Mitochondrial Cytochrome c Oxidase Deficiency

Abstract

As with other mitochondrial respiratory chain components, marked clinical and genetic heterogeneity is observed in patients with a cytochrome c oxidase deficiency. This constitutes a considerable diagnostic challenge and raises a number of puzzling questions. So far, pathological mutations have been reported in more than 30 genes, in both mitochondrial and nuclear DNA, affecting either structural subunits of the enzyme or proteins involved in its biogenesis. In this review, we discuss the possible causes of the discrepancy between the spectacular advances made in the identification of the molecular bases of cytochrome oxidase deficiency and the lack of any efficient treatment in diseases resulting from such deficiencies. This brings back many unsolved questions related to the frequent delay of clinical manifestation, variable course and severity, and tissue-involvement often associated with these diseases. In this context, we stress the importance to study different models of these diseases, but also discuss the limitations encountered in most available disease models. In the future, with the possible exception of replacement therapy using genes, cells or organs, a better understanding of underlying mechanism(s) of these mitochondrial diseases is presumably required to develop efficient therapy.

1. Cytochrome oxidase, an overview of structure, function, and regulation

Cytochrome oxidase (COX; EC 1.9.3.1) is the unique terminal oxidase of the mitochondrial respiratory chain (RC) in mammals (Fig. 1). COX also referred to as complex IV is made up of thirteen subunits that catalyze the transfer of electrons from ferro-cytochrome c to molecular oxygen. This exergonic reaction is coupled to proton transfer across the inner membrane, which contributes to the electrochemical gradient used for ATP synthesis (Fig. 1A). The electrochemical gradient is also crucial for preserving the capacity of mitochondria to exchange metabolites and ions with the surrounding cytosol and other organelles.

Figure 1. Cytochrome oxidase location in the respiratory chain and activity assay in human skin fibroblasts.

A. Schematic representation of the respiratory chain in the inner mitochondrial membrane showing the interaction of cytochrome oxidase (complex IV) with complexes I and III in a super-complex (respirasome). The site of action of specific inhibitors is indicated in red. The green arrow shows the alternative oxidase (AOX) by-pass, which when expressed in COX-defective human mitochondria or flies rescues their various phenotypes. The assay of COX with externally added cytochrome c requires the permeabilization of the outer membrane. B. Cytochrome oxidase is assayed spectrophotometrically by measuring using a double-wavelength spectrophotometer (550–540 nm) the oxidation of reduced cytochrome c in skin fibroblasts permeabilized by 2 successive freeze/thaw cycles. The reaction is first order with respect to substrate concentration and is thus diminished by half when half of the reduced cytochrome c is consumed. Subsequent sequential addition of rotenone, cyanide, oxidized cytochrome c and succinate measures reduction of cytochrome c, first by the succinate-cytochrome c reductase (CII plus CIII). The activity is essentially rate controlled by CII and can be inhibited by malonate, a competitive inhibitor of CII. Further addition of glycerol-3 phosphate measures the activity from the glycerol-3 phosphate dehydrogenase (G3Pdh) to CIII. This activity can be selectively inhibited by iGP1(143). Finally, addition of decylubiquinol in the presence of EDTA is used to measure antimycin-sensitive CIII activity. Abbreviations: The RC complexes are abbreviated as, CI, CII, CIII, CIV, and the ATP synthase as CV; c, cytochrome c; COX, cytochrome oxidase; Ddh, the dihydroorotate dehydrogenase which catalyze the production of uridine, an essential step for the synthesis of nucleic acids; EDTA, ethylenediamine tetraacetic acid; ETF, the electron transfer flavoprotein involved in the oxidation of fatty acids; G3Pdh, the glycerol 3-phosophate dehydrogenase; GCCR, iGP1-sensitive glycerol 3-phosphate; IM, inner membrane; KCN, potassium cyanide; OM, outer membrane; QCCR, antimycin-sensitive decylubiquinol-cytochrome c reductase; SCCR, malonate-sensitive cytochrome c reductase; UQ, ubiquinone 50, or coenzyme Q₁₀.

