Saccharomyces cerevisiae has three distinct inner mitochondrial membrane NADH dehydrogenases mediating the transfer of electrons from NADH to CoQ (coenzyme Q): Nde1p, Nde2p and Ndi1p. The active site of Ndi1p faces the matrix side, whereas the enzymatic activities of Nde1p and Nde2p are restricted to the intermembrane space side, where they are responsible for cytosolic NADH oxidation. In the present study we genetically manipulated yeast strains in order to alter the redox state of CoQ and NADH dehydrogenases to evaluate the consequences on mtDNA (mitochondrial DNA) maintenance. Interestingly, nde1 deletion was protective for mtDNA in strains defective in CoQ function. Additionally, the absence of functional Nde1p promoted a decrease in the rate of H2O2 release in isolated mitochondria from different yeast strains. On the other hand, overexpression of the predominant NADH dehydrogenase NDE1 elevated the rate of mtDNA loss and was toxic to coq10 and coq4 mutants. Increased CoQ synthesis through COQ8 overexpression also demonstrated that there is a correlation between CoQ respiratory function and mtDNA loss: supraphysiological CoQ levels were protective against mtDNA loss in the presence of oxidative imbalance generated by Nde1p excess or exogenous H2O2. Altogether, our results indicate that impairment in the oxidation of cytosolic NADH by Nde1p is deleterious towards mitochondrial biogenesis due to an increase in reactive oxygen species release.
- coenzyme Q (CoQ)
- mitochondrial DNA (mtDNA)
- oxidative stress
- respiratory chain
As a petite-positive yeast, Saccharomyces cerevisiae is able to survive even without any mtDNA (mitochondrial DNA). Indeed, S. cerevisiae mtDNA instability has been studied for more than 60 years, since the pioneering work of Boris Ephrussi and the characterization of respiratory-incompetent “petite colonie” mutants . These mutants present large deletions (rho−) or even complete absence (rho0) of mtDNA. Other yeasts such as Kluyveromyces lactis and Schizosaccharomcyes pombe are petite-negative and do not survive in the absence of full-length mtDNA [2,3].
The percentage of rho−/0 cells in S. cerevisiae culture can range from <1% to 100% depending on the nuclear genotype . Interestingly, mtDNA integrity depends on ATP synthase assembly . Impairments in ATP synthase Fo formation, either due to nuclear or mitochondrial point mutations, lead to rapid mtDNA loss, generating rho− cells [3,5]. Consistently, loss of the adenine nucleotide carrier or impairments in ATP synthase F1 formation convert S. cerevisiae to a petite-negative status , whereas K. lactis can become petite-positive when expressing specific atp1, atp2 and atp3 alleles .
mtDNA codes for eight essential polypeptides for the biogenesis of the oxidative phosphorylation apparatus in S. cerevisiae, and three of them (ATP6, ATP8 and ATP9) encode ATP synthase Fo subunits. Considering the importance of a functional ATP synthase for the integrity of mtDNA, mutations that alter mitochondrial protein synthesis apparatus are expected to affect the intactness of mtDNA in the cell . On the other hand, nuclear mutations that lead to dysfunctional cytochrome c oxidase and ubiquinol cytochrome c reductase are not known to significantly affect the stability of mtDNA . Other aspects of mitochondrial metabolism can also indirectly affect mtDNA stability as observed in mutants dysfunctional in fatty acid metabolism, mitochondrial outer membrane structure and ion homoeostasis [3,7–9].
CoQ (coenzyme Q) is responsible for electron transfer from succinate dehydrogenases and NADH dehydrogenases to complex III in the respiratory chain and is also an important cellular antioxidant . Recently, CoQ10 deficiencies have been described in patients with genetic mutations in genes required for CoQ10 function . CoQ10 disorders can also result from the use of drugs such as statins (3-hydroxy-3-methylglutaryl CoA reductase inhibitors) . However, little is known about the pathophysiological consequences of CoQ10 deficiency. Analyses of cell lines from CoQ10 patients exhibit elevations in ROS (reactive oxygen species) production, fragmentation of mitochondria and cell death that correlate with the level of CoQ10 deficiency . The understanding of redox imbalance in the pathological mechanism of CoQ10 deficiency may lead to more rational therapeutic strategies. In this sense, yeast models can be highly useful.
