Mutations in SCO2, a protein required for the proper assembly and functioning of cytochrome c oxidase (COX; complex IV of the mitochondrial respiratory chain), cause a fatal infantile cardioencephalomyopathy with COX deficiency. We have generated mice harboring a Sco2 knock-out (KO) allele and a Sco2 knock-in (KI) allele expressing an E→K mutation at position 129 (E129K), corresponding to the E140K mutation found in almost all human SCO2-mutated patients. Whereas homozygous KO mice were embryonic lethals, homozygous KI and compound heterozygous KI/KO mice were viable, but had muscle weakness; biochemically, they had respiratory chain deficiencies as well as complex IV assembly defects in multiple tissues. There was a concomitant reduction in mitochondrial copper content, but the total amount of copper in examined tissues was not reduced. These mouse models should be of use in further studies of Sco2 function, as well as in testing therapeutic approaches to treat the human disorder.
Mammalian cytochrome c oxidase (COX), also known as complex IV of the mitochondrial respiratory chain, is a multi-subunit holoprotein composed of three subunits encoded by mitochondrial DNA (mtDNA) and 10 subunits encoded by nuclear DNA, as well as two hemes (a and a3), three coppers, one magnesium and one zinc as prosthetic groups (1). The mtDNA-encoded subunits (COX I–III) form the catalytic core of the enzyme, and contain three redox centers that oxidize cytochrome c and reduce oxygen to form water, while at the same time pumping protons from the matrix to the intermembrane space. In this reaction, electrons from cytochrome c are transferred first to the CuA site in subunit II (containing two copper atoms), then to the heme a site in subunit I, then to the CuB-heme a3 binuclear center, also in subunit I, and finally to molecular oxygen (2).
The assembly of the COX holoprotein is a complicated and regulated process, requiring more than 20 ancillary assembly factors, including proteins required for processing of the structural subunits and insertion of the holoprotein into the mitochondrial inner membrane, for heme biosynthesis and maturation, and for metabolism and insertion of copper into the holoprotein (2–7).
At least eight proteins are required for insertion of copper into human COX, including CMC1, COX11, COX17, COX19, COX23, PET191, SCO1 and SCO2 (2). SCO1 and SCO2 (SCO stands for synthesis of cytochrome coxidase) are homologous proteins related to bacterial prrC, a member of the prrBCA operon associated with the photosynthetic regulatory response in Rhodobacter sphaeroides (8,9). As such, SCO may function not only to transport copper to the CuA site (10), but may also function as a redox sensor (9,11,12). While both SCO1 and SCO2 contain a highly conserved CXXXC motif that presumably binds copper, and therefore are thought to be metallochaperones that are essential for the assembly of the catalytic core of COX (13,14), the two proteins apparently have non-overlapping functions (15,16) and may participate differentially in the regulation of cellular copper homeostasis (16). To date, however, the precise function of SCO proteins is still unclear.
Pathogenic mutations in human SCO1 and SCO2 cause mitochondrial diseases associated with fatal infantile COX deficiency. Although both proteins appear to be expressed in all tissues (17), pathogenic mutations in SCO1 lead to fatal infantile hepatoencephalomyopathy (18), whereas those in SCO2 lead to hypertrophic cardiomyopathy, encephalopathy and myopathy soon after birth (17). Why mutations in the two genes cause such different presentations is under investigation.
To date, over 25 patients with autosomal recessive COX deficiency caused by mutations in SCO2 have been described (17,19–27). With but one exception (27), all patients harbored a missense mutation that converts Glu-140 to Lys (i.e. E140K) on at least one allele, at a position located only three residues away from the CXXXC copper-binding domain (9). Interestingly, patients homozygous for the E140K mutation had a delayed onset of symptoms and a more prolonged course of the disease, with death during mid- to late-infancy, when compared with compound heterozygotes who had E140K on one allele and a null mutation (either a truncation or a frameshift mutation) on the other allele (20).
