the reaction between cytochrome oxidase and O₂. The binuclear site, which contains heme a₃, Cu_B, and the cross-linked, H240 - Y244 (H-Y) structure, is shown. Reduction and protonation of the oxidized form of the center produces the reduced site. This binds O₂ to form initially the oxy species, which reacts further to produce P and F intermediates, before regenerating the oxidized form of the enzyme. The reduction of P and F are limited by proton transfer reactions, as indicated. The steps between P and the reduced form of the site have been implicated in proton pumping processes, which are indicated by red arrows. The stoichiometry of these steps is a matter of current investigation, although up to four protons can be pumped during the complete cycle.

A continuing issue in unraveling the oxygen chemistry at the binuclear center in cytochrome oxidase and its linkage to the proton pump is to establish the molecular structures of the intermediates in the scheme above. There is consensus that the F intermediate involves a ferryl-oxo intermediate at heme a₃, a₃⁴⁺⩵O (3, 6, 11, 12), but the structure of P has been a matter of considerable controversy. The initial assignments of this species postulated that it contained a bond intact, a₃³⁺—O₂^⩵ species, hence its designation as P for “peroxy” (e.g., refs. 3, 8, and 13). Weng and Baker, however, interpreted their optical data to indicate that O⩵O bond cleavage had already occurred at P and that this species, as well, had an a₃⁴⁺⩵O structure at the binuclear center (14). This conclusion was subsequently supported by several spectroscopic investigations (15–17). Kitagawa, Proshlyakov, and their coworkers succeeded in using Raman spectroscopy to detect the a₃⁴⁺⩵O stretching motion (18, 19) in a form of P generated by adding peroxide to the oxidized enzyme. Subsequent work showed that the same vibration could be observed when oxygen is added to a two-electron reduced form of the enzyme, confirming that oxygen chemistry and peroxide chemistry in oxidase proceed through common intermediates (20). Moreover, the time course of the appearance of P in this work showed that this species is kinetically competent (also see refs. 21 and 22). Thus, from the spectroscopic work, and from recent computational work as well (23), the emerging view is that P is indeed an O⩵O bond-cleaved species.

The work reported by Fabian et al. (1) provides novel, independent, and convincing evidence that the O⩵O bond is cleaved in cytochrome oxidase at the P level. In their experiments, they reasoned that neither oxygen atom in a bond-intact peroxy structure is likely to exchange with solvent water. If P does occur as the a₃⁴⁺⩵O species, however, then one expects that the second oxygen atom is probably at the level of hydroxide or water and that this oxygen may well exchange with water in the aqueous buffer. Using ¹⁸O₂ as the substrate in an aqueous buffer that contained H₂¹⁶O, they trapped the P intermediate and assayed for the appearance of H₂¹⁸O. Their mass spectrometric results show clearly that a single oxygen atom from the ¹⁸O₂ substrate is exchangeable with solvent water, in excellent agreement with their analysis above and the assignment of P as a bond-cleaved, ferryl-oxo species.

The realization that P has an a₃⁴⁺⩵O structure has a number of important implications. The transformation of bound O₂ in the oxy species to hydroxide (or water) and a ferryl-oxo in P requires a total of four electrons. Only three, however, are readily available in the binuclear center—two from heme a₃ as it goes from the +2 to the +4 valence state and one from Cu_B as it is oxidized from cuprous to cupric. The source of the fourth electron is unclear. Oxidation of the heme macrocycle, as occurs in Compounds I in some peroxidases, can be eliminated on the basis of Raman and optical data (6, 7), and Cu³⁺ has not been detected in biological milieu. The most likely candidate, then, is a redox-active protein side chain, as occurs in cytochrome c peroxidase, in which tryptophan is redox active, or in prostaglandin synthase, which contains an oxidizable tyrosine residue (24). Yoshikawa and coworkers (25) provided striking crystallographic evidence that strongly supports the occurrence of a redox-active side chain. They showed that Y244 in the binuclear center is cross-linked to one of the Cu_B ligands, H240, and that the phenol head group is oriented so that the −OH group points directly into the O₂-binding cavity (Fig. 1). Michel has reported similar crystallographic data (26), and Buse and coworkers have recently reported biochemical data that support the occurrence of the H240-Y244 crosslink (27). Recent EPR data have also been reported that indicate the presence of tyrosyl radicals when peroxide is added to the resting enzyme, although the specific side chain(s) involved have not been identified (28, 29). Taken together, these results strongly suggest that the cross-linked tyrosine is the source of the fourth electron in the activation and reduction of O₂ by cytochrome oxidase. This conjecture leads to the simplified reaction cycle in Fig. 2, in which the cross-linked H-Y structure is shown explicitly and proposed to be oxidized to the neutral tyrosyl radical in the P intermediate.

The scheme in Fig. 2 highlights the analogies between cytochrome oxidase and the peroxidases and catalases in terms of oxygen–oxygen bond cleavage chemistry and in terms of the products that result from the reaction. In oxidase, the enzyme extracts three electrons from metals in the active site and a fourth electron from an organic moiety to reduce O₂ in one step to O^⩵ and OH^—. Both of these products are at the level of water, although further protonation and release only occur in later steps of the reaction. In peroxidases and catalases, the enzyme extracts one electron from a metal in the active site and a second electron from an organic moiety to reduce H₂O₂ in one step to O^⩵ and OH^—. In peroxidases and catalases, the immediate product of this chemistry is Compound I, which contains a ferryl-oxo species and an organic radical. These structures are exactly analogous to the a₃⁴⁺⩵O/radical structure that occurs in P in cytochrome oxidase. The organic radical in Compound I is reduced in a subsequent step in the peroxidase and catalase enzymes to produce Compound II, which maintains the ferryl-oxo structure. In oxidase, the same chemistry occurs to produce the F intermediate. The similarity in chemistry of the oxygen-metabolizing heme proteins has only emerged with the realization of the a₃⁴⁺⩵O structure for P and suggests that other oxygen-metabolizing enzymes may follow the same sort of chemistry in activating and reducing oxygen and peroxides.

An interesting strategy emerges from Fig. 2 in terms of how oxidase couples oxygen chemistry to the proton pump. The pumping steps only occur after P has been formed (8–10), which means that the enzyme first activates and reduces O₂ to fully reduced but incompletely protonated product water molecules; the enzyme completes the four-electron transfer of electrons to oxygen, and stores the free energy that results as highly oxidizing a₃⁴⁺⩵O and radical species, before engaging the pump. Recent calculations on the bond-cleavage chemistry support this idea, as the results indicate that reduction of O₂ to oxo and hydroxo with formation of a radical and a ferryl-oxo is close to thermoneutral (23). This represents a remarkably effective strategy for avoiding toxic, partially reduced oxygen species, as none occur in the reaction cycle. Moreover, by transferring the free energy that will be used to drive the pump from substate oxygen products to the protein, it appears as if oxidase has maximized the control and efficiency with which it can operate the proton-translocating apparatus.