Mitochondrial Structure and Bioenergetics in Normal and Disease Conditions

The mature forms of SDHA and SDHB are generated independently before the complex assembly, while SDHC and SDHD are able to form an intermediate subcomplex [182]. SDHA is initially imported into the matrix as an apo-protein and the FAD cofactor is inserted thanks to the interaction with the assembly factor SDHAF2/Sdh5 [183]. Then, mature SDHA binds to SDHAF4/Sdh8, a chaperone that protects the subunit from auto-oxidation and facilitates the assembly with SDHB. Mature SDHB contains Fe/S clusters, which are inserted by SDHAF1 [184,185]. SDHB stability is then maintained by the association with an LYR motif protein recently identified in yeast, Sdh7 (SDHAF3/ACN9/LYRM10, human ortholog), which shields one or more of the prosthetic centers from oxidants [186]. Mature SDHA and SDHB are then able to assemble together and join SDHC and SDHD and insert into the membrane via a still uncharacterized mechanism.

Patients presenting with a specific CII defect are quite rare, less than 10% of OXPHOS deficiency cases [187]. Two main phenotypes can originate from mutations in CII subunits or assembly factors. Mutations in SDHAF1 and SDHA lead to encephalomyopathy and leukoencephalopathy [184,188], while variants in SDHA, SDHB, SDHC, SDHD and SDHAF2 are responsible for hereditary paraganglioma and pheochromocytomas, rare neuroendocrine tumors [183,189,190,191,192]. Moreover, other genes involved in FAD and Fe/S cluster synthesis can impair CII activity and stability [193]. The main pathological mutations found in CII subunits or assembly factors are summarized in Table 3.

The ubiquinol:cytochrome c oxidoreductase (cytochrome bc1, complex III, CIII) is the central element of the respiratory chain. In yeast, it is formed of 10 different subunits, while in mammals, an additional subunit was identified, corresponding to the mitochondrial targeting sequence of the Rieske protein/UQCRFS1, which remains anchored to the complex after the proteolytic cleavage [200]. However, recent studies proposed that the latter is not a stoichiometric subunit and that the N-terminal UQCRFS1 peptide needs to be eliminated in order to maintain the functionality of CIII [201,202]. All CIII subunits are encoded by nuclear DNA except cytochrome b (MTCYB), which is mitochondrial-encoded [92,203]. CIII is always dimeric and high-resolution crystal structures of the bovine, chicken and yeast bc1 complexes have been resolved [91,204,205,206] (Figure 10).

Both in yeast and mammals, CIII contains three protein subunits with redox prosthetic groups: cytochrome b, which contains both the high-potential bH (b562) and the lower-potential bL (b565) heme centers, cytochrome c1 (CYC1), containing the c-type heme c1, and the Rieske iron-sulfur protein (Rip1 in yeast, UQCRFS1 protein in mammals) with a 2Fe–2S cluster [208]. The di-heme cytochrome b polypeptide forms eight transmembrane helices and contains two histidine residues in each of the second (helix B) and fourth (helix D), forming the binding site for quinone [209]. The low-potential heme bL is located on the intermembrane space side of the IMM, while the high-potential heme bH is positioned in a cavity accessible from the matrix, where it can receive electrons from bL and pass them to the Qi site, where it reduces one bound ubiquinone to semiquinone [208]. CYC1 has a wedge-like structure containing the heme group and is anchored in the membrane through a C-terminal transmembrane anchor next to helix E of cytochrome b [93]. The heme group binds a CXXCH domain, highly conserved in c-type cytochromes. UQCRFS1 contacts MTCYB on one of the two CIII heterodecamers with its N-terminal transmembrane domain, where it receives one electron and undergoes a conformational change that makes it reach CYC1 on the other one [210]. The maturation of UQCRFS1 has been studied in detail in simpler organisms, such as Neurospora crassa and Saccharomyces cerevisiae [211]. The newly synthesized protein undergoes two post-translational modifications: the cleavage of a targeting pre-sequence and the insertion of the iron-sulfur cluster into the mitochondrial matrix. Initially, the MTS is cleaved by a mitochondrial matrix protease (MPP), and finally, an extra eight-amino acid-long sequence is removed by a mitochondrial intermediate protease (MIP).