A yeast mutant defective in CoQ respiratory function, coq10, has previously been shown to display an unusually high mtDNA instability, with approximately 50% of cells presenting a rho− phenotype after 30 h in culture . Coq10 differs from other coq mutants with respect to CoQ content, which is equal to wild-type cells . However, electron transfer is interrupted before cytochrome c reduction, and is restored by the addition of synthetic CoQ2, a hallmark of CoQ deficiency .
In the present study we investigated the reasons that lead to mtDNA instability in S. cerevisiae coq10 mutants  through genetic manipulation of NADH dehydrogenase genes. We hypothesized that oxidative stress generated by imbalance in the ubiquinol/ubiquinone ratio and CoQ diysfunction may lead to poor maintenance of mtDNA.
MATERIALS AND METHODS
Yeast strains and growth media
Construction of nde1-, nde2- and ndi1-null alleles
NDE1, NDE2 and NDI1 genes were PCR amplified from yeast nuclear DNA with the respective primers: NDE1 forward, 5′-GCGGATCCCGTCGATCGCATTG-3′, reverse, 5′-GCGGATCCCGCGCTTTCTCTTCG-3′; NDE2 forward, 5′-GCGGATCCCGGATGGCCGGGTAAA-3′, reverse, 5′-GCGGATCCCTCTAGCTACTATATC-3′; NDI1 forward, 5′-GCCTGCAGCGGCAATTGCCACTGGGC-3′, reverse, 5′-GGCCTGCAGGAAAAAACGGTGCC-3′. The 2.6 kb fragment containing the NDE1 reading frame and the 2.2 kb fragment containing NDE2 were digested with BamHI and cloned in YEp352 , whereas the 2.1 kb fragment containing the NDI1 reading frame was digested with PstI and also cloned in YEp352. The three recombinant plasmids were named pNDE1/ST1, pNDE2/ST1 and pNDI1/ST1. The recombinant pNDE1/ST1 plasmid was digested with a combination of KpnI and BclI and ligated to a 1.7 kb BamHI/KpnI fragment containing the yeast HIS3 gene. pNDE2/ST1 was digested with BglII and ligated to a 1.6 kb BamHI fragment containing the yeast LEU2 gene, and finally pNDI/ST1 was digested with a combination of BamHI and SalI and ligated to a 1.1 kb BamHI/SalI fragment containing the yeast URA3 gene. Null alleles, nde1::HIS3, nde2::LEU2 and ndi1::URA3, isolated as linear fragments of DNA, were substituted for the native gene in W303-1A and W303-1B by the one-step gene replacement method .
Construction of NDE1 and Nde1p tagged with HA (haemagglutinin)
A hybrid NDE1 gene coding for Nde1p–HA with nine C-terminal residues constituting the HA tag was obtained by PCR amplification with 5′-GGCGGATCCATGATTAGACAATCATTA-3′ and 5′-GGCCTGCAGTCAAGCGTAGTCTGGGACGTCGTATGGGTAGATAGATGAATCTCTACC-3′ primers. The 1728 bp fragment was digested with BamHI and PstI and fused to GAL10 promoters in YIp351-GAL . The resultant plasmid containing the GAL10-NDE1-HA fusion was linearized and integrated at the chromosomal LEU2 locus in the wild-type W303-1A strain by the one-step gene insertion method.
In order to estimate the rate of spontaneous rho− colony formation, purified rho+ (intact mitochondrial DNA) cultures of the tested strains were first grown in 10 ml of liquid YPD. After overnight growth, cell cultures were diluted into 10 ml of fresh YPD to a D600 of 0.01 for another 16 h of growth at 30°C. Aliquots from every culture were diluted and plated on to YPD plates. The respiratory-competent strains were directly replicated on to selective medium for respiratory activity (YPEG) and the number of viable colonies on each plate was counted manually and compared. YPEG-negative colonies were also crossed on YPD with the tester strain KL14 devoid of mtDNA [19,20] in order to discriminate nuclear Mendelian mutations and cytoplasmic rho−/0. For strains that did not grow on YPEG, e.g. coq2, coq4, bcs1, rip1, cox11 and cox15 mutants, the reported rho− frequency was determined by crossing the colonies on YPD with the tester strain KL14. The diploids were first selected in minimal medium and then replicated on YPEG. coq10 mutants presented slow growth on YPEG, with a doubling time of 12 h . This slow growth allowed the assessment of rho− formation by simple replication on YPEG.
