Mitochondria are organelles that have their own DNA (mitochondrial DNA, mtDNA) whose maintenance is necessary for the majority of ATP production in eukaryotic cells. Defects in mtDNA maintenance or integrity are responsible for numerous diseases. The DNA polymerase γ (POLG) ensures proper mtDNA replication and repair. Mutations in POLG are a major cause of mitochondrial disorders including hepatic insufficiency, Alpers syndrome, progressive external ophthalmoplegia, sensory neuropathy and ataxia. Mutations in POLG are also associated with parkinsonism. To date, no effective therapy is available. Based on the conservation of mitochondrial function from yeast to human, we used Saccharomyces cerevisiae and Caenorhabditis elegans as first pass filters to identify a chemical that suppresses mtDNA instability in cultured fibroblasts of a POLG-deficient patient. We showed that this unsuspected compound, clofilium tosylate (CLO), belonging to a class of anti-arrhythmic agents, prevents mtDNA loss of all yeast mitochondrial polymerase mutants tested, improves behavior and mtDNA content of polg-1-deficient worms and increases mtDNA content of quiescent POLG-deficient fibroblasts. Furthermore, the mode of action of the drug seems conserved as CLO increases POLG steady-state level in yeast and human cells. Two other anti-arrhythmic agents (FDA-approved) sharing common pharmacological properties and chemical structure also show potential benefit for POLG deficiency in C. elegans. Our findings provide evidence of the first mtDNA-stabilizing compound that may be an effective pharmacological alternative for the treatment of POLG-related diseases.
Mitochondrial disorders are a large and heterogeneous group of different orphan diseases for which no satisfactory therapy is presently available (1,2). These diseases are characterized by mitochondrial respiratory chain (RC) deficiency with a wide variety of genetic causes and represent one of the major causes of metabolic disorders, with a frequency of 1/5000 births. A subgroup of mitochondrial disorders is related to mutations in the mitochondrial DNA (mtDNA) polymerase γ (POLG) gene (3). Mutations in the POLG gene lead to the accumulation of base substitutions, mtDNA deletions or mtDNA depletion that result in an impaired energy production via oxidative phosphorylation. To date, more than 250 pathogenic mutations in the POLG gene have been reported (http://tools.niehs.nih.gov/polg). They are associated with a broad spectrum of mitochondrial disorders including dominant, recessive or sporadic progressive external ophthalmoplegia (PEO) (OMIM #157640 and #258450), the fatal neonatal hepatic failure, Alpers syndrome (#203700), mitochondrial recessive ataxia syndrome (including sensory ataxic neuropathy, dysarthria, and ophthalmoparesis and spinocerebellar ataxia with epilepsy) (#607459) and the mitochondrial neurogastrointestinal encephalopathy syndrome (#613662). In addition, POLG pathogenic variants have been shown to be associated with myopathy, Parkinsonism, premature menopause, psychological disorders, ataxia, encephalopathy and some forms of Charcot–Marie–Tooth syndrome and of Leigh syndrome, diseases that were generally related to defects in other genes (reviewed in 4,5). Dominant POLG mutations result in PEO associated with multiple mtDNA deletions, whereas recessive mutations result mainly either in PEO with multiple mtDNA deletions or in fatal neonatal hepatic failure or Alpers syndrome with mtDNA depletion (2–10% of normal mtDNA content).
Multiple alignments of the POLG amino acid sequences from human and other vertebrates, arthropods, worms, protozoa and fungi proteins indicate highly conserved regions in both the polymerase and exonuclease domains. Based on the similarity of yeast mitochondrial polymerase Mip1 and human POLG, Saccharomyces cerevisiae has been used as a model organism to mimic human mitochondrial diseases associated to POLG mutations reviewed in (6). This yeast has allowed the discrimination between pathogenic mutations versus single-nucleotide polymorphisms (7). A good correlation between the impaired mtDNA maintenance in yeast mip1 mutants and disease severity in humans has been observed. Furthermore, human POLG can efficiently complement the yeast mip1 deletion mutant, showing growth rate and spontaneous mtDNA mutation rate similar with that of MIP1 wild-type (wt) cells (8). Thus, despite the absence of an accessory subunit in the yeast POLG and the differences in the mtDNA structure compared with humans, common fundamental mechanisms of replication are retained to support mtDNA replication.