In order to better understand the biological role of SCO2, and to test potential therapeutic approaches, we have created mice harboring both a Sco2 knock-out (KO) allele and a Sco2 knock-in (KI) E129K allele, corresponding to the common E140K mutation in humans. Although homozygous KO mice are embryonic lethals, both homozygous KI mice and compound heterozygous KI/KO mice show COX deficiency in tissues affected in the human disease.
Generation of Sco2 mutant mice
Using a 5 kb KpnI fragment containing the entire Sco2 gene and portions of the flanking Tymp and Ncaph2 genes as a targeting vector, we generated Sco2 KO and E129K KI constructs (Supplementary Material, Fig. S1). These were transfected into ES cells, and positive clones were injected into the blastocysts from C57BL/6J mice. Heterozygous WT/KO and WT/KI mice were verified by polymerase chain reaction (PCR) analysis (Supplementary Material, Fig. S2) and sequencing of tail DNA. After germline transmission was obtained, we removed the NeoR cassette and one of the flanking LoxP sites by crossing the heterozygous mice with cre mice. The progeny were then backcrossed with WT 129Sv/J mice. The resulting heterozygous mice were healthy and fertile, and did not show any apparent differences when compared with their wild-type (WT) littermates.
Inbreeding of heterozygous KI (E129K) mice resulted in the production of 80 mice, harboring Sco2 E129/E129 (WT), homozygous K129/K129 and heterozygous E129/K129 alleles in the expected Mendelian ratio (20:38:22 animals, respectively). Adult weight, postnatal growth and development were indistinguishable from their WT littermates (data not shown).
Further matings, including those with heterozygous KO mice, allowed us to generate homozygous KI/KI (hereafter called KIKI mice) and compound heterozygous KI/KO (KIKO) mice, but we were unable to obtain viable homozygous KO/KO (KOKO) mice: upon inbreeding WT/KO heterozygotes, we observed a Mendelian ratio of progeny at 7.5 dpc (days postcoitum) (6 WT/WT, 14 WT/KO and 7 KO/KO), but observed no viable KO homozygotes at birth (42 WT/WT, 87 WT/KO and 0 KO/KO). The embryonic lethality in the KOKO mice occurred during early gestation; the latest KOKO embryo that we could obtain was at day 8.5 postcoitum, and had severe morphological abnormalities (Fig. 1). Histochemistry to detect COX activity showed profound COX deficiency in the embryo compared with an age-matched WT embryo, even though both WT and mutant embryos had mitochondria, as determined by the positive staining for succinate dehydrogenase (SDH) activity (Fig. 1B).
In contrast to the KOKO mice, the KIKI and KIKO mice were viable. They had longevities similar to that of WT littermates, and their body weights were similar to that of controls at every age tested (not shown). The mice did not develop overt symptoms of the cardioencephalomyopathy seen in the human disease, but upon further investigation, we found significant biochemical, morphological and behavioral differences in the Sco2-mutant mice compared with their normal counterparts.
We assessed the activities of respiratory chain complexes I, III and IV (COX), normalized to that of citrate synthase (CS), in crude mitochondria isolated from the brain, heart, liver and muscle of WT, KIKI and KIKO mice at 4 months of age. In all cases, the normalized complex I activities in KIKI and KIKO mice were comparable to the values in WT controls (Fig. 2). However, complex IV activities were reduced in all examined tissues from the KIKO mice (by approximately 20–60%), with the lowest values in liver (by approximately 60%; Fig. 2). Complex IV activity was also reduced in all tissues of the KIKI mice, but to a lesser extent (by approximately 15–45%) (Fig. 2). Surprisingly, we also detected reduced complex III activities in all tissues from the mutant mice (by approximately 15–45%; Fig. 2).
We analyzed the amount of copper in the same tissues and in crude mitochondria. There was essentially no difference in the total copper content of the tissues (Fig. 3A and B), but there was a noticeable trend towards reduced amounts of copper in the crude mitochondrial fractions isolated from all examined tissues from the KIKI and KIKO mice when compared with WT animals (Fig. 3C).
