Sco1 and Sco2 are mitochondrial copper-binding proteins involved in the biogenesis of the CuA site in the cytochrome c oxidase (CcO) subunit Cox2 and in the maintenance of cellular copper homeostasis. Human Surf1 is a CcO assembly factor with an important but poorly characterized role in CcO biogenesis. Here, we analyzed the impact on CcO assembly and tissue copper levels of a G132S mutation in the juxtamembrane region of SCO1 metallochaperone associated with early onset hypertrophic cardiomyopathy, encephalopathy, hypotonia, and hepatopathy, assessed the total copper content of various SURF1 and SCO2-deficient tissues, and investigated the possible physical association between CcO and Sco1. The steady-state level of mutant Sco1 was severely decreased in the muscle mitochondria of the SCO1 patient, indicating compromised stability and thus loss of function of the protein. Unlike the wild-type variant, residual mutant Sco1 appeared to migrate exclusively in the monomeric form on blue native gels. Both the activity and content of CcO were reduced in the patient's muscle to ∼10–20% of control values. SCO1-deficient mitochondria showed accumulation of two Cox2 subcomplexes, suggesting that Sco1 is very likely responsible for a different posttranslational aspect of Cox2 maturation than Sco2. Intriguingly, the various SURF1-deficient samples analyzed showed a tissue-specific copper deficiency similar to that of SCO-deficient samples, suggesting a role for Surf1 in copper homeostasis regulation. Finally, both blue native immunoblot analysis and coimmunoprecipitation revealed that a fraction of Sco1 physically associates with the CcO complex in human muscle mitochondria, suggesting a possible direct relationship between CcO and the regulation of cellular copper homeostasis.
cytochrome c oxidase (CcO) is the terminal heterooligomeric complex of the energy-transducing mitochondrial electron transport chain. This enzyme, which belongs to the superfamily of heme-copper containing aa3-type terminal oxidases, is embedded in the inner mitochondrial membrane, where it catalyzes the transfer of electrons from reduced electron carrier cytochrome c to dioxygen, coupling this reaction with vectorial proton pumping across the inner mitochondrial membrane. The biogenesis of eukaryotic CcO is a complicated, sequential process requiring the coordinated synthesis and import of 3 mitochondria- and 10 nuclear-encoded structural subunits with subsequent assembly and insertion into the inner mitochondrial membrane. This process also requires the concurrent import/synthesis and incorporation of several prosthetic groups, including two heme a moieties, three copper ions, and zinc, magnesium, and sodium ions (16).
A number of nuclear-encoded accessory factors essential for CcO biogenesis have been identified in eukaryotes, and mutations in six of them have been shown to be involved in human pathologies (16). Human Sco1 and Sco2 are closely related inner mitochondrial membrane copper-binding proteins encoded by paralogous genes. They have been demonstrated to exert nonoverlapping, cooperative roles in copper delivery to CcO (7). In addition, they have been shown to be involved in the maintenance of cellular copper homeostasis, presumably by controlling cellular copper export (6, 8). Mutations in both SCO1 and SCO2 cause severe tissue-specific CcO assembly impairment accompanied by marked copper deficiency (11, 17, 18). However, both genes have been shown to be ubiquitously expressed, displaying a similar expression pattern across human tissues. Mutations of SCO1 have thus far only been reported in a single pedigree, where the two patients were compound heterozygotes carrying a nonsense mutation on one allele and a P174L missense mutation on the second allele (18). In contrast, mutations of SCO2 are more common, with all reported patients carrying at least one E140K missense allele (19). Mutations in SCO2 cause fatal infantile encephalomyopathy and hypertrophic cardiomyopathy, whereas the two reported SCO1 patients presented with fatal infantile encephalomyopathy and hepatopathy. SCO2 patients homozygous for the E140K substitution have a delayed onset and slightly prolonged course of the disease compared with compound heterozygotes.
Human Surf1 is an integral protein of the inner mitochondrial membrane composed of two transmembrane domains with a central loop region facing the mitochondrial intermembrane space (22). Mutations of SURF1, which account for the majority of nuclear-encoded, isolated CcO deficiencies in humans, are characterized by the development of Leigh syndrome, a subacute necrotizing encephalomyopathy (14, 23). Most of the identified SURF1 mutations are predicted to lead to loss of protein function. Despite numerous attempts, the molecular function of human Surf1 in CcO assembly still remains to be elucidated (12, 13).
