Function and redox state of mitochondrial localized cysteine-rich proteins important in the assembly of cytochrome c oxidase

Summary

The cytochrome c oxidase (CcO) complex of the mitochondrial respiratory chain exists within the mitochondrial inner membrane (IM). The biogenesis of the complex is a multi-faceted process requiring multiple assembly factors that function on both faces of the IM. Formation of the two copper centers of CcO occurs within the intermembrane space (IMS) and is dependent on assembly factors with critical cysteinyl thiolates. Two classes of assembly factors exist, one group being soluble IMS proteins and the second class being proteins tethered to the IM. A common motif in the soluble assembly factors is a duplicated Cx₉C sequence motif. Since mitochondrial respiration is a major source of reactive oxygen species, control of the redox state of mitochondrial proteins is an important process. This review documents the role of these cysteinyl CcO assembly factors within the IMS and the necessity of redox control in their function.

1. Mitochondrial Organization

The mitochondrion consists of a continuous reticulum that makes up nearly 10% of the cell volume in respiring yeast cells. The tubular network is highly dynamic and changes size and shape through fission and fusion events [1]. A double membrane forming two internal spaces encloses mitochondria. The space between the two membranes is called the intermembrane space (IMS) and the volume enclosed within the inner membrane (IM) is designated as the matrix compartment. The two membranes differ significantly in their composition with the IM more highly enriched in protein content. The outer membrane (OM) envelope contains 25 Å nm pores that allow diffusion of small ions such as glutathione. The IM is a barrier to diffusion, so passage of metabolites requires transporters. As such, the import of glutathione across the IM is believed to occur through dicarboxylate and 2-oxoglutarate transporters [2].

Mitochondria maintain their genome within the matrix compartment. This genome codes for a limited number of polypeptides, 13 and 8 in humans and yeast, respectively. All, but one, of these proteins are components of the oxidative phosphorylation system.

The double membrane of mitochondria is interrupted by junction points of contact between the IM and OM [3]. The junction points are likely formed by protein assemblies involved in protein import [4, 5]. The mean distance across the OM and IM is 20 nm, although the distance narrows to only about 14 nm at junction points. In cells with high respiration rates, the IM is invaginated, folding into tubular structures designated cristae. Cristae are enriched in the enzyme complexes involved in oxidative phosphorylation. Electron microscopy tomography revealed that the cristae tubules are 30–40 nm in diameter but narrow to about 28 nm at junction points with the boundary IM, designated inner boundary membrane (IBM) [3, 5]. The constriction of the cristae junctions at the IBM resolves the soluble IMS into separate volumes that appear to be in equilibrium only for small molecules [6]. Cristae can exist as tubular structures or merged to form flattened lamellar compartments. Stacked lamellar cristae remain connected to the IBM by tubular cristae junctions. The cristae junctions are believed to be dynamic and modulated by the energetics of the organelle as well as by the fusion/fission process.

In this review, the lumen of the cristae will be referred to as the cristae lumen and the space between the boundary IM and OM as the boundary IMS. The generic term IMS will specify the generalized volume between the OM and IM without sub-compartmentization specification.

Oxidative phosphorylation occurs predominantly, but not exclusively, on the cristae membrane (CM). The enrichment of cytochrome c oxidase (CcO) in cristae is well established [7], but recent studies confirm the abundance of complexes III (cytochrome bc₁ complex) and V (ATP synthase) within the CM [5]. Respiratory complexes I, III and IV are largely present as supercomplexes within the CM [8–11]. The yeast III/IV supercomplex consists of a dimeric complex III species at the core with one or two CcO complexes at opposite ends [11]. Synthesis and assembly of the respiratory complexes occurs preferentially on the CM as mitochondrial ribosomal proteins are associated with the CM [5].

2. Generation of reactive oxygen species in mitochondria

Mitochondrial respiration is a major source of reactive oxygen species. Respiring mitochondria convert 1–2% of the oxygen consumed to superoxide anion [12, 13]. The bulk of superoxide anion produced on a daily basis in most organisms comes from ubisemiquinone of coenzyme Q within the respiratory chain [14]. CoQ shuttles electrons from complexes I and II to complex III. The CoQ semiquinone generated either at complex I or during the Q-cycle in complex III can react with oxygen generating superoxide anions. Based on the sidedness of the Q-cycle in complex III, superoxide is generated in both the matrix and IMS, although the bulk of the superoxide is generated within the matrix [15, 16]. Superoxide may itself cause damage or may react further to yield other reactive species such as hydrogen peroxide or the hydroxyl radical. The presence of reactive oxygen species can induce oxidation events such as modification of protein thiols on both sides of the IM [17]. Endogenous reactive oxygen species generated at complex I and III were shown to modify nine distinct mitochondrial proteins [17]. Several of the modified proteins function in fatty acid oxidation.