Regulation of mtDNA replication is poorly understood. Unlike nuclear DNA, which only replicates once during cell division, mtDNA is continuously recycled (9); however, it does not seem to occur randomly. The mtDNA is organized into nucleoprotein structures called nucleoids, which are distributed throughout the mitochondrial network. In yeast, most of the division events result in a nucleoid present in each new mitochondrial tubule tip generated upon fission (10). Nevertheless, the exact mechanism that drives mtDNA replication remains unsolved.
Recently, Caenorhabditis elegans has been proved to be an excellent model animal to study mechanisms of mtDNA inheritance and maintenance in both post-mitotic and proliferative tissues (11). A C. elegans model of POLG deficiency demonstrated that the activity of POLG-1 is critical for development and reproduction (11,12). POLG-1 deficiency causes severe mtDNA depletion, but no developmental arrest. In the absence of a functional POLG-1, a high-maternal contribution of both mtDNA and POLG-1 in the embryo ensures normal development. However, those animals are sterile (progenies are arrested at different embryonic stages) and die sooner due to a fatal gonadal and intestinal protrusion (11).
No effective treatment for POLG mutation-associated disorders exists to date. In order to identify molecules that could represent therapeutic drugs active against these mitochondrial diseases, we established a yeast-based screening assay. The relevance of such assay has been previously demonstrated in the case of mtDNA ATP6 mutations related to NARP syndrome (13,14). In yeast, the deletion of the MIP1 gene results in an extremely rapid and complete loss of mtDNA, hampering any drug screening. However, some pathogenic human POLG mutations introduced in the MIP1 gene lead to a temperature sensitive (ts) phenotype on the respiratory medium due to the induction of mtDNA defects, enabling drug-screening assay. A collection of 1500 compounds from two drug-repurposing libraries (Prestwick and TEBU-BIO) was used and allowed the identification of several molecules able to rescue the mip1-respiratory defect. Among them, clofilium tosylate (CLO) was able to rescue mip1 mtDNA instability. We further showed that CLO relieved POLG defect consequences of a polg-1-deficient worm, an animal model offering the possibility to measure the drug impact on physiological homeostasis at the scale of an entire organism. Finally, CLO was found active in fibroblasts of a patient with POLG mutations and mtDNA depletion. Our data indicate that CLO could stabilize the POLG protein in budding yeast, but also in human cells suggesting a common mechanism of CLO rescue. Altogether, our results showed that CLO prevents mtDNA instability in all yeast mip1 mutants tested, improves the reproductive defects and the mtDNA content of polg-1-deficient worms, and increases the mtDNA content of quiescent POLG-deficient fibroblasts.
Identification of CLO, a molecule rescuing the effects of mip1 mutations in yeast
Taking advantage of the high level of conservation between mtDNA polymerases from yeast to human (S. cerevisiae Mip1, C. elegans POLG-1 and human POLG), a wide battery of mip1 mutant alleles was previously constructed by introducing human POLG mutations in the corresponding position of the MIP1 gene (15–17). Yeast strains harboring these mutant alleles displayed growth defects with varying severity on the ethanol-based medium and different levels of mtDNA instability, i.e. point mutations, deletions (rho−) or loss (rho0) of the mtDNA (used mip1 mutations in this study are listed in the Supplementary Material, Table S1). This results in the appearance of small colonies (i.e.petite colonies) on differential medium (low glucose and non-fermentable carbon source). Some of these mip1 mutant strains exhibited temperature sensitive (ts) growth defect on ethanol. At permissive temperature (28°C), the low frequency of petite colonies in some of the ts strains allowed the growth on respiratory carbon sources. At the non-permissive temperature (37°C), the frequency of petite drastically increased, thus impairing growth on ethanol. The ts phenotype of these mutants allowed us to develop a yeast-based screening assay in order to select molecules able to suppress the respiratory growth defect at 37°C. For chemical libraries screening, we chose the ts mip1G651S recessive mutation, located in the polymerase domain and equivalent to human POLG G848S mutation, which exhibits in yeast a mild mtDNA instability and presents the lower growth background at 37°C, making it easy to identify molecules which rescued growth.
