Structure and Organization of Mitochondrial Respiratory Complexes: A New Understanding of an Old Subject

1. Complex I

Complex I (NADH-coenzyme Q reductase, E.C. 1.6.5.3) catalyzes the first step of the electron-transport chain of mitochondria and several bacteria (32). The reaction is accompanied by translocation of four protons from the matrix to the intermembrane space.

Notably, some organisms lack complex I, including the most-investigated yeast species, Saccharomyces cerevisiae (92) and some other yeasts, as S. carlsbergensis and Kluyveromyces lactis (143), and the malarial parasite Plasmodium yoelii yoelii (313).

a. Structure and mechanism

The bovine enzyme is a heteromultimer consisting of 45 subunits for a molecular mass of ∼1,000 kDa, making complex I by far the largest enzyme of the respiratory chain. Seven subunits are the products of the mitochondrial genome and correspond to hydrophobic components named ND1-ND6 and ND4L. The minimal active form of the enzyme is that found in bacteria, composed of 14 “core” subunits, all of which are homologous to their mitochondrial counterparts, whereas all other “accessory” subunits still have an undefined role. From structural and phylogenetic considerations, the enzyme is envisaged to consist of three different sectors: a dehydrogenase unit and a hydrogenase-like unit, constituting the peripheral arm protruding into the matrix, and a transporter unit deeply embedded in the membrane and involved in proton translocation (91). The dehydrogenase domain contains the NADH oxidizing site, whereas the hydrogenase domain binds and reduces CoQ; both domains contain prosthetic groups, whereas the transporter unit appears to be devoid of cofactors.

Trypanosomes have a complex I slightly larger than that of bacteria, but lacking some of the core subunits: four subunits normally encoded by the mitochondrial genome. This deficiency results into loss of proton translocation, presumably an adaptation to parasitic life (231). Besides the classic NADH-ubiquinone oxidoreductase reaction performed by all mitochondrial complex I (see later), additional enzymatic activities have been found or proposed to be associated with complex I from plants, bovines, or fungi. However, the physiological significance of these activities, displayed by unique noncore subunits like γ-carbonic anhydrase, l-galactono-1,4-lactone dehydrogenase, acyl carrier protein, or thiosulfate-cyanide sulfurtransferase, is still not fully understood (259).

Figure 1 is a general scheme illustrating the electron-transfer pathway from NADH to CoQ; major sites for reduction of artificial acceptors [ferricyanide, 2,4-dichloro-phenol-indophenol (DCIP)] are shown in the figure.

OPEN IN VIEWER

FIG. 1.

Complex I subunit topology. The L-shaped enzyme complex can be dissected into fragments Iα, Iβ, Iλ, and Iγ, whose composition allows a basic arrangement of the 45 subunits currently described for human complex I (partly redrawn from Fig. 1 in ref. 322, ©2007, with permission from Elsevier). A schematic representation of the electron pathway from NADH through FMN and the iron–sulfur clusters (N1 to N6) and hence toward the physiologic acceptor CoQ is shown, according to the scheme by Hinchliffe and Sazanov (126). Major sites for reduction of artificial acceptors [ferricyanide, 2,4-dichloro-phenol-indophenol (DCIP)] are shown as well.

Several prosthetic groups contribute to electron transfer within the enzyme: FMN is the entry point for electrons that are then transferred to a series of iron–sulfur clusters. Two clusters present different characteristics: N1a, of the type Fe₂S₂, has the lowest midpoint potential (E_m = −370 mV), whereas N2, that is of the type Fe₄S₄ and resides at the interface between the PSST and the 49-kDa subunits, has the highest midpoint potential (E_m between −150 mV and −50 mV), presenting EPR magnetic interactions with the ubisemiquinone radicals; for these reasons, it is considered to be the direct electron donor to ubiquinone. N2 iron–sulfur cluster is most likely located in the connection between the peripheral and the membrane arm. The magnetic interaction with the semiquinone radical, corresponding to a distance of ∼10 Å (194), suggests that the ubiquinone headgroup could somehow reach up into the peripheral arm, as assumed by Brandt et al. (33), who hypothesized an amphipathic “ramp” guiding ubiquinone into the catalytic site. The arrangement of iron–sulfur clusters in the hydrophilic domain of complex I from Thermus thermophilus has been determined by x-ray crystallography, showing a linear chain of all clusters except N1a and N7 (275).

Complex I is inhibited by more than 60 different families of compounds from rotenone, the prototype of this series, to a number of synthetic insecticides/acaricides (192), which have been grouped into three classes (57). It is commonly accepted that they share the same hydrophobic large pocket in the enzyme (81) (Table 2).

OPEN IN VIEWER

Table 2.

Functional Classification of Complex I Inhibitors

Inhibitor type and action	Representative inhibitors	Likely inhibitors
Type A: quinone antagonists	Rolliniastatin-2	Other annonaceous acetogenins
	Piericidin A (first site)	Erythrosin iodoacetamide
	Idebenone	Aurachin A
		Phenalamid
		Ranolazine
Type B: semiquinone antagonists	Rotenone	Other rotenoids
	Piericidin A (second site)	Ubicidins
	Piericidin B	Reduced Q-2
	Aureothin	4-Alkyl-quinazolines (SAN 547A)
	Amytal	Pyrazole acaricides
	4-alkyl-acridones	Squamocin
	4′-alkyl-MPP+ analogues	Otivarin
	Phenoxan	Quinolones
Type C: quinol antagonists	Quinol products	4′-Alkyl-MPP+ analogues
	Reduced Q-2	4′-Alkyl-pyridines
	Myxothiazol	Myxalamid
	Stigmatellin	Vacor
	TDS	Carbocyanine dyes
	2M-TIO	Acridine dyes
	Meperidine (Demerol)	DNP and other uncouplers
	NP

Data are taken from ref. 57.

The pathway of electrons is now well defined (178). The primary acceptor of electrons from NADH is FMN bound to the 51-kDa subunit; because iron–sulfur cluster N1a has a very negative potential and is situated too far from the other iron–sulfur clusters, it is not likely to reside in the main pathway of electrons. Thus, electrons would flow from FMN to N3 in the same 51-kDa subunit, and to N4 and N5 in the 75-kDa subunit, and then to N6a and N6b in the TYKY subunit and to N2 in the PSST subunit shared with the 49-kDa subunit. N2 is the direct electron donor to bound ubiquinone, and probably this step is linked to proton translocation, although the mechanism is still debated (32). Because all redox groups in the enzyme appear to be located in the hydrophilic arm or at least at the interface with the hydrophobic arm, direct coupling mechanisms appear unlikely; this implies that the driving force for proton translocation must be transduced over a considerable distance to the actual pumping process in the membrane arm via conformational coupling (32, 341). Because a mutation completely abolishing the pH dependence of cluster N2 redox potential has no effect on proton translocation, it is likely that the conformational change driving proton translocation is linked exclusively to ubiquinone reduction.

The mechanism of CoQ reduction is particularly intriguing, because more than one bound quinone species has been assigned to the enzyme; three ubisemiquinone signals are detectable in the enzyme (186). The findings in our laboratory that two different classes of inhibitors have opposite effects on oxygen reduction to superoxide during forward electron transfer (see also Section III.B), together with other observations, allowed us to draw a tentative scheme of electron transfer in complex I (78) (Fig. 2). In the bifurcated scheme shown in the figure, an iron–sulfur cluster located upstream of the N2 center might act as a “switch” for electron delivery.

OPEN IN VIEWER

FIG. 2.

Proposed two-step mechanism for electron transfer from NADH to quinone in complex I (A), in the presence of class A inhibitors (B) and in the presence of class B inhibitors (C). All experiments with complex I inhibitors were performed in bovine heart submitochondrial particles with complex III preinhibited with 1.8 μM mucidin to detect only reactions occurring in complex I (78). The role of hydrophilic (CoQ₁) and hydrophobic (DB) quinones is highlighted. CoQ₁ can react with the physiologic ubiquinone reducing site and, because of its higher water solubility, it also can react with the electron-escape site, increasing superoxide production. A mechanism of bifurcated electron transfer is depicted, in which an iron–sulfur cluster located upstream of the N2 center would act as a “switch” for electron delivery in such a way that one-electron quinone reduction to semiquinone and semiquinone reduction to quinol would be accomplished by two different electron donors. Because it is highly unlikely that quinone can reach iron–sulfur clusters other than N2 (the only center not deeply buried in the protein), the delivery of both electrons by N2 requires that the switch between the two gated states be represented by a suitable conformational change. The presence of oxidized CoQ₁₀ in the Q-pocket induces an enzyme conformation, allowing electron delivery to reduce CoQ₁₀ to semiquinone. The semiquinone formation induces a conformational change, now allowing the delivery of the second electron to the semiquinone to produce the fully reduced form. Class A inhibitors (B), not allowing access of the quinone to the active site, would block the enzyme in a conformation that does not allow quinone reduction but permits only electron delivery from N2 to oxygen. Conversely, class B inhibitors (C) would block the enzyme in a conformation allowing the first electron delivery to form the semiquinone, but the incapability for further reduction to quinol. Such conformation would not allow reaction of N2 with oxygen. (Reprinted from ref. 78, ©2009, with permission from Elsevier.)

The findings of Fato et al. (78) have to be reconciled with the linear pathway of electrons along the series of iron–sulfur clusters, as demonstrated by the crystallographic study of Sazanov (276); our interpretation is not in contrast with the existence of a linear pathway, because the two electrons delivered to CoQ for its complete reduction could be provided by the same cluster (N2) consecutively, if a suitable conformational change occurs after the first electron delivery to provide a gating mechanism for the second electron.

2. Complex III

The cytochrome bc₁ complex or complex III (ubiquinol-cytochrome c oxidoreductase, E.C. 1.10.2.2) from mitochondria of several species has been crystallized, and its structure solved to atomic resolution (139). The mechanism of the enzyme is generally well understood, although some questions remain.