The three largest subunits of COX are encoded in mitochondrial DNA (mtDNA). These core subunits contain all the heme and metal prosthetic groups needed for catalysis. The remaining ten subunits are products of nuclear genes that are translated as pre-proteins on cytosolic ribosomes, imported to different compartments of mitochondria by the TIM and TOM transport machineries and possibly modified before entering the COX assembly pathway (1, 2). A large number of factors, sometimes specific to COX assembly and in other cases with broader specificity, are known to facilitate the various steps of COX assembly and its incorporation into the RC super-complexes, also referred to as the respirasome (Fig. 2 and 3).

Figure 2. Synthesis and assembly of COX subunits.

A scheme summarizing what is presently known about the pathways for the integrated synthesis and assembly of COX subunits expressed from the nuclear and mitochondrial genomes. Protein subunits translated on cytosolic ribosomes with N-terminal presequences are first transported by the outer membrane TOM complex and are subsequently matured and sorted by the TIM and MITRAC machineries to the intermembrane space, inner membrane, or matrix where they interact with partner proteins to form assembly intermediates (144–146). Subunits encoded by mtDNA genes are translated on mitochondrial ribosomes attached to the matrix side of the inner membrane. Following insertion into the inner membrane by Oxa1 they interact with their nucleo-cytoplasmic partners to form subcomplexes that subsequently assemble into COX. The overall process is assisted by numerous proteins acting in transport, translation, chaperoning of different assembly steps. Oxa1 is also involved in the biogenesis of other respiratory chain complexes. Some of the genes coding for ancillary factors (indicated in red) have been found to be mutated in COX deficient patients. IM, inner membrane; OM, outer membrane; 1, 2, 4, 5a, COX subunits (purple)

Figure 3. Maturation and insertion of COX into the respiratory chain.

Mammalian COX exists as a dimer. Each monomer consists of 13 different subunits. At present human mutations leading to a COX deficiency have been identified in six structural subunits including the three mtDNA-encoded core proteins (in red) and in 9 ancillary proteins (also indicated in red). The catalytic activity of COX depends on heme a, a₃ and two copper centers (Cu_A and Cu_B) linking COX biosynthesis to both copper and heme metabolism. Maturation of COX active centers involves a number of factors, some of which have also been found mutated in COX deficient patients (denoted in red). IM, inner membrane; OM, outer membrane; 1, 2, 4, 5a, 5c, 6a, 6b, 6c, 7a, 7b, 8, the different COX subunits; CI, CII, CIII, CIV, the various complexes of the respiratory chain

Complex IV of the mammalian RC has been shown to interact with complexes I and III with variable stoichiometry to form the respirasome (3, 4). This organization of the respiratory chain into a respirasome, de facto, isolates kinetic pools of electron carriers, including mobile carriers such as ubiquinone(5), to efficiently channel electrons supplied by various dehydrogenases of mitochondria to the appropriate segments of the RC (6). In the context of mitochondrial diseases, a primary loss of COX may have secondary effects on the organization of the respirasome thereby eliciting more complex biochemical phenotypes (7, 8).

In mammals, the composition of tightly-bound subunits of the COX core is constant. However, the expression of several nuclear-encoded isoforms of one or more imported subunits may vary depending on the developmental stage and the tissue (Table I; Fig. 4). Indeed, extensive SDS-PAGE analyses together with immunological and/or sequencing data by the Kadenbach and Grossman groups have demonstrated the occurrence of several isoforms of the COX4, COX6A, COX6B, COX7A and COX8 (1, 2) subunits that are variably expressed during fetal development in mammals (9) (Table I; Fig. 4). COX activity is regulated by the ATP/ADP ratio, which affects its phosphorylation status (10), and association of COX into dimers, as well as its interactions with cardiolipin (11) and other proteins (2). Finally, beside the thirteen tightly bound subunits observed by the Kadenbach group and later confirmed by the Yoshikawa group by X-ray crystallography (12), a more loosely bound 14th subunit product of the NDUFA4 gene was detected in stoichiometric amounts when analyzed by BN-PAGE (13). This idea gained support from the finding that a mutation in NDUFA4 results in a specific COX deficiency (14).

Table I.