Here, we studied the impact on CcO biogenesis and tissue copper content of a novel homozygous missense mutation in SCO1 associated with early onset hypertrophic cardiomyopathy, encephalopathy, hypotonia, and hepatopathy. We also measured the total copper levels of various SURF1 and SCO2 tissues and investigated the possible physical association between CcO and Sco1. We found that the G132S substitution compromises the stability of the mutant protein, presumably by preventing its oligomerization, leading to a CcO assembly defect, accumulation of Cox2-containing subcomplexes, and severe copper deficiency in the patient's skeletal muscle. We further demonstrate that, similar to SCO1 and SCO2 tissues, the SURF1 samples exhibited marked tissue-dependent copper deficiency, indicating that Surf1 is required, in a tissue-specific manner, to maintain proper cellular copper homeostasis. The fact that the Cox2 subunit accumulates in SCO1 muscle mitochondria in the form of low-molecular-weight subcomplexes suggests that Sco1 is very likely required for a different posttranslational step in Cox2 maturation than Sco2. Finally, both blue native (BN) electrophoresis and coimmunoprecipitation indicated that a fraction of Sco1 associates with fully assembled CcO in human muscle mitochondria. This is of particular interest since the Sco1-dependent metallation of Cox2 is thought to occur on membrane-embedded, yet unassembled subunit, and the functional relevance of this interaction might therefore relate to the role of Sco1 in copper homeostasis maintenance rather than CcO metallation.
MATERIALS AND METHODS
The present study was carried out in accordance with the Declaration of Helsinki of the World Medical Association and was approved by the Committees of Medical Ethics at all collaborating institutions. Informed parental consent, in accordance with guidelines of the participating institutions, was obtained for all biopsies and autopsies.
Tissue samples and cell cultures.
Tissue samples and/or cell cultures were obtained from one SCO1 patient, five SCO2 patients, and four SURF1 patients. The three SCO2 patients homozygous for the common g.1541G>A (E140K) substitution presented with progressive encephalopathy since the third month of age, and they died at <1 yr of age. The two SCO2 compound heterozygotic patients (g.1280C>T/g.1541G>A and g.1518delA/g.1541G>A) presented with hypertrophic cardiomyopathy since birth and died at the ages of 7 and 16 wk, respectively (17, 19, 20). All SURF1 patients (c.790_792delCT/c.845_846delCT, c.312insATdel10/c.821del18, c.688C>T/c.688C>T, and c.469C>T/c.845_846delCT) presented with failure to thrive and progressive hypotonia at the end of the first year of life, followed by a total arrest of psychomotor development. They all died between the third and fifth year of life (12).
Primary skin fibroblast cultures were established from forearm skin biopsies and grown in Quantum 333 complete medium for fibroblasts (PAA). Open muscle biopsies were obtained from the tibialis anterior muscle, frozen in liquid nitrogen, and stored at −80°C. Postmortem heart, liver, and brain (frontal cortex) tissue specimens were obtained at autopsy and frozen in liquid nitrogen at <2 h after death.
Mutation detection in the SCO1 gene.
Genomic DNA was isolated from the peripheral blood. The six exons of the SCO1 gene were amplified by PCR using SCO1-specific intronic primers. PCR products were purified using a QIAquick Gel Extraction Kit (QIAGEN) and directly sequenced using the ABI Prism 3100 AVANT genetic analyzer using the recommended chemistry. The detected mutation was confirmed by restriction fragment length polymorphism analysis using PCR with a mismatch primer changing the thymine at c.392 to cytosine to create a restriction site for HpaII endonuclease (BioLabs). The mismatch primer corresponded to c.379-393. The 17-bp universal M13 primer was added to the 5′-end to enlarge the cleaved fragment. Thus, HpaII cuts the wild-type allele, whereas the mutated allele retains its full length.
Copper content analysis.
The total copper content of skeletal muscle, heart, brain (frontal cortex), and liver tissue specimens was measured by flame atomic absorption spectrometry on a Perkin-Elmer 3300 Atomic Absorption Spectrophotometer.
Enzyme activity assays.
The enzyme activities of CcO and citrate synthase (CS) were measured spectrophotometrically in isolated muscle mitochondria essentially as previously described (17).
