Supercomplexes in the respiratory chains of yeast and mammalian mitochondria

Abstract

Around 30-40 years after the first isolation of the five complexes of oxidative phosphorylation from mammalian mitochondria, we present data that fundamentally change the paradigm of how the yeast and mammalian system of oxidative phosphorylation is organized. The complexes are not randomly distributed within the inner mitochondrial membrane, but assemble into supramolecular structures. We show that all cytochrome c oxidase (complex IV) of Saccharomyces cerevisiae is bound to cytochrome c reductase (complex III), which exists in three forms: the free dimer, and two supercomplexes comprising an additional one or two complex IV monomers. The distribution between these forms varies with growth conditions. In mammalian mitochondria, almost all complex I is assembled into supercomplexes comprising complexes I and III and up to four copies of complex IV, which guided us to present a model for a network of respiratory chain complexes: a 'respirasome'. A fraction of total bovine ATP synthase (complex V) was isolated in dimeric form, suggesting that a dimeric state is not limited to S.cerevisiae, but also exists in mammalian mitochondria.

Fig. 1. Saccharomyces cerevisiae complex IV is quantitatively associated with complex III. The amounts of complex III in free dimeric form (III₂) or associated with one (III₂IV₁) or two complex IV monomers (III₂IV₂) vary with growth conditions. (A) BN-PAGE of mitochondria from strain W303-1A grown on glucose for 24 h, and two-dimensional resolution by SDS–PAGE. The subunits of complexes III and IV were assigned after N-terminal protein sequencing. (B) Growth on lactate increases the amount of III₂IV₂ complex, and decreases the fraction of free complex III. (C) Prolonged growth on glucose reduces the amounts of the III₂IV₂ complex, and increases the fraction of free complex III. V_Dim and V_Mon are dimeric and monomeric ATP synthase (complex V), respectively.

Fig. 2. Functional association of complexes III and IV in digitonin-solubilized yeast mitochondria. Detergent dependence (A) of ubiquinol cytochrome c reductase (mol cytochrome c reduced/mol complex III/s), (B) of cytochrome c oxidase (mol cytochrome c oxidized/mol complex IV/s) and (C) of ubiquinol oxidase (complexes III+IV; mol DBH oxidized/mol complex III/s) after addition of 5 µM yeast cytochrome c. Arrows indicate the solubilization range. A decrease in ubiquinol oxidase rates after solubilization by DDM indicates a separation of complexes III and IV. No decay of rates and no dissociation of complex III–IV occur after solubilization by digitonin (cf. text).

Fig. 3. Complex I from bovine heart mitochondria is associated with complexes III and IV. (A) BN-PAGE of bovine heart mitochondria after solubilization by digitonin. Most complex I and complex III was found assembled into two major supercomplexes a and b, and two minor supercomplexes c and d. The 200 kDa mass differences indicate the presence of varying copy numbers of monomeric complex IV. (B) Solubilization by DDM was used as a reference for quantitative solubilization of all OXPHOS complexes.

Fig. 4. Bovine complex I has binding sites for complex III and complex IV, and ATP synthase can be isolated as a dimer. (A) Complexes a–d and dimeric ATP synthase (V_Dim) separated by BN-PAGE (Figure 3A; 4 g/g digitonin) were dissociated by 2D BN-PAGE using addition of DDM to the cathode buffer. Direct interaction of complexes I and III was apparent from the dissociation of complex a (I₁III₂) into monomeric complex I and dimeric complex III. Complexes b–d comprised complex IV in addition (silver staining required for c and d). Dimeric complex V dissociated into the monomeric form V_Mon. (B) The line of complex b from 2D BN-PAGE (brackets in Figure 4A) was resolved by 3D SDS–PAGE and Coomassie Blue stained. The stoichiometry within this complex (I₁III₂IV₁) was determined by densitometric analysis using purified complexes I, III and IV for calibration. (C) 2D BN-PAGE similar to (A), but using addition of Triton X-100 to the cathode buffer dissociated complex b (I₁III₂IV₁) in a different way (two additional spots larger than complex I; less complex IV dissociated). (D) The line of complex b from 2D BN-PAGE (brackets in Figure 4C) was resolved by 3D SDS–PAGE. The additional spots were identified as undisociated I₁III₂IV₁ complex and as I₁IV₁ complex. COXI and COXII, subunits of cytochrome c oxidase (complex IV).

Fig. 5. Identification of dimeric complex IV, direct complex III–IV interactions, and complex e (I₁III₂IV₄) comprising four copies of complex IV. (A) BN-PAGE of bovine heart mitochondria using various Triton X-100/protein ratios for solubilization. Low Triton X-100(1.0 g/g) selectively solubilized complexes V and II. High Triton X-100 (2.4 g/g) solubilized all OXPHOS complexes quantitatively, but retained supercomplexes a (I₁III₂) and c (I₁III₂IV₂). Intermediate Triton X-100 (1.4 g/g) retained supercomplexes a–e (I₁III₂IV_X) comprising0–4 copies of complex IV. (B) 2D BN-PAGE of the boxed sector from (A). Arrowheads mark the lines of dissociated complexes a–e. The identification of dimeric complex IV in lines c–e, but not in b (only one copy of complex IV), indicates the presence of complex IV dimers within the supercomplexes. (C) BN-PAGE of bovine heart mitochondria using high (1.6 g/g) and low (0.6 g/g) DDM/protein ratios for solubilization. Low DDM solubilizes all OXPHOS complexes quantitatively, but retains ∼20% of total complex V in dimeric form (V_Dim). (D) 2D BN-PAGE of the lane from (C) using low DDM. Arrowheads mark the lines of complexes that were retained after first dimension BN-PAGE but dissociated by 2D BN-PAGE.

Fig. 6. Functional association of complexes I and III in digitonin-solubilized bovine heart mitochondria. (A) Dergent dependence of NADH ubiquinone reductase (mol NADH oxidized/mol complex I/s). Arrows indicate the solubilization range. (B) Detergent dependence of ubiquinol cytochrome c reductase (mol cytochrome c reduced/mol complex III/s). (C) Detergent dependence of NADH cytochrome c reductase (complexes I+III; mol cytochrome c/mol complex I/s). A decrease in NADH cytochrome c reductase rates after solubilization by DDM indicates a separation of complexes I and III. The absence of a decay of rates after complete solubilization by digitonin indicates a retained association of complexes I and III (cf. text).

Fig. 7. Model for of a network of mammalian respiratory chain complexes. This model is based on the identification of direct interactions of complexes and on the overall 1:3:6 stoichiometry of complex I:III:IV by Hatefi (1985). It postulates two copies of a large building block comprising complexes I, III and IV, and one smaller building block without complex I. Comparison of the different solubilization by DDM and Triton X-100 indicated that these building blocks can interact to form a network of respiratory chain complexes.