The normal production of superoxide anion during respiration and subsequent generation of hydrogen peroxide and other oxidants in both the IMS and matrix necessitates control of transition metal ion availability to minimize Fenton chemistry to generate more potent oxidants. Copper ions used in metallation reactions are protein bound to minimize the deleterious effects of unbound Cu(I) ions. Copper metallation of CcO and superoxide dismutase (Sod1) within the IMS occurs by metallochaperone proteins Cox17 and Ccs1, respectively. Only a small fraction of the cellular Sod1 exists within the IMS [18]. The transfer reactions are protein-mediated. However, the Cu(I) sites on Ccs1 are partially solvent accessible and this is likely also true for Cox17. Although the level of copper complexes of these two proteins isn’t known within the IMS, it is possible that the level of Cu(I) within the IMS may be regulated to minimize chances of copper-induced oxidation reactions. The Cu(I) ions used in the IMS metallation reactions derives from a storage pool within the matrix [19, 20]. It is conceivable that the Cu(I) transporter within the IM that translocates Cu(I) to the IMS is regulated such that the Cu(I) transported is coupled to the biogenesis of CcO and Sod1.

Another defense against deleterious oxidative processes is the availability of redox systems to maintain redox homeostasis. The effectiveness of the matrix redox system is highlighted by the observation that the apparent redox potential of the mitochondrial matrix is more negative (−360 mV) relative to the cytoplasm (−320 mV) in HeLa cells using redox-sensitive fluorescent GFP variants [21, 22]. No information is available on the redox potential of the IMS compartment.

The mitochondria matrix contains well-defined redox components including glutathione, thioredoxin and glutaredoxin systems. A fraction of the cellular glutathione reductase Glr1 exists within the matrix in yeast. The thioredoxin system involves the mitochondrial-specific Trx3 thioredoxin and the Trr2 thioredoxin reductase. Both monothiol (Grx5) and dithiol (Grx2) glutaredoxins exist within the matrix. Initiation at an upstream ATG in the transcripts for Grx2 and Glr1 generate the mitochondrially-targeted variants. The presence of multiple protein reductants and the overlapping functions of Glr1 and Trr2 reductases illustrates that a robust redox pathway exists within the matrix. Apart from an initial report of mammalian thioredoxin reductase-I existing within the IMS [23], it is not clear whether other redox components are present within the IMS compartments. The porous OM suggests that GSH may equilibrate across the membrane, but no evidence exists for the IMS presence of Glr1 to maintain GSH/GSSG redox homeostasis.

3. Role of Cysteines in IMS Proteins

The mitochondrial proteome is expected to contain nearly 850 proteins in yeast [24] but closer to 1000 distinct proteins in humans [25]. About 14% of the proteins are involved in oxidative phosphorylation, whereas 25% are predicted to be involved in maintaining and expressing the mitochondrial genome [26]. A subset of the proteome resides within the IMS either as soluble proteins or as molecules tethered to the IM. Many of these proteins are either cysteine-rich or have functionally important cysteine residues that can exist as within disulfide bonds or as reduced thiolates. The abundance of disulfide-containing molecules within the IMS suggests that redox control within this compartment differs from that within the cytoplasm where disulfide bond formation is rare due to the high reducing potential of the cytoplasm [27]. The constriction of the cristae junctions creating both cristae lumen and the boundary IMS opens the possibility that redox pathways may be distinct within the two subcompartments.

A common structural motif of cysteine-rich proteins within the IMS is a helical hairpin conformer. The structural paradigm for the helical hairpin motif in IMS proteins is small Tim proteins (Tim9/Tim10) characterized by a conserved twin Cx₃C sequence motif. Tim9 and Tim10 each form a helix-loop-helix conformer held together by paired disulfide bonds in the twin Cx₃C motif [28] (Fig. 1A). Tim9 and Tim10 form a heterohexameric complex that functions as a chaperone for incoming proteins as they are delivered to the TIM22 complex for IM insertion and SAM complex for OM insertion [29]. A second heterohexameric complex Tim8/Tim13 exists within the IMS that also mediates import and insertion of polytopic IM proteins [30]. Within the twin Cx₃C motif, one disulfide consists of the most N-terminal and C-terminal Cys residues (designated proximal pair), whereas the second disulfide consists of the two internal Cys residues (distal pair). The hexameric complex (Fig. 1B) is dependent on the disulfides within each subunit [31]. Reduction of the disulfides disassembles the complex.