Immunoprecipitation of the CcO complex from skeletal muscle mitochondria and HEK-293 cell mitochondria was performed using the Complex IV Immunocapture Kit (Mitosciences, Eugene, OR). Mitochondria were solubilized with 1% n-dodecyl-β-d-maltoside (Sigma-Aldrich), and, after a clarifying spin (50,000 g), the respective supernatants supplemented with protease inhibitor cocktail (Sigma-Aldrich) were incubated overnight with antibody-loaded agarose beds under gentle agitation at 4°C. To partially correct for the lower CcO content in HEK-293 cell mitochondria, immunoprecipitation was performed at a protein concentrations of 6 mg/ml (HEK-293 cell mitochondria) and 2 mg/ml (muscle mitochondria). After an incubation step, immunocomplexes were washed five times with a buffer containing 0.05% (wt/vol) n-dodecyl-β-d-maltoside, and the bound material was then eluted with a buffer containing 1% SDS and routinely processed for SDS-PAGE.
Electrophoresis and immunoblot analysis.
Tricine SDS-PAGE was carried out under standard conditions with 12% polyacrylamide-0.1% (wt/vol) SDS gels. Mitochondrial fractions were dissociated in 50 mM Tris·HCl (pH 6.8), 12% (vol/vol) glycerol, 4% SDS, 2% (vol/vol) 2-mercaptoethanol, and 0.01% (wt/vol) bromophenol blue for 30 min at 37°C, and ∼10 μg of protein were loaded for each lane.
BN-PAGE was performed with 8–16% and 10–20% (wt/vol) polyacrylamide gradient gels using a MiniProtean 3 System (Bio-Rad Laboratories). Mitochondrial proteins were extracted with 1% (wt/vol) n-dodecyl-β-d-maltoside at a protein concentration of 2 mg/ml in a buffer containing 1.5 M aminocaproic acid, 2 mM EDTA, and 50 mM bis-Tris (pH 7.0) at 4°C. Serva Blue G (Serva) was added to solubilized mitochondrial proteins to a final concentration of 0.1 mg/mg of detergent, and 10–50 μg of protein were loaded for each lane. Electrophoresis was performed at 40 V and 4°C for 1 h and then at 100 V.
For two-dimensional BN-SDS-PAGE, strips of the first-dimension gels were incubated for 15 min in a buffer containing 1% 2-mercaptoethanol and 1% SDS and then for 2 × 10 min in 1% SDS, and denatured proteins were then resolved in the second dimension on 12% polyacrylamide-0.1% SDS gels.
Proteins were electroblotted from the gels onto Immobilon-P polyvinylidene difluoride membranes (Millipore) using the TE 70XP Semi-Dry Transfer Unit (Hoefer, Holliston, MA). Blots were finally developed using West Femto Chemiluminescent substrate (Pierce). Signal acquisition was performed using a VersaDoc 4000 imaging system (Bio-Rad Laboratories), and the resulting digital images were analyzed using the Quantity One application (Bio-Rad Laboratories).
Histology, immunohistochemistry, and electron microscopy.
All analyses were performed in paraffin-embedded sections obtained from the autopsy of the SCO1 patient. Paraffin sections were treated using routine histological stains. Mitochondria were detected essentially as previously described (20). For electron microscopy, paraffin-embedded heart, liver, and skeletal muscle samples were deparaffinized in xylol, gradually hydrated, osmicated, reembedded into Araldite Epon mixture, cut with a diamond knife, double contrasted, and examined using a Tesla BS-500 transmission electron microscope (Tesla, Brno, Czech Republic).
Clinical presentation of the SCO1 patient.
A girl with intrauterine growth retardation was born to unrelated healthy parents at term with a birth weight of 2,200 g and length of 46 cm. The early postnatal adaptation was uneventful with an Apgar score of 10 in the fifth and tenth minutes. However, poor feeding, failure to thrive, progressive hypotonia, and hepatomegalia developed since the second month of life. Echocardiography revealed left ventricular hypertrophy, and sonography showed brain atrophy. Biochemical investigation revealed a mild increase in aminotransferase levels (alanine aminotransferase: 1.37–1.78 μkat/l and aspartate aminotransferase: 1.28–1.69 μkat/l) with normal levels of creatine kinase and blood copper (Cu: 18 μmol/l). The dominant metabolic finding since the age of 5 mo was hyperlactacidemia with a fasting B-lactate level of 3.6 mmol/l (reference: <2.3 mmol/l) and a postprandial B-lactate level of 7.7 mmol/l. The fasting lactate-to-pyruvate ratio (L/P) was normal (L/P: 13; reference: 10–20) with a postprandial increase (L/P: 25). Blood alanine was markedly elevated at 738 μmol/l (reference: <500 μmol/l), the free B-carnitine level was reduced at 18 μmol/l (reference: <24 μmol/l), and the ratio between acylcarnitine and free carnitine was increased at 1.5 (reference: 0.23–0.50). The urinary organic acid profile showed increased excretion of lactate (103 mg/g creatinine; reference: <60 mg/g) and the Krebs cycle intermediates 2-oxoglutarate (1,150 mg/g creatinine; reference: <200 mg/g), fumarate (185–193 mg/g creatinine; reference: <15 mg/g), and malate (75 mg/g creatinine; reference: <15 mg/g).