Contrary to yeast and birds, mammalian UQCRFS1 maturation generates a 78-amino acid-long fragment, which remains temporally bound to CIII as an eleventh subunit, Subunit 9 (Su9) [200]. This additional subunit localizes between the two core subunits UQCRC1 and UQCRC2 and it has been proposed that these two are responsible for the cleavage of UQCRFS1, due to the conservation of their MPP function [212,213,214].

The remaining subunits are accessory and their function is mainly to support and stabilize the complex [215,216].

The CIII assembly pathway has been studied in depth in S. cerevisiae [217,218,219,220,221], while the human CIII assembly model has been initially constructed by homology, as some of the steps have been shown to be analogous to yeast [222], and recently updated following our study on CIII-deficient cybrids [223].

The first step of CIII assembly, both in yeast and in mammals, is the synthesis and the insertion into the IMM of cytochrome b. Yeast cytochrome b contains introns and requires processing [224], while mammalian MTCYB is transcribed as a polycistronic segment. MtDNA is organized so that mRNAs coding for proteins are divided by tRNAs, which assume a specific secondary structure, and function as punctuation marks between the genes. tRNAs are then cleaved by mitochondrial RNase P at the 5′ ends and by RNase Z at the 3′ ends, and mRNAs are then translated [225]. Moreover, studies in mice suggested the involvement of PTCD2 (pentatricopeptide repeat domain protein 2) in processing the pre-processed ND5-CYTB RNA transcript [226].

The transcription and translation of cytochrome b must be coordinated with the synthesis of nuclear-encoded proteins. This mitochondrial–nuclear communication is possible thanks to a group of proteins called translational activators. These nuclear-encoded factors regulate the expression of mitochondrial genes and their own expression in relation to the OXPHOS activity, in order to limit the accumulation of unused subunits, which can have toxic effects on the organelle. In yeast, four translational activators of COB have been identified: Cbp1, Cbs1, Cbs2 and the complex Cbp3/Cbp6 [220]. These factors interact with mitochondrial ribosomes and the mitochondrial organization of gene expression (MIOREX) complex [227]. The primary role of Cbp1 is to protect COB mRNAs and to transfer them to the translational apparatus [228]. Cbs1 and Cbs2 have been found associated in the same high-molecular weight complex with mitochondrial ribosomes, but they might also form other subcomplexes including those with different activator proteins, such as COX-specific activators [229,230]. Recent studies on Cbs1 determined that this protein binds to a segment of the 5′ UTR of COB mRNA, sequestering it and repressing the translation. Cbs1 is then replaced by the complex Cbp3/Cbp6 liberated during assembly, which activates the translation [231]. These proteins, necessary for the stability of COB mRNA and its translation, do not have orthologs in mammals [220].

The Cbp3-Cbp6 complex, instead, plays a role in the second phase of translation, binding the nascent polypeptide exiting the ribosome, and it does not leave the protein until the incorporation of the bL heme group [232,233]. This complex has orthologs in mammals named ubiquinol-cytochrome c reductase complex assembly factors 1 and 2 (UQCC1 and UQCC2), with the same function [234]. Cbp3 interacts directly with Cbp4 (human ortholog: UQCC3), an assembly factor anchored in the IMM and protruding into the intermembrane space. Cbp4 is not necessary for complete translation and release from ribosomes of cytochrome b, but it has a role in the stabilization of the semihemylated intermediate that contains bL [232,233]. Together, Cbp3–Cbp6, Cbp4 and cytochrome b compose intermediate I. In yeast, the now mature cytochrome b forms a subcomplex with the subunits Qcr7 and Qcr8, called intermediate II. Deletion of any of the genes encoding cytochrome b, Qcr7 or Qcr8 leads to the almost complete loss of the other two subunits and Qcr6, while the other subunits are only partially reduced [235]. Similarly, the mammalian orthologs UQCRB and UQCRQ are incorporated in the early stages of CIII assembly, provoking the release of the UQCC1-UQCC2 complex.