Figure 1. Structural organization of the Tim9-Tim10 complex.

(A). Structure of the oxidized form of Tim9. Conserved cysteine residues forming twin Cx₃C motif are shown in red. The two pairs are designated as the proximal pair (closest to chain termini and distal pair (furthest from chain termini). (B). Structure of the Tim9-Tim10 hexameric complex. Each subunit is shown in different color. The four conserved cysteines in disulfide linkages within the twin Cx₃C motif are shown in red in one subunit colored yellow.

Yeast contains two IMS cytochrome c heme lyases, Cyc3 and Cyc7, for the assembly of cytochrome c and cytochrome c₁, respectively [32]. The two lyases attach heme to their respective cytochromes within the IMS. Human cells use only a single heme lyase to assemble cytochrome c and cytochrome c₁. Whereas the two lyases are not cysteine-rich molecules, their substrates cytochrome c and cytochrome c₁ have essential cysteinyl residues that become covalently attached to the bound heme. Thus, the cysteines in cytochrome c and cytochrome c₁ must be maintained in their reduced state for covalent attachment of hemes. The prevalence of respiratory complexes within the cristae suggests that cytochrome maturation occurs predominantly within cristae lumen. Nothing is known how the cytochrome cysteines are maintained in their reduced state.

Three subunits in the cytochrome bc₁ complex and CcO that project into the IMS, primarily the cristae lumen, contain disulfide bonds. The Rieske iron-sulfur protein Rip1 and Qcr2 of the bc₁ complex contains disulfide linkages [33]. The CcO subunit VIB (yeast Cox12) contains two disulfide bonds in the bovine CcO structure [34] (Fig. 2).

Figure 2. Disulfide bond in subunit VIB (yeast Cox12) of cytochrome c oxidase.

The structure of bovine CcO is shown with the disulfide bonds in subunit VIB shown in red. The four cysteines are arranged in a Cx₉C…..Cx₁₀C motif. This subunit projects to the IMS side of the inner membrane. Subunit VIB is shown in grey.

Sod1 contains a disulfide bond that is essential for its enzymatic function. The copper metallochaperone Ccs1 activates Sod1 during its folding by inserting the catalytic Cu(I) ion and in addition catalyzes formation of the essential disulfide in Sod1 [35]. Ccs1 itself has CxC and Cx₂C motifs that may be redox active. Ccs1 co-exists with Sod1 within the IMS. The distribution of Sod1 between the boundary IMS and cristae lumen is not known, but its presence in the cristae lumen is expected as the respiratory complexes within the cristae IM are the major source of superoxide anions. Recent studies on the IMS Sod1 in rat mitochondria suggest that the redox status of Sod1 is more complex [23, 36]. Sod1 appears to exist within the IMS as an inactive, reduced enzyme. It is transiently activated either by exogenous peroxide or superoxide anions or disruption of the OM [23]. Pretreatment of mitochondria with alkylating agents quenches activation of the enzyme. Thioredoxin reductase-I is shown to exist within the IMS and to be competent to reduce and inactivate Sod1. Inarrea et al. [23] postulate that the redox state, and therefore the enzymatic activity, of Sod1 is highly regulated within the IMS.

A redox proteomic study attempted to identify oxidized proteins in yeast using an affinity capture protocol [27]. Of the 64 proteins identified, two IMS proteins Ccs1 and Sod1 were identified, although as mentioned, both proteins also exist within the cytoplasm. The other mitochondrial proteins identified, Prx1, Hsp60, Yhb1, Ilv5 and Aco1, are matrix proteins.