One thousand and five hundred different molecules from two chemical libraries (TEBU-BIO and Prestwick for which toxicity and bioavailability studies have been carried out in humans) were tested for their ability to suppress the respiratory growth defect of the mip1G651S mutant strain, by drug drop test (13,14). Mutant cells were spread on rich ethanol plates and exposed to filters spotted with the compounds. Among the 30 primary hits identified by a halo of enhanced growth around the filter after 5 days of growth on ethanol plates at 37°C (not shown), we chose to study in more details the compound CLO, an anti-arrhythmic drug that allowed a strong enhanced growth with low toxicity (Fig. 1A). Further studies showed that CLO was able to suppress, in a dose-dependent manner, the respiratory growth defect of the mip1G651S mutant strain (Fig. 1B).
Analyses of CLO-rescuing effects on the mip1G651S mutant strain
To determine the CLO-optimal working concentration, the mip1G651S mutant strain was grown with increasing concentrations of CLO. The highest non-toxic concentration of CLO is 128 µm as increasing concentrations resulted in reduced growth of the mip1G651S cells (Supplementary Material, Fig. S1). The petite frequency in growing cells is influenced by two factors: the frequency of petite production from rho+ cells and the different growth properties of petite and rho+ cells (18). We found that the optimal CLO concentration was 32 µm for which no negative effect on petite cell growth (fitness) was observed, so that the difference between the petite frequency in the presence or absence of CLO was only due to a different petite production.
We further investigated whether the rescue of the mip1G651S growth defect by CLO might be due to a decrease of the mip1G651S mtDNA instability. We measured the frequency of petite in the haploid Δmip1 strain harboring the mip1G651S mutant allele. The strain was grown at 28°C in the glucose rich medium for 48 h with either DMSO (dimethyl sulfoxide, compound vehicle) or 32 µm of CLO. Only 32.6 ± 4.7% petite cells were observed when mip1G651S cells were grown with CLO whereas 57.3 ± 1.6% of petite cells appeared in DMSO (Fig. 1C). Therefore, CLO treatment resulted in a 43.1% decrease of petite cells production.
Using ethanol as an electron donor, we then measured the oxygen consumption rate of the wt MIP1 and mip1G651S whole cells grown in ethanol at 28°C with or without CLO. The mip1G651S mutant presented a 1.25-fold decreased of O2 consumption rate (VO2) when compared with the wt strain (Fig. 1D). Treatment with CLO resulted in a significant increase (2-fold) in the respiration rate of mip1G651S. Noteworthy, CLO also increased by 1.8-fold the O2 consumption rate of the wt strain. We also measured the maximal respiration rate (uncoupled state) in whole cells, in the presence of the mitochondrial potential uncoupler, carbonyl cyanide m-chlorophenylhydrazone (CCCP). In this condition, the respiration rates of mip1G651S and of wt strains, treated or not with CLO, were similar and significantly (2-fold) increased. This result indicates that CLO does not change or dissipate the mitochondrial membrane potential (passive permeability to protons).
We, next, analyzed the Mip1 steady-state level in cells grown with or without CLO. Mip1 was tagged with a HA epitope in wt and mip1G651S genetic contexts. The Mip1 steady-state level determined by western blot was significantly increased after CLO exposure in both genetic backgrounds (Fig. 1E). This was not related to an overall increase of mitochondrial biogenesis since the level of the mitochondrial porin (Por1, Fig. 1E) was unchanged and as the expression of the β-galactosidase reporter gene under the control of two different promoters responding to the global respiratory gene regulators (HAP2/3/5/4 complex) was not modified after the addition of CLO (Supplementary Material, Fig. S2). In addition, the amount of MIP1 mRNA quantified by real-time quantitative polymerase chain reaction (RT-qPCR) and normalized by the ACT1 mRNA level was not modified by CLO addition (Supplementary Material, Fig. S3), suggesting that CLO may stabilize the Mip1 protein.
Finally, we investigated the effect of CLO on the number of mitochondrial replicating nucleoids (replisome) in wt yeast cells. To this end, we analyzed the distribution of Mip1-Green Fluorescent Protein (GFP) foci in cells grown with or without CLO. As previously shown (19), signals of Mip1-GFP appeared in the cytosol as small dots (foci). After the addition of increasing concentrations of CLO to the medium, the number of Mip1-GFP foci increased (Fig. 1F).