The enzyme represents a confluence point for reducing equivalents from various dehydrogenases: it can catalyze the transfer of electrons from hydroxyquinones (ubiquinol, reduced CoQ) to a water-soluble c-type cytochrome, and it can, concomitantly, link this redox reaction to translocation of protons across the membrane (52).

All cytochrome bc₁ complexes contain three protein subunits with redox prosthetic groups, a di-heme cytochrome b containing a relatively high-potential b _H (or b₅₆₆) heme and a lower potential b _L (or b₅₆₂) heme, cytochrome c₁ and an iron–sulfur protein (Rieske protein) with a 2Fe-2S cluster (21). As many as seven or eight supernumerary subunits also are present in the mitochondrial enzymes. These nonredox subunits are not required for electron-transfer and proton-translocation activities of the enzyme; their possible functions include structural stability and regulation of coordinated activity of the dimeric enzyme, and docking sites for ternary complex formation with the dehydrogenase and oxidase complexes (277). Table 3 shows the subunit composition of mammalian complex III.

The mitochondrial complex III is a symmetrical, oligomeric dimer; it has been demonstrated that the iron–sulfur protein spans the dimer structure because it is anchored in one monomer, whereas its peripheral domain is located in the other monomer, where it forms part of the ubiquinol oxidation site. Evidence exists that the dimer behaves as a functional monomer, on the basis of the stoichiometry of inhibitor action on enzyme activity (51). Striking evidence exists that the peripheral domain of the Rieske protein moves back and forth between positions close to cytochrome b and cytochrome c₁ (225), facilitating electron transfer within the enzyme. Crystal structures established the location of the ubiquinol oxidation and ubiquinone reduction at topographically separated sites within each monomer and demonstrated the transmembrane disposition of the b hemes. These structural details provide a final confirmatory evidence of the proton-motive Q-cycle mechanism of the enzyme, with protons being carried across the inner mitochondrial membrane, whereas electrons from ubiquinol are transferred through the bc₁ complex (211, 232) (Fig. 3).

OPEN IN VIEWER

FIG. 3.

The Q-cycle in complex III. (A) Ubiquinol (QH₂) delivers the first electron at the outer positive site (called site o or P) of the inner membrane to the Rieske iron–sulfur protein and hence to cytochromes c₁ and c; the result is release of two protons in the intermembrane space and the formation of an unstable semiquinone anion (Q^·-) at the Qo site, which is immediately oxidized to ubiquinone (Q) by the low-potential cytochrome b₅₆₆ (b_L). The electron is then delivered to the high-potential cytochrome b₅₆₂ (b_H) at the internal negative site (site i or N), and then b_H is reoxidized by Q at the Qi site, forming another semiquinone. (B) The cycle is completed by oxidation of a second molecule of QH₂.

A detailed survey of the effects of complex III inhibitors has been the basis for establishing the Q-cycle as the molecular mechanism of electron transfer and proton translocation. Two major classes of inhibitors have been individuated, acting at two separate sites, denominated site or center i (inner) or N (negative), and site or center o (outer) or P (positive); site o inhibitors have been subdivided into two further subclasses of compounds (proximal and distal inhibitors, with reference to the block of the first or of the second electron released from ubiquinol (Table 4).

Being water insoluble, reduced CoQ₁₀ as well as other long isoprenoid chain ubiquinols cannot be used in the assay of complex III; widely used short-chain homologues are CoQ₁ and CoQ₂ and analogues such as duroquinol or decyl-ubiquinol (100); the hydroxyl derivative of decyl ubiquinone, idebenone, in its reduced form, is a good electron donor to complex III (58). The donor substrates interact with the enzyme by previously partitioning in the membrane lipids, so that their partition coefficients must be taken into consideration for determining the true kinetic constants of the enzyme (80). Weiss and Wingfield (327) studied the enzymology of complex II and of the bc₁ complex embedded in detergent micelles by using CoQ₁₀/ubiquinol₁₀ as the connecting substrate, and found that the transfer of the reduced quinone from one micelle to another was the rate-limiting step of the integrated activity.

Steady-state kinetic analysis by two-substrate titrations indicates, for complex III, a two-site ping-pong mechanism; the kinetic analysis suggests that the enzyme is not controlled by ubiquinol diffusion to the active reduction site, but may be controlled by cytochrome c diffusion to the oxidation site.

3. Complex IV

Complex IV (cytochrome c oxidase, EC. 1.9.3.1) belongs to the heme-copper oxygen reductase superfamily whose members catalyze the complete reduction of dioxygen to water and promote proton translocation across the mitochondrial or periplasmic membrane, further contributing to the difference in electrochemical potential. These enzymes have in common the same general structural fold of the catalytic subunit (subunit I), which contains a low-spin heme and a bimetallic center, composed of a high-spin heme and a copper iron (Cu_B), retained in the protein by ligation with histidine residues. Dioxygen reduction takes place at this binuclear site. Stoichiometries for the redox-driven proton pumping are variable among the different members of the superfamily, and even the same enzyme has no fixed stoichiometry in all conditions, as generally considered in the literature (13). The maximal stoichiometry of the mitochondrial oxidase is translocation of 2H⁺/2e⁻.

Heme-copper enzymes are classified according to the amino acid residues of their proton-conducting channels. The mitochondrial enzyme is a member of the type A1 family (Fig. 4), having Asp-124 (D124, amino acid numbering as in Paracoccus denitrificans, after which the D-channel proton pathway is called) close to the negative side of the membrane and besides hydrophilic amino acid residues (Asn-199, Asn-113, Asn-131, Tyr-35, Ser-134, Ser-193) ending at Glu-278, considered a key residue for the operating mechanism of the enzyme. The residues Lys-354 (K), Thr-351, Ser-291, and Tyr-280 are part of a second proton pathway (K,channel) and have also been demonstrated to play a crucial role in the catalytic cycle (187).

OPEN IN VIEWER

FIG. 4.

Proton-conducting pathways (dark grey arrows) together with the redox centers of cytochrome c oxidase. The two key subunits, COXI and COXII, are depicted in the membrane together with docked cytochrome c; the D pathway begins from Asp124 and leads to Glu278, from where, depending on the orientation of water molecules (not indicated), protons can be directed toward the P-side of the membrane or to the catalytic center of the enzyme (heme a₃/CuB). The K pathway leads to the conservative Lys354 and on farther to the binuclear center. The electron-transfer path is indicated in light grey. (Partly redrawn using data from Fig. 1 in ref. 13, ©2007 National Academy of Sciences, U.S.A.)

Electron transfer through complex IV occurs from ferrocytochrome c to the Cu_A center (which acts as a single-electron receptor), then to heme a onto the heme a₃ /Cu_B center, and finally to oxygen bound to heme a₃ . H⁺/e⁻ cooperative linkage at Fe(a₃ )/Cu_B is envisaged to be involved in proton-pump mechanisms confined to the binuclear center (330). Models have also been proposed that involve a role in proton pumping of cooperative H⁺/e⁻ linkage at heme a/Cu_A (205, 243).

Cytochrome c oxidase is potently inhibited by cyanide, azide, and carbon monoxide, which bind at the oxygen-binding site (48). The molecular mechanism of inhibition by NO is more complex: (a) a major component is competitive with oxygen and, presumably, requires the presence of electrons in the binuclear center, but (b) an additional interaction occurs at the oxidized enzyme by binding to Cu_B ²⁺ (103). In the latter case, the enzyme becomes reduced (Cu_B ⁺), and NO is oxidized to nitrite (NO₂ ⁻). Kinetically, this results in uncompetitive inhibition with respect to the oxygen kinetics. Recent kinetic models have successfully incorporated both modes of inhibition (49); therefore, depending on the balance of the two effects, the activity of cytochrome oxidase toward NO may result in the strong inhibition of cell respiration or in the removal of NO from the cell.

As extensively reviewed in a recent article by Belevich and Verkhovsky (14), a real breakthrough in the understanding of the function of cytochrome oxidase was achieved when the first crystallographic structures of the enzyme were resolved, in both oxidized and reduced states, but solving the structures of all intermediates in the catalytic cycle is still a difficult task, as those intermediates are quite unstable. At present, five x-ray crystallographic structures of heme-copper oxygen reductases have been determined, with resolution up to 1.8 Å (cf. ref. 27 for a detailed list of references), shedding light also on the structure of additional protein subunits that can compose the functional unit of cytochrome oxidase, besides the catalytic subunit I. The mammalian cytochrome oxidase has a molecular mass of ∼200 kDa and consists of 13 subunits originating both from nuclear and mitochondrial DNA, whereas the bacterial enzyme is simpler in structure because it contains three core subunits, whose sequence homology highly corresponds to the three subunits that, in the majority of eukaryotes, are encoded in the mitochondrial DNA (COXI, COXII, COXIII); one extra subunit is present in P. denitrificans only (Table 5).

Because none of the nuclear-encoded subunits is associated with the active site, it was formerly assumed that they were not important in the functional mechanism of the enzyme. However, it is now demonstrated that some of those additional subunits are involved in the stabilization of a dimer state of the oxidase (88) and might participate in the interaction of complex IV with its partner complexes within a respiratory supercomplex, or they are suggested to regulate complex IV activity either by chemical modification like glycosylation and phosphorylation (120) or by binding effectors, such as ADP/ATP or protein kinase A (18). Lee and colleagues (169) proposed that the physiological meaning of such feedback regulation of the respiratory chain by its end product, ATP, is in keeping with the low membrane potential and, consequently, reduces ROS production of mitochondria. The allosteric ATP-inhibition is lost when the enzyme is dephosphorylated.

1. Complex II

Besides its functional role as succinate dehydrogenase in the Krebs cycle, the enzyme (EC 1.3.5.1) is involved in aerobic metabolism by the respiratory chain because it can couple the two-electron oxidation of succinate to fumarate with the electron transfer directly to the quinone pool; hence complex II is more precisely termed succinate:quinone oxidoreductase (SQR) (164).