While not strictly quantitative, histochemical detection of COX activity in muscle, heart, brain, liver and kidney from the mice showed a greater apparent degree of COX deficiency in the KIKO mice than in the KIKI mice when compared with WT, with skeletal muscle and brain most noticeably affected (Supplementary Material, Fig. S3), similar to the two tissues most affected in human SCO2 deficiency (17). More detailed histochemical examination of skeletal muscle showed that the pattern of COX activity was similar to the pattern of the Sco2 immunostaining in both WT and mutant animals (Fig. 4).
We performed one-dimensional blue-native gel electrophoresis (1D-BNGE) (28,29) of mitochondria from liver and brain, probing with representative antibodies to each of the five OxPhos complexes. We found reduced levels of fully assembled complex IV in both KIKO and KIKI mice, together with the apparent disappearance of a band in KIKO mice corresponding to a supercomplex (SC) consisting of complexes III and IV that was present in WT and KIKI mice (Fig. 5A). To further investigate complex IV assembly, we subjected the samples to two-dimensional BNGE (2D-BNGE) (28) followed by western blotting to detect different subunits of complex IV. When antibodies to COX I and COX II were used, we observed the appearance of a subcomplex, approximately 80–100 kDa in size, in KIKI and KIKO samples, in both liver and brain, that were absent in the WT controls (Fig. 5B). Using antibodies to COX Va, we observed the appearance of a different subcomplex, approximately 150 kDa in size, again in both liver and brain mitochondria from KIKI and KIKO, but not WT, mice (Fig. 5B). We saw no evidence of free unassembled subunits in the low-molecular weight portion of the gels (i.e. between 40 and 66 kDa), but we cannot exclude the possibility that subunits smaller than 40 kDa were present but ran off the gel. On the other hand, we did not observe any aberrant assembly intermediates when blotting with antibodies to COX IV (Fig. 5B). These results reveal a role for Sco2 not only in complex IV function but also in its assembly. Finally, using a cocktail of five antibodies that detect one representative polypeptide from each of the five OxPhos complexes in western blotting of proteins from brain and liver crude mitochondria separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE), we found decreased amounts of COX I in KIKI and KIKO mice compared with WT (Fig. 5C). This decrease did not appear to correlate with the amount of Sco2 (or Sco1, for that matter) in these tissues (Fig. 5D).
We conducted evaluations of motor function, coordination and neuromuscular strength. There was no difference in neuromuscular coordination between mutant and WT mice, as determined by performance on a vertical pole test. However, when the mice were subjected to a running test on a treadmill, the endurance of 4-month-old KIKO male mice was significantly poorer than that of their WT littermates (Fig. 6B). The KIKO mice also had muscle weakness, as measured by the hanging wire test (Fig. 6A). Interestingly, while both male and female mice developed muscle weakness, its onset was delayed in the females: whereas males had obvious loss of strength at 4 months, the females were normal at that age, and did not show loss of strength until 8 months of age (Fig. 6A). We note, in this regard, that there was a gender difference in the degree of response to upregulation of peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α) of mice whose muscle lacked Cox10, another COX assembly protein (30).
We found no significant differences in cardiac function between WT and Sco2-mutated mice, as measured by transthoracic M-mode and two-dimensional echocardiography (31) (not shown).
We describe here the first mouse models of COX deficiency owing to mutations in Sco2, a COX-assembly protein. While homozygous Sco2 KO mice were embryonic lethals, both the homozygous E129K KIKI and compound heterozygous KIKO mice showed biochemical, morphological and functional defects reminiscent of, and compatible with, loss of COX activity in affected tissues in human SCO2 deficiency. However, while the Sco2-mutated mice recapitulated many of the biochemical and functional features of the human disease, they did not replicate all of them. In particular, whereas most SCO2-mutant patients die in infancy of a combined cardiopathy and myopathy, both the KIKI and KIKO mice showed no evidence of cardiomyopathy and no reduction in lifespan.