The human cytochrome c oxidase subunits and isoforms

Subunit name and symbol	Alternative names	OMIM n°
CYTOCHROME c OXIDASE SUBUNIT I; MTCO1	COI; COX1	516030
CYTOCHROME c OXIDASE SUBUNIT II; MTCO2	COII; COX2	516040
CYTOCHROME c OXIDASE SUBUNIT III; MTCO3	COIII; COX3	516050
CYTOCHROME c OXIDASE, SUBUNIT IV, ISOFORM 1; COX4I1	COX4	123864
CYTOCHROME c OXIDASE, SUBUNIT IV, ISOFORM 2; COX4I2	COX IV-2; COX4-2	607976
CYTOCHROME c OXIDASE, SUBUNIT Va; COX5A		603773
CYTOCHROME c OXIDASE, SUBUNIT Vb; COX5B		123866
CYTOCHROME c OXIDASE, SUBUNIT VIa, POLYPEPTIDE 1; COX6A1	LIVER ISOFORM; COX6AL	602072
CYTOCHROME c OXIDASE, SUBUNIT VIa, POLYPEPTIDE 2; COX6A2	MUSCLE ISOFORM; COX6AH; COX6AM	602009
CYTOCHROME c OXIDASE, SUBUNIT VIb POLYPEPTIDE 1; COX6B1		124089
CYTOCHROME c OXIDASE, SUBUNIT VIb POLYPEPTIDE 2; COX6B2	Cancer/Testis antigen 59 CT59	Not available
CYTOCHROME c OXIDASE, SUBUNIT VIc; COX6C		124090
CYTOCHROME c OXIDASE, SUBUNIT VIIa, POLYPEPTIDE 1; COX7A1	MUSCLE ISOFORM; COX7AM	123995
CYTOCHROME c OXIDASE, SUBUNIT VIIa, POLYPEPTIDE 2; COX7A2	LIVER ISOFORM 1; COX7AL; COX7AL1	123996
CYTOCHROME c OXIDASE, SUBUNIT VIIb; COX7B		300885
CYTOCHROME c OXIDASE, SUBUNIT VIIc; COX7C		603774
CYTOCHROME c OXIDASE, SUBUNIT VIII; COX8		123870

Figure 4. The non-overlapping clinical symptoms of COX deficiency and expression territories of COX subunits.

On the left are shown ubiquitously expressed subunits which when mutated result in a constellation of symptom but sparing numerous organs. The representations in the middle and on the right show that mutation in genes encoding subunits with more specific organ/tissue expression does not necessarily result in symptoms affecting the predicted organs. Subunits encoded in mtDNA are depicted in green. Genes with identified mutations in patients with COX deficiency are shown in yellow (or green).

2. Genetic diseases characterized by a COX deficiency

Patients with a COX deficiency can present surprisingly variable clinical phenotypes. As of now mutations in more than 20 mitochondrial and nuclear genes associated with COX deficiency are known to lead to a constellation of phenotypes. Parameters such as lactic acidosis in the blood or cerebro-spinal fluid, ragged-red fibers in adult skeletal muscle biopsies are hallmarks of mitochondrial diseases but they are not specific to COX deficiencies. The following circumstances may modulate the severity and clinical manifestations of a COX and more generally an RC mutation; a) the organs being affected (e.g. a heart specific COX defect is unlikely to change the general metabolic equilibrium of the organism), b) the severity of the deficiency (e.g. partial COX defect can have deleterious consequence, yet not significantly changing metabolic equilibrium under basal conditions), and c) the capacity to cope with a COX defect (e.g. course of disease resulting from a given mutation often differ between patients). As a result, even the absence of typical manifestation of COX (or RC) deficiency should not exclude a clinical anamnesis by advertised clinician, which may justify proceeding to further investigation.

3. New methods, old questions

The large scale screening with micro-chips identifying both mtDNA and nuclear gene mutations has been partially responsible for the steady increase in the number of identified mutations in patients suspected to have dysfunctional mitochondria (see MITOMAP for an update)(15). Unfortunately, except for known or non-ambiguous mutations affecting COX-related genes, there currently is no simple way to ascribe a clinical phenotype to the reported base change even when the latter is in a gene of known function. Examples abound showing that the nature of the amino acid change, its location in the protein, or its evolutionary conservation are not sufficient to establish pathogenicity. Consequently, a biochemical screen of the RC complexes function is usually required. COX activity can be conveniently assayed in virtually all human tissues and cells (16) (Fig. 1B). It should be kept in mind, however, that the results obtained with tissue samples do not always lead to a correct identification of the underlying lesion. Indeed, errors can occur when the mutation shows a high degree of tissue specific expression and therefore is not necessarily detectable in the particular sample analyzed. Another source of ambiguity may arise with mutations causing an accumulation of mitochondria, thus compensating for the decrease of the enzyme and giving the impression of a normal activity. This is not uncommon, especially in the skeletal muscle but can be avoided by normalizing the activity of the deficient enzyme to another RC component or to a marker for mitochondrial mass (Fig. 1B) (17).