Mammalian complex II is part of a class of ubiquinone-reducing enzymes containing a single b heme and anchored to the inner mitochondrial membrane by two hydrophobic subunits, SdhC (14.2 kDa) and SdhD (12.8 kDa). Although the membrane anchor domain shows low sequence identity (less than 20%) and varies in composition between organisms, the primary sequence of the soluble domain of complex II is highly conserved (30 to 50% sequence identity) and consists of a flavoprotein subunit (SdhA, Fp, 64 to 79 kDa) containing covalently linked FAD and an iron–sulfur protein subunit (SdhB, Ip, 27 to 31 kDa), both located on the matrix side of the membrane (for a review, see ref. 40). One feature of the complex II structure is a linear electron-transport chain that extends from the flavin and Fe-S redox cofactors in the extrinsic domain to the quinone and heme b cofactors in the membrane domain (40). The interaction of quinones with complex II is an area located at the fringe of a hydrophobic pocket comprising residues from subunits SdhB, SdhC, and SdhD (132). Besides two electrons from the oxidation of succinate, the full reduction of the quinone in SQR would require two protons to be donated by the protein environment of the Q-site followed by re-protonation of the site after catalytic turnover. In the native structure of SQR from Escherichia coli, Horsefield et al. (132) identified a proton-uptake pathway suitable for such purpose that crosses the membrane anchor arriving at the Q-site. The high homology between the SQR Q-sites in E. coli and in mammalians, based on absolute conservation of amino acids in contact with ubiquinone, suggests the same mechanism for electron transfer to ubiquinone, thus making E. coli an excellent model system for mitochondrial complex II research (301). This is of particular interest in humans, because mutations in complex II result in various physiological disorders (269). The structure of complex II is shown in Fig. 5.

OPEN IN VIEWER

FIG. 5.

Overall structure of SQR monomer of Escherichia coli. SdhA, SdhB, SdhC, and SdhD subunits are shown in orange, green, red, and blue, respectively. FAD, oxaloacetate (OAA), heme b, and ubiquinone (UQ) are shown in light green, light blue, yellow, and pink, respectively. Fe and S atoms of Fe–S clusters are painted red and grey, respectively. On the right, schematic representation of the chain-like arrangement of the redox cofactors. Picture taken from www.thehorsefields.com/RobHorsefieldThesis.htm, ©Rob Horsefield 2005, with permission from the author. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article at www.liebertonline.com/ars).

Although the natural acceptor of complex II is hydrophobic CoQ in the membrane, the enzyme is not usually assayed with short-chain quinones as acceptor substrates, but by using 2,6-dichlorophenol-indophenol, which has a higher midpoint redox potential and accepts electrons directly from endogenous CoQ (87); alternatively, especially in studies from muscle biopsies in the case of mitochondrial genetics diseases, the activity of complex II+III is measured cumulatively as succinate cytochrome c reductase (e.g., in ref. 9). “Soluble” succinate dehydrogenase is assayed by using as acceptor phenazine methosulfate, which accepts electrons upstream from the physiologic site (87). Complex II is typically inhibited by thenoyl trifluoroacetone and carboxin (5,6-dihydro-2-methyl-1,4-oxathin-3-carboxanilide) that bind to the same binding site situated in the SdhB iron–sulfur protein subunit (201).

The competitive inhibition by malonate reflects a physiologic inhibition by oxaloacetate (4): this is the reason that succinate dehydrogenase and related activities are usually low in isolated mitochondria, and incubation with succinate is needed to remove tightly bound oxaloacetate. In mitochondria from rat brain, but not heart or liver, the oxaloacetate inhibition is released by allowing mitochondria to oxidize pyruvate plus glutamate and malate, probably by removing the inhibitor through metabolic reactions (241).

Recent research has shown that the mitochondrial complex II plays an important role in the anerobic energy metabolism of parasites: often, the parasite uses aerobic metabolism during the free-living stage outside the host, but adapts to hypoxic environments and uses systems other than oxidative phosphorylation for ATP synthesis when inhabiting host mammals. Many adult parasites perform fumarate respiration by expressing a stage-specific isoform of complex II that catalyzes the reduction of fumarate (quinol-fumarate reductase, QFR), which is the reverse of the reaction catalyzed by SQR (152).

2. Mitochondrial glycerol-3-phosphate dehydrogenase

Glycerol-3-phosphate dehydrogenase (mtGPDH, EC 1.1.99.5) shuttles reducing equivalents from cytosol through the respiratory chain to molecular oxygen. This metabolic shuttle was first discovered in insect flight muscle (337) and in brown adipose tissue (133), where the enzyme has highest activity. In pancreatic islet β cells, many studies support the significant participation of the shuttle in the events signaling the release of insulin in response to increased glucose (196). The mtGPDH is a very hydrophobic protein of the inner membrane; its catalytic center is accessible from the outer surface of the inner membrane. Besides containing FAD as a prosthetic group, the presence of an iron–sulfur cluster has been suggested on the basis of ascorbic acid stimulation and inhibition by di-iron metallo-enzymes inhibitors (144); however, the putative center has not been characterized. The activity of mtGPDH is inhibited by acyl-CoA esters and free fatty acids (257) and induced by hormones (299). The enzyme is calcium dependent: because its glycerol phosphate-binding site faces the outer surface of the inner membrane, it is exposed to fluctuations in cytoplasmic calcium concentrations (226); the binding site is part of the polypeptide chain of the enzyme, contrary to other calcium-sensitive mitochondrial dehydrogenases in which calcium appears to bind separate subunits (226). A specific activation by short-chain CoQ homologues and by the CoQ analogue idebenone was related to the release of the inhibitory effect of free fatty acids. This competition suggested that the inhibitory effect of free fatty acids is exerted by occupying the CoQ-reducing site of the enzyme, thus preventing transfer of reducing equivalents to the CoQ pool. The mechanism of CoQ reduction by mtGPDH is not well understood; progress in this direction has derived from studies of ROS production by the enzyme (section III.E). Glycerol-3-phosphate dehydrogenase appears to interact directly with the CoQ pool, therefore not forming a supramolecular aggregate with complex III, as suggested by the convex hyperbolic curve of inhibition of glycerol phosphate cytochrome c reductase by myxothiazol, indicating the existence of a mobile intermediate between mtGPDH and complex III (258).

3. ETF-ubiquinone oxidoreductase

The electron-transfer flavoprotein (ETF)-ubiquinone oxidoreductase (EC 1.5.5.1) is a globular protein located on the matrix surface of the inner mitochondrial membrane. The enzyme can accept reducing equivalents from a variety of dehydrogenases (12), including those involved in fatty acid oxidation, in amino acid oxidation, and in choline catabolism (dimethylglycine dehydrogenase and sarcosine dehydrogenase), and is oxidized by ubiquinone.

Crystal structures of the enzyme (338) indicate that the molecule forms a single structural domain where three closely packed functional regions bind FAD, the 4Fe4S cluster, and ubiquinone. The ubiquinone molecule penetrates deep into its binding pocket, which consists mainly of hydrophobic residues. Only five units of the 10 isoprenes in the flexible tail of CoQ could be seen in the structure of the ubiquinone-containing protein (338). Studies of site-directed mutagenesis in Rhodobacter sphaeroides indicated that FAD is involved in electron transfer to ubiquinone but not in electron transfer from ETF, demonstrating that the iron–sulfur cluster is the immediate acceptor from ETF.

4. Choline dehydrogenase

Choline dehydrogenase (EC 1.1.99.1) catalyzes the oxidation of choline to betaine aldehyde. The enzyme is localized at the matrix side of the inner mitochondrial membrane; because its oxidation through the respiratory chain was shown to yield a P/O ratio approaching 2, it was suggested that it feeds electrons to CoQ at a similar position to that of respiratory complex II (134).

The enzyme contains FAD and an iron–sulfur cluster. The sequence predicted by computer analysis from the rat liver gene sequence in the NCBI database was confirmed by cloning a full-length cDNA; expression of the recombinant gene in S. cerevisiae led to enrichment of the active target protein in the inner mitochondrial membrane (134).

5. Dihydroorotate dehydrogenase

Dihydroorotate dehydrogenase (DHODH, EC. 1.3.3.1) is an iron-containing 43-kDa flavoprotein (FMN) that catalyzes the oxidation of dihydroorotate to orotate, the fourth step in de novo pyrimidine biosynthesis (76).

Biochemical and microscopic studies (128, 209) showed that the mammalian DHODH and that isolated from Neurospora crassa (class 2 enzymes) are integral membrane proteins localized in the inner mitochondrial membrane (325) with the active site facing the intermembrane space. The enzyme is functionally linked to the electron-transport system of the respiratory chain because it uses ubiquinone as co-substrate electron acceptor (209); thus, it is classified as a dihydroorotate:ubiquinone oxidoreductase. According to the catalytic properties described by Hines and Johnston (127), it seems reasonable that FMN functions as the proximal electron acceptor, experiencing two-electron reduction concomitant with dihydroorotate oxidation. Reduced flavin would then become reoxidized by passing electrons, perhaps one at a time, to a putative iron–sulfur cluster that, in turn, would be exposed to the quinone. Crystallographic studies, however, failed to detect iron–sulfur clusters in the enzyme (325). In situ, the reduced ubiquinone would be expected to equilibrate with the membrane CoQ-pool and to be reoxidized by complex III. By contrast, the rat liver DHODH lacks flavin, contains iron and zinc as the two apparent redox-active cofactors, and, like the cytosolic enzymes isolated from parasitic protozoa (class 1 enzymes) (245), delivers electrons directly to molecular oxygen (89).

The high-resolution crystal structure of human DHODH (189) shows a small domain that forms the opening of a tunnel leading to the bound FMN and that provides access to ubiquinone, whereas it is unlikely that orotate may enter via the same tunnel. Insight into the structure of the enzyme has been useful to design drugs active against protozoan parasites like Plasmodium falciparum and also in human diseases (7).