The relatively mild phenotype observed in our Sco2 mouse models when compared with that in the human disease stands in stark contrast to the embryonic lethal phenotype observed in the homozygous Sco2 KO mice, which itself contrasts with mice lacking Surf1, another COX assembly gene, which were essentially normal (32,33). In particular, homozygous disruption of the mouse Sco2 gene led to embryonic lethality prior to 8.5 dpc, implying that Sco2 plays a crucial role during early development. Further analysis of cells derived from these embryos will be required to better determine the nature of the lethality.
Our data also support the view that although SCO1 and SCO2 are duplicated genes in both mammals and yeast (17), they probably have non-overlapping functions (15,34). We note that Sco1 mRNA has been detected in mouse embryos as early as the 2-cell stage (Genbank AK139512.10), as well as in 8-, 9-, and 10-day embryos (AK017794.1, AK083729.1 and AK011251.1, respectively), implying that Sco1 protein is present during early development. If so, our results imply that Sco1 was not able to rescue the loss of Sco2 function in homozygous KO mouse embryos, supporting the idea that Sco2 is not only critical for embryonic viability, but also that Sco1 and Sco2 have at least partially non-overlapping functions at this stage of development (and possibly in postnatal life as well). The lack of complementation between Sco1 and Sco2 has also been reported in cell culture, where reduced copper levels and COX activity owing to mutations in either SCO1 or SCO2 could be suppressed by overexpression of SCO2, but not SCO1 (16).
The two viable mouse models created in this study carried the E129K mutation (the homolog of the human E140K mutation), either as a homozygous KIKI or as a compound heterozygote, with one E129K allele and one KO allele (KIKO). Both mice were viable, fertile and appeared to live as long as their WT littermates, and we found no difference between the mutant mice and their WT littermates in body weight, appearance, grooming, nesting and sleeping behaviors. This apparent good health of the KIKI and KIKO mice is in striking contrast to the analagous situation in human SCO2 patients with E140K homozygous (20) and compound heterozygous (17) mutations, who die within approximately 1 year of life, typically of cardioencephalomyopathy.
On the other hand, the KIKO mice performed poorly on a treadmill endurance test, and they were more susceptible to fatigue when compared with their WT littermates. They also had impaired motor function, as measured by the hanging wire grip test. The defects in motor activity and muscle endurance were consistent with, and were likely due to, a significant deficiency in COX activity in muscle. In fact, we found a reduction in COX activity in all tissues examined, including brain, heart, liver and skeletal muscle, with the lowest values in liver. The reduction in COX activity as measured biochemically was also seen in the COX histochemistry of these tissues. However, the reduction in COX biochemical and histochemical activity in the mice was less severe in muscle than in, for example, liver or heart, and stands in contrast to what has been observed in human SCO2 patients, where the COX deficiency appears to be most severe in cardiac and skeletal muscle (17). We do not have an explanation for these differences, but they likely explain the relatively mild phenotypes that we observed.
It has been proposed that independent of its roles in COX assembly, SCO2 acts upstream of SCO1 to help maintain cellular copper homeostasis via regulation of the rate of copper efflux from the cell (16), which would explain the reduction in total cellular Cu found in SCO-mutant patient cells and tissues (16,35). However, while the amounts of copper found in the crude mitochondrial fractions in the KIKI and KIKO mice were lower than in WT (consistent with the reduction in COX activity), the total amount of copper in the tissues that we examined did not decline in the Sco2-mutant animals. We do not know why the data in the mice are discrepant from those from human tissues.