The progressive course of the disease and pathological results of metabolic analyses suggested mitochondrial etiology; thus, the patient was recommended for muscle biopsy. The patient died at the age of 6 mo due to cardiac failure. The dominant finding at autopsy was marked cardiac hypertrophy (55 vs. 29 g), with uniform concentric thickening of the ventricles, predominantly the left ventricle.
Structural analysis of SCO1 tissues.
Cardiocytes were hypertrophic with enlarged and often doubled nuclei (Fig. 1A). There was a uniform increase in the number of mitochondria, with many of them enlarged. Mitochondria were strongly stained with MU213-UC antibody and an antibody against prohibitin (Fig. 1B). Ultrastructurally, there was a numerical increase in the mitochondrial population, which displayed remarkable variation in the size of individual mitochondria (some reached 3–4 μm in diameter) with frequent intermitochondrial contacts and lucent vacuoles in the matrix. Occasionally, there were densified linear rigid cristae and discrete granular deposits of medium density mainly surrounded by a limiting membrane (Fig. 1, C and D).
The liver showed moderate microvesicular steatosis without other obvious structural abnormalities. Skeletal muscles were free of pronounced regressive or regenerative changes. The number of mitochondria showed a only slight tendency to increase, without evidence for ragged-red fibers. Ultrastructural findings of mitochondria in both locations were limited to slight variations in size. Megamitochondria were absent. There was no evidence of any histological changes in other organs (kidney, gastrointestinal tract, and adrenal glands).
Mutation detection in the SCO1 gene.
Direct sequencing revealed the novel homozygous mutation c.394G>A in the third exon of the SCO1 gene. This nucleotide substitution was not found in 200 healthy controls. The mutation is predicted to change the highly conserved glycine residue at position 132 to serine within a juxtamembrane region of Sco1.
Steady-state level and oligomeric state of Sco1 in wild-type and SCO1-deficient muscle mitochondria.
Skeletal muscle mitochondria were resolved using SDS-PAGE, and the resulting immunoblots were probed with rabbit polyclonal antiserum raised against human Sco1 (7) and with an antiserum against Sco2 (17). Equal loading was verified using an antibody targeting succinate dehydrogenase complex subunit A (SDHA) of respiratory complex II. The steady-state level of mutant Sco1 was decreased to ∼10% of control values in the SCO1 patient sample (Fig. 2A). Conversely, no significant alterations in Sco1 content were observed in the SCO2 patient background. Probing of the immunoblots with the Sco2 antiserum revealed that the steady-state level of Sco2 was severely reduced in the SCO2 patient sample but virtually unaffected in the Sco1 patient background (Fig. 2A). Consistent with the known functional involvement of SCO gene products, the levels of copper-containing mitochondrially encoded CcO subunits Cox1 and Cox2 were found to be severely reduced in both SCO1 and SCO2 patient backgrounds (Fig. 2A).
Since the SCO1 mutation occurs within a region presumably involved in protein dimerization (1), we examined the oligomeric state of G132S mutant Sco1 in patient muscle mitochondria using two-dimensional BN-SDS-PAGE immunoblot analysis. Mitochondrial fractions were solubilized with 1% (wt/vol) n-dodecyl-β-d-maltoside and fractionated on a 10–20% (wt/vol) polyacrylamide gradient in the first (BN) dimension. Immunoblots were subsequently probed with an antibody targeting human Sco1. Wild-type Sco1 migrated as part of two minor high-molecular-weight and two predominant lower-molecular-weight assemblies (Figs. 2B and 3A). The 30- and 70-kDa estimated molecular weight of the faster-migrating predominant Sco1 complexes suggests that they represent monomeric and homooligomeric forms of Sco1 protein, respectively. Indeed, the 70-kDa apparent molecular weight of the larger, fast-migrating assembly suggests that it represents the homodimeric form of Sco1, which has been previously described as a 60-kDa complex (1, 7). The 10-kDa difference in apparent molecular weight likely reflects binding of dodecyl maltoside to the dimer. Furthermore, detection with antibodies against Cox1 and Cox2 revealed that the two minor high-molecular-weight (200–300 kDa) Sco1 complexes specifically comigrated with the CcO holoenzyme complex, suggesting a possible physical association of Sco1 with CcO (Fig. 3A).