The role of quinone reductases in mammalian metabolism is depicted schematically in Fig. 6.

OPEN IN VIEWER

FIG. 6.

Schematic picture of the metabolic role of mitochondrial inner membrane ubiquinone reductases. GPDH, glycerol-3-phosphate dehydrogenase, is involved in reoxidation of glycolytic NADH through the glycerol-3-phosphate shuttle; DHA-P, dioxyacetone phosphate. DHODH, dihydroorotate dehydrogenase, catalyzes the formation of orotate in pyrimidine nucleotide biosynthesis. ETFDH, electron-transfer flavoprotein dehydrogenase; the enzyme can accept reducing equivalents via ETF from a variety of dehydrogenases, including those involved in fatty acid and amino acid oxidation and in choline catabolism (dimethylglycine dehydrogenase and sarcosine dehydrogenase). CHDH, choline dehydrogenase, oxidizes choline to betaine aldehyde, which acts as a methyl-group donor. Most of these oxidations taking place in the matrix yield acetyl CoA, which may then be further oxidized in the tricarboxylic acid cycle. Hexagons, CoQ molecules in the bilayer.

6. Alternative NADH dehydrogenases

Alternative NADH dehydrogenases (NDs) designate a family of proteins located in the inner membrane of eukaryotic mitochondria, which catalyze oxidation of NAD(P)H from either the cytosol (external enzymes) or the mitochondrial matrix (internal enzymes) and enable quinone reduction. The greatest functional difference from complex I is that their oxidoreductase activity is rotenone insensitive and is not coupled to proton pumping.

The number and specificity of alternative NADH dehydrogenases vary considerably when comparing different organisms: none was described in humans, whereas plants may have up to four proteins (two in each side of the membrane, but they were not yet conclusively identified), suggesting that they may have organism-specific roles.

Alternative dehydrogenases are present in bacteria as well as in the mitochondria of fungi. In Neurospora crassa mitochondria, the presence of both internal and external rotenone-insensitive alternative NADH dehydrogenases has been reported since the early 1970s (38). In the yeast S. cerevisiae, which lacks complex I, an internal and two external enzymes have been quite well characterized: NDI1, NDE1, and NDE2, respectively (234). Another yeast, Yarrowia lipolytica, contains only one external enzyme in the inner mitochondrial membrane in addition to complex I (147).

Alternative NADH dehydrogenases are encoded by a single nuclear gene and have a mature peptide molecular mass of 50 to 60 kDa. The only prosthetic group is FAD, by contrast with the FMN and multiple FeS centers of complex I. Gene cloning has established that NDA and NDB in potato mitochondria are markedly similar to the yeast NDI1, with sequence identity of ∼30–40% (256). Both NDA and NDB have NADH- and FAD-binding motifs, whereas neither has any indication of membrane-spanning α-helices. Both proteins bind to the inner mitochondrial membrane, but their different targeting leads to locations on opposite sides, as shown in Fig. 7.

OPEN IN VIEWER

FIG. 7.

Organization of alternative NADH dehydrogenases. The internal (NDI1) and external (NDE1) NADH dehydrogenases of yeast are depicted as similar proteins with opposite topologies in the inner mitochondrial membrane. The experimental evidence relates to potato mitochondria, as discussed in the text. (Reprinted from ref. 143, ©2001, with permission from Elsevier.)

The cellular role and need for alternative NADH dehydrogenases remains mostly unclear; it was suggested that they provide the organisms with plasticity to adapt to different environmental conditions. A pivotal question to be elucidated is how these enzymes interact with other mitochondrial dehydrogenases. The alternative NDs and the proton-pumping complex I have overlapping roles in oxidoreductase reactions and, in some cases, it was demonstrated that alternative NDs are not essential proteins, given the fact that mutants are viable (38). Evidence also suggests that alternative NADH dehydrogenases can complement complex I defects in different situations. Disruption of complex I genes in Paracoccus denitrificans was possible only after introduction in the organism of the NDH-2 gene of E. coli (86). Likewise, the segregation of complex I mutants in Y. lipolytica required the previous targeting of its external single alternative NADH dehydrogenase to the matrix face of the inner mitochondrial membrane (147). Moreover, the complementation of complex I defects in mammalian cells with the NDI1 gene of S. cerevisiae is quite amazing and points to a possible strategy for gene therapy in human mitochondrial diseases (333, 247).

In the yeast S. cerevisiae, NDE1 and NDE2 were described to be associated in a membrane-bound supramolecular complex with both other known intermembrane space- and matrix-facing dehydrogenases (glycerol-3-phosphate dehydrogenase, d- and l-lactate-dehydrogenases, NDIp) and with four tricarboxylic acid cycle enzymes (malate dehydrogenase, citrate synthase, succinate dehydrogenase and fumarate hydratase), suggesting that such dynamic interactions may have a function in the regulation of mitochondrial metabolism (35, 106).

Because alternative NADH dehydrogenases do not pump protons, they may be useful to keep reducing equivalents at physiologic levels and to avoid the production of reactive oxygen species associated with complex I. Moreover, data on the characterization of the expression and activity regulation of these enzymes are emerging: alternative components respond to factors ranging from oxidative stress to the stage of fungal development. For instance, their direct involvement in oxidative stress in yeast (55) or in development and light responses in plants (303) was described. Their variability among species is a sign that they accomplish specific requirements of the different organisms.

7. Malate-quinone oxidoreductase

Bacteria possess a malate dehydrogenase (EC 1.1.99.16) catalyzing the oxidation of malate to oxaloacetate by the respiratory chain without using NAD as the intermediate acceptor; the enzyme is a membrane-associated protein containing FAD as a prosthetic group and donates electrons directly to coenzyme Q (214).

The enzyme, however, was found also in mitochondria of some eukaryotes. Genes encoding for a malate quinone oxidoreductase have been detected in the genomes of P. falciparum and P. yoelii; moreover, malate was shown to stimulate rotenone-insensitive respiration and ADP phosphorylation in the parasites Toxoplasma gondii and Plasmodium yoelii (313).

8. Alternative quinol oxidases

Although complex III is the only ubiquinol-oxidizing enzyme in mammalian mitochondria, most plants and some yeasts and fungi posses a cyanide- and antimycin-insensitive alternative oxidase (AOX) that catalyzes the aerobic oxidation of ubiquinol in addition to the cytochrome pathway (155). The enzyme is nonprotonmotive and its activity does not contribute to the conservation of energy that can therefore be dissipated as heat (217). However, the ubiquitous presence of AOX in plants, including nonthermogenic species, suggested a more general physiologic role of the enzyme as an overflow mechanism. It has been predicted that AOX allows Krebs-cycle turnover when the energy state of the cell is high and that it protects against oxidative stress. In transgenic tobacco cells, the antisense suppression of AOX resulted in cells with a significantly higher level of ROS compared with wild-type cells, whereas the overexpression of AOX resulted in cells with lower ROS abundance (202). Conversely, in a long-lived respiration mutant of the fungus Podospora anserina overexpression of AOX enhanced ROS production (190); in an analysis of supercomplex arrangement of the respiratory chain in wild-type and long-lived mutants of P. anserina overexpressing AOX, Krause et al. (159) found two different types of supramolecular organization of complex I and complex III and concluded that it is the supramolecular arrangement of the respiratory chain to dictate the overall properties of the respiratory system (see section VI.A.1).

Despite the difficulty of purifying the enzyme to homogeneity in a stable, active form, recent models considered AOX as a homodimeric interfacial protein, the functional unit being a single polypeptide of around 32 kDa, peripherically associated with the matrix side of the inner mitochondrial membrane (289). Few studies addressed the problem of possible protein–protein-specific interactions of AOX with respiratory chain complexes or supercomplexes (cf. section V.A).

The structure of the active site of the oxidase comprises a nonheme di-iron center that is reduced by two electrons delivered from ubiquinol. Moore and Albury (216) proposed a model for the ubiquinol-binding site in AOX, which identifies a hydrophobic pocket, between helices II and III, leading from a membrane-binding domain to the catalytic domain; this crevice could act as a channel through which the substrate gains entry to the active site (Fig. 8).

OPEN IN VIEWER

FIG. 8.

Diagrammatic representation of the structure of the alternative oxidase in plant mitochondria. Numbers 1 through 4 denote the four diiron-binding helices of the active site. Grey circles, the iron atoms. The C and N indicate the C- and N-terminal ends of the protein, respectively. The AOX is present in the inner membrane as a dimer in which the two subunits can be covalently linked through a disulfide bridge between highly conserved cysteine residues in the N-terminus region of the enzyme (white) that has not been modeled to date. The reduced (SH) non-covalently linked dimer is the active form, and this form appears to be the predominant one in vivo (318). (Created using data from refs. 289 and 312.)

A significant engagement of the alternative pathway is not apparent until the reduction level of the CoQ pool reaches 40%; to explain this kinetic characteristic, Siedow and Moore (288) proposed a detailed model, based on CoQ pool behavior, that predicts the changing affinity for oxygen with changes in CoQ-pool reduction. Various regulatory phenomena that affect the amount and activity of the alternative oxidase have been reported in the literature (288). For example, induction of AOX in N. crassa occurs only when mutations or chemicals inhibit the cytochrome pathway (60), and no active AOX is present under normal growth conditions, whereas the conventional and alternative respiratory pathway can operate simultaneously in other fungi (121).

Moreover, not only the amount of alternative oxidase, but also its kinetic characteristics vary with tissue conditions; a clear example is afforded by mitochondria isolated from young and mature spadices of Arum maculatum, in which the plot of AOX activity rate versus Q-pool reduction changes from nonlinear to nearly linear as a function of tissue-growth stage. Biochemical regulation is known to occur at the highly conserved cysteine residue CysI located in the structurally undefined N-terminus (Fig. 8). When the CysI residues of the AOX dimer interact with α-keto acids, perhaps forming a thiohemiacetal, the enzyme becomes activated through a charge-induced conformational change. When this conformational change is prevented, either by oxidation of CysI residues in the native homodimer to form an intermolecular disulfide bond or by substitution of CysI with a hydrophobic amino acid residue, an inactive enzyme results. These regulatory features allow the plant AOX activity to be influenced by intermediates of carbohydrate metabolism and cellular redox state, consistent with its hypothesized functions listed earlier (312).