4. Role of cysteines in the MIA import pathway

Cysteinyl residues in Tim proteins are additionally important for their import into the IMS. The import of Tim proteins is dependent on a disulfide relay system involving the MIA machinery consisting of at least Mia40 and Erv1 [37–41]. Transit of the Tim proteins through the TOM complex in the outer membrane results in transient capture of the imported molecules by Mia40 through disulfide bonding. Disulfide interchange between oxidized Mia40 and reduced thiolates on the imported protein generate intermolecular disulfides that trap the molecules within the IMS. The most N-terminal Cys residue in the Cx₃C motif of Tim9 and Tim10 was recently shown to be important for efficient capture by Mia40, whereas a mutant variant lacking all four Cys residues in not imported at all [42]. The importance of the most N-terminal Cys residue in both Tim9 and Tim10 suggests that recognition by Mia40 is precise and specific [42]. Replacement of only the distal Cys pair (Cys 2 & 3 in Fig. 1A) enabled import, but the process was inefficient. The mechanism of Tim10 release from Mia40 may involve intramolecular disulfide exchange with the fourth Cys (the Cys in the proximal pair) [43]. The third Cys in Tim9 appears to be more important than the fourth Cys for Mia40 release [37]. The distal Cys disulfide pair is important for assembly of the Tim hexameric complex but doesn’t appear essential for the release of Tim from the Mia40 import complex [43].

Erv1 is a flavin-containing sulfhydryl oxidase that generates disulfides in Mia40 for IMS protein import [44] (Fig. 3). Erv1 has a limited substrate specificity, and Mia40 is its only known in vivo substrate [45]. Erv1 has a redox-active Cx₂C motif close to the FAD that forms an initial disulfide [46]. Through disulfide exchange, the disulfide is likely transferred to an N-terminal Cx₂C motif for transfer to its target Mia40 [47]. In addition, Erv1 appears to contain a structural disulfide in the flavin domain. In the absence of the N-terminal Cx₂C motif, Erv1 can induce disulfide bond formation in small molecules, but this truncate is nonfunctional in vivo [47]. All three cysteine pairs are essential for normal function that also includes a role in cytosolic Fe/S cluster biogenesis [48]. Oxidants for Erv1 include cytochrome c and molecular oxygen [49]. Sulfhydryl oxidases that use oxygen as the electron acceptor generate one hydrogen peroxide for every disulfide bond formed [50]. Thus, oxidative folding may contribute to reactive oxygen species.

Figure 3. Role of the MIA machinery in the import of IMS proteins.

(A). Mia40 captures cysteinyl proteins imported by the TOM complex in a transient intermolecular disulfide. The Erv1 sulfhydryl oxidase generates the active disulfide in Mia40 used to capture the imported protein. Disulfide exchange reactions resulting in intramolecular disulfide bond formation within the imported protein releases it from Mia40. (B). Import and maintenance of twin Cx₉C motif proteins require the same MIA import machinery and the ill-defined role of downstream Pet191.

Protein substrates of the MIA pathway identified to date include the Tim proteins with a conserved twin Cx₃C motif, Erv1 with a Cx₂C motif and a series of proteins with twin Cx₉C motifs [51]. The transient capture of Tim9/Tim10 proteins by Mia40 was recently shown to be specific [42], suggesting that recognition may involve more than an exposed thiolate. Within the class of twin Cx₉C motif protein are the IMS proteins Cox17, Cox19, Cox23, Mdm35, Mic14 and Mic17 (Fig. 4). Import of Sod1 may be MIA dependent as it is absent in the IMS of erv1–1ts cells cultured at the non-permissive temperature [52]. Sod1 has only two conserved cysteinyl residues separated by 88 residues. However, the import of Sod1 is dependent on its chaperone Ccs1 that itself has Cx₂C and CxC motifs. Ccs1 may be imported into the IMS through one or both of those conserved motifs and the apparent MIA-dependency of Sod1 import may arise from an indirect effect through Ccs1 [53].

Figure 4. Mitochondrial IMS proteins with the twin Cx₉C motifs.

Schematic representation of the yeast IMS proteins possessing twin or quadruple Cx₉C motif. All conserved cysteines and their relative spacing are shown in red. Positions of the Cx₉C motifs are indicated.

The presence of a twin Cx₉C sequence motif does not ensure that import is MIA-mediated, since the CcO assembly protein Pet191 has a twin Cx₉C motifs and is imported in a Mia40-independent manner [53]. Likewise, Mia40 has a twin Cx₉C motif yet is imported through the classical mitochondrial presequence pathway.

Proteins transiently trapped by Mia40 during import are released by disulfide exchange reactions resulting in disulfides in the imported proteins. It is not clear whether the oxidative folding of Tim proteins is a paradigm for the twin Cx₉C motif class of IMS proteins. For proteins with multiple disulfide bonds, Erv1 may have a role for disulfide bond formation analogous to its oxidative role on Mia40. Alternatively, an additional mechanism may generate the other disulfide linkages.