Mainly due to the low abundance of residual mutant Sco1 in the sample, the complementary immunoblots of SCO1 patient mitochondria produced relatively high levels of nonspecific signal in addition to the specific one. Nevertheless, it was possible to clearly recognize the presence of the monomeric form of Sco1 on the immunoblot. In contrast, no significant signals corresponding to the ∼70-kDa complex as well as to higher-molecular-weight forms of Sco1 were discernible on SCO1 patient immunoblots (Fig. 2B).
Coimmunoprecipitation of Sco1 with the CcO complex from human muscle mitochondria.
To investigate the possible physical association of Sco1 with the CcO complex, we immunoprecipitated CcO from human muscle mitochondria and HEK-293 cell mitochondria solubilized with the nonionic detergent dodecyl maltoside and analyzed the immunoprecipitate using denaturing immunoblots. Consistent with the higher CcO content in muscle mitochondria compared with HEK-293 cell mitochondria, the former immunoprecipitate showed substantially higher CcO recovery (Fig. 3B). Immunodetection of Cox1 confirmed that immunoprecipitation was similarly effective for both samples (Fig. 3B). However, probably due to the higher protein concentration, slightly higher nonspecific binding was observed for the HEK-293 cell immunoprecipitate (Fig. 3B). Despite the high Sco1 content in HEK-293 samples, the muscle immunoprecipitate showed substantially increased recovery of Sco1 that matched the severalfold increase in the recovery of CcO (Fig. 3B). These results indicate that a fraction of Sco1 physically associated with the CcO complex in human muscle mitochondria. Furthermore, the fact that we were not able to precipitate Sco1 with CcO from HEK-293 cell mitochondria might suggests the tissue-specific nature of this interaction.
Steady-state level, activity, and assembly pattern of CcO in SCO1 muscle mitochondria.
The residual content of CcO in mitochondria from muscle biopsy of the SCO1 patient was assessed using BN-PAGE immunoblot analysis. Equal loading was verified using an antibody against the core 2 subunit of respiratory complex III. The content of the CcO holoenzyme was found to be reduced to ∼10–20% of control values (Fig. 4A); this reduction was comparable with that observed in the SCO2 patient sample.
In accord with the reduced holoenzyme content, the enzyme activity assay showed a marked reduction of the CcO-to-CS ratio in SCO1 skeletal muscle mitochondria to 8% (0.047) of the mean reference value (0.6 ± 0.21). However, when the highly increased CS activity (240%) in the sample is taken into account, the normalized CcO activity was reduced to 19% (11.42 nmol·min−1·mg−1) compared with age-matched controls (58 ± 33 nmol·min−1·mg−1, n = 18).
Using two-dimensional BN-SDS-PAGE immunoblot analysis with a cocktail of antibodies against Cox1, Cox2, Cox4, and Cox5a subunits, mitochondria from the muscle biopsy of the SCO1 patient were found to accumulate several CcO subcomplexes concomitantly with the reduction of the holoenzyme level. The highest increase in steady-state levels was observed for the previously identified Cox1-containing subcomplexes c, d, and e (17, 21). However, ∼110-kDa subcomplex c was found to lack both Cox4 and Cox5a subunits. Unexpectedly, prolonged exposure of the immunoblot revealed that this subcomplex instead contained a significant amount of the Cox2 subunit. Moreover, the Cox2 subunit was found in a ∼60- to 70-kDa assembly that comigrated with Cox1-containing subcomplexes e and f (Fig. 4B).
Distinct pattern of CcO subcomplexes in SCO2 and SURF1 primary fibroblasts.
Mitochondria-enriched fractions from three different SURF1 fibroblast cultures and one SCO2 culture (E140K/Q53X) were resolved using BN-PAGE and subsequently probed with the antibody against Cox1. The relative amount of mitochondrial protein loaded as well as the molecular weight of detected CcO subcomplexes were estimated using an antibody targeting the SDHA subunit of respiratory complex II. All SURF1, and to a lesser extent SCO2, patient cells were found to accumulate distinct Cox1-containing subcomplexes in addition to showing reduced holoenzyme levels. As expected, all SURF1 cultures showed an accumulation of the previously identified 110-kDa subcomplex S2 (21). However, the Cox1-containing assembly from SCO2 cells displayed markedly distinct mobility, with an estimated molecular weight of at least 130–140 kDa (Fig. 5). This is intriguing since both SCO2 and SURF1 cells are thought to accumulate an identical subcomplex (S2) composed of Cox1, Cox4, and Cox5a subunits (7, 21).