AOX expression is well tolerated in cultured mammalian cells; cotransforming rho-0 cells with the NADH dehydrogenase of S. cerevisiae and the alternative oxidase of Emericella nidulans, NDI1 and AOX recovered full NADH oxidation without proton pumping (247). Furthermore, the ectopic expression of the alternative oxidase from Ciona intestinalis was able to complement cytochrome oxidase defects in Drosophila (82). These studies highlight the potential use of AOX for gene therapy of respiratory chain deficiencies.

9. Sulfite oxidase

Sulfite oxidase (EC 1.8.3.1) is the only enzyme, besides complex III, to be able to deliver electrons at the level of cytochrome c. The enzyme catalyzes the reduction of cytochrome c by sulfite and is involved in liver in the final steps of degradation of the sulfur-containing amino acids cysteine and methionine and in detoxification of sulfite from environmental sources (142).

Sulfite oxidase is localized in the intermembrane space; it is a homodimer composed of a large molybdenum domain linked to a small heme b domain. The model of ionic interaction between cytochrome c and its reaction partners predicts a cluster of specifically oriented carboxyl groups; such a cluster has indeed been found for other partners but not for sulfite oxidase (263).

A schematic drawing of the respiratory chain complexes and their relation with the inner mitochondrial membrane is represented in Fig. 9.

OPEN IN VIEWER

FIG. 9.

A schematic drawing of the respiratory chain depicting the protein complexes and their substrates in relation with the inner mitochondrial membrane. I, NADH:ubiquinone oxidoreductase; II, succinate:ubiquinone oxidoreductase; III, ubiquinol:cytochrome c oxidoreductase; IV cytochrome oxidase; NDi and NDe, internal and external alternative NAD(P)H dehydrogenases; AOX, alternative oxidase; αGP, glycerol-3-phosphate; ETF, electron-transfer flavoprotein; DHO, dihydroorotate, CoQ, Coenzyme Q; C, cytochrome c. See text for details. (Reprinted from ref. 183, ©2009, with permission from Elsevier.)

1. Coenzyme Q

The natural coenzyme Q (CoQ, ubiquinone) is 2,3-dimethoxy-5-methyl-6-polyprenyl-1,4-benzoquinone, in which the polyprenylated side chain is six to 10 units long, depending on the species. Within mammals, only CoQ₉ and CoQ₁₀ are found, with CoQ₉ distributed only among rodents.

Mitochondria from very few eukaryotes (Tetrahymena, Euglena) have CoQ₈, whereas this homologue is present in many bacteria including E. coli (255); plant mitochondria may have either CoQ₉ or CoQ₁₀. Saccharomyces cerevisiae, among other peculiarities, has CoQ₆ as the only ubiquinone species.

Parasite mitochondria may contain redox-active quinones not present in mitochondria from other animals. The reduction of fumarate to succinate, representing an adaptation to anaerobic conditions, is the opposite reaction to succinate oxidation catalyzed by complex II; prokaryotes contain two distinct enzymes and two different quinones, menaquinone and ubiquinone, for fumarate reductase and succinate CoQ reductase. Likewise, mitochondria from parasitic helminths and some marine organisms adapted to low oxygen tension also use two different quinones, rhodoquinone (in which an amino group substitutes the methoxy group in the 3-position) and ubiquinone, for fumarate reduction and succinate oxidation, respectively (315). In the widely investigated nematode Caenorhabditis elegans, not a parasite, in addition to CoQ₈ taken up from the diet consisting of E. coli bacteria, contains as a major species both ubiquinone-9 and rhodoquinone-9 (306). The organic structural specificity of CoQ homologues and analogues was investigated in beef heart mitochondria after pentane extraction and reconstitution. As widely discussed by Lenaz (175), the homologue specificity based on the number of isoprenoid units in the 6-position is critical for the reduction of the quinone ring in the active pocket in complex I.

Because of its extreme hydrophobicity, natural CoQ can be present in three physical states only: forming micellar aggregates, dissolved in lipid bilayers, and bound to proteins. The former state is very important working with CoQ in cell-free systems (77); however, in the living cell, CoQ is distributed among the other two states.

The extent to which CoQ is bound to mitochondrial proteins is an important parameter in relation to its function. If we consider bound CoQ in a 1:1 stoichiometry with the complexes interacting with the quinone (C_I, C_II, C_III), in beef heart mitochondria, we come up to ∼0.5 nmol/mg protein, that would increase to ∼0.8 nmol, assuming more than one site to be fully occupied in complex I and complex III. Because the total CoQ content is higher than 3 nmol/mg (37, 80), we must assume that most CoQ (more than 75%) is free in the bilayer. A direct study (166) of the amount of CoQ bound to mitochondrial proteins in five different mammalian species showed that the protein-bound aliquot ranges between 10 and 32% of total CoQ.

It has been assumed for long time that the shape of the CoQ molecule is linear, with some possibility of rotation allowed for the long isoprenoid tail. Bending of the molecule is required in a model proposed by us (174), on the basis of previous evidence and of theoretic considerations, and confirmed by linear dichroism studies (274) of the location of CoQ₁₀ in the hydrophobic midplane of the lipid bilayer, with the polar head oscillating about the third isoprene unit between the midplane (wholly linear shape) and the polar heads of the phospholipids (maximal bending of 90 degrees).

Contrary to these predictions, a computer-simulation study of the molecular dynamics of CoQ homologues in the vacuum starting from different initial configurations showed that the conformation with the lowest energy level is a folded one, in which the polar head is in tight contact with the last isoprenoid unit of the hydrophobic tail (61). Within the series of homologues, the cut-off for the folded conformation is four isoprenoid units.

Important implications of a folded structure exist. First, the similar size of short and long homologues would explain the similar high rates of lateral diffusion for all quinone homologues (61, 77). In addition, protein binding during electron transfer may require unfolding, contributing to the high activation energy and low collision efficiency observed for electron transfer (e.g., 80).

The cyclohexane/water partition coefficients of different quinones are good parameters of their hydrophobicities and are known from the literature (260). The membrane/water partition coefficients of CoQ₁ and pentyl-ubiquinone (PB), determined by fluorescence quenching, agree with the cyclohexane/water corresponding values, but more hydrophobic quinones are underevaluated because their partition from water to the membrane competes with their micellization in water (77). An additional consequence of the high hydrophobicity of ubiquinones, related to their partition coefficients, is their extent of solubility in monomeric state (77); only quinones with very short chains (as CoQ₁ or PB) are monomeric in the concentration ranges used in complex I assays, whereas CoQ₂ and decyl-ubiquinone (DB) form micelles at or below micromolar concentrations in the assay medium. If the micelle-to-monomer transition is rate limiting with respect to the enzymatic kinetic steps, then any rate determination would become meaningless.

Water insolubility is a particularly serious phenomenon for oxidized quinones, as in complex I activity determination, because the hydroquinone forms used in complex III activity determination are significantly less hydrophobic (58).

The lateral diffusion of quinones in lipid bilayers has received particular attention in relation to their role in the electron-transfer processes in the mitochondrial respiratory chain; according to the “random collision model” of the electron transfer proposed by Hackenbrock et al. (114), all components of the mitochondrial respiratory chain are randomly distributed in the plane of the membrane and undergo independent lateral diffusion. The mobility of the smaller components, such as coenzyme Q (CoQ) and cytochrome c, is faster than that of the macromolecular complexes and assures electron transfer by random collisions with the latter. In addition, Hackenbrock and co-workers (114) suggested that CoQ diffusion in the mitochondrial membrane is the rate-limiting step in the whole electron-transfer process.

We discuss in the following sections that the random collision model is, at least in part, invalidated by the findings of direct electron channeling between complexes I and III. For the fraction of CoQ that is mobile in the mitochondrial inner membrane and is required for electron transfer between other complexes, it is likely that high diffusion rates make CoQ diffusion not rate limiting for electron transfer, as amply discussed in previous publications (174, 179).

2. Cytochrome c

Cytochrome c is a water-soluble ∼12-kDa heme-containing protein, encoded by the nuclear genome, that first forms as apocytochrome c in the cytosol. The basic assumption that cytochrome c binds to the mitochondrial inner membrane through electrostatic attraction to the phospholipid head groups has been challenged by experimental findings that demonstrate the presence of hydrophobic interactions between cytochrome c and phospholipid acyl chains that extend outward from the lipid bilayer (146). A fraction of total cytochrome c (10%) persists as membrane-bound molecules, even after treatment of mitochondria with digitonin (a mild nonionic detergent) (50), thus supporting the idea of a spatial and functional repartition of cytochrome c in the IMS between (a) a soluble, loosely bound pool that is sensitive to electrostatic alterations, such as ionic strength and surface charge density; and (b) a pool that binds more strongly to the inner membrane and, possibly, is in closer contact with the complexes of the respiratory chain. It is well known that cardiolipin is functionally relevant for the energy-transduction process. It may increase the surface concentration of cytochrome c close to the respiratory complexes so that the binding of cytochrome c to cytochrome oxidase may be facilitated (1); moreover, it modulates the catalytic activity of the major proteins of the mitochondrial OXPHOS apparatus (68, 93, 104, 264). In addition, cardiolipin was found to be required for the organization of the respiratory chain into supramolecular assemblies (340). A mechanistic model for the lipid anchorage of cytochrome c to cardiolipin-containing membranes was proposed by Kalanxhi and Wallace (146), who identified a crevice in the protein structure as a route of entry for one pivoting acyl chain of cardiolipin. The same authors indicated the electrostatic association of cytochrome c with the membrane surface as a necessary prerequisite step (146).