CLO rescues other mip1 mutants
Analysis of petite frequency was carried out in the haploid Δmip1 strain carrying seven other ts recessive mip1 mutant alleles. In all strains, a reduction of petite frequency was observed after CLO addition (Fig. 2A). The effect of CLO was quite modest for mutations G259R, C261R, H734Y (2–10% of petite reduction), moderate for mutations A692 T and P829L (30–40% of petite reduction) and stronger for mutations R467W, G651S, G807R (43–63% of petite reduction). We noticed an inverse correlation between the frequency of petite production due to mip1 mutations and the rescuing effect of CLO (Fig. 2A), i.e. the greatest CLO effect was observed for G807R that showed a mild mtDNA instability (23.8% of petite) while the lowest CLO effect was observed for the harmful H734Y change (99.6% of petite). Tested mutations were located in the various domains of the Mip1 protein (polymerase, linker or exonuclease domains) and the efficiency of CLO-rescue displayed no relation to mutation location.
The analysis was also carried out in heteroallelic strains harboring the wt MIP1 gene as well as the mutant alleles mip1E698G, mip1K749R or mip1Y757C, corresponding to dominant mutations in human. As shown in Figure 2A, a rescue of petite production by CLO was observed for each mutant allele, though at different extent, ranging from ≈30% of petite decrease for Y757C to ≈70% of petite decrease for K749R. Again, an inverse correlation was observed between the petite frequency induced by these dominant mutations and rescue efficacy. Altogether these results clearly show that CLO strongly reduces the mtDNA instability due to all mip1 pathological mutations tested.
CLO does not change the polymerase fidelity
In yeast, the mitochondrial protein synthesis is specifically sensitive to the antibiotic erythromycin, whose target is the mitochondrial ribosome. Erythromycin resistant (EryR) mutations may arise by spontaneous point mutations in the mitochondrial 21S rRNA gene, consequently to the incorporation of an incorrect nucleotide by the Mip1 protein during replication. Thus, frequency of EryR mutations can be considered as an index of mtDNA replication fidelity (20,21). As expected (15,16), an increased frequency of EryR mutants was observed in all mip1 mutants compared with the wt strain (Fig. 2B). The addition of CLO did not change significantly the mtDNA mutagenesis frequency of all the mip1 mutant strains tested. Therefore, even though CLO increased mtDNA stability as shown by the lower production of petite colonies, it did not change the mtDNA replication fidelity of the mip1 mutator alleles, underlying that CLO is not mutagenic.
CLO-rescue is specific to mtDNA replication defect
In S. cerevisiae, hundreds of genes are directly or indirectly involved in faithful maintenance of the mtDNA (18). We investigated the potential rescue effect of CLO on three mtDNA instability mutants associated with mutations in the MIR1 gene (encoding a mitochondrial phosphate carrier) and genes involved in mtDNA repair and recombination such as CCE1 and PIF1. The CCE1 gene encodes a mitochondrial cruciform cutting endonuclease that cleaves Holliday junctions formed during recombination of the mtDNA. The PIF1 gene encodes a DNA helicase, with both nuclear and mitochondrial isoforms, which one is involved in repair and recombination of the mtDNA (22). The cce1, pif1 and mir1 mutants present a high frequency of mtDNA rearrangements and a high level of petite accumulation [rho− (23)]. The frequency of petite accumulation in Δcce1 and Δmir1 cells was not modified by exposure to CLO (Fig. 2C). This result suggests that CLO could not significantly rescue alteration in the mitochondrial recombination process as well as the mtDNA instability due to impaired mitochondrial inorganic phosphate import. However, a modest but significant decrease (13.8%) of petite mutants was noticed in the Δpif1 strain treated with CLO, suggesting that the drug can moderately restore a mtDNA repair defect, which requires mtDNA synthesis.