In order to study erythromycin resistance, cells from each strain pre-cultured in YPEG medium were inoculated into 10 ml of YPGal medium and grown at 30°C to a D600 of 10. Aliquots of the cultures were diluted and plated on to YPEG plates where respiration-competent colonies were counted. Leftover cells were plated on to solid YPEG medium containing erythromycin (1 mg/ml). Erythromycin-resistant colonies were scored after incubation at 30°C for 4–5 days for the number of erythromycin-resistant colonies/total number of respiratory-competent cells plated.
Mitochondrial oxygen consumption was monitored using a computer-interfaced Clark-type electrode at 30°C with 10 μmol of NADH as the substrate in the presence of mitochondria at 100 μg/ml of protein concentration. KCN (1 mM) was added at the end of the trace to measure non-mitochondrial oxygen consumption, which was subtracted from all traces.
H2O2 formation in mitochondria was monitored for 10 min at 30°C in a buffer containing 50 μM Amplex Red (Molecular Probes), 0.5 unit/ml horseradish peroxidase (Sigma), 2% ethanol, 1 mM malate, 6 mM glutamate or 30 μmol of NADH and 100 μg/ml of mitochondrial protein. Resorufin formation was recorded using a fluorescence spectrophotometer operating at 563 nm (excitation) and 587 nm (emission) wavelengths. Known amounts of H2O2 were used to obtain a calibration curve, which was used to convert fluorescence units into concentration of H2O2.
Graph generation and statistical analysis
Graphs were generated and statistical analysis was performed using GraphPad Prism 5.00 software. The results are expressed as means±S.E.M. Student's t test (for paired comparisons) or two-way ANOVA (for multiple comparisons) were used for paired and multiple comparisons respectively.
Mitochondria were isolated following the method of Faye et al. , except that zymolyase 20T was used instead of glusulase. Total mitochondrial proteins were separated by SDS/PAGE (12% gels) in the buffer system described by Laemmli  and Western blots were probed with antibodies against HA followed by a second reaction with anti-rabbit IgG conjugated to horseradish peroxidase (Sigma). Antibody–antigen complexes were visualized by the SuperSignal chemiluminescent substrate kit (Pierce).
coq mutants are more susceptible to mtDNA loss than complex III and IV mutants
We evaluated the number of rho− cells in strains defective in CoQ synthesis (coq2 and coq4 mutants) and CoQ transport (coq10 mutant) after 16 h growth in rich glucose medium. COQ2 encodes p-hydroxybenzoate:polyprenyl transferase, the second step of CoQ biossyntesis . Coq4p is also essential for CoQ synthesis . Coq10p has a structured tunnel which probably mediates CoQ binding and transport [14,25]. As observed previously, mtDNA in Δcoq10 cells is significantly unstable . On the other hand, the group of coq mutants presents significantly less stable mtDNA than complex III and IV mutants (Figure 1). Δbcs1 and Δrip1 cells present defective assembly of complex III [26,27], whereas Δcox11 and Δcox15 cells present an arrest in complex IV assembly [28,29].
NDE1 deletion decreases the rate of mtDNA loss in coq mutants
We previously observed that expression of an alternative oxidase from Aspergillus fumigatus  in Δcoq10 mutants partially rescues respiratory growth capacity and improves mtDNA stability of the transformed strains . Alternative oxidase receives electrons directly from ubiquinol , and a possible explanation for the phenotypic improvements observed would be a decrease of ubiquinol in the CoQ pool. In the present study we investigate this hypothesis through the removal of S. cerevisiae NADH dehydrogenases, decreasing CoQ reduction.