In the late 1980s, the modes and rates of cytochrome c diffusion were extensively investigated in both purified inner membranes and intact mitochondria, showing that the highest rate of diffusion is measured at physiologic ionic strength (100 to 150 mM), where the diffusion mode is three-dimensional and cytochrome c has the lowest affinity (concentration near the surface) for the inner membrane while it mediates the highest rate of electron transport through maximum collision efficiency with its redox partners, complex III and complex IV (110). Although largely debated, the existence of different physical pools of cytochrome c is consistent with the demonstration of a physical sub-compartmentalization of the mitochondrial interior and supports the notion of a “molecular reservoir” that could influence the modalities of respiratory-chain substrate use at different energy states, as suggested by Benard et al. (17).

1. Phosphorylation of complex I

Early evidences on the phosphorylation of complex I subunits showed cAMP-dependent phosphorylation of 18- and 42-kDa proteins associated with complex I from bovine heart mitochondria. Newly developed procedures based on BN-PAGE, and complemented by in-gel digestion, phosphopeptide enrichment by titanium dioxide (TiO₂), and phosphopeptide-directed triple MS analysis, besides confirming serine phosphorylation of the 42-kDa ESSS and B14.5a subunits, also revealed threonine phosphorylation of the B14.5b (human gene NDUFC2) and B16.6 (GRIM-19) subunits of bovine complex I (43, 238). The mass spectrometric analyses detected both the phosphorylated and nonphosphorylated peptides from these subunits, indicating that they may undergo a dynamic condition of phosphorylation/dephosphorylation.

Investigation of mammalian and human cell cultures has provided evidence that modulation of subunit phosphorylation by intramitochondrial protein kinase A and phosphoprotein phosphatase contributes to the stability of complex I and regulates its functional activity (59). Moreover, activation of the cAMP-dependent protein kinase prevents the formation of oxygen free radicals in mitochondria by affecting the ROS-generating capacity of complex I (244, 254).

Direct evidence of the potential effects of phosphorylation of the MWFE and ESSS subunits on complex I activity and assembly was recently reported by Yadava et al. (332) by mutational analysis of the phosphorylation sites in two Chinese hamster respiratory-deficient cell lines showing, respectively, null mutations in the NDUFA1 and NDUFB11 genes encoding the MWFE and ESSS subunits, respectively: complementation with mutant cDNAs appeared to cause low levels of mature protein or a complete failure for complex I to assemble. These observations strongly support the rationale that electron transport and oxidative phosphorylation might be regulated by a mechanism involving the covalent modification by phosphorylation of the respiratory complexes and clearly indicate that if phosphorylation occurs in vivo, the effects on complex I activity are significant.

2. Phosphorylation of complex IV

Lee et al. (170) examined the cAMP-dependent phosphorylation of mitochondrial complex IV isolated from fresh bovine liver and heart in the presence of theophylline, a phosphodiesterase inhibitor that induces high cellular cAMP levels. Under the conditions applied, they were able to identify Tyr-304 of subunit I (COX-I) as the target site of phosphorylation and to demonstrate that such phosphorylation leads to strong decrease of V_max in the isolated enzyme while decreasing the K _m for cytochrome c. In particular, the phosphorylated enzyme shows sigmoidal behavior such that up to 12 μM cytochrome c (i.e., in the range of physiologic content for heart muscle mitochondria) the activity is less than 20%, compared with saturating substrate concentrations, both in normal and in ADP-stimulated preparations. The same authors (334) indicated that cytochrome c is also targeted for phosphorylation in vivo and that it produces enhanced sigmoidal kinetics with cytochrome c oxidase. Other treatments that would increase phosphorylation (e.g., elevation of cAMP levels by glucagon addition or by forskolin activation of adenylyl cyclase) were tested in a cell-culture system and found to have the same effect on cytochrome oxidase kinetics as predicted by the theophylline results: Complex IV is switched off when it is tyrosine-phosphorylated in the catalytic subunit I.

Phosphorylation at COX-I appears to be rapidly lost during storage at −80°C or during freezing and thawing of the bovine heart, or both, and is obtained only under certain isolation conditions of the enzyme, including the use of protein phosphatase inhibitors in the medium, whereas more-stable phosphorylation sites are detectable at tyrosine, serine, and threonine in subunits II and III, as well as at specific amino acid residues in subunits IV, Vab, VIabc, VIIabc, and VIII (120). However, in the latter cases, the identified phosphorylated amino acids are located at the matrix side of complex IV and are apparently not related to the allosteric kinetics of the enzyme in the presence of ADP and ATP, which appears to involve phosphorylation of subunit I. Their function could be to change the binding affinity of complex IV to specific proteins [i.e., EGFR-pY845, the viral protein HBx, PKCe, NO synthase, subunit RIa of PKA, the androgen receptor; for a review, see Vogt et al. (323)]; it seems worthwhile to test whether phosphorylation at those sites may also modify supercomplex formation by cytochrome oxidase.

1. Flux-control analysis of the respiratory complexes

As explained in section II.A, metabolic flux-control analysis allows a quantitative measurement of the control exerted on a composite pathway by its individual enzymatic steps. It is assumed by the theory of this analysis that the individual steps consist of separate enzymes joined by the diffusion of common intermediates. In any such system, the control would be differently exerted by one or more steps in the pathway, but the sum of the control coefficients of all steps would approach 1 and not overcome unity.

The situation would be different, however, in enzymatic supercomplexes in which the interactions between active sites are fixed and substrates and intermediates are channeled from one defined site to another one without leaving the protein environment; in such a supercomplex, the metabolic pathway would behave as a single enzyme unit, and inhibition of any one of the enzyme components would elicit the same flux control. In particular, in a system in which the respiratory chain is totally dissociated from other components of the OXPHOS apparatus (i.e., ATP synthase, membrane potential, and carriers), such as open nonphosphorylating submitochondrial particles, the existence of a supercomplex would elicit a flux-control coefficient near unity at any of the respiratory complexes, and the sum of all apparent coefficients would be greater than 1 (148). The two alternatives (random diffusion or channeling) in terms of flux-control analysis are depicted for the mitochondrial respiratory chain in the scheme in Fig. 14.

OPEN IN VIEWER

FIG. 14.

Scheme of flux control in two models of NADH-dependent respiration by the mitochondrial respiratory chain. The flux-summation theorem, a central principle of MCA, describes how the control of flux through any metabolic pathway of arbitrary complexity is distributed among the component reaction steps. (A) In a linear pathway composed of a series of enzymes connected by mobile intermediates (i.e., CoQ and cytochrome c), the relative control of each respiratory complex may be different, and the sum of all the flux coefficients (C_i) is equal to unity. (B) In a supercomplex, any step in the obligatory path is regarded as a component of a single enzyme and must be completely rate controlling; therefore, each component elicits a maximal flux-control coefficient (C_i = 1), and the sum of all coefficients results in a value higher than 1.

The use of flux-control coefficients for the assessment of pool behavior, although based on different theoretic grounds, has a practical exploitation very similar to that of inhibitor titrations: the presence of a lag in the inhibition of the integrated activity, as an indication that the inhibited step is not rate limiting, should be common to both types of analysis.

We addressed the problem in mammalian and in plant mitochondria. The flux-control coefficients of the respiratory complexes (I, II, III, IV) were investigated by using bovine heart mitochondria and submitochondrial particles devoid of substrate permeability barriers and by performing titrations with specific inhibitors of each complex (23). Both complexes I and III were found to be highly rate controlling over NADH oxidation (C_I = 1.06; C_III = 0.99), with the strong kinetic evidence suggesting the existence of a functionally relevant association between the two complexes. On the contrary, complex IV appears to be randomly distributed (C_IV = 0.26), although it is possible that if any stable interaction with complex IV exists in mammalian mitochondria, it escaped detection, most likely because of a pronounced abundance of molecules in nonassembled form. Moreover, complex II is fully rate limiting for succinate oxidation (C_II = 0.88; C_III = 0.34; C_IV = 0.20), clearly indicating the absence of substrate channeling toward complexes III and IV (23).

In permeabilized mitochondria from freshly harvested potato tubers, in which no activity of the so-called alternative oxidase, AOX, is present at the level of ubiquinone (5), inhibitor-titration experiments indicate that complexes III and IV are involved in the formation of a supercomplex assembly comprising complex I (98), whereas the alternative dehydrogenases, as well as the molecules of complex II, are considered to be independent structures within the inner mitochondrial membrane (Genova et al., unpublished data).

The electrophoretic demonstration of the existence of respiratory supercomplexes requires the use of digitonin that alters the membrane and destroys the permeability barriers; Conversely, our initial demonstration with flux-control analysis that a complex I to III aggregate exists in respiring mitochondrial membrane fragments (23) was done under nonphosphorylating conditions in a system lacking possible rate-limiting steps external to the respiratory chain per se (dehydrogenases, substrate carriers, membrane potential, ATP synthase, adenine nucleotide carrier).

Flux-control analysis in intact mitochondria under phosphorylating or uncoupled conditions usually exhibits low flux-control coefficients for respiratory complexes in mitochondria isolated from various tissues (cf. 54, 136, 219, 319) because the control is distributed among other components. We obtained recent evidence of the existence of a complex I to III aggregate in mitochondria under phosphorylating conditions (state 3) by exploiting flux-control analysis in freshly prepared liver mitochondria from old rats (98), in which, being respiration-rate limiting over other accessory activities because of aging, the old rats showed high control by both complex I and complex III. No information about the presence of supramolecular assemblies can be inferred based on the results obtained in state 4 (controlled respiration) because, in this condition, the proton leak across the inner membrane becomes the rate-limiting step, and therefore, a very low metabolic control over respiration is exerted by the enzyme complexes regardless of their assembly status.