The following steps of CIII assembly, instead, have been proposed to differ between yeasts and mammals. The order of incorporation in S. cerevisiae was determined by generating yeast mutants for single CIII subunits and studying the stability of the remaining components of the complex [217,218,219]. The third intermediate step involves the insertion of four subunits: Qcr6 (UQCRH in humans), the two large structural core subunits Cor1 and Cor2 (UQCRC1 and UQCRC2), and cytochrome c1 (Cyt1). At this stage, dimerization occurs by joining assembly intermediate II and the Cor1/2 modules [236]. Interestingly, Cor1, Cor2 and cytochrome c1 were found associated in a subassembly module, even if the contacts between the core proteins and the catalytic subunits in the fully assembled enzyme are minimal [217]. Surprisingly, the complex Cor1/Cor2 was detected in various subcomplexes in two-dimensional electrophoresis. This behavior might be due to the association of these subunits with other proteins or ETC complexes in the IMM or to the formation of Cor1/Cor2 aggregates. However, these subassemblies were detectable only in mutant strains and disappeared when the complex was assembled correctly; therefore, they might not represent a physiological intermediate. The last assembly factor that might have a role in the early or intermediate phases of CIII assembly is Bca1, an inner membrane protein found only in fungi [237]. However, its function is not clear yet.

Taking advantage of CIII-deficient transmitochondrial cybrids carrying a mutation in MTCYB, however, we recently highlighted two important differences between these steps of CIII biogenesis in yeasts and in humans [223]. Firstly, we observed the formation of subassemblies containing CYC1, UQCR10 and potentially UQCRH, while our data did not suggest any interaction of CYC1 with the core subunits UQCRC1 and UQCRC2. Secondly, we identified CIV subunits, mainly belonging to the MTCO2 module, consistently interacting with this intermediate in the CIII-mutant cells. These results suggest that CIII might use CIV or CIV modules as a structural scaffold in a physiological context, or sequester CIV-specific subunits or intermediates when supercomplex formation is impaired, as a control mechanism to inhibit the complete biogenesis of the enzyme. The yeasts-based and the updated CIII assembly models are shown in Figure 11.

A requirement for the generation of intermediate III is the synthesis, the import and the maturation of Cyt1. Cyt1 contains a single heme center and is anchored to the IMM via a single transmembrane segment near its C-terminus [91]. The precursor of this subunit is translated in the cytosol and transported through TOM and TIM complexes into the mitochondria. The cytochrome c1 precursor protein contains an N-terminal cleavable bipartite pre-sequence [238]. The first of the two independent sequences, a strongly basic region of 35 amino acids, is a mitochondrial targeting signal and it is proteolytically removed by MPP in the matrix. The second sequence is a hydrophobic sorting sequence, which targets Cyt1 to the IMM. However, two models have been proposed to explain this process. The first one proposes that the whole protein, and therefore both the targeting sequences, reach the mitochondrial matrix and that only later is the second sequence re-located into the membrane, allowing the proteolytic cleavages [239]. Instead, in the second model, only the first segment reaches the mitochondrial matrix, while the second internal hydrophobic sequence remains anchored in the membrane, stopping the import. In the matrix, the positive-charged mitochondrial targeting sequence is cleaved by MPP. At the same time, the C-terminal alpha-helix gets inserted into the membrane and the heme center is attached to the protein. The hemylation is mediated by holocytochrome c1 synthetase (Cyt2 or HCCS1 in mammals) [240]. This modification provokes a conformational change that allows the exposure and the cleavage of the second targeting sequence by Imp2 (inner membrane peptidase 2), leaving the N-terminus of the mature Cyt1 soluble in the intermembrane space [241].

The late assembly stages involve the incorporation of Qcr9 (mammalian UQCR10) in yeasts, and Qcr10 (UQCR11) and Rip1 (UQCRFS1) both in yeasts and mammals. Firstly, Qcr9 is inserted [242]. This small (7.3 kDa) accessory subunit is necessary for the functionality of the complex and its deletion results in the formation of a nearly inactive enzyme. Indeed, it has been observed that, lacking Qcr9, the conformation of Rip1 is altered and the Fe/S cluster is not incorporated correctly [243]. The last two proteins to be inserted are Qcr10 and Rip1. Qcr10 is an 8.5 kDa supernumerary subunit incorporated before Rip1 and required for its stabilization. However, it is not clear yet how and when it is recruited [244]. Many studies, instead, have been published about the maturation and insertion of the Rieske protein both in yeast and mammals.