The CLO-rescue of mip1-induced mtDNA instability is not related to dNTP pool availability
It has been previously demonstrated that mtDNA instability caused by mip1 mutations can be partially suppressed by increasing the dNTP pool availability by either overexpressing the RNR1 gene that encodes the large subunit of the ribonucleotide reductase (RNR), or by deleting the SML1 gene that encodes a RNR inhibitor (17,24,25). Increased concentration of dNTPs is beneficial because the basal level of mitochondrial dNTP pool is a limiting factor to maintain mtDNA copy number in yeast (26). Yeast strains harboring mip1G651S and mip1A692T mutant alleles were deleted for the SML1 gene and the frequency of petite colonies was determined after growth with or without CLO. As expected, a drastic reduction of petite production for both mip1 mutations was observed when the dNTP pool availability increased due to the sml1 deletion (Fig. 2D). CLO exposure induced a further reduction of petite frequency. In addition, a similar petite frequency was observed in the mip1G651SΔsml1 strain (13.2 ± 0.9 and 6.4 ± 0.5 with DMSO and CLO, respectively) and in the mip1G651SΔsml1 strain overexpressing RNR1 (14.5 ± 0.7 and 6.1 ± 0.4 with DMSO and CLO, respectively) suggesting that the maximal levels of dNTPs necessary to fully sustain the mtDNA replication in wt or mutant mip1 cells were reached in the absence of SML1. Overall, these results indicate that CLO has an additive effect to the increased dNTP pool availability, strongly suggesting that CLO may act through a different pathway.
Effect of CLO on a C. elegans POLG-1-deficient animal
Caenorhabditis elegans offers the possibility to evaluate the impact of compounds at the scale of an entire organism. Therefore, we evaluated the CLO effect on a worm model of POLG deficiency. In C. elegans, only very few mitochondrial genomes are sufficient to support life, but more genomes are required for energy-intensive processes such as reproduction (11,12). To investigate the in vivo CLO effect on polg-1-deficient worms, we used a heterozygous deletion mutant polg-1(ok1548/+) carrying a 2149 bp deletion that removes Exons 8–10 encoding part of the POLG-1 polymerase domain. These heterozygous polg-1(ok1548/+) mutant animals behave as wt animals, but exhibit a lower brood size compared with N2 control animals (11). CLO was provided to the worms through the media. Non-toxic dose of CLO was first determined on wt animals as the highest concentration that did not lead to developmental delays compared with DMSO treatment. Exposure of embryos until adulthood to 5, 10 and 50 µm of CLO did not reveal abnormal development schedule. CLO concentrations higher than 100 µm were toxic. Therefore, the highest CLO concentration used in the following experiments was 50 µm.
Positive effects of CLO on wt animals
We first analyzed the effect of CLO on mitochondrial homeostasis of wt worms. For that purpose, we looked at the mitochondrial unfolded protein response, a stress response that activates transcription of nuclear encoded mitochondrial chaperone genes to promote protein homeostasis within the organelle (27). We used a reporter animal expressing GFP under the control of the mitochondrial chaperone hsp-6 (Hsp70 family) gene promoter. In this animal GFP expression is low in basal condition, but increases significantly when mitochondrial stress is induced. CLO treatment resulted in a slight increase of GFP expression compared with DMSO treatment, but was considerably lower than the huge GFP expression after animal exposure to 10 µg/ml of ethidium bromide (EtBr) that is known to induced a mitochondrial stress [Fig. 3A (28)]. This suggests that CLO has a minor effect on mitochondrial proteostasis, but negligible compared with that induced by EtBr. We next analyzed the effect of CLO on mitochondrial nucleoid. The mitochondrial single-strand DNA-binding protein, encoded by the mtss-1 gene, coats newly synthesized mtDNA molecules and expression of GFP fused to MTSS-1 in the worm has been shown to mark replicating nucleoids in living animals (29). The effect of CLO on in vivo mtDNA replication was studied by the use of a transgenic animal expressing a MTSS-1::GFP fusion protein under the control of its own promoter (29). As expected, the GFP signal in MTSS-1::GFP transgenic animals appeared as punctuated foci. Animals treated with 50 µm CLO showed an increased number of MTSS-1::GFP foci compared with the untreated controls (Fig. 3B). This suggests that CLO may increase mtDNA replication.