Total copper content of various tissues harboring mutations in SCO1, SCO2, and SURF1.
The total copper content of skeletal muscle, heart, brain (frontal cortex), and liver tissue specimens was measured by flame atomic absorption spectroscopy. SCO2 and SURF1 liver samples were the most severely affected, showing an average reduction in total copper levels to 23% (12.5 ± 5.4 μg/g, n = 4) and 20% (10.7 ± 10.3 μg/g, n = 3) of reference values (53.6 ± 26.8 μg/g, n = 10), respectively. This is contrasted by the lack of clinical liver involvement in these patients (12, 17). Moreover, all SCO2 liver samples contained normal levels of the CcO holoenzyme (17). Consistent with their severe clinical and biochemical involvement, SURF1 skeletal muscle samples showed an average copper content reduction to 26.5% (0.5 ± 0.1 μg/g, n = 4) of control values (1.9 ± 0.4 μg/g, n = 7). In contrast, despite marked CcO deficiency and severe clinical presentation (17), SCO2 skeletal muscles displayed rather moderate copper deficiency (51.2% of control values, 1.3 ± 0.6 μg/g, n = 3). Indeed, an additional SCO2 muscle sample from the E140K homozygote displayed completely normal copper content. In the skeletal muscle from the SCO1 patient with profound CcO deficiency, the total copper level was found to be decreased to 21.2% (0.4 μg/g) of control values. Consistent with the severe clinical and biochemical presentation (19), SCO2 heart specimens showed a marked reduction of total copper levels to 31.4% (1.03 ± 0.05 μg/g, n = 3) of control values (3.28 ± 0.76 μg/g, n = 8). In contrast, clinically unaffected SURF1 heart samples with only moderately reduced CcO content showed a reduction of copper levels to 53.2% (1.75 ± 0.25 μg/g, n = 2) of control values. The copper content of SCO2 frontal cortex specimens was reduced to an average of 58.2% (1.3 ± 0.3 μg/g, n = 3) of control values (2.23 ± 1.0 μg/g, n = 6). Finally, the two available SURF1 frontal cortex samples showed moderately reduced (1.1 μg/g) and normal copper levels (Fig. 6). Intriguingly, all SCO2 and SURF1 patients showed severe central nervous system involvement, and all the studied frontal cortex specimens showed marked CcO deficiency (12, 17, 19).
In the present study, we showed that the G132S mutation in SCO1, which is associated with early-onset hypertrophic cardiomyopathy, encephalopathy, hypotonia, and hepatopathy, compromised the stability of the protein, presumably by preventing its oligomerization, leading to an impairment of CcO assembly, accumulation of Cox2-containing subcomplexes, and severe copper deficiency in the skeletal muscle of the SCO1 patient. We further showed that, similar to SCO1 and SCO2 tissues, SURF1 samples displayed a marked tissue type-dependent copper-deficient phenotype, indicating a role for Surf1 in copper homeostasis maintenance. Finally, we demonstrated that a fraction of Sco1 associates with fully assembled CcO in human muscle mitochondria, suggesting a possible direct relationship between CcO and the regulation of cellular copper homeostasis.
The described SCO1 mutation is predicted to change the highly conserved glycine residue at position 132 to serine within a juxtamembrane region separating the NH2-terminal transmembrane helix from the globular domain. Substitution of the strongly hydrophilic serine for neutral and highly compact glycine is likely to significantly alter the character of this region. We showed that the steady-state level of G132S mutant Sco1 was substantially attenuated in the muscle mitochondria of the SCO1 patient, indicating severely compromised stability and loss of function of the protein. In contrast, the only SCO1 missense mutation reported to date (P174L) does not affect the stability of the protein but leads to altered functional properties, namely, compromised copper loading by the Cox17 metallochaperone and transduction of an aberrant copper overload signal (2, 6). Based on our inability to detect mutant Sco1 in its ∼70-kDa, most likely homodimeric, form on BN immunoblots, we speculate that the mutation might abrogate the ability of Sco1 to oligomerize, eventually leading to reduced stability of the monomeric form. This is supported by the fact that the amino acid substitution occurs within a region shown to be required for dimerization of Sco1 (1), which exists exclusively in the homooligomeric form in human cells (7).