4. Human diseases caused by genetically-determined COX deficiency

A sensible and useful way of classifying COX deficiencies takes into consideration their genetic origin. Indeed mutations in the same gene are expected to affect a common step or activity in the same tissue and consequently have a similar outcome. However, what is observed in patients is an impressive diversity of phenotypes independently of whether the mutation affects the biosynthesis/maturation of a mitochondrial or nucleus-encoded COX subunit or of its assembly into the mature complex and ultimately super-complex. A number of excellent reviews have been written on various aspects of COX deficiencies (18–20) and databases are regularly updated incorporating the rapidly growing body of new information (15).

4a. Mutations in mitochondrial COX genes

Numerous diseases resulting from mutations in the mitochondrial genes encoding MTCO1, MTCO2 and MTCO3 have been reported. The mutations in each of these three COX subunits have been associated with a variety of more or less severe phenotypes (Table II). In addition, COX3 mutations have also been connected to Alzheimer disease (21) but this has been disputed as the underlying cause may be an overproduction of superoxides by the mitochondrial RC rather than by the COX deficiency (22).

Table II.

Mutations in COX assembly and structural genes and their phenotypic consequences in human and animals (As in the text, for all cited species, human nomenclature has been used (HGNC) (92)

Mutant gene	Known clinical phenotype in human	Phenotype in animal
MTCO1	MELAS syndrome (93), myopathy (94), rhabdomyolysis(95), prostate cancer (96), myoglobinuria (97), motoneurone disease (98), exercise intolerance (99), epilepsy (100), acquired idiopathic sideroblastic anemia (101), multisystem disorders (102), deafness, LHON, or mitochondrial sensorineural hearing loss (103)	Mus musculus no overt phenotype (62)
MTCO2	Encephalomyopathy (104) LHON (105) myopathy (106), hypertrophic cardiomyopathy (107), Alpers-Huttenlocher like disease (108), encephalomyopathy (109), pseudoexfoliation glaucoma (110), asthenozoospermy (111), rhabdomyolysis (112)
MTCO3	MIDD (113), LHON (99), myopathy (114), asthenozoospermia (115), Leigh disease (116), myoglobinuria (117), sporadic bilateral optic neuropathy (118), rhabdomyoloysis (119), encephalopathy (120), progressive encephalopathy, MELAS or non-arteritic ischemic optic neuropathy (121), hypertensive end-stage renal disease (122)
COX6B1	Severe infantile encephalomyopathy (23)
COX6A1	Recessive axonal or mixed form of Charcot-Marie-Tooth disease (24)	Drosophila melanogaster, COX deficiency and premature death (58).
COX7B	Microphthalmia with linear skin lesions (25)
COX4I2, (COX6A1 COX6A2, COX7A1, COX7A2	Failure to thrive, psychomotor delay, progressive leucodystrophy, encephalomyopathy, epilepsy, hypotony, hepatopathy, anemia, lactic acidosis, and visual impairment (123)	Caenorhabditis elegans, knock-down of COX4 and COX5a homologs using RNA interference shortened lifespans (60). Mus musculus COX4I2 KO lung pathology (63)
COX6A2,		Mus musculus KO led to cardiac dysfunction (65)
COX7A1,		Mus musculus; KO of the heart/skeletal muscle-specific COX7A1 exercise intolerance reminiscent of a mild myopathy (66) or dilated cardiomyopathy (67)
COX7A2
AIFM1	Encephalomyopathy (124), prenatal ventriculomegaly (125), hearing loss, external ophthalmoplegia, ataxia and muscle wasting (126), infantile motor neuron disease (127)
COA3	Neuropathy, exercise intolerance, obesity, and short stature (128)
COA5	Cardiomyopathy (129)
COX10	Neonatal tubulopathy and encephalopathy, Leigh syndrome, or cardiomyopathy (130, 131)	Mus musculus transient liver KO caused increase of hepatocytes apoptosis (68), muscle-specific KO led to myopathy (69); forebrain-specific Cox10 deletion resulted in astroglyosis and inflammation (70)
COX15	Early-onset cardiomyopathy, or Leigh syndrome (132, 133)
FAM36A	Ataxia and muscle hypotonia (134)
FARS2	Infantile-onset epilepsy (69)
FASTKD2	Encephalomyopathy in (135)
PET100	Infantile lactic acidosis (136); Leigh syndrome (137)
SCO1	Neonatal-onset hepatic failure and encephalopathy (138)
SCO2	Neonatal cardioencephalomyopathy; Myopia (36, 139)	Mus musculus A heterozygous KI/KO for SCO2 developed muscle weakness(71, 72) podocyte-specific KO of KLF6 (Krüppel-like factor 6) acting on SCO2-transcription increased focal segmental glomerulosclerosis induced by adryamicin (73)
SURF1	Leigh syndrome (140); villous atrophy and hypertrichosis, without central nervous system pathology(141)	D. melanogaster. Ubiquitous post-transcriptional silencing mostly larvae death some reaching pupal stage dying as early imagos (59). Silencing in the central nervous system, cephalic low COX activity associated with behavioral and electrophysiological abnormality (59). Danio rerio. COX reduced expression induced by using morpholinos, tissue-specific consequences with increased apoptosis in neural tissues but not in the heart that however showed time-increased poor performance (61) Mus musculus SURF1 KO did not manifest disease phenotype despite 30–40% decreased COX activity
TACO1	Leigh syndrome (142)

