Dysfunctions of Cellular Oxidative Metabolism in Patients with Mutations in the NDUFS1 and NDUFS4 Genes of Complex I*

The pathogenic mechanism of a G44A nonsense mutation in the NDUFS4 gene and a C1564A mutation in the NDUFS1 gene of respiratory chain complex I was investigated in fibroblasts from human patients. As previously observed the NDUFS4 mutation prevented complete assembly of the complex and caused full suppression of the activity. The mutation (Q522K replacement) in NDUFS1 gene, coding for the 75-kDa Fe-S subunit of the complex, was associated with (a) reduced level of the mature complex, (b) marked, albeit not complete, inhibition of the activity, (c) accumulation of H2O2 and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{{\dot{2}}}^{-}\) \end{document} in mitochondria, (d) decreased cellular content of glutathione, (e) enhanced expression and activity of glutathione peroxidase, and (f) decrease of the mitochondrial potential and enhanced mitochondrial susceptibility to reactive oxygen species (ROS) damage. No ROS increase was observed in the NDUFS4 mutation. Exposure of the NDUFS1 mutant fibroblasts to dibutyryl-cAMP stimulated the residual NADH-ubiquinone oxidoreductase activity, induced disappearance of ROS, and restored the mitochondrial potential. These are relevant observations for a possible therapeutical strategy in NDUFS1 mutant patients.

The nuclear NDUFS4 gene codes for an 18-kDa subunit of the complex (11), which in high eukaryotes contains potential phosphorylation sites for cAMP-dependent protein kinase in both the presequence and the carboxyl-terminal region (EMBL Data Bank). In mammalian (16 -18) and human (19) cell cultures cAMP promotes the phosphorylation of the NDUFS4 protein and enhances the functional capacity of complex I. Three recessive mutations in the nuclear NDUFS4 gene have been identified in three unrelated children affected by Leigh syndromelike syndrome with deficiency of complex I, including an AAGTC duplication at position 466 -470 in exon 5 (20), a single base deletion at position 289/290 in exon 3 (21), and a G44A nonsense mutation in the first exon of the gene, introducing a premature termination codon in the sequence coding for the mitochondrial leader peptide (13). All three mutations resulted in the disappearance of the 18-kDa subunit and defect in both the activity and assembly of the complex (22). In the 289/290 deletion in exon 3, which predicts the synthesis of an aberrant and prematurely truncated protein, we found almost complete absence of the NDUFS4 transcript (22), apparently degraded by Nonsense Mediated Decay (23). The G44A nonsense mutation in the first exon resulted in stabilization of three alternatively spliced transcript variants of the NDUFS4 gene (24). Recently, large-scale deletion and point mutations in the NDUFS1 gene coding for the 75-kDa Fe-S subunit of complex I have been found (6,25) in children with mitochondrial encephalopathy.
In this report a detailed study is presented on the functional consequences of the G44A mutation in NDUFS4 and of a C1564A mutation in the NDUFS1 gene (Q522K replacement) (6). The assembly and catalytic activity of complex I, mitochondrial energy-transfer, and oxygen-free radical balance were investigated in primary fibroblast cultures of two patients.
Case Report-The patient with the homozygous G44A nonsense mutation in the NDUFS4 gene was a baby diagnosed as a case of Leigh syndrome. The patient with severe lactic acidosis died at the age of 7 months. The homozygous C1564A mutation in the NDUFS1 gene was found in two brothers diagnosed for leukodystrophy; at six months a lactic acidosis appeared in both children, who had a progressive disease course.
Cell Culture and Mitoplast Preparation-Control fibroblasts (neonatal human dermal fibroblasts-neo) and patient's fibroblasts were grown in DMEM supplemented with 10% fetal bovine serum, 2 mM glutamine, 100 IU/ml of penicillin, and 100 IU/ml of streptomycin. The growth of the fibroblasts from the NDUFS4 mutant was depressed (doubling time Ͼ72 h) as compared with the fibroblasts from the NDUFS1 mutant, whose growth was comparable with that of controls (doubling time ϳ24 h).
For mitoplast preparation fibroblasts were harvested with 0.05% trypsin and 0.02% EDTA and washed in phosphate-buffered saline, pH 7.4, with 5% calf serum. Cells in PBS were exposed 10 min on ice to 0.2 mg of digitonin/mg cellular protein. Mitoplasts were pelleted at 14,000 ϫ g and resuspended in PBS.