A number of studies have shown that high residual levels of COX assembly intermediates are present in SCO2-mutant patient tissues (23,36). The SCO2 mutations in patients are thought to reduce the efficiency by which the CuA center of COX II is formed (15,36), implying that the presence of copper in the CuA site is required for the assembly or stability of COX II (37). This idea is supported by studies in which COX activity was rescued in SCO2-mutant cells by the addition of copper (34,36,38,39). However, the direct transfer of copper to COX has still not been demonstrated and is still speculative, and nothing is known regarding the spatial arrangement of SCO2 vis-à-vis COX II.
We investigated the defect of COX in our mice by 2D-BNGE, which is a sensitive way to detect assembly intermediates of the respiratory chain complexes. In brain and liver mitochondria from the KIKI and KIKO mice, we found an accumulation of aberrant assembly intermediates containing COX I and COX II, implying that in addition to any defect of copper transport to the CuA site in COX II (10), the E129K mutation might also impair indirectly the transport of copper to COX I (40), thereby compromising COX assembly and reducing COX activity in these tissues. We also detected an aberrant assembly intermediate containing COX Va. The defects in COX assembly observed in our mice had similarities to those observed in cells and tissues from human SCO patients, but it is noteworthy that the composition of these sub-assemblies has varied among patients. Taanman's group described two sub-assemblies in one SCO1 patient, one containing COX I, IV and Va, and another containing just COX I (41). Houstek's group described a sub-assembly in a SCO1 patient containing both COX I and II but neither COX IV nor Va (35). They also found a SCO2 patient with altered sub-assemblies containing COX I, IV and Va, but not COX II; interestingly, the assembly defects varied among tissues (37). Finally, Shoubridge's group found a SCO1 patient with a sub-assembly containing COX I and IV, but not COX II (COX Va was not examined) (15). Thus, the aberrant sub-assemblies found in the Sco2-mutant mice—containing Cox I, II and Va, but not Cox IV—mimic closely, but not exactly, those found in the human patients. It may be that mutations in SCO1 cause different assembly defects than those in SCO2, or that SCO mutations perturb the COX assembly pathway in a tissue-specific manner, or that different specific SCO mutations have different effects on the COX assembly process. Finally, it is also possible that some of the subassemblies that we (and perhaps others) have observed are degradation products rather than authentic assembly intermediates, since the COX holocomplex is more unstable when SCO2 is mutated.
According to the current model of COX assembly, the process starts with the interaction of COX I with either COX IV or COX Va (29,41,42). The intermediates containing COX I in our 2D-BNGE analysis indicate that the E129K mutation impairs COX assembly at a very early step in the assembly pathway, and is consistent with observations of COX deficiency in patients with mutations in COX I who have defective assembly of the COX holoenzyme (43,44). Similarly, immunoblot analysis of cells harboring a mutation in COX II also revealed decreased amounts of both COX I and COX II, again accompanied by a reduction in COX activity (45,46). In spite of the fact that SCO1 and SCO2 may have non-overlapping functions, their roles in the COX assembly pathway appear to be similar, as assembly intermediates containing COX I and COX II, but not COX IV, were found both in our Sco2-deficient mice and in cells from a SCO1-deficient patient (35).
We also detected reduced complex III activity in both the KIKO and KIKI mice. This finding was somewhat unexpected, as complex III deficiency had not been observed in the initial reports on human patients with SCO2 mutations (17,19,20), but we note that reduced complex III activity in a SCO2 patient was reported recently (27). While patients with mitochondrial diseases in which there is combined complex I + III (47) and I + III + IV (48) deficiency have been reported, III +IV deficiency is relatively rare, although we note that some SCO2 patients have been reported with I + III deficiency (23,27). Of course, since the respiratory chain can exist as a SC (49), it is possible that a mutation in a complex IV assembly protein, such as SCO2, could also affect the activity of complex III.