Altogether, mutations in the 3 mtDNA encoded COX subunits can result in more than twenty different phenotypes. The degree of heteroplasmy (co-existence of variable levels of mutant and non-mutant mtDNA) in different tissues has been invoked to account for the large number of phenotypes associated with mutations in these mitochondrial genes. Although heteroplasmy may be a contributing factor, it cannot be the entire explanation as a similar clinical variability is observed in patients harboring mutations in nuclear genes encoding RC components and assembly factors.

4b. Mutations in nuclear COX genes

Relatively few mutations have been reported in nuclear genes encoding COX subunits and prior to the discovery of the first mutation in COX6B1 (23), it was suggested that such mutations might be incompatible with life. A mutation in COX6B1 was associated with severe infantile encephalomyopathy, a quite typical presentation for a mitochondrial disease (23). In a more recent study a mutation in COX6A1 was shown to cause a neurological disorder characterized by a recessive axonal or mixed form of Charcot-Marie-Tooth disease (24). In addition, a mutation in COX7B was identified in a patient presenting microphthalmia with linear skin lesions, an unusual phenotype for a mitochondrial disease (25). Loss-of-function mutations in nuclear-encoded subunits, although based on still limited data, appear to result in a loss of COX activity and accumulation of COX partially assembled complexes (26), however, the differences in clinical phenotype cannot be related to the function of a particular COX subunit or to its level of expression. A possible factor contributing to this phenomenon is the existence in human mitochondria of isoforms of some COX subunits with differential tissue-specific expression (Table 1 and Fig. 4). COX6B1 like COX6A1 is ubiquitously expressed, while COX6B2 and COX6A2 are expressed mainly in testes and muscle tissue, respectively (Fig 4). Thus, isoform expression by itself may sometimes explains the phenotypic variability associated with mutations in these genes.

4c. Mutant COX assembly genes

Much more frequent mutations affecting COX occur in genes encoding protein factors involved in the biosynthesis and assembly of this enzyme (Fig. 2 and 3, in red). Mutations in 15–20 genes, depending on whether they elicit a singular COX or a predominantly COX deficiency in patients, have been identified. Some of these genes products are factors known to act in pathways other than COX assembly (e.g. AIF1M, or LRPPRC)(27–33).