Positive effects of CLO on polg-1 mutant animals
To investigate the effect of CLO on mtDNA stability in polg-1 mutant worms, we first tested the EtBr sensitivity of polg-1(ok1548/+) animals in the presence or absence of CLO. A complete blockage of mtDNA replication and transcription induced by high dose of the DNA-intercalating dye EtBr has been reported to result in developmental arrest at the L3 stage (12). We previously showed that an increased sensitivity of polg-1 RNAi worms toward EtBr exposure was specific to a low mtDNA content of those animals (29). Exposure of embryos to 30 µg/ml of EtBr for 4 days resulted in 86% of polg-1(ok158/+) mutants arrested at the L3 stage compared with only 50% for N2 (wt) worms, indicating a lower mtDNA content in polg-1(ok158/+) mutants than in N2 (Fig. 3C). In order to evaluate a possible effect of CLO, synchronized L1 polg-1(ok158/+) larvae were grown on plates containing 30 µg/ml of EtBr with or without 50 µm of CLO. Only 22% of the polg-1 mutant worms were arrested at the L3 stage when exposed to CLO compared with 86% in its absence, the remaining worms reaching the adulthood. Noteworthy, CLO also increased the EtBr tolerance of wt animals since 5% of N2 CLO-treated animals were arrested at the L3 stage compared with 50% for untreated N2 animals. A dose–response experiment showed that the effect of CLO on polg-1(ok158/+) EtBr sensitivity was visible down to 0.1 µm and optimal at 50 µm for CLO (Fig. 3D). CLO reduced EtBr sensitivity of polg-1(ok158/+) mutants indicating that the EtBr dose used was not sufficient to inhibit the mtDNA replication required to reach adulthood in the presence of CLO, thereby suggesting that the mtDNA content of these CLO-treated animals was higher than the untreated ones (see below). Finally, to investigate CLO consequences on the whole nematode organism, O2 consumption rate of polg-1(+/−) mutant worms was measured in the presence and absence of CLO (Fig. 4E). The O2 consumption rate of polg-1(+/−) mutant worms was 1.8-fold less than wt animals and addition of CLO was able to improve the O2 consumption rate deficiency of the polg-1(+/−) mutant worm near wt level (80%).
Homozygous polg-1(ok1548) animals have been reported to develop rather normally, due to a high maternal contribution of POLG-1, but to produce very few eggs (11). Indeed, polg-1(ok158) mutant animals produced an average number of only 17 eggs/worm (Fig. 4A) whereas wt N2 animals laid ∼300 eggs/worm (not shown). Treatment of polg-1(ok158) mutants with 50 µm CLO increased by 2-fold the number of laid eggs (Fig. 4A).
It has been previously shown that all eggs produced by polg-1(ok158) mutants arrested at early stages of embryonic development. Rarely, the first embryo of a young adult may hatch, but the larva was not viable (11). We analyzed the progeny of polg-1(ok158) mutants treated or not with CLO from the L4 stage (Fig. 4B). For untreated animals, an average of only one embryo hatched in the progeny of five young polg-1(ok158) adults. In the presence of CLO, at least one embryo per polg-1(ok158) worm hatched. However, in all cases, those larvae did not pass the L2 stage. These differences in the embryonic development are determined by the maternally inherited mtDNA levels, indicating that CLO may have increase the mtDNA content of the polg-1(ok158) embryos, even though the larvae were not viable due to the complete absence of a functional polymerase γ.
Marked mtDNA depletion in polg-1(ok158) adult worms leads also to a shortened lifespan due to an early fatal gonadal and/or intestinal protrusion through the vulva (11). The animals die sooner due to starvation because their movements are impaired. Gonadal/intestinal protrusion was proposed to be the result of vulva muscles weakening due to mitochondrial dysfunction (11). We indeed observed a protrusion phenotype, starting from Day 6 of adulthood to Day 9, for >70% of the untreated polg-1(ok158) adults (Fig. 4C). CLO treatment from the L4 stage resulted in an important decrease of protruding animals in the population (46% in the presence of CLO compared with 70% with the compound vehicle, DMSO).
It has been evaluated that the mtDNA copy number in polg-1(ok1548) somatic cells of adult worms was 10-fold reduced compared with wt N2 animals (11). In order to evaluate the effect of CLO on mtDNA content in somatic cells of polg-1(ok1548) mutants, N2 and polg-1(ok1548) L1 larvae were treated or not with 50 µm of CLO until Day 6 of adulthood and mtDNA amount was then determined by qPCR. As expected, polg-1(ok1548) worms showed a 10-fold reduction of mtDNA levels compared with N2 animals (Fig. 4D). Meanwhile, a 2.3-fold increase of mtDNA content in polg-1(ok1548) worms treated with 50 µm of CLO was observed. Interestingly, a 1.9-fold increase of mtDNA content in wt animals treated by CLO was also observed.