5. Function of cysteinyl CcO assembly factors within the IMS

One class of CcO assembly factors within the IMS consists of soluble proteins with a conserved twin Cx₉C motif. These include Cox17, Cox19, Cox23 and Pet191 (Fig. 4). Three other members Mic14, Mic17 and Mdm35 have no defined function in CcO biogenesis. In addition, one subunit of CcO, Cox12, projecting into the IMS contains a related twin Cx₉C structural motif. The four cysteines in Cox12 are present in two disulfides stabilized in a helical hairpin conformation (Fig. 2). The structure of Cox17, like Cox12, exist in a helical hairpin [54, 55] (Fig. 5A). A second class of cysteinyl proteins important for CcO biogenesis includes the Sco protein family and Cox11. These proteins are tethered to the IM.

Figure 5. Structures of Cox17 and Sco1.

(A). Solution structure of the oxidized form of apo-Cox17. The four conserved Cys residues of the twin Cx₉C motif are shown in disulfide linkage in red. Two additional Cys residues participating in Cu(I) ligation are also shown. (B). Structure of the C-terminal globular domain of human Sco1 with a bound Cu(I) ion (colored in cyan). The two liganding Cys and His residues are shown in red and brown, respectively. Although the site appears on the surface, the coordinated Cu(I) is solvent shielded.

5a. Cox17

Cox17 is a soluble copper metallochaperone within the IMS acting as a Cu(I) donor to two accessory proteins Sco1 and Cox11 implicated in the copper metallation of the Cu_A and Cu_B sites of CcO, respectively [56] (Fig. 6). The importance of Cox17 in CcO assembly is highlighted by the embryonic lethality of embryos homozygous for COX17 disruption [57]. Cox17 does not form a stable interaction with either Sco1 or Cox11, yet appears to use distinct interfaces to transfer Cu(I) to each target protein [56]. The C⁵⁷Y Cox17 mutant is capable of Cu(I) transfer to Cox11, but not Sco1 [56]. One known conformer of Cu-Cox17 is a helical hairpin monomer stabilized by two disulfide bonds with Cys residues in the twin Cx₉C motif [54, 55]. Cox17 molecules have two additional conserved Cys residues present just upstream of the first Cys of the twin Cx₉C motif. Those three Cys residues in a C²³CxC²⁶ sequence motif are essential for in vivo function, only Cys²⁶ is part of the twin Cx₉C motif [58]. The single copper is digonally coordinated by Cys²³ and Cys²⁶ or Cys²³ and Cys²⁴. The redox couple of the double disulfide configuration to the fully reduced state has a midpoint potential of −340 mV consistent with the dual disulfide molecule being a likely species in vivo [59]. However, a mutant form of Cox17 lacking the remaining three conserved twin Cx₉C motif cysteines is functional, suggesting that Cox17 is functional without either of the two disulfides in the twin Cx₉C structural motif. Cu(I) coordination is, therefore, not dependent on the disulfide-bonded helical hairpin configuration.

Figure 6. Copper metallation of CcO subunits.

Formation of the Cu_A and Cu_B sites in Cox2 and Cox1, respectively is mediated by chaperone proteins. Cox17 is the Cu(I) donor to both Sco1 and Cox11 for subsequent transfer to Cox2 and Cox1, respectively. Cox17 can exist in two Cu(I) binding states, a monomeric monocopper configuration and a polycopper oligomeric conformer. The significance of the two Cu(I) conformers of Cox17 is unknown.

The metal-free conformer is also a disulfide-bonded helical hairpin. The apo-molecule is capable of forming three disulfides, yet the midpoint redox potential of redox couple of the triple disulfide form to the double disulfide form is −197 mV suggesting that the triple disulfide form of Cox17 probably doesn’t stably exist in vivo [59].

A second Cu(I) conformer is an oligomeric protein complex containing reduced thiolates that is capable of binding a polycopperthiolate cluster [60, 61]. The polycopper protein is in a dimer/tetramer equilibrium, with the polycopper cluster likely existing at the dimer interface [60]. The tetracopper cluster conformer necessitates that multiple cysteine residues are in the reduced thiolate state. An unresolved major question is whether the physiological state of Cox17 is monomeric or oligomeric and what the Cu(I) binding stoichiometry is within the IMS. Cox17 purified from the IMS is largely in the copper-free state (unpublished observation).