The temporally and perhaps spatially coordinated biogenesis of different RC complexes and their assembly into supercomplexes is likely to rely on factors that are not specific to a single RC complex. For example the product of OXA1, which was initially thought to be a specific COX factor (34), turned out to function as a more general inner membrane insertase of mitochondrial gene products. However, most COX factors that have been described, mainly from studies of Saccharomyces cerevisiae, and for which there are human homologues, have functions confined to COX biogenesis. With a few exceptions COX assembly factors are ubiquitously expressed in humans and when mutated affect multiple organs. The SCO1 and SCO2 proteins are such exceptions as they display a tissue-specific pattern of expression that has been claimed to fit with the phenotype observed in patients with mutations in these proteins (35). Even in this case, however, it is not always easy to correlate the presentation of the disease with tissue distribution of the protein. A case in point is the highly specific presentation as an autosomal-dominant high-grade myopia in patients with mutations in SCO2. This phenotype is difficult to reconcile with the restricted expression territory of the gene (36, 37).

Mutations in COX assembly factors, similar to those in the structural genes exhibit a multiplicity of phenotypes (see Table II). A number of studies have concluded that COX activity is also critical for cell proliferation in lung, breast cancers, nasopharyngeal carcinomas and gliomas, perhaps by favoring metabolic reprograming required for cell proliferation (38–42).

More than twenty phenotypes observed in patients with mutations in nuclear genes encoding COX subunits and assembly factors are not unlike those described for mutations in the mitochondrial genes. As already mentioned, the fact that these genes have a nuclear origin and are inherited by Mendelian laws excludes heteroplasmy as a contributing factor to the large diversity of phenotypes. Tissue-specific expression of some isoforms can also be discounted as there are numerous examples of gene products that have no isoforms yet also showing a range of different phenotypes.

A number of factors may contribute to the phenotypic variability. The function(s) of at least some proteins encoded by these nuclear genes is far from being completely understood and might be more diversified than presently recognized. Indeed, some of these genes products attributed to play a role in the assembly of COX could also functions in the biogenesis of other RC complexes. Mitochondria, depending on the organ and its energy needs (43, 44), could be differently affected despite expressing a similar biochemical defect. Additionally, because mitochondria are intimately connected with and strongly influence other cellular activities (45), the capacity of organs and the whole organism to cope with mitochondrial dysfunction could depend on the mobilization of genetic resources, expected to be unique to each individual. Evidence is only beginning to emerge now showing that genetic background plays a crucial role in determining the consequence of mitochondrial dysfunctions (46, 47).

5. Studies of COX deficiency in cultured cells, and micro-organisms

Because of their low invasiveness, biopsies of skin fibroblasts or lymphocytes from patient blood samples have been (and still are) extensively used in studies of mitochondrial diseases(16). They have been instrumental in clarifying questions related to the biochemistry of mitochondrial dysfunction and the levels of mtDNA heteroplasmy as a function of cell divisions (48, 49). Additionally, primary cell cultures have proved to be useful to establish the deleterious character of a number of COX gene mutations, especially exploiting their requirement for glucose in culture media. These cells have been in particular useful in devising rational pharmacological (50) and genetic (51) approaches to fight the consequences of COX deficiencies. Exciting recent studies offer the promise that COX deficiencies in patients might one day be ameliorated by an AOX (alternative oxidase) by-pass. Human COX-defective fibroblasts unable to grow in the absence of glucose as a result of COX15 silencing were rescued by ectopic expression of AOX from Ciona intestinalis (52). This non-proton motive oxidase restores NADH and succinate oxidation in mitochondria without increasing their ATP generating capacity (Fig. 1A). The rescue of COX-defective fibroblasts by AOX suggests that the deleterious effect of the COX deficiency in these cells and under the culture conditions used is not the result of lowered ATP production. This is also supported by recent data showing AOX rescue of the different phenotypes seen in COX deficient fruit flies (53).

A number of important questions about COX deficiency remains unanswered, and, for obvious reasons they cannot be solved by studies of cultured cells. Foremost are tissue specificity and development-related phenotypes. In view of this shortcoming several COX defective models have been created to unravel COX subunit/isoform function, hopefully to model human mitochondrial diseases, and ideally to find ways to counteract disease-causing mutations.