Prior to the insertion, Rip1 is imported into mitochondria and receives its 2Fe/2S center, likely by the resident iron-sulfur cluster (ISC) system. In yeast, the import is followed by two proteolytic steps that eliminate the N-terminal MTS. The precursor form is first processed by a matrix MPP protease into an intermediate form [245]. The second cleavage generates the mature form of the protein and is catalyzed by the mitochondrial intermediate peptidase (MIP). At this point, Rip1 is transported back across the IMM into the intermembrane space, where it is assembled in the complex. In mammals, however, the UQCRFS1 N-terminal import signal is cleaved in a single step when the protein is already incorporated in the complex and the cleaved segment remains attached to the enzyme [200].

Two assembly factors are necessary for the Rieske protein assembly in both mammals and yeast: Bcs1 (BCS1L in mammals) and Mzm1 (LYRM7). Bcs1 is a 456 amino acid protein formed by three different domains: a positively charged 126 amino acid N-terminal targeting signal [246], a central Bcs1p-specific domain and a highly conserved C-terminal AAA-ATPase domain [247]. It has been initially proposed that Bcs1p might have a role in Fe/S cluster insertion, act as a chaperone [248] or bind to the partially formed CIII in an ATP-dependent manner, keeping it in a state that allows the incorporation of the Rieske protein [249]. The most recent theory is that Bcs1 is responsible for the export of the Rieske Fe/S domain from the matrix into the intermembrane space [250]. Bcs1, indeed, is able to recognize the correctly folded Rieske protein and act as a protein translocase. This model has been confirmed by the determination of the cryogenic electron microscopy (cryo-EM) structure of Bcs1 in yeast [251] and mice [252], which suggested an airlock-like mechanism for Rip1/UQCRFS1 translocation. Bcs1, indeed, seems to form two large aqueous vestibules, a bigger one on the matrix side and a smaller one in the inner membrane, through which the Rieske protein is transported.

Mzm1 is a 14 kDa protein located in the mitochondrial matrix. It was initially thought to be involved in the modulation of the zinc pool and to this function it owes its name (mitochondrial zinc maintenance 1) [253]. However, in addition to the reduced zinc pool, cells lacking Mzm1 have a defect in CIII due to faulty insertion of Rip1 [254]. Its role is to stabilize Rip1 in the matrix before the translocation to the IMM. The same function is shared by the human ortholog, LYRM7 or MZM1L [255].

Finally, the third factor necessary for UQCRFS1 metabolism, which does not have a yeast ortholog, is TTC19 [256]. TTC19 binds to CIII after the incorporation of UQCRFS1 and is involved in the clearance of UQCRFS1 fragments, a process that is necessary to keep the complex in a functionally competent state [201]. A complete list of CIII subunits is indicated in Table 4 and the latest model of the CIII biogenesis pathway is represented in Figure 11.

Pathologies due to deficiencies in CIII activity are relatively infrequent and most of them derive from mutations in MTCYB, the only mtDNA-encoded subunit. Mutations in this protein are generally associated with myopathy and exercise intolerance [193]. Defects in nuclear-encoded subunits are rarer, but a handful of mutations have been found in several patients (Table 5). The majority of the pathological variants associated with mitochondrial CIII deficiency of nuclear origin are found in BCS1L [273]. The genes found mutated in patients with CIII deficiency and the relative clinical phenotypes are summarized in Table 5.

Cytochrome c oxidase (COX, complex IV, CIV) is the terminal step of the ETC. The enzyme has a molecular mass of about 200 kDa and in mammals it is composed of 13 subunits, 10 nuclear-encoded and 3 encoded by the mtDNA (MTCO1, MTCO2 and MTCO3), which form the functional core of the complex [296] (Figure 12). However, recently a 14th subunit, NDUFA4, previously attributed to CI, has been described [297,298] and was found to be incorporated in the structure of monomeric human CIV [299].

MTCO1 contains three prosthetic groups: cytochrome a3 and CuB, which form the bi-nuclear center that binds oxygen, and cytochrome a. MTCO2 incorporates the CuA center [300] and MTCO3 does not have catalytic activity. The remaining subunits (COX4, 5A, 5B, 6A, 6B, 6C, 7A, 7B, 7C, 8A) are thought to have a structural role in the stabilization of the complex. Interestingly, CIV is the only ETC complex that evolved tissue-, developmental- and species-specific isoforms for COX subunits 4, 6A, 6B, 7A, 7B and 8A [301], probably in order to regulate ATP and energy production in different conditions [302].