The mononuclear Cu(I) conformer was identified by in vitro Cu(I) titration studies, whereas the polycopper conformer was identified from recombinant Cox17 purified from E. coli cultured in copper-supplemented medium. In the absence of copper supplementation, the recombinant protein is devoid of bound Cu(I), but the E. coli cytoplasm has essentially no free Cu(I). The K_d value for the Cu₁Cox17 was reported to be between 10⁻⁶–10⁻⁷ M, as measured by isothermal titration calorimetry (ITC) [54]. In contrast, the dissociation constant for the tetracopper Cox17 complex was determined to be 1 × 10⁻¹⁴ M [61]. Although the two complexes appear to differ markedly in Cu(I) affinity, it is likely that the ITC determination is an underestimate.

5b. Cox19 and Cox23

Two additional twin Cx₉C proteins exist in the IMS that are relevant to CcO assembly. Cox19 and Cox23 have conserved twin Cx₉C structural motifs and resemble Cox17 in being soluble proteins, although both Cox17 and Cox19 are functional when tethered to the IM. Cells lacking either Cox19 or Cox23 are respiratory deficient and have diminished CcO activity. Cox19 resembles Cox17 in its ability to coordinate Cu(I) [62]. Cysteinyl residues in Cox19 are important for Cu(I) binding as well for the in vivo function of Cox19 in CcO assembly. Cox19 isolated from the IMS contains titratable thiolates, suggesting the protein is largely in the reduced state. The correlation of its ability to bind Cu(I) and in vivo function suggests redox control of cysteines in Cox19 is important for its function [62]. If Cox19 folds into a helical hairpin analogous to Cox17; the mutational analysis of Cox19 reveals that, as with Cox17, the protein is functional without disulfide bonding. Mutations of cysteinyl codons that would be expected to make a disulfide in a helical hairpin do not abrogate function. Little is known about the function of Cox23 or its redox state.

5c. Pet191

Another twin Cx₉C motif protein localized within the IMS is Pet191. The motif in Pet191 is a variant from the motif in Cox17, Cox19 or Cox23 in that the twin motifs are separated by 22–30 residues in Pet191 molecules from different species, unlike the short 9–11 linkers in Cox17, Cox19 and Cox23. In addition, two additional conserved Cys residues exist in the candidate Pet191 linker, assuming it also folds in a helical hairpin. Yeast cells lacking Pet191 are respiratory deficient and have a specific defect in CcO assembly [63]. Steady state levels of Cox1, Cox2 and Cox3 are markedly diminished. The respiratory deficiency of pet191Δ cells correlates with the attenuation in the IMS protein level of Cox17, Cox19 and Cox23 [53]. Although the three Cox proteins are imported into the mitochondrion by the MIA pathway, the MIA import pathway remains functional in pet191Δ cells, as the import of other MIA targets including Tim13, Sod1 and Ccs1 is normal [53]. Pet191 does not have a specialized role in MIA-dependent import of the Cox proteins, as in vitro import of radiolabeled Cox19 is normal in pet191Δ cells [53]. Pet191 is tightly associated with a membrane, presumably the IM, and is facing the IMS. We are currently assessing the role of Pet191 in the maintenance of twin Cx₉C motif proteins within the IMS.

5d. Sco1/Sco2

Sco1 is implicated in formation of the mixed valent Cu_A site in Cox2. Yeast lacking Sco1 are devoid of CcO activity and show greatly attenuated Cox2 protein levels [64, 65]. Human cells have two functional Sco molecules that are required for viability [66]. Yeast have a second Sco, Sco2, but cells lacking Sco2 are not respiratory deficient. Mutations in either hSCO1 or hSCO2 lead to decreased CcO activity and early death. Patients with mutations in SCO2 have a clinical presentation distinct from that of SCO1 patients [67]. SCO2 patients present with neonatal encephalocardiomyopathy, whereas SCO1 patients exhibit neonatal hepatic failure. The distinctive clinical presentation is not a result of tissue-specific expression of the two genes, as SCO1 and SCO2 are ubiquitously expressed and exhibit a similar expression pattern in different human tissues. Studies with immortalized fibroblasts from SCO1 and SCO2 patients suggest that Sco1 and Sco2 have non-overlapping but cooperative functions in CcO assembly [66].