Most of our basic understanding of COX and its assembly comes from studies of microorganisms, particularly of the unicellular yeast, Saccharomyces cerevisiae. In addition to the large arsenal of genetic tools for manipulating the mitochondrial and nuclear genes of this organism, it is also extremely useful in validating the pathogenic impact of a particular mutation identified in a constituent of the RC (54). A requirement for such heterologous complementation tests is the presence of a yeast homolog for a mutated human gene. Although this is usually the case with genes coding for RC complexes II, III, and IV, importantly human complex I has not homolog in S. cerevisiae, which uses instead two different dehydrogenases completely unrelated to human complex I. The absence of CI in S. cerevisiae mitochondria implies a different organization of the respirasome that could also influence the deficit related phenotype. There are also important differences in the genetic systems of mammalian and yeast mitochondria. The absence of introns in mammalian mtDNA, and the existence of different sets of translational activators and regulatory proteins than those present in yeast implies that the balanced output of nuclear and mitochondrial gene products of COX and other complexes of the RC must be attained by regulatory mechanisms different from those described in yeast (55, 56),(57). Obviously, yeast is also not the best system to study questions related to tissue specificity and the effects on developmental parameters of COX and RC deficiencies.

6. Higher organisms to study COX deficiency

Several groups have exploited the fruit fly (Drosophila melanogaster), the worm (Caenorhabditis elegans), the Zebrafish (Danio rerio) or the mouse as models for genetic modification of COX-related genes to evaluate their consequent phenotypes in a higher organism. Interestingly, the phenotypes observed in these model organisms have not always been consistent with those of human patients. A mutation in COX6A of Drosophila caused a COX deficiency and premature death, while different mutations in the human gene, similarly causing COX deficiency, result in encephalomyopathy (58). Body-wide post-transcriptional silencing of the fly homolog of the COX-assembly gene SURF1 caused extensive larvae death as early as the imago stage of pupal development (59). Flies with SURF1 silenced in the central nervous system only, reached adulthood with low cephalic COX activity and with behavioral and electrophysiological abnormality (59). Knock-down of the COX4 and COX5A homologs using RNA interference shortened the lifespan of worms. Surprisingly, the authors reported that the COX defect was accompanied by a loss of complex I function (60). Reduced expression of either COX5A or SURF1 induced with morpholinos in Zebra fish elicited a panoply of tissue-specific abnormalities, including increased apoptosis in neural tissues but not in the heart. Heart function, however, decreased with time (61).

Several genes for COX subunits have also been targeted in the mouse. No obvious phenotype could be linked to a mild COX1 point mutant, a missense mutation at nucleotide 6589 (T6589C) converting a highly conserved valine at codon 421 to alanine (V421A) (62). A KO of the nuclear COX4I2, however, caused a lung pathology stressing the importance of the gene product for normal lung function (63). In human, a mutation in COX4I2 causes exocrine pancreatic insufficiency, dyserythropoeitic anemia, and calvarial hyperostosis (64) affecting organs/tissues supposedly not expressing this protein, which might suggest an indirect effect of the mutation. A KO of COX6A2 led to cardiac dysfunction (65) and a KO of the heart/skeletal muscle-specific COX7A1 to exercise intolerance reminiscent of a mild myopathy (66) or dilated cardiomyopathy (67). Known assembly factors for COX have also been inactivated in the mouse. A transient liver KO of COX10 induced hepatocytes apoptosis (68) and a muscle-specific KO of COX10 led to myopathy (69). A forebrain-specific COX10 deletion resulted in astrogliosis and inflammation (70). A SURF1 KO did not manifest a disease phenotype despite 30–40% decreased in COX activity, suggesting that the mitochondrial concentration of this enzyme is higher than required to sustain cell growth and survival. Heterozygous mice with SCO2 KI/KO developed muscle weakness (71, 72), while podocyte-specific KO of KLF6 (Krüppel-like factor 6) acting on SCO2 transcription, increased focal segmental glomerulosclerosis induced by adryamicin (73). A heterozygous KO of CHCHD4 exhibited a COX deficiency and reduced weight gain (28).

Genetic manipulation of the model organisms produce in general rather well defined phenotypes.. This contrasts sharply with the large variation in phenotypes frequently observed in patients with mutations in the homologous genes. This discrepancy can be explained to some extent by the use of clonally-selected animals and by organ-specific targeting (46). The major impact of genetic background in the phenotypic expression of an RC dysfunction can also account for the variability (74). This is clearly demonstrated by the large number of phenotypes observed in a genetically heterogeneous Harlequin mouse population harboring a proviral insertion that reduces expression of the AIF1M gene by 80% (75), resulting only in a significant complex I deficiency (27, 74).