2. Electron-transfer activity of isolated supercomplexes

Very recently, Acin-Perez et al. (3), in a detailed study of mouse liver mitochondria, confirmed the presence of different forms of supercomplexes after solubilization in different detergents and BN-PAGE; different from previous studies, some supercomplexes also contained complex II and ATP synthase (complex V). One particular subfraction (band 3) contained all complexes I, II, III, and IV and, in addition, cytochrome c and CoQ₉. The band was excised from the gel, and respiration was measured with a Clark electrode, showing full respiratory activity from either NADH or succinate, which was sensitive to the specific respiratory inhibitors of all involved complexes. The simultaneous inclusion in the oxygen electrode chamber of the individual respiratory complexes did not allow an integrated respiration. Therefore, the oxygen consumption shown by the supercomplex bands not only is the consequence of the presence of all the respiratory complexes and “mobile” carriers needed, but also reflects the proper arrangement into a functional structure.

This finding directly confirms that supercomplex organization is compatible with electron transfer; however, it does not discard the idea that electron transfer is possible in the membrane in the absence of supercomplex organization in accordance with the random-collision model. Acin-Perez et al. (3) confirmed that most or all complex I is indeed in the form of a supercomplex, but most of complexes II, III, and IV appear to be free as isolated complexes. The authors propose a “plasticity model” in which both types of organization are possible and functional, depending on the different mitochondrial systems and on the particular physiological states: for example, a supercomplex I to III devoid of complex IV may still be useful under brief hypoxic conditions to generate a local membrane potential and ATP synthesis, in absence of cytochrome oxidase activity but in presence of excess cytochrome c.

The plasticity model well suits the information obtained by us with flux-control analysis suggesting that electron transfer between complex I and complex III is effected only by CoQ channelling, whereas that between complexes II and III and complexes III and IV seems to occur mostly by the pools of CoQ and cytochrome c, respectively (at least in beef heart and rat liver mitochondria).

A special case may be represented by electron transfer between complexes III and IV: the adherence to pool behavior shown in beef heart mitochondria by flux-control analysis may be dictated by the high percentage of individual free complexes in excess with respect to the III to IV supercomplex, so that overall flux-control analysis detects only pool behavior; thus, substrate (cytochrome c) channeling is present but negligible. In addition to that, however, the behavior in that region may depend on the strength of binding of cytochrome c to the supercomplex; unpublished studies from our laboratory suggest that in potato tuber mitochondria, where channeling occurs between complex III and complex IV, cytochrome c is tightly bound and detected in the III to IV supercomplex with BN-PAGE, whereas in beef heart and rat liver mitochondria (where no channeling is detected between complex III and cytochrome oxidase), cytochrome c is not found in the supercomplex, unless forced by a high concentration of exogenous cytochrome c.

1. Kinetic advantage: substrate channeling and metabolic consequences

The physiological implications of the specific interaction between complex I and complex III to form a supercomplex are not yet fully understood. It was speculated that they include enhancing of electron flow by direct channeling of ubiquinone. We now examine this question in more detail.

The key question concerns the compatibility/incompatibility of the stoichiometric channelling of CoQ between complex I and complex III with the existence of pool behavior by the bulk of CoQ molecules free in the bilayer. First, the problem appears to be confined to the interaction between complex I and complex III, because no clear demonstration exists that complex II is part of a supercomplex. Complex II kinetically follows pool behavior in mitochondrial membranes, in reconstituted systems, and in the double-inhibitor titration experiments in intact mitochondria (cf. 182), in accordance with the lack of supercomplexes found with both BN-PAGE (281) and MCA (23). Nevertheless, the isolation of discrete units having succinate-cytochrome c reductase activity (335), the biophysical studies by Gwak et al. (113), and the antimycin-titration experiments in yeast (28) appear in strong contrast to this indication. The only possible explanation so far is in the existence of very loose but specific contacts between complex II and complex III, not giving rise to any kind of channelling, even in phosphorylating mitochondria. Significantly, in the recent article by Acin-Perez et al. (3), minute amounts of functional supercomplexes containing complex II and endowed with succinate oxidase activity were detected.

2. Stability and assembly of individual complexes

The first chromatographic isolation of a complete respirasome (I₁III₄IV₄) from digitonin-solubilized membranes of Paracoccus denitrificans indicated that complex I is stabilized by assembly into the NADH oxidase supercomplex, because attempts to isolate complex I from mutant strains lacking complexes III or IV led to the complete dissociation of complex I under the conditions of BN-PAGE. Reduced stability of complex I in those mutant strains was also apparent from an almost complete loss of NADH-ubiquinone oxidoreductase activity when the same protocol as that for parental strain was applied (298).

Analysis of the state of supercomplexes in human patients with an isolated deficiency of single complexes (279) and in cultured-cell models harboring cytochrome b mutations (2, 56) also provided evidence that the formation of respirasomes is essential for the assembly/stability of complex I. Genetic alterations leading to a loss of complex III prevented respirasome formation and led to secondary loss of complex I; therefore, primary complex III–assembly deficiencies were seen as complex III/I defects. Conversely, complex III stability was not influenced by the absence of complex I.

D'Aurelio et al. (56) studied mtDNA complementation in human cells by fusing two cell lines, one containing a homoplasmic mutation in the COXI subunit of complex IV, and the other, a distinct homoplasmic mutation in cytochrome b of complex III. With cell fusion, respiration was recovered in hybrids cells, indicating that mitochondria fuse and exchange genetic and protein materials. The recovery of mitochondrial respiration correlated with the presence of supercomplexes containing complexes I, III, and IV; critical amounts of complexes III or IV are therefore required for supercomplexes to form and provide mitochondrial functional complementation. From these findings, supercomplex assembly emerged as a necessary step for respiration, its defect setting the threshold for respiratory impairment in mtDNA mutant cells. Several other studies in mutants and in patients having specific defects in a single respiratory complex have suggested that complex III and, to a lesser extent, complex IV are involved in the assembly and stabilisation of complex I in mammals (62, 203). This is not the case in fungi, because a P. anserina mutant lacking both complexes III and IV possesses a normal complex I, presumably as a consequence of special fungal features such as the presence of AOX and a dimerized complex I (193).

Conversely, mutations of complex I had controversial effects, because in some studies, they did not affect the amounts of other complexes, whereas in others, they significantly reduced the amounts of complexes III and IV (105, 279, 311). The reason for this discrepancy is not known, but might be related to the specificity of the mutation affecting subunits of complex I involved in the contacts with the other complexes.

3. Possible role of supercomplex organization in limiting ROS formation

Although no direct study is available on the effect of supramolecular organization on ROS production by the respiratory chain, indirect circumstantial evidence suggests that supercomplex assembly may limit the extent of superoxide generation by the respiratory chain. Zhang et al. (339) showed that reconstitution of succinate CoQ reductase with the bc ₁ complex to yield an active succinate cytochrome c reductase inhibited superoxide formation. Although the relevance of a II to III supercomplex in vivo is questionable in light of that amply discussed in this review, the principle that a tighter organization of the respiratory enzymes may hide autooxidizable prosthetic groups hindering their reaction with oxygen may have a wider application.

Similarly, it was discussed in section III.B that two potential sites for oxygen reduction exist in complex I, represented by FMN and iron–sulfur cluster N2; controversial results from different laboratories working either on isolated complex I (72, 95) or on mitochondrial membranes (79, 180) would generally indicate that N2 as a source of ROS would be predominant in membrane particles, whereas FMN might become available after complex I isolation. A reasonable hypothesis is that FMN becomes exposed to oxygen only when complex I is dissociated from complex III. Although the molecular structure of the individual complexes does not allow us to envisage a close apposition of the matrix arm of complex I, where FMN is localized, with either complex III or IV (66, 248, 277), the actual shape of the I₁III₂IV₁ supercomplex from bovine heart (277) suggests a slightly different conformation of complex I in the supercomplex, with a smaller angle of the matrix arm, with the membrane arm showing a greater bending toward the membrane (and presumably complex III), in line with the notion that complex I may undergo important conformational changes (253). Moreover, the observed destabilization of complex I in the absence of supercomplex (section VI.A.2) may render the 51-kDa subunit containing the FMN more “loose,” allowing it to interact with oxygen. The elevated ROS production observed in P. anserina respiring on AOX, in which the major form of complex I is a I₂III₂ supercomplex rather than the usual I₁III₂ supercomplex (159), is in line with this reasoning, because it is likely that the complex I dimer may undergo a looser interaction than a complex I monomer with the complex III dimer.

On the basis of their studies on rat brain mitochondria oxidizing different substrates, Panov and collaborators (240) suggested that the supercomplex organization of complex I within the chain prevents excessive superoxide production on oxidation of NAD-linked substrates because the efficient channeling helps to maintain the chain in the oxidized state, whereas on succinate oxidation, the backward electron flow keeps the centers more reduced in complex I, favoring production of superoxide; it is interesting to note that complex II does not form a respirasome.

Additional circumstantial evidence on the role of supercomplex organization may come from the observation that high mitochondrial membrane potential elicits ROS generation, whereas uncoupling strongly reduces ROS production (176); although likely explanations have been given for these observations, they are compatible with the suggestion by Piccoli et al. (251) that high membrane potential may dissociate the supercomplexes into the individual units (described in section VI.B.3).

1. Protein concentration

At low protein concentrations in lipid bilayers, proteins are usually randomly dispersed, whereas at high concentrations, they tend to aggregate. Integral membrane proteins become surrounded by a lipid domain of boundary lipids; the possibility that the wetting lipid layer is shared by two or more proteins also gives rise to aggregation phenomena that are at a longer-range scale than those involved in direct protein–protein interactions (168).

In the mitochondrial inner membrane, integral proteins of the OXPHOS system are densely packed so that the average distance between complexes may be calculated to be a few nanometers (174); in addition, the presence of different types of lipids would enhance wetting and capillary condensation, as well as the possibility of phase separations. Thus, the presence of protein aggregates of the transmembrane respiratory complexes would not be at all unexpected on theoretic grounds. The immobilization of the proteins would be also favored by the presence of outer membrane–inner membrane contacts, by the narrow tubular connections of the cristae (197), and by the high viscosity of the matrix, which would exert a slowing effect on the diffusion of proteins spanning the inner membrane.