Sco1 and Sco2 localize to the IM and are tethered by a single transmembrane helix. A globular domain of each exhibiting a thioredoxin fold protrudes into the IMS [68, 69, 76]. A single Cu(I) binding site exists within the globular domain of Sco1 and Sco2 consisting of two cysteinyl residues within a Cx₃C motif and a conserved histidyl residue. Mutation of the Cys or His residues abrogates Cu(I) binding and leads to a non-functional CcO complex [68, 69]. The structure predicts that the single Cu(I) ion coordinated to Sco1 is solvent-exposed and poised for a ligand exchange transfer reaction (Fig. 5B). The structures of the metal-free human Sco1 and Cu₁Sco1 complex are similar with only one loop showing significant rearrangements [70] (loop marked by arrow in Fig. 5B). The movement of this loop orients the Cu(I) binding His residue in the proper orientation for metal binding. Although the Cu binding site is somewhat disordered in the apo-conformation, the site is largely preformed poised for Cu(I) binding [71–73]. The structural dynamics of loop 8 suggests it may be important interface for interactions with Cox17 and/or Cox2 [70]. In addition to binding Cu(I), Sco proteins bind Cu(II) [74]. The Cu(II) site has a higher coordination number than the three-coordinate Cu(I) site [75]. It is not clear whether Sco1 transfers both Cu(I) and Cu(II) ions to build the mixed valent, binuclear Cu_A site in Cox2. The human Sco2 conformer resembles human Sco1, although Sco2 shows greater conformational dynamics [76]. It remains unclear whether human Sco2 participates in Cu(I) transfer reactions during CcO assembly.

Cu(I) binding to Cox17 and Sco1 necessitates that the Cu(I)-binding cysteines be maintained in the reduced state during the Cu(I) transfer reactions. Mutations that alter the redox state or Cu(I) binding capacity of either protein are expected to attenuate CcO assembly. Although no respiratory deficient mutations in human COX17 have been described, missense mutations identified in SCO1 (P174L) and SCO2 patients (E140K and S240F) are located near the conserved Cx₃C and essential His residues, suggesting that the loss of function in both proteins may relate to aberrant copper binding. However, the P174L substitution in hSco1 does not affect the ability of the protein to bind and retain Cu(I) or Cu(II) [77], yet its function in vivo is severely compromised as evidenced by the pronounced CcO assembly defect in SCO1 patient tissues and fibroblasts [66, 78]. The molecular defect in the P174L mutant Sco1 is an impaired ability to be copper metallated by Cox17 using both in vitro and in vivo assays [77, 79]. The defect is attributed to either an attenuated interaction with Cox17 [77], or to an attenuated Cu(I) binding affinity due to a structural defect [79]. Defective Cox17-mediated copper metallation of Sco1, and subsequent failure of Cu_A site maturation, is the basis for the inefficient assembly of the CcO complex in SCO1 patient fibroblasts.

Cu ions associated to Sco1 are postulated to be delivered to Cox2 thereby forming the binuclear Cu_A (Fig. 6). The Cu_A site in Cox2 is formed within a ten-stranded β-barrel and two cysteine residues within a Cx₃C motif bridging the two Cu ions [80]. The domain of Cox2 containing the Cu_A site protrudes into the IMS with the Cu_A site 8 Å above the membrane surface [34]. Cu_A site formation requires the cysteines to be reduced thiolates. If the two copper sites are filled sequentially, it is conceivable that the Cys residues within the Cx₃C motif in Cox2 may be involved in ligand exchange reactions to move Cu(I) from Sco1 to Cox2. An initial Cu(I) coordination in Cox2 may be a distorted two-coodinate complex involved the two Cys residues prior to rearrangements to form the final coordination sites.

The thioredoxin-like fold of Sco1 suggested that Sco1 may have a redox function, perhaps as a thiol:disulfide oxidoreductase to maintain the Cu_A site cysteines in the reduced state ready for metallation [71, 72]. Alternatively, Sco1 may function as a redox switch, in which oxidation of the two Cys residues in the Cu(I) binding Cx₃C motif may faciliate Cu(I) transfer to Cox2 [73]. In support of a redox role for Sco1, sco1Δ yeast cells were observed to be sensitive to hydrogen peroxide [71]. Subsequently, we showed that the hydrogen peroxide sensitivity of sco1Δ cells arises from the transient accumulation of a pro-oxidant heme a₃:Cox1 stalled intermediate and not an oxidoreductase function [81] (Fig. 7). Heme a/a₃ insertion occurs in Cox1 prior to the addition of Cox2 in the biogenesis of CcO. A partially accessible channel exists on the IMS side of Cox1 through which hemes a/a₃ may be inserted. The candidate channel is sterically blocked upon insertion of Cox2. The globular domain of Cox2 packs onto Cox1 occluding accessibility to the hemes. Peroxide sensitivity is also observed in cox2Δ cells as well as cox20Δ cells that fail to process Cox2. The lack of Cox2 or the inability to fold or mature Cox2 results in varying extents of peroxide sensitivity from accumulation of the pro-oxidant Cox1 intermediate [81]. Thus, the structural similarity of Sco1 to thioredoxin may not necessarily imply a redox function for Sco1. If oxidation of Cys residues within Sco1 facilitates Cu(I) transfer, further evidence is needed to support this postulate.