CLO increases the mtDNA content in fibroblasts of a POLG patient
We then studied the effect of CLO on the mtDNA content of cultured skin fibroblasts from a patient with compound heterozygous POLG mutations (R807P/D1184N). This patient presented with hepatic failure and severe mtDNA depletion in the liver (25% of normal mtDNA amount). We used fibroblasts with moderate phenotype, namely normal RC activities (Supplementary Material, Table S2), decreased POLG amount and moderate mtDNA depletion (50% of normal mtDNA amount), as we hypothesized that it should be hopeless to increase POLG activity and therefore mtDNA content in cells with too severe phenotype.
The efficiency and toxicity of CLO on human fibroblasts has been first determined with the xCELLigence system by following the proliferation of cells incubated with increasing CLO concentrations. Control and POLG patient's fibroblasts incubated for 5 days in 10 µm of CLO did not show any cell proliferation suggesting that at 10 µm CLO is toxic. In contrast, 2.5 µm of CLO allowed increased cell growth of control and POLG patient's fibroblasts compared with DMSO treated cells therefore indicating a positive effect on cell proliferation and allowing to consider 2.5 µm of CLO as the highest non-toxic concentration (Supplementary Material, Fig. S4).
Although mtDNA synthesis is cell-cycle independent in control fibroblasts, it has been previously reported that it mainly occurs during S phase in fibroblasts of patient with mutations in the DGUOK gene. DGUOK encodes the mitochondrial deoxyguanosine kinase that allows the first step of the dNTP salvage pathway (30). Moreover, when cell cycle is inhibited by serum deprivation in control fibroblasts, mtDNA replication relies only on the dNTP salvage pathway as source of dNTP (31,32). We, therefore, investigated the mtDNA copy number of control and POLG fibroblasts maintained for 18 days in the low-serum medium (0.1%) designed as quiescent condition. The mtDNA copy number in control fibroblasts showed a 20% decrease after 7 days of culture in quiescent condition and remained then stable over the next 11 days (Fig. 5A). This decrease is probably related to the lack of import of cytosolic dNTP pool that in turn results in a slight decrease of the mitochondrial dNTP pool availability. In POLG fibroblasts, the mtDNA copy number was 55% of the control values and progressively decreased to 20% after 18 days in quiescent culture condition (Fig. 5A). The addition of 1 µm of CLO to the culture medium induced an increase of mtDNA copy number in POLG mutant fibroblasts that reached 80% of control values (Fig. 5B). We then tested the effect of various CLO concentrations and observed a clear dose–response of CLO on mtDNA copy number in POLG patient cells. Indeed, 0.5 µm of CLO has a higher effect than 1 µm whereas increased CLO concentration (2.5 µm) not only failed to increase the mtDNA content, but rather drastically reduced it (Fig. 5C). Examination of the POLG steady-state level showed a clear increase of POLG upon CLO addition in patient and control fibroblasts (Fig. 5D). In addition, ATP5β level, a subunit of the respiratory complex V and TFAM, reflecting cell mtDNA content, were also increased. Furthermore, the amount of MTCO1, subunit 1 of the cytochrome oxidase (complex IV) encoded by the mtDNA was clearly increased upon addition of CLO.
Altogether, these results indicate that mtDNA copy number responds to CLO concentration in a dose-dependent manner and that treatment of POLG-deficient fibroblasts with low CLO concentration can increase their mtDNA cellular content by increasing the POLG steady-state level.
Two drugs, structurally and pharmacology related to CLO, rescue polg-1-deficient worms toward EtBr toxicity
CLO is an anti-arrhythmic agent belonging to Class III as it acts as a potassium (K+) channel blocker. Other known Class III anti-arrythmic agents are amiodarone, sotalol, ibutilide, dofetilide and dronedarone. Among them, ibutilide and dofetilide share the common chemical structure with CLO (Fig. 6A). The three compounds are charged amines with a single phenyl ring except for dofetilide, which contains two phenyl rings. Clofilium is a quaternary amine with a chlorine on the phenyl ring. Ibutilide and dofetilide are tertiary amine with a methanesulfonanilide group on the phenyl ring. EtBr toxicity tests were performed on polg-1(ok1548/+) worms in the presence or absence of different concentrations of ibutilide or dofetilide (Fig. 6B). Both drugs allowed polg-1-deficient animals to reach adulthood in the presence of 30 µg/ml of EtBr, but in no case these compounds were as efficacious as CLO.