Rotenone Titration of Cellular Respiration-Respiration was monitored by oxygen polarography.

JOURNAL OF BIOLOGICAL CHEMISTRY 10375
Enzymatic Spectrophotometric Assay-NADH-UQ oxidoreductase activity: mitoplasts were exposed to ultrasound energy for 15 s at 0°C, and V max and K m were determined as in Ref. 22. NADH ferricyanide oxidoreductase activity: mitoplasts were added to 700 l of 50 mM potassium phosphate buffer, MgCl 2 5 mM, pH 7.4, in the presence of 1 g/ml of Antimycin A and 3 mM KCN, and the reduction of 250 M ferricyanide was measured spectrophotometrically (26). Cytochrome c oxidase activity was determined on mitoplasts in 700 l of 10 mM phosphate buffer, pH 7.4, following the oxidation of 10 M ferrocytochrome c (27). Glutathione peroxidase activity was determined as described in Ref. 28. Cells were suspended in PBS and disrupted by exposure to ultrasound.
Electrophoretic Procedures and Antibodies-Two-dimensional gel analysis (blue native PAGE/SDS-PAGE) of the mitoplasts and immunoblotting were carried out as in Ref. 22. Rabbit antiserum against the 20 residues of the carboxyl terminus with phosphorylated Ser-131 of the NDUFS4 protein (18 kDa) was produced by Neosystem, Strasbourg, France. Monoclonal antibodies against other subunits of complex I (39, 20, 17, 12 kDa), the core II subunit of complex III, the subunit IV of complex IV, and the ␤ subunit of complex V were from Molecular Probes (Eugene, Oregon).  (31) and assayed by bioluminescence using a luciferin-luciferase system (Promega).

Laser Scanning Confocal Microscopy Analysis (LSCM)-Cells
Measurement of Glutathione-Cells harvested from Petri dishes with trypsin-EDTA were washed in PBS supplemented with 5% serum, suspended in PBS, and homogenized. After protein precipitation with 2% sulfosalycilic acid, glutathione was determined as described in Ref. 32. For measurement of reduced glutathione, proteins were precipitated with 10% perchloric acid and the supernatant analyzed by high performance liquid chromatography. The samples were loaded onto a Synergi 4 Hydro-RP column (150 ϫ 4.6 mm) and eluted with an isocratic mobile phase of 220 mM potassium phosphate, pH 2.7/acetonitrile (99:1).

RESULTS
Assembly and Catalytic Activities of Complex I-The assembly of complex I in the inner mitochondrial membrane was examined by Western blot of several complex I subunits on two-dimension blue native/SDS-PAGE of the mitoplast fraction isolated from the patient's fibroblasts. The blots in Fig. 1 show that the NDUFS4 G44A null mutation, which causes the disappearance of the 18-kDa subunit, results in the incomplete assembly of complex I, because other subunits of the complex appear adjacent to a complex V subunit, i.e. in a position corresponding to a complex of lower molecular mass (see also Ref. 22). The C1564A mutation in the NDUFS1 gene resulted in decreased amount of a normally assembled complex I and the appearance of a lower molecular mass form of the complex. Interestingly, the 18-kDa NDUFS4 subunit was present in the normally assembled complex but absent in the lower molecular mass form. Table 1 summarizes data on the catalytic activity of complex I and IV (cytochrome c oxidase) in mitoplasts. The NDUFS4 mutation was associated with complete suppression of the forward, rotenone-sensitive NADH-ubiquinone oxidoreductase activity. The NDUFS1 mutation was associated with 80% decrease of the V max of this activity and an apparent increased affinity of NADH. The NADH-ferricyanide, rotenone-insensitive activity of complex I was increased in both mutations, whereas the activity of cytochrome c oxidase was in the normal range.
Titration of the inhibitory effect of rotenone on fibroblast respiration (Fig. 2) showed in the NDUFS1 mutant a decrease of the functional content of the complex (the concentration of rotenone causing 50% inhibition of respiratory rate decreased from 2.4 to 1.5 attomol/cell). The NDUFS4 mutation resulted in loss of the rotenone sensitivity of the residual respiration (Fig. 2).