An equally intriguing result was the quantitative nature of the COX deficiency observed in the KIKI versus KIKO mice. The phenotype of the KIKI mice, with two E129K alleles, was less severe than that of the KIKO mice, with one E129K allele and one null allele. This result parallels what has been observed with the corresponding E140K mutation in humans, where patients carrying two E140K alleles survive longer than those with one E140K allele and one null allele (20). The quantitative behavior of this missense mutation in both mice and humans implies that it is a hypomorph that still allows for residual function, either in copper transport, or more speculatively, in redox sensing (9). If E140K is indeed hypomorphic, unaffected heterozygous parents carrying this mutation ought to have more than half-normal SCO2 function (and hence >50% COX activity), and thereby might have a selective advantage over those carriers with more deleterious alleles, thereby explaining why the E140K mutation is so common in the population. Thus, while our KIKI and KIKO mice reproduce only some aspects of the human disease, they model quite nicely the quantitative nature of the common E140K mutation.
In summary, mutations in mouse Sco2 decrease COX activity through impairment of the stability of the COX holoprotein, with an unexpected impact on the activity of complex III as well. Since these Sco2-mutant mice also recapitulate many of the functional features of the human disease, including muscle weakness, and yet are viable and fertile, they will be of value in testing the effects of potential therapies (30) for Mendelian-inherited COX deficiency, a group of currently incurable and fatal disorders.
MATERIAL AND METHODS
All experiments were performed according to a protocol approved by the Institutional Animal Care and Use Committee of the Columbia University Medical Center, which is consistent with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Mice were housed and bred according to international standard conditions, with a 12-h light, 12-h dark cycle.
Generation of Sco2 mouse models
The detailed procedures to generate and analyze the Sco2-mutant mice are described in the Supplementary Material.
Enzyme activities of respiratory complexes
Crude mitochondria were obtained by homogenization and differential centrifugation of brain, heart, liver and muscle tissues. Tissues were homogenized in four volumes of SETH buffer (250 mm sucrose, 2 mm EDTA, 10 mm Tris–HCl, 50 U/ml heparin, pH 7.4) in a glass-glass homogenizer (5–8 strokes) on ice. The homogenate was centifugated at 600g for 5 min at 4°C. The supernatant was centrifugated a second time at 8000g for 15 min at 4°C. The pellet containing the crude mitochondria was resuspended in SETH buffer (approximately 10–25 mg/ml). Protein concentration was determined by the method of Bradford using bovine serum albumin as a standard.
Histochemical and immunohistochemical analyses
For histochemistry, brain, heart, liver and muscle tissues were frozen in isopentane cooled in liquid nitrogen. The tissues were cryo-sectioned (8 µm thick) and stained for SDH and COX activities, as described (53). For immunohistochemistry, tissue sections were fixed for 10 min at room temperature (RT) in 4% paraformaldehyde in phosphate-buffered saline (PBS). Tissues were rinsed in PBS followed by an antigen retrieval step in citrate buffer (10 mm citric acid, pH 6.0, at 100°C for 45 min). Slides were then incubated in 0.4% Triton X-100 in PBS for 20 min at RT and blocked for 1 h at RT in SuperBlock blocking buffer (Pierce) containing 160 µl/ml Avidin Block (Vector Labs). A cross-reacting rabbit polyclonal antibody to human SCO2 (see below) was diluted 1:100 in SuperBlock blocking buffer containing 160 µl/ml Biotin Block (Vector Labs) and was added to the tissue sections overnight at 4°C. Secondary biotinylated rabbit antibody (Amersham) was added for 1 h at RT, followed by incubation with streptavidin fluorescein (Amersham) for 1 h at RT.
We assayed for copper using the inductively coupled plasma-mass spectrometer (ICP-MS) method. We digested tissue homogenates in concentrated HNO3 for 20 h at RT, diluted to a final acid concentration of 10%, and analyzed for Cu concentration using a Perkin-Elmer Elan Dynamic Reaction Cell (DRC) II ICP-MS equipped with an AS 93+ autosampler along with appropriate calibration standards. We corrected for matrix-induced interferences by using an internal standard matched to the mass and ionization properties of copper, namely gallium. We suppressed polyatomic interferences in the instrument's DRC using ammonia as a second gas. The intraprecision coefficient of variation for standard serum samples with known Cu concentrations was 2.8%, and for selected replicate tissue samples was 1.6%.