Two lines of experimental evidence indicate that a fraction of Sco1 physically interacts with fully assembled CcO in human muscle mitochondria. First, two Sco1 complexes were found to specifically comigrate with the CcO holoenzyme in human muscle mitochondria solubilized with 1% dodecyl maltoside, and, second, a fraction of Sco1 was specifically coimmunoprecipitated with the CcO complex from an identical sample (Fig. 3). Using size-exclusion chromatography, Leary et al. (7) showed that in the presence of 1% sodium deoxycholate, wild-type Sco1 is exclusively found in the mitochondria of HEK-293 cells as a complex of ∼60 kDa. Consistent with this result, we were not able to coimmunoprecipitate Sco1 with CcO from dodecyl maltoside-treated mitochondria isolated from HEK-293 cells despite their relatively high Sco1 content. This might suggest a tissue-dependent nature of this interaction that should be further investigated in detail. The fact that, in contrast to immunoprecipitation, only a relatively minor fraction of Sco1 remained associated with CcO after native electrophoresis suggests that the interaction is relatively weak. Given that the Sco1-dependent metallation of Cox2 presumably occurs during an early stage of CcO assembly, most likely in the unassembled polypeptide, we speculate that the functional relevance of this interaction might relate to the role of Sco1 in copper homeostasis signaling rather than in CuA site formation (8). The association of Sco1 with the holoenzyme complex might enable CcO, as an important cellular copper recipient, to participate in the maintenance of cellular copper homeostasis. Together with the potential tissue-specific nature of this interaction, this effect might provide an alternative mechanistic explanation for the severe, tissue-dependent copper deficiency of tissues with CcO defects arising from mutations in CcO assembly factors such as Cox10, Cox15, and Surf1 (Fig. 6) (6). Recently, it has been reported that Shy1, the yeast homolog of human Surf1, physically interacts with different assembly forms of CcO, including its supercomplexes (9). Intriguingly, similar to Sco1, human Surf1 also appears to play a role in cellular copper homeostasis maintenance (Fig. 6).
Consistent with the crucial role of Sco1 in CcO biogenesis, mitochondria from the muscle biopsy of the SCO1 patient showed the accumulation of several CcO subcomplexes in addition to severely reduced holoenzyme levels. However, the subunit composition and relative abundance of these species markedly differed from those found in P174L SCO1 fibroblasts and myoblasts and in SCO2 muscle mitochondria (17, 21). We could not detect the common Cox1-Cox4-Cox5a subcomplex (S2) in SCO1 muscle mitochondria, as ∼110-kDa Cox1-containing subcomplex c apparently lacked both nuclear-encoded subunits. Intriguingly, a minor but significant amount of subunit Cox2 was found in G132S SCO1 mitochondria in the form of two distinct, faster-migrating assemblies. The larger one contained at least Cox1 and Cox2 subunits, whereas the precise subunit composition of the faster-migrating species remains elusive. Indeed, it appeared too small to contain a Cox1-Cox2 heterodimer and was probably too large to represent free, unassembled apoCox2.
Newly synthesized Cox2 was shown to undergo accelerated turnover in P174L SCO1 fibroblasts, presumably due to a failure to incorporate the CuA-lacking subunit into the Cox1-Cox4-Cox5a subcomplex (2). This is thought to cause an impairment of enzyme assembly, characterized by the accumulation of the S2 subcomplex, which appears to be resistant to inner membrane proteases. Indeed, consistent with a cooperative role for Sco2 in copper trafficking to CcO, a subcomplex pattern identical to that of P174L SCO1 fibroblasts was observed in both heart and skeletal muscle mitochondria carrying loss-of-function mutations of SCO2 (17). Given the inability to detect even trace amounts of free unassembled Cox2 in these samples, it was deduced that the function of Sco2 is strictly required for the assembly/stability of Cox2 (5, 17). Therefore, the accumulation of Cox2 in the form of oligomeric subcomplexes in Sco1-deficient mitochondria indicates that Sco1 is very likely responsible for a different posttranslational aspect of Cox2 biogenesis than Sco2. This is further supported by the fact that the steady-state level of Sco2 was virtually unaffected in the G132S Sco1 patient background (Fig. 2). The rather unusual character of CcO subcomplexes found in G132S SCO1 mitochondria suggests that they might represent misassembled off-path products. Indeed, it is possible that Cox2 prematurely associates with Cox1, thereby compromising the subsequent association of the Cox4-Cox5a subcomplex and leading to stalled CcO assembly. This may result either directly from misfolding of Cox2 or from a lack of specific assembly chaperone activity.