Mitochondrial Membrane Potential and Cellular ATP Level-The mitochondrial membrane potential was monitored by LSCM on fibroblasts stained with the MitoCapture probe. The cationic form of the probe is accumulated by the potential in the organelle where it emits a red fluorescence in the aggregate high potential state and a green fluorescence in the monomeric low potential state. As shown in Fig. 3 the NDUFS1 mutation caused a significant depression of the mitochondrial membrane potential, whereas the NDUFS4 mutation had no effect on the potential.
Exposure to H 2 O 2 results in depolarization of mitochondria (33). This effect was exacerbated by the NDUFS1, but not by the NDUFS4, mutation (Fig. 3). No significant decrease in cellular ATP was found for either mutation when fibroblasts were grown in the presence of 10% fetal bovine serum (Fig. 4). Under conditions of serum limitation a higher content of cellular ATP was found in both mutant fibroblasts, in particular in the NDUFS4 mutant, as compared with control fibroblasts.
Cellular Reactive Oxygen Species and Scavenger Systems-In the NDUFS1 mutant fibroblasts we found markedly high levels of H 2 O 2 detected by the green fluorescence of 2Ј-7Ј-dichlorofluorescein. However, we found no difference in the H 2 O 2 content of the NDUFS4 mutant fibroblasts compared with control cells (Fig. 5). In the NDUFS1 mutant fibroblasts the H 2 O 2 -DCF green fluorescence merged with the red fluorescence of the mitochondrial probe, Mito Tracker Red, giving an orange/yellow fluorescence (Fig. 5). The MitoSOX probe, a fluorochrome specific to oxygen superoxide (O 2 . ) produced in the inner mitochondrial compartment, showed higher staining in the NDUFS1 fibroblasts as compared with both control and NDUFS4 mutant fibroblasts (Fig. 5).
In contrast with the NDUFS4 mutation, which caused a marked depression, the NDUFS1 mutation did not affect the fibroblast growth (see "Experimental Procedures"). Marked cell shrinkage was, however, detected in NDUFS1 mutant fibroblasts when exposed to H 2 O 2 . No such effect was observed in the NDUFS4 mutant fibroblasts (Fig. 6).  Reverse transcription PCR analysis showed that the transcript levels of both cytosolic CuZn-superoxide dismutase (SOD1) and mitochondrial Mn-superoxide dismutase (SOD2) were within the control range in both NDUFS1 and NDUFS4 mutant fibroblasts (Fig. 7). Moderate increase in both transcript and specific activity of glutathione peroxidase was detected in the NDUFS1 mutant fibroblasts (Fig. 7). In the same cells, the total content of glutathione was reduced, especially the reduced fraction. No such changes were observed in the NDUFS4 mutant fibroblasts (Fig. 7).
Effect of Dibutyryl-cAMP on Functional Parameters of NDUFS1 Mutant Fibroblasts-Exposure of the NDUFS1 mutant fibroblasts to dibutyryl-cAMP resulted in stimulation of the residual forward  control fibroblasts (black bars, a), fibroblasts  from the NDUFS1 mutant patient (gray bars, b), and fibroblasts from the NDUFS4 mutant  patient (empty bars, c). Transcript levels (A) were determined by reverse transcription PCR on extracts of total cellular RNA. Glutathione peroxidase activity (B) and levels of total glutathione and reduced glutathione (C) were determined on cellular extracts as described under "Experimental Procedures." Student's t-test analysis b versus a: *, p Ͻ 0.05; **, p Ͻ 0.01. NADH-ubiquinone oxidoreductase activity of complex I and restored the mitochondrial potential, whereas the H 2 O 2 virtually disappeared (Fig. 8).