CIV assembly is now understood as a modular process. MTCO1 was classically considered the “seed” around which the rest of the complex assembles [304,305]. However, recent evidence indicates that the first CIV subassembly is formed by the association between two nuclear-encoded subunits, COX4I1 and COX5A [306]. This module contains also HIGD1A, a protein initially proposed to be involved in the regulation of CIV activity during hypoxia [307].

In parallel, the MTCO1 module, also known as “MITRAC” (MItochondrial TRanslation Regulation Assembly intermediate of Cytochrome c oxidase) [308,309], is formed, composed of the CIV subunit and a series of assembly factors necessary for its maturation and stabilization. The first level of regulation of this module is translational, with the activity of the mitochondrial RNA-binding protein LRPPRC [310] and the translational activator TACO1 [311]. After translation, the newly synthesized protein has to be inserted in the IMM. The first factors binding MTCO1 are COX14/C12ORF62 and COA3/CCDC56/MITRAC12 [308,312,313,314]. These two chaperones assist MTCO1 during and after its insertion in the membrane and avoid the aggregation of MTCO1 subunits. The transient complex MTCO1/COX14/COA3 is stabilized by another assembly factor, CMC1 [315].

At this point of the pathway, MTCO1 needs the incorporation of the three prosthetic groups. Heme a biosynthesis is carried out by COX10 and COX15 catalyzing the conversion of heme b to heme o, and then heme o to heme a [316,317]. On the other hand, the assembly factor SURF1 has been proposed to participate in its delivery [305]. In addition, PET117 might also have a role in this process because it was found interacting with COX15 in yeast, but its involvement still has to be shown in mammals [318]. CuB incorporation is mediated by the metallochaperone COX11 [319], which is maintained in the correct redox state by COX19 [320]. The coppers are donated by COX17 [321].

MTCO2 requires binding with COX18 in order to be inserted in the IMM [322] and with COX20/FAM36A and TMEM177 for stabilization [323,324]. Secondly, the CuA center must be inserted in MTCO2. This process is regulated by the copper-binding proteins COX17, SCO1 and SCO2 [325,326,327], together with COA6 [328,329] and COX16 [330,331]. The MTCO2 module (MTCO2 + COX5B + COX6C + COX7C + COX8A and, most probably, COX7B) is incorporated in intermediate steps of the assembly process by joining the COX4I1-COX5A and the MTCO1 modules, forming the “S3” intermediary. This intermediary binds three assembly factors, PET100 [332], PET117 [318] and MR-1S [306]. MR-1S is a vertebrate-specific chaperone that interacts with the highly conserved factors PET100 and PET117. An additional assembly factor, APOPT1 or COA8, was proposed to have a role in the intermediate steps of CIV assembly in mouse and human mitochondria (Signes et al., 2019). Finally, the MTCO3 module (MTCO3 + COX6A1 + COX6B1 + COX7A2) is incorporated [306], followed by NDUFA4 initially described as a CI subunit and later assigned to CIV [305]. All known CIV assembly factors are summarized in Table 6.

After CI-related pathologies, defects in CIV are the most common OXPHOS defects associated with mitochondrial disease [193]. The most frequent clinical phenotypes associated with CIV deficiencies are myopathy, affecting the skeletal muscles, and systemic pathologies, such as Leigh’s disease and multiorgan failure (https://rarediseases.org/rare-diseases/cytochrome-c-oxidase-deficiency/). As in most of the mitochondrial diseases, symptoms can be very heterogeneous and the severity of the pathology can vary greatly [354]. While most of the pathological mutations found in patients are associated with assembly factors of the enzyme or with mitochondrial tRNAs (mutations in tRNALys, tRNAAla, mt-tRNAPhe, tRNALeu, tRNATrp, tRNAAsp and tRNAGlu were found associated with COX deficiency [354], only few cases of mutations in CIV structural subunits have been reported. This observation suggests that mutations in CIV subunits might be incompatible with life. The genes found mutated in patients with CIV deficiency and the relative clinical phenotypes are summarized in Table 7.