Based on a single report of two affected siblings, encephalopathy, hypotonia, and hepatopathy have been considered to be the primary symptoms of SCO1 deficiency (18). Here, we report that, similar to SCO2 compound heterozygotes, our SCO1 patient presented with early onset hypertrophic cardiomyopathy, in addition to hypotonia, encephalopathy, and hepatopathy. Given the similar functional involvement of SCO proteins and their ubiquitous expression with a similar expression pattern across human tissues, such clinical phenotypes were rather expected. The lack of apparent cardiac involvement in previously published SCO1 cases likely resulted from either the considerably reduced survival time of both siblings or the distinct nature of the missense allele expressed in these patients. Indeed, P174L mutant Sco1 shows markedly altered functional properties and almost normal polypeptide levels (6), whereas the G132S allele appears to lead to a simple, yet almost complete loss of protein and function.
Mitochondrial ultrastructural abnormalities in the SCO1 heart resembled those described in the myocardium of SCO2 compound heterozygotes (20). The finally granular hemidense deposits occasionally seen in mitochondria are suggestive of glycogen origin (4). They were not lipid containing, as they resisted dehydration during embedding in paraffin. They did not resemble any of the crystalloid inclusions described mainly in skeletal muscles in a variety of mitochondrial disorders (10).
Finally, the SCO1 and SCO2 tissues analyzed displayed a marked copper-deficient phenotype, the extent of which significantly varied according to tissue type. Intriguingly, the various SURF1 samples studied also showed marked tissue type-dependent copper deficiency, suggesting a role for human Surf1 in copper homeostasis maintenance. Interestingly, yeast cells lacking Shy1, the yeast homolog of human Surf1, are deficient in mitochondrial copper, whereas the total cellular copper content remains normal (15). In addition, supplementation of shy1Δ cultures with exogenous copper partially rescues the respiratory capacity of these cells (3, 15). Recently, it has been reported that both Sco1 and Sco2 play important roles in the regulation of cellular copper export in addition to their roles in copper trafficking to CcO. It was further demonstrated that the corresponding copper defect was fully separable from CcO deficiency, at least in immortalized human fibroblasts (6). Consistent with these results, we found that the residual copper content was negatively correlated with the extent of CcO deficiency in the livers and brains of SCO2 and SURF1 patients. Intriguingly, a similar discrepancy between residual copper and CcO levels was also found in all studied SCO2 muscles. In contrast, a positive relationship was observed in SCO2 and SURF1 hearts and in SCO1 and SURF1 muscles. Collectively, these results once again indicate tissue-specific functional differences of these ubiquitously expressed proteins that likely evolved to meet the profound tissue-specific requirements of mitochondrial function and biogenesis.
In summary, we demonstrate here that the G132S substitution in Sco1 compromises the stability of the protein, presumably by preventing its oligomerization, leading to severe copper deficiency and CcO assembly impairment. We further show that a fraction of Sco1 physically interacts with the CcO complex in human muscle mitochondria and, based on a dissection of the Sco1-deficient CcO assembly pattern, suggest distinct functional involvement of SCO proteins in posttranslational Cox2 maturation. Finally, based on the marked tissue-specific copper deficiency of analyzed SURF1 tissues, we hypothesize that this CcO assembly factor plays an important role in cellular copper homeostasis maintenance.
NOTE ADDED IN PROOF
Since the submission of this article, we have performed immunoprecipitation of Sco1 from dodecyl maltoside-solubilized HEK-293 and skeletal muscle mitochondria using the Sco1 antiserum. Surprisingly, a significant amount of CcO was found to copurify with Sco1 from HEK-293 mitochondria in this experiment. Our inability to pull down a significant amount of Sco1 from HEK-293 mitochondria using the anti-CcO antibody was thus likely caused by the low CcO content in HEK-293 mitochondria and its resulting low recovery.
This work was supported by institutional projects MSM 0021620806 and 1M6837805002 and Grant Agency of the Ministry of Health of the Czech Republic Grant NS/10581-3.
The rabbit polyclonal antiserum raised against human Sco1 was generously provided by Scot E. Leary and Eric A. Shoubridge (Montreal Neurological Institute).
- Copyright © 2009 the American Physiological Society