DISCUSSION
The homozygous NDUFS1 and NDUFS4 mutations are associated with remarkably different patterns of biochemical dysfunction. These differences can explain the different clinical features in the two patients and may possibly suggest rational therapeutical strategies. The G44A nonsense NDUFS4 mutation was associated with earlier onset and more severe disease course. Similar to other mutations of this gene (22), the mutation caused the disappearance of the corresponding protein (18-kDa subunit), the incomplete assembly of complex I, the complete abolition of the rotenone-sensitive NADH-ubiquinone oxidoreductase activity of the complex, and the marked depression of both respiration and cell growth. The C1564A NDUFS1 mutation, which determines a Q522K replacement in the 75-kDa Fe-S protein of complex I, was associated with a later onset, more progressive disease course. The mutation did not abolish the rotenone-sensitive NADH-ubiquinone oxidoreductase completely, caused little decrease of fibroblast respiration, and did not affect their growth rate. However, the NDUFS1 mutation determined a decrease of the mitochondrial membrane potential, which was not observed in the NDUFS4 mutant fibroblasts. In the latter, complete abolition of complex I activity and marked depression of respiration failed to produce a decrease in the cellular ATP content (Fig. 4), possibly because in cell culture ATP is produced in high amount by both glycolysis and the mitochondrial oxidation of glycolytic NADH mediated by the glycerophosphate shuttle, which bypasses complex I (34). Mitochondrial hydrolysis of glycolytic ATP and/or electron flow through the second and third sites of the respiratory chain can contribute to maintaining the mitochondrial potential. Consistent with this possibility is the observation that under serum limitation growth conditions the cellular ATP content increased, particularly in the NDUFS4 mutant fibroblasts. This compensatory mechanism is clearly insufficient to prevent the progression of the disease in the whole organism and can indeed explain the severe lactic acidosis documented in the patient.
Direct monitoring of ROS level and glutathione showed ROS overproduction in the NDUFS1 mutant fibroblasts. The mutation was associated with a large increase in the level of H 2 O 2 in/around mitochondria, accumulation of O 2 . in the inner mitochondrial compartment, increased expression of glutathione peroxidase, and a decrease in the cellular reserve of glutathione and in its reduction level. No such changes were observed in the fibroblasts from the NDUFS4 patient. These observations substantiate the view that complex I is the major source of O 2 . and ROS derivatives in human fibroblasts (1,35,36). The complete abolition of the normal rotenone-sensitive NADH-ubiquinone oxidoreductase caused by the deletion of the NDUFS4-encoded 18-kDa subunit is likely to result from inhibition of a redox step that is also involved in the direct reduction of O 2 to O 2 . (37). This step, which remains to be identified, might also be controlled by cAMP-dependent phosphorylation of the NDUFS4 18-kDa subunit of complex I (16,19).
In the case of the NDUFS1 mutation, the inhibition of the NADHubiquinone oxidoreductase can be attributed to altered function of the 75-kDa Fe-S protein encoded by this gene. The Q522K substitution can promote direct oxidation by molecular oxygen of the NDUFS1 Fe-S center once it is reduced by NADH. Both mutations are likely to involve redox step(s) below the site where ferricyanide accepts electrons from the complex, because both were associated with increased rotenoneinsensitive NADH-ferricyanide oxidoreduction activity.
In the NDUFS1 mutant fibroblasts, dibutyryl-cAMP stimulated the residual rotenone-sensitive NADH-ubiquinone oxidoreductase activity, determined the disappearance of ROS, and restored the mitochondrial potential. The pattern of complex I assembly in NDUFS4 and NDUFS1 mutant fibroblasts provides clues to explain our biochemical results (cf. Ref. 38). The absence of the 18-kDa subunit blocked a late step in the assembly of a mature functional complex, determining the formation of an inactive subcomplex whose molecular mass was ϳ100 kDa lower than normal (see also Ref. 22). The Q522K substitution in the 75-kDa Fe-S subunit of complex I was associated with reduced level of the fully assembled complex and the presence of a subcomplex of similar molecular mass to the subcomplex observed in the NDUFS4 mutant. The NDUFS4 18-kDa protein detected in the residual amount of the mature complex was absent in the subcomplex. The mutation of the 75-kDa Fe-S protein could have impaired the last step in the assembly of the complex and/or induced oxidative degradation of the complex as a consequence of enhanced ROS production (cf. Ref. 39).
In conclusion, the lack of a completely assembled, functional NADHubiquinone oxidoreductase complex I and the consequent severe acidosis due to accumulation of pyruvate/lactate and other NAD-linked substrates could explain the early onset, fatal course of the disease in the NDUFS4 mutant patient. In the NDUFS1 mutant patient, the partial depression of the NADH-ubiquinone oxidoreductase activity of complex I could explain the less severe clinical course. In this mutation an additional adverse event, however, results from the enhanced production of ROS, which could in turn trigger oxidative stress in a "vicious circle" leading to amplification of biochemical damage and disease progression. Selected antioxidants and ␤-agonists could offer a rational therapeutical strategy in patients carrying mutations in NDUFS1 or in other complex I genes that result in biochemical dysfunction centered on excess of ROS production and oxidative stress.