Individual deletions of mitochondrial NADH dehydrogenases do not impair respiratory growth, since matrix NADH or cytosolic/intermembrane NADH can be exchanged through the ethanol/acetaldehyde redox shuttle . On the other hand, the double Δnde1Δndi1 mutant displays compromised respiratory growth and the triple Δnde1Δndi1Δnde2 mutant has no respiratory activity (Figure 2A). Consistently, mitochondria isolated from Δnde1 cells showed a diminished capacity to oxidize exogenous NADH and the double mutant Δnde1Δnde2 had almost none (Figure 2B) [33–35]. Nde1p is clearly predominant over Nde2p in the oxidation of cytosolic NADH. The analysis of the steady state level of these two proteins as well as Ndi1p demonstrated that Nde1p is much more abundant than the others (Figure 2C). Nde1p contains 560 amino acids and Nde2p has 545. They are homologous with 58% identity . Curiously, the analysis of the Nde2p–HA C-terminal fusion revealed a prominent band of approximately half the expected size of that observed in Nde1p–HA, which in turn also presents some degradation products.
Through genetic crosses we combined NADH dehydrogenase mutants with Δcoq4 and Δcoq10 mutants, selecting segregants with appropriate auxotrophies. In all crossings, the combination of coq10 deletion with nde1 or ndi1 deletions decreases the percentage of rho− colonies, showing that mtDNA maintenance can be increased under these conditions. This effect on mtDNA maintenance of Δcoq10 cells promoted by NADH dehydrogenase depletion was also observed in the double Δcoq4 Δnde1 mutant, but not in Δcoq4 Δndi1. Deletion of nde2 did not show any effect in mtDNA maintenance in the tested strains (results not shown). The different effects promoted by nde1 and nde2 deletions are consistent with their physiological differences in NADH oxidation and interactions with other dehydrogenases .
Overexpression of NDE1 dramatically enhances mtDNA instability
Since nde1 deletion improves mtDNA stability in Δcoq10 cells, we reasoned that overexpression of NDE1 would promote the opposite effect. In fact, the expression of NDE1-HA under GAL10 promoter control slowed the growth of Δcoq10 and Δcoq4 transformed cells in galactose but not in glucose (Figure 3A). The GAL10 promoter is active in galactose and repressed in glucose . Consequently, mitochondria isolated from cells expressing the GAL10-NDE1 fusion showed a large increase in Nde1p–HA (Figure 3D). Moreover, NDE1 overexpression resulted in a strong increase in the percentage of rho− cells in the parental strain transformed with this construct (Figure 3B). In Δcoq10 cells, GAL10-NDE1 expression practically abolished the presence of rho+ cells. On the other hand, we also tested the effects of overexpressing NDI1 as well as NDE2. Excess Ndi1p or Nde2p did not impair growth on galactose or affect the rate of mtDNA loss. A possible explanation for Nde1p excess leading to mtDNA instability is that Nde1p is a source for superoxide radical generation in yeast mitochondria . In fact, superoxide radicals are produced in mammalian complex I by reduced flavin sites , and the same seems to be applicable to yeast Nde1p.
Since CoQ has well-known antioxidant effects , we questioned whether extra CoQ could rescue cells containing large amounts of Nde1p. We observed previously that yeast mitochondria produce more CoQ if COQ8 is overexpressed . Coq8p is a putative kinase  believed to be involved in the regulation of CoQ synthesis from protein complexes [40,41]. If excess of electron leakage and consequent ROS formation occurs in mitochondria harbouring extra Nde1p, an increase in CoQ content by means of COQ8 overexpression should improve mtDNA stability. This was true in both the parental (wild-type) and Δcoq10-transformed strains (Figure 3B). Indeed, extra COQ8 slightly improved growth of the Δcoq10-overexpressing NDE1 on galactose (Figure 3A).
mtDNA stability in cells overexpressing NDE1 was also evaluated by checking the rate of spontaneous erythromycin resistance appearance in these cells (Figure 3C). This measurement can be used as an estimate of mitochondrial mutation frequency, since erythromycin resistance can arise from mutations in the mtDNA RNA21S gene, through alterations in positions 1050, 1951 and 3993 [42,43]. Consistently, cells with extra NDE1 are more prone to resistance to erythromycin (Figure 3C).