ATP synthase (Complex V, CV) is the enzyme that catalyzes the synthesis of ATP from ADP and phosphate. It is composed of two distinct domains: the F1 domain, which faces the mitochondrial matrix, and the Fo domain, located in the IMM [390,391]. The human CV is composed of 29 proteins of 18 kinds, only two of which are encoded by the mtDNA (ATP6 and ATP8) [392]. A schematic representation of the main domains of CV is shown in Figure 13. The F1 domain is composed of three copies of subunits α and β, and one copy of subunits γ, δ and ε. γ, δ and ε subunits form the central stalk of the complex, while α and β are the subunits that physically interact with the ADP and ATP molecules [390]. Fo, instead, is composed of a ring of c subunits and one copy each of subunits a, b, d, F6 and the oligomycin sensitivity-conferring protein (OSCP). The c-ring stoichiometry is not constant, but can vary, ranging from 8 copies in vertebrates to 15 in photosynthetic organisms [393].

Each c subunit is able to bind one proton in the IMS, which interacts with a conserved carboxylate group from a glutamate or aspartate side chain. The protonation of these subunits provokes the rotation of the c-ring, until the final dissociation of the proton at the matrix side favored by the positive charge on a conserved arginine residue (A210) of subunit a [56]. The c-ring is structurally linked to the γ and ϵ subunits and its rotation provokes the turn of these subunits inside the α3β3 hexamer unit of F1. On the external side, the α3β3 hexamer is prevented from rotating by the peripheral stalk formed by the two b chains and the d subunit. The result of the proton transport, therefore, is first the rotation of the c-ring, followed by the rotation of the γ subunit, and the consequent synthesis of ATP through the binding change mechanism of α3β3. The binding change mechanism is based on the fact that the interactions between the γ subunit and the three β subunits are not identical. The result is three different conformations for the three β subunits: T (tight), L (loose) and O (open). The subunit in T conformation binds ATP very strongly and its affinity for the molecule is so high that it will induce the conversion of ADP + Pi into ATP. The subunit in the L conformation, instead, is able to bind ADP and Pi but it cannot release the nucleotides. Finally, the O conformation allows the release of the formed ATP. The result of the γ subunit rotation is the change in these subunit conformations allowing the passage through all the stages and the generation of ATP [56].

As for the other complexes described, the CV assembly is also modular. Three sub-assemblies are formed individually and then put together: the F1 module, the c-ring and the peripheral stalk [395]. The F1 subcomplex formation requires the activity of the chaperones ATPAF1/ATP11 and ATPAF2/ATP12, which bind ATP5B and ATP5A1, respectively [396]. Initially, the F1 and the c-ring modules assemble. The peripheral stalk, instead, is incorporated in two additional steps: the incorporation of b/ATP5F1, d/ATPH, F6/ATP5J and OSCP/ATP5O first and the addition of e/ATP5I, g/ATP5L and f/ATPJ2 in a second step [397,398]. Different assembly factors involved in this process have been identified in yeasts. Atp25 stimulates the synthesis and assembly of subunit c of the c-ring [399], while the protease Atp23 processes and stabilizes the membrane-inserted yeast Atp6, unprocessed in mammals [400]. Finally, the inner membrane assembly complex (INAC), composed of Ina17 and Ina22 [401], binds and stabilizes two distinct assembly intermediates of the yeast ATP synthase: the newly assembled c-ring and an assembly intermediate composed of the F1 domain and the peripheral stalk.

Patients presenting with CV defects are rare and generally associated with neonatal-onset hypotonia and hypertrophic cardiomyopathy, lactic acidosis and 3-methylglutaconic aciduria [193]. Only few pathological mutations in CV subunits or assembly factors have been found in patients so far. The majority of these mutations were identified in MT-ATP6, responsible for neurogenic muscle weakness and ataxia and retinitis pigmentosa (NARP) syndrome [402,403], and MT-ATP8 [404], while rarer cases were found in ATP5E, ATP5A1, ATPAF2 and TMEM70 [404,405,406,407,408,409]. The CV subunits or assembly factors found mutated in patients with CV deficiency and the relative clinical phenotypes are summarized in Table 8.