Figure 7. A pro-oxidant intermediate in the CcO assembly pathway.

A pro-oxidant intermediate consisting of heme A₃-containing Cox1 accumulates when the assembly pathway is blocked subsequent to Cox1 maturation. The deleterious effects of the intermediate can be suppressed by overexpression of Cox11 or Sco1 suggesting that these molecules can cap the Cox1 intermediate thereby precluding peroxide access to the heme A₃ catalytic center. Formation of the mature core of CcO with Cox2 inserted on Cox1 resolves the potentially deleterious Cox1 intermediate.

The thioredoxin family of proteins often contains a conserved cis-Pro in juxtaposition to the redox active cysteinyl residues. The cis-Pro was shown recently to preclude metal ion binding [82]. In the absence of the Pro residue, thioredoxin proteins are poised for metal ion binding either Fe/S cluster or Cu(I)/Zn(II) ions. The corresponding position in Sco1 is the Cu(I) ligand His, consistent with its Cu(I) binding function. The replacement of the Pro in human thioredoxin with a His residue enables the protein to bind Zn(II) or Cu(I) [82]. Thus, the presence of the conserved His in Sco proteins is consistent with the postulated role of Sco proteins in Cu(I) transfer.

5e. Cox11

Formation of the Cu_B site in Cox1 requires Cox11 [83]. CcO isolated from Rhodobacter sphaeroides cox11 Δ cells lacked Cu_B but contained both hemes, however the environment of heme a₃ was altered [83]. S. cerevisiae cells lacking Cox11 have impaired CcO activity and have lower levels of Cox1 [84]. Cox11 is tethered to the IM by a single transmembrane helix with a C-terminal domain protruding into the IMS [85, 86]. The Cu(I) binding cysteinyl residues lie within this C-terminal domain. The cysteinyl residues in this C-terminal domain that are important for both Cu(I)-binding correlate with in vivo function, such that when mutated abrogate Cu(I) binding and CcO activity. Removal of the transmembrane domain yields a soluble protein that dimerizes upon Cu(I) binding [85]. The Cu(I) sites in each monomer are closely juxtaposed as the dimeric complexes can form a binuclear Cu(I) thiolate cluster at the dimer interface [85].

Cu(I) transfer from Cox11 to Cox1 forming the Cu_B site that lies 13 Å below the membrane surface appears to occur in nascent Cox1 chains during its insertion and folding within the IM [86, 87]. The Cu_B site is a heterobimetallic site with heme a₃, so Cu_B site formation is likely concurrent with heme a₃ insertion. Heme a₃ insertion requires Shy1 [88]. Mammalian cells containing a mutant Shy1 (SURF1) accumulate a Cox1, Cox4 and Cox5A subcomplex suggesting that heme a₃ insertion occurs in the subcomplex. The Cu_B site in Cox1 consists of histidyl ligands, whereas the Cu(I) site in Cox11 consists of cysteinyl residues. Thus, the redox state of Cox11 must be maintained within the IMS to ensure successful Cu(I) transfer.

6. Outlook

Redox control of CcO assembly within the IMS is an important process. Maintanence of reduced thiolates in Cox17, Sco1, Sco2, Cox2 and Cox11 is important for the proposed copper metallation of Cu_A and Cu_B sites. It remains to be established whether disulfide bond formation in Cox17 or Sco1 is important to direct Cu(I) to the Cu_A site in Cox2. The situation may be more complex if the initial report that the IMS Sod1 is in a dynamic redox equilibrium [23]. If the Sod1 postulate is substantiated, the possibility exists that CcO assembly factors may also exist in redox equilibrium that serves a regulatory role in CcO biogenesis. Further work is needed to identify the mechanism of disulfide bond formation within the IMS independent of the role of Erv1 in redox control of Mia40. In addition, the identity and regulation of disulfide reductants needs to be addressed. It remains to be determined whether the redox potential of the boundary IMS or cristae lumen differs from that of the cytoplasm or matrix. A redox proteomic study of the two compartments would be insightful.