H2O2 release in isolated mitochondria is decreased in CoQ mutants with defective nde1 and elevated with Nde1p excess
One possible explanation for the elevated loss of mtDNA in CoQ mutants is the generation of higher amounts of ROS . We thus measured the rate of H2O2 release from mitochondria isolated from different yeast strains, including those depleted or overexpressing NADH dehydrogenases (Figure 4). H2O2 release was detected using the Amplex Red fluorescence detection system (see the Materials and methods section), which provides a steady-state estimate of electron leakage through the formation of superoxide radical anions and consequent dismutation to H2O2 . In fact, our previous results showing higher H2O2 release in coq mutants relative to wild-type isolated mitochondria  were confirmed with preferential substrates for Ndi1p supplementation (Figure 4A). Under these conditions, the removal of NADH dehydrogenases slightly alters the level of H2O2 release, with the exception of the Δcoq4Δndi1 double mutant, which presented half the H2O2 release in comparison with Δcoq4 mitochondria. However, when NADH (a preferred substrate for Nde1p) was used in the measurements, we observed a dramatic decrease in H2O2 release in mitochondria isolated from cells containing the nde1 deletion, as well as a striking increase in H2O2 release in mitochondria from cells overexpressing NDE1 (Figure 4B). Accordingly, nde1 deletion resulted in 10-fold less H2O2 release, whereas NDE1 overexpression elevated H2O2 release 4-fold in the strains tested. Interestingly, Δcoq10 cells have less Nde1p than wild-type and Δcoq4 cells (Figure 3D). Consequently, in the presence of NADH as the respiratory substrate, Δcoq10 mitochondria released less H2O2. Although the effect of ROS on mtDNA stability of the yeast strains studied needs to be further explored, H2O2 release measurements correlated with the rate of rho− cell formation in strains overexpressing NDE1. Figure 5 presents a scheme of our results in representative tested strains and substrates.
Addition of exogenous H2O2 promotes mtDNA instability in a manner partially rescued by COQ8 overexpression
In order to corroborate our hypothesis that mtDNA instability can be due to elevated ROS in cells defective in CoQ function, we challenged wild-type cells, Δcoq4 and Δcoq10 mutants with 5 mM H2O2 for 6 h. In parallel, the same strains transformed with pTEF1-COQ8  were also analysed in order to verify a possible benefit when CoQ synthesis is stimulated [14,40,41]. All strains tested showed a significant decrease in cell viability after the treatment (Figure 6A), and the presence of extra Coq8p improved the respiratory growth of wild-type and Δcoq10 transformants after H2O2 treatment. This observation was confirmed by counting the number of rho+ colonies after treatment in comparison with non-treated cells (Figure 6B). COQ8 excess partially protected mtDNA in the wild-type and Δcoq10 cells, but not in Δcoq4, as expected owing to the complete absence of CoQ in this mutant.
Petite mutations are characterized by their high rate of spontaneous occurrence. The proportion of petite cells in a culture is highly dependent on the genetic background of the strain, even when the genetic markers are not directly, or indirectly, required for mtDNA replication [3,4]. Δcoq10 mutants present unexpected mtDNA instability and are defective in CoQ function even though they have normal amounts of mitochondrial CoQ . In the present study we observed that coq mutants are more prone to mtDNA loss than complex III and IV respiratory mutants (Figure 1). This can be related to the antioxidant role of CoQ , as well as the exacerbation of electron leakage in CoQ reduction sites at mitochondrial NADH dehydrogenases, more specifically Nde1p, in S. cerevisiae. Accordingly, isolated Δnde1 mitochondria release roughly 10-fold less H2O2 in assays using NADH as a substrate, suggesting that Nde1p is an important point of electron leakage in yeast. Indeed, electron leakage and superoxide radical formation is well documented in mammalian complex I flavin reduced sites such as those found in yeast NADH dehydrogenases . Indeed, in the respiratory electron transport chain, complexes III and I have been recognized as important intracellular sources of ROS . Moreover, we previously observed that complex III-specific inhibitors such as antimycin A and myxothiazol elevated the rate of H2O2 release in S. cerevisiae , as well as mutations in genes required for complex III assembly [31,46]. mtDNA maintenance of bcs1 mutants were analysed in the present study, and the rate of rho− formation was not significantly different from wild-type (Figure 1), although the measured rate of H2O2 release was previously detected to be 3-fold higher than the wild-type, and similar to the coq10 mutant .