7. Therapeutic hopes

Since our last review on cytochrome oxidase, more than 10 years ago (76), very few approaches have demonstrated efficacy in counteracting COX deficiency in patients. Yet in a small number of cases (77, 78), without human intervention and mostly for unknown reason, a partial reversal of the disease has been seen. This gives hope that reversing disease phenotype owing to an efficient therapy is not out of reach. This particularly holds true when considering mtDNA mutations where partial change of mutant load might be sufficient to counteract at least disease progression. This has been shown possibly reachable in cultured human cells by using mitochondria-targeted nuclases (mitoTALEN) specifically identifying mutant mtDNA(79). Furthermore, thanks to the rapid progress being made in gene vectorization, gene therapy for a number of human diseases including COX deficiency has become an attainable goal. Accordingly, gene therapy has shown some promise in alleviating Leber Hereditary Optic Neuropathy (LHON), which was established to be in part related to a COX deficiency (80). In a mouse model of LHON stemming from a mutation in the mitochondrial ND4 gene, reversal of the disease was observed by optimizing the allotopic expression in the nucleus of an adeno-associated virus-harboring a version of ND4 modified for mitochondrial import and translation (81–83). LHON disease is particularly suitable for this approach as the retinal ganglion cell layer, which as a consequence of its degeneration is instrumental for the loss of vision, can be easily accessed. Gene therapy as a means of treating patients with LHON is still in the early stage with trials being done in France and the US. Unfortunately, this approach is hampered by problems of accessibility for most COX-related disorders affecting other organs or tissues.

Approaches alternative to gene therapy have also been explored to fight COX deficiency and other RC defects (84). A ketogenic diet has been claimed to increase energy metabolism in the brain by enhancing mitochondrial biogenesis, which in turn raises the cellular concentrations of adenosine and ATP, enhances neuron–glia interactions and may even shift the level of heteroplasmy (84). Numerous studies in which mixture(s) of non-toxic dietary supplements (creatine, lipoic acid, CoQ10, etc.) have been tested in patients, claimed to ameliorate some symptoms. Definitive evidence of their efficacy, however, is lacking. On the other hand, treating the symptoms such as strokes with arginine in MELAS patients with a COX-defect may help (85). Some drugs should however be avoided. For example, valproate, but possibly other antiepileptic drugs as well, have been shown to trigger hepatic failure in COX-defective patients (86, 87). Because of a lack of evidences for increased oxidative stress in COX impaired cells (88, 89), the use of antioxidants is also unlikely to be effective for COX deficient patients.

The pan-PPAR agonist bezafibrate has been shown to rescue RC deficiency in COX-defective human cells (50) (90) and in COX-defective mouse (69). However, this effect of bezafibrate could not be reproduced in a SURF KO, a SCO2 KO/KI, and in a muscle-restricted mouse with a COX15 KO. On the other hand, treatment of these three COX-defective mice with the AMPK agonist AICAR led to a partial recovery of COX (91). At this time it is important that these observations be corroborated by additional experiments and hopefully confirmed in patients.

8. Conclusion

Much progress has been made in our understanding of the molecular basis for COX deficiencies in patients, thereby reducing diagnostic wavering and allowing the clinician to better inform family members. The numerous mutations in structural and assembly genes identified in COX deficient patients has also served as an incentive to better understand their functions. Research along this line has revealed that in some cases genes products involved in COX biogenesis also play a more general role in maintaining the respiratory integrity of mitochondria by participating in the assembly pathways for other RC complexes. These studies have contributed to the significant progress made in recent years in deciphering mechanisms responsible for the biogenesis of the respiratory chain complexes and the role of the factors involved in this process.

The same cannot be said of the efforts to find ways of slowing down the course of these often devastating diseases. The emergence of gene therapy gives hope that development of vectors allowing targeting of a specific organ will be paralleled by equal strides in the treatment of diseases including those stemming from mitochondrial disorders.

Abbreviation list

COX

cytochrome c oxidase

mtDNA

mitochondrial DNA

RC

respiratory chain