POLG encodes the mtDNA polymerase and mutations of this gene have been associated with several subgroups of mitochondrial disorders such as myocerebrohepatopathy spectrum disorders, Alpers–Huttenlocher syndrome, myoclonic epilepsy myopathy and sensory ataxia, ataxia neuropathy spectrum and PEO. Most often, recessive POLG mutations are associated with mtDNA depletion and dominant POLG mutations with multiple mtDNA deletions. Whereas it is still unknown why these mutations result in quantitative or qualitative mtDNA defects, in both cases they are always associated with mtDNA replication defects.
Despite several studies focused on clinical and biochemical aspects of POLG mutations, until now no therapy targeted to the associated diseases could be developed. This feature is not restricted to diseases related to POLG mutations, but also extends to the vast majority of mitochondrial disorders. Indeed, treatments remain largely symptomatic and include cofactor supplementation, prevention of oxygen-radical damage to mitochondrial membranes, dietary recommendations and avoidance of drugs or procedures known to have a detrimental effect on mitochondria. Yet none of the current therapies has been convincingly shown to hinder or prevent the progression of the disease, except for rare patients carrying a primary deficiency of coenzyme Q10, who have been treated successfully with oral ubiquinone (33). The lack of therapeutics is partly linked to the difficulty of using human cell models for testing a large number of candidate drugs targeted to mitochondrial dysfunction. Indeed, most of the time the only available cells are patient fibroblasts, which grow relatively slowly and have only a limited number of generations. To overcome these difficulties, the use of a fast growing, cheap and simple organism like the yeast S. cerevisiae, has been shown to be a powerful tool to screen chemical libraries as it has been proven in the case of the mtDNA ATP6 mutations related to NARP syndrome (13,14).
The mtDNA polymerase amino acid sequences are highly conserved from yeast to human allowing the use of S. cerevisiae as a model organism to mimic human POLG mutations. Here, we exploited this yeast as a suitable tool for the initial screening of drug libraries. Through the use of an mtDNA polymerase-deficient yeast strain (mip1), we identified CLO as a putative therapeutic molecule for POLG-related defect since this compound was shown to rescue the respiratory growth defect of the mip1 mutant on respiratory medium. We then demonstrated that CLO could also alleviate the petite production of eleven pathogenic mip1 mutations affecting residues localized in the polymerase, the linker or the exonuclease domains of the mtDNA polymerase, which suggests a general effect of CLO on the Mip1 protein. Furthermore, the CLO effect was found to be specific of mip1 defects since it does not rescue the growth defect of other mutant genes affecting mtDNA stability, but not related to mtDNA replication defects. We then investigated the effect of CLO in more details and found that it increases the oxygen consumption in both wt and mip1 mutant strains, thus rescuing the oxygen consumption defect in the mutant. We observed similar amounts of the Mip1 protein in both wt and mip1G651S mutant strains suggesting that the G651S modification, located in the polymerase domain of the protein, does not modify mtDNA polymerase stability but rather affects its enzymatic activity. Mip1 protein amounts were significantly increased by CLO treatment in wt and mutant strains whereas the mitochondrial biogenesis seemed to be unchanged. We can, therefore, hypothesize that CLO specifically stabilizes the Mip1 protein and that the increase amount of mutant protein at least partially compensates its defective polymerase activity. This hypothesis is supported by the fact that two copies of mip1 mutants in the Δmip1 strain produce half frequency of petite compared with one copy (7). Interestingly, the polymerase fidelity was not modified by CLO treatment. Moreover, the CLO rescuing effect on Mip1 activity is additive to the increased dNTP pool availability, suggesting that CLO may act through a different pathway. Finally, the reduced frequency of petite colonies in mip1 mutants and the increased number of replicating nucleoids in wt yeast highly suggest that CLO increases mtDNA replication. Whether the increase in mtDNA replication is directly due to Mip1 stabilization by the drug deserves more studies.
These results prompted us to test CLO on C. elegans animals deficient for POLG, which offers the possibility to study the effect of chemical compounds at the scale of an entire organism. POLG defect in worm results in a severe phenotype detected only in adulthood that consists in severe compromised gonadal function