On the other hand, deletion of Δnde1 and Δndi1 altered mtDNA maintenance and H2O2 release in coq mutants (Figures 2D and 4). Decreases in H2O2 release have been described before in nde1 and nde2 mutants subjected to heat stress . However, mtDNA maintenance in the double mutants needed better scrutiny, since the genetic crosses employed to construct new strains allow the onset of new polymorphisms that may affect, by other means, mtDNA maintenance . Actually, Δcoq4Δndi1 cells have less stable mtDNA than the single mutants and this result did not correlate with H2O2 release of the respective mitochondria (Figure 4A). This can be explained by a possible change in the NADH/NAD+ ratio in the mitochondrial matrix of these cells and the consequent increase in superoxide radical formation at other matrix sites [20,48]. The combined disruptions of nde1 and ndi1 also promoted higher mtDNA instability (results not shown), further supporting the negative effect of ndi1 absence. However, it is important to note that the amount of substrates (30 μM NADH, 1 mM malate or 6 mM glutamate) added to isolated mitochondria in the H2O2 measurements may be saturating and may not correspond to the physiological conditions in which the rho− cells arise. Furthermore, H2O2 measurements were conducted in isolated mitochondria under non-phosphorylating conditions, which tend to maximize ROS production , but may not relate directly to oxidant release rates within cells.
Consistent with the hypothesis of Nde1p as a source of ROS, overexpression of NDE1 under GAL10 promoter control was notably deleterious for Δcoq10 and Δcoq4 cells, postponing growth on galactose and promoting the complete loss of mtDNA 48 h after promoter activation (results not shown). On the other hand, the increase of CoQ in these cells, obtained through overexpression of COQ8 [14,40,41] was beneficial for growth, protective against oxidant challenge growth and improved mtDNA maintenance (Figures 3 and 6).
CoQ is a key component of the mitochondrial respiratory chain, fundamental as an antioxidant and in the maintenance of the redox homoeostasis of the cell. In addition to respiratory disfunction, S. cerevisae coq mutants present hypersensitivity to oxidized polyunsaturated fatty acids and elevated formation of lipid hydroperoxides . Enhanced lipid peroxidation promoted by exacerbated ROS production also changes mitochondrial lipid composition, affecting mtDNA maintenance [3,7] and promoting apoptosis in human cells . ROS overproduction was also detected in CoQ yeast mutants [31,39,47], and in cell lines obtained from patients with primary CoQ deficiencies . In the present study we show that Nde1p electron leakage due to accumulation of ubiquinol (Δcoq10 mutants), total absence of CoQ (Δcoq4 mutants) or Nde1p excess potentiates ROS formation, which correlates with the higher generation of rho− cells in these strains.
It is possible that ROS generated in the intermembrane space, through Nde1p electron leakage, are more deleterious for mtDNA maintenance than those primarily generated at the bc1 complex matrix sites , perhaps because of the antioxidant system present in each compartment  (Figure 5). ROS are widely known to promote oxidative damage to proteins, lipids and DNA. It is likely that the cause for mtDNA loss in CoQ mutants is a conjunction of factors. ROS excess seems to play a central role in this phenotype by affecting mitochondrial lipid composition or by direct damage to DNA. Our proposed correlation between oxidative stress derived from CoQ impairments and mtDNA instability in S. cerevisiae could model a pathological condition in patients with primary CoQ deficiency, which worsens their clinical condition.
Mario H. Barros conceived the project, Fernando Gomes, Mario H. Barros and Cleverson Busso performed the research, and all authors analysed the results and participated in preparation of the paper.
This work was supported by the Fundação de Amparo a Pesquisa de São Paulo (FAPESP) [grant number 2011/07366-5], Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), INCT de Processos Redox em Biomedicina-Redoxoma (CNPq-FAPESP/CAPES) and Nucleo de Apoio à Pesquisa de Processos Redox em Biomedicina (NAP-Redoxoma).
Abbreviations: CoQ, coenzyme Q; HA, haemagglutinin; mtDNA, mitochondrial DNA; ROS, reactive oxygen species; rho0, complete abscence of mitochondrial DNA; rho−, mitochondrial DNA with large deletions; rho+, intact mitochondrial DNA
- © The Authors Journal compilation © 2013 Biochemical Society