In addition, direct binding of matrix enzymes with respiratory chain complexes may occur, as proposed for complex I with several NAD-linked matrix dehydrogenases, including pyruvate dehydrogenase and α-ketoglutarate dehydrogenase complexes, malate dehydrogenase, and β-hydroxyacyl-CoA dehydrogenase (233, 300).

Early experiments reported by Heron et al. (123) provided evidence that purified complex I and complex III, when mixed as concentrated solutions in detergent and then co-dialyzed, combine reversibly in a 1:1 molar ratio to form a complex I to III unit (NADH-cytochrome c oxidoreductase) that contains equimolar FMN and cytochrome c₁ (clearly at difference with the supercomplexes found by BN-PAGE, where complex III is present as a dimer, cf. section V.A) and 2 to 3 moles of ubiquinone-10 per mole of protein unit. Activation-energy measurements for NADH-cytochrome c oxidoreductase activity showed that oxidoreduction of endogenous ubiquinone-10 proceeds somewhat differently from the oxidation and the reduction of exogenous quinones, supporting the idea that the mobility of ubiquinone-10 in the complex I to III unit is highly restricted and suggesting that CoQ₁₀ is effectively trapped between the component complexes in an environment that may be partly protein and partly derived from the lipid annuli of those complexes. However, CoQ-pool behavior could be restored, and complex I and complex III could be made to operate independent of each other by increasing the concentrations of phospholipid and ubiquinone (approximately a twofold and a sixfold increase, respectively) in the concentrated mixture (123). Inclusion of phospholipids in the reconstituted system may have a number of effects on the physical state of the system. Heron and co-workers (123) proposed that, when phospholipids in excess of those needed to form an annulus are absent, relative mobility is lost, and complexes are frozen in their complex I to III assembly, favoring a stable orientation of the site of reduction of ubiquinone with respect to the site of oxidation. Heron et al. (123) also reported that endogenous ubiquinone-10 leaks out of the complex I to III unit when extra phospholipids are present, causing a decrease in activity that could be alleviated by adding more ubiquinone. It is likely that the function of the large amount of ubiquinone in the natural membrane may be, therefore, to maintain the ubiquinone-10 content in the supercomplex unit when it is formed (cf. section VII.A).

An analogous system, obtained by fusing a crude mitochondrial fraction (R₄B) (262) enriched in complex I and complex III with different amounts of phospholipids and CoQ₁₀ (179), was used in our laboratory to discriminate whether the reconstituted protein fraction behaves as individual enzymes (CoQ-pool behavior) or as assembled supercomplexes, depending on the experimental distances between the intramembrane particles. The comparison of the experimentally determined NADH-cytochrome c reductase activity with the values expected by theoretic calculation applying the pool equation showed overlapping results at phospholipid dilutions (wt/wt) from 1:10 on (i.e., for distances >50 nm), whereas at shorter distances between complex I and complex III, resembling the mean nearest-neighbor distance between respiratory complexes in mitochondria, pool behavior was no longer effective (22, 179). In the two experimental models, kinetic testing according to the metabolic flux-control analysis (Table 6) validated the hypothesis of a random organization and of a functional association between complex I and complex III, respectively (98)

2. Lipid composition

The possibility that a protein exhibits preference for a given lipid species so that a lipid annulus wets the protein surface may be another source of protein aggregation (168); changing the nature of the lipids would dramatically change their matching conditions to the proteins. Changes in membrane protein aggregation induced by lipid peroxidation may well be expected on theoretic grounds as a consequence of profound biophysical changes in the lipid bilayer. All purified preparations of mitochondrial electron-transfer complexes are isolated as lipoprotein complexes, the extent of associated lipid depending on the particular method used for isolation, and the phospholipid composition reflecting that for the mitochondrial inner membrane. Predominant phospholipids present include cardiolipin, phosphatidyl choline, phosphatidyl ethanolamine, and lesser amounts of neutral lipids and phosphatidyl inositol.

Two roles of phospholipids have been distinguished: (a) a dispersive solubilization effect that can be duplicated by appropriate detergents; and (b) a catalytic effect that can be specifically fulfilled only by tightly bound cardiolipin, more likely buried within the protein complexes (93, 163, 168, 264). Two additional possible roles may have to be met, particularly in the cases of complex I and complex III. These roles might be to provide a sufficiently lipophilic environment for the interaction of the lipophilic electron carrier, ubiquinone, and to participate in linking together components of the respiratory chain.

The absolute requirement of cardiolipin (CL) for the activity of cytochrome oxidase, complex I, and complex III, as well as for that of several mitochondrial carriers, suggests that this phospholipid plays a crucial role in the coupled electron-transfer process (93), but recent results seem also to indicate that cardiolipin stabilizes respiratory-chain supercomplexes as well as the individual complexes. The availability of a CL-lacking yeast mutant (Δcrd1 null) provided the opportunity to demonstrate that mitochondrial membranes still contained the III₂–IV₂ supercomplex, but that it was significantly less stable than supercomplexes in the parental strain. Other phospholipids that are increased in the mutant, including phosphatidyl-ethanolamine and phosphatidyl-glycerol, could not substitute for cardiolipin and could not prevent dissociation of supercomplexes, showing 90% of the individual homodimers of complex III and IV not organized into supercomplex under BN-PAGE conditions (250, 340).

The putative direct protein–protein interaction of cytochrome oxidase and complex III in yeast (250) is proposed also to involve two phospholipid molecules (i.e., cardiolipin and phosphatidyl-ethanolamine), tightly bound in a cavity of the membrane-imbedded domain of complex III (165), suggesting that they can provide a flexible linkage between the interacting subunits of complexes III and IV. It also was found that, in the absence of cardiolipin, most supramolecular associations, including those involving the adenine nucleotide carrier, are destabilized or undetectable (47).

In a different study, the stability and assembly of complex IV was found to be reduced in yeast cells lacking Taz1 (31), the hortologue of human Tafazzin, an acyl transferase involved in the synthesis of mature tetralinoleyl cardiolipin; mutations of Tafazzin in humans result in Barth syndrome, a cardioskeletal myopathy with neutropenia, characterized by respiratory-chain dysfunction. It was later found (203) that the cardiolipin defect in Barth syndrome results in destabilization of the supercomplexes by weakening the interactions between respiratory complexes. Tafazzin is found in the inner membrane in a complex including ATP synthase and the adenine nucleotide carrier; the absence of this complex due to Taz1 mutations also induces altered cristae morphology (46).

It is well documented that exposure of mitochondria to reactive oxygen species (ROS) can affect the respiratory activity through oxidative damage of cardiolipin (249). We have demonstrated with flux-control analysis that the maintenance of a I–III supercomplex after reconstitution of a protein fraction enriched with complex I and III (R₄B) into phospholipid vesicles at high protein-to-lipid ratios (see earlier) is abolished if lipid peroxidation is induced by 2,2′-azobis-(2-amidinopropane) dihydrochloride (AAPH) before reconstitution (22). Evidently, the distortion of the lipid bilayer induced by peroxidation and the alteration of the tightly bound phospholipids determine dissociation of the supercomplex originally present in the lipid-poor preparation (Table 6).

In addition to lipid peroxidation, changes in membrane shape and lipid composition, as caused by modification of lipid biosynthesis enzymes, can be putative actors in the regulation of the oxidative phosphorylation system through supercomplexes stabilization. In their recent review (94), Furt and Moreau described the role of different lipids in mitochondrial morphodynamics and also illustrated how the lipid-induced membrane curvature can affect the activity of various protein partners by conditioning their recruitment and segregation into microdomains. The challenge in the future will be to unravel the detailed mechanisms.

3. Functional state

The impact of the mitochondrial transmembrane potential (Δμ_H+) on the flux control exerted by cytochrome c oxidase on the respiratory activity in intact cells and isolated mitochondria was evaluated by Piccoli et al. (251). The results indicate that, under conditions mimicking the mitochondrial state 4 respiration, the control strength of the oxidase is decreased in respect to endogenous state 3 respiration. Although the interpretation of the results in such a complex system is very difficult, the authors suggest that such a change in control strength might be featured in terms of equilibrium between a random organization of cytochrome oxidase with respect to other complexes (at high membrane potential, state 4) and a respirasome organization (at lower membrane potential, state 3). Because the respiratory rate is high in state 3 conditions, the supercomplex organization would produce an extra advantage by increasing the rate with channeling. Although the driving forces leading to the assembly/disassembly of the supercomplexes have been not yet defined, it is not inconceivable that, given the membrane-integrated nature of the single complexes, electrostatic/hydrophobic interactions may enter into play in response to Δμ_H+. No direct demonstration of the association state of respiratory complexes as a function of the physiological states of the OXPHOS system is available.

The regulatory role of complex I and complex IV phosphorylation has been treated in section II.B. No direct study is available investigating the supramolecular organization of the respiratory chain as a function of phosphorylation/dephosphorylation events, but it is tempting to speculate that the increase of activity of complex I and the decrease of ROS generation elicited by phosphorylation of the NDUFS4 subunit may be the result in part of enhanced stability of the I to III supercomplex.

In cytochrome oxidase, one may speculate on the role of phosphorylation of subunit COXI on tyrosine-304. This amino acid is located in the intermembrane-space domain of subunit I, on a helix that is in contact with the other catalytic subunit (COXII), and facing the interface region of the two monomers. The enhancement of the monomer–monomer interaction might be a plausible effect of phosphorylation (170). Interestingly, the concave face of complex IV, which is the dimer interface in the x-ray structure, is also the contact surface of the monomer with its protein partners when it is assembled in the I₁III₂IV₁ bovine supercomplex (277).