BNGE and western blotting
For western blotting of proteins separated by SDS–PAGE, 15 µg of crude mitochondria were electrophoresed through a 12% polyacrylamide gel and electroblotted onto polyvinylidene difluoride (PVDF) filters (Bio-Rad). For protein detection, we used the MitoProfile Total OxPhos rodent antibody cocktail (Mitosciences MS604). Anti-cytochrome c (Invitrogen no. 456100; 1:1000 dilution) was used as a loading control. For western blotting of Sco proteins, 30 µg of mouse crude mitochondrial proteins were loaded onto a 15% polyacrylamide gel, separated by SDS–PAGE, transferred to a nitrocellulose membrane, and probed with rabbit antisera to human SCO1 or SCO2 that cross-react with the corresponding mouse proteins. Antisera were raised against amino acids nos 272–285 (EFLDYFGQNKRKGE) of human SCO1 and amino acids nos 236–249 (LFTDYYGRSRSAEQ) of human SCO2 (Zymed Laboratories, Invitrogen). After affinity purification of the antisera, the antibodies were diluted 1:400 for immunoblotting. Anti-porin (Abcam no. ab15895; 1:500 dilution) was used as a loading control.
For BNGE, 50–75 µg of mitochondrial proteins that had been extracted with lauryl maltoside were applied on a 5–13% gradient BN gel followed, in the case of 2D-BNGE, by separation through a 12.5% denaturing gel (54). After electophoresis, proteins were electroblotted onto PVDF filters and sequentially probed with specific antibodies. For protein detection, we used antibodies against NDUFA9 (complex I), Fp (complex II), core 2 (complex III), COX I (complex IV), COX II (complex IV), COX IV (complex IV), COX Va (complex IV) and ATPase-β (complex V). All antibodies were from Invitrogen.
Mouse functional assays
To measure endurance capacity, the mice were run on a treadmill (Columbus Instruments) at a speed of 22 m/min until they reached fatigue. Fatigue was defined as the inability of the mouse to maintain an appropriate pace despite continuous hand prodding for 1 min, at which time the mouse was removed from the treadmill and its run time recorded.
The coordination and balance of the mice were measured by a vertical pole test. A plastic pole, 2 cm in diameter and 40 cm long, was wrapped with cloth tape for improved traction. The mice were placed in the center of the pole, in a horizontal position. The pole was then gradually lifted to a vertical position. Normal coordination and balance was defined as the ability of the mouse to stay on the pole beyond a 45° angle.
Muscular abnormalities in the mice were detected by a standard hanging wire test (55). The mice were placed on a wire cage lid, and the lid was slowly inverted. The lid was held at a height approximately 20 cm above the bench. Masking tape placed around the perimeter of the lid prevented the mice walking off the edge. The time till the mice fell off the cage lid was recorded, with a 2-min cut-off time. The test was repeated five times, with a 20 s rest period between trials.
Cardiac function was measured by transthoracic M-mode and two-dimensional echocardiography, as described (31).
This work was supported by grants from the National Institutes of Health (HD83062, NS11766 and AG08702 [to E.A.S.], K02NS047306 [to G.M.], HL73029 [to I.J.G.], T3207343 [to R.K.] and P42ES10340 and P30ES09089 [to J.G.]), the Muscular Dystrophy Association (to E.A.S. and G.M.), the United Mitochondrial Disease Foundation (to R.A.-P.), and the Marriott Foundation (to E.A.S.).
We thank Victor C.S. Lin (Columbia Transgenic/Gene-Targeting Facility) for assistance with obtaining germline transmission of the Sco2 constructs, and Virginia Papaioannou (Columbia Department of Genetics and Development) for helpful advice and discussions.
Conflict of Interest statement. The authors have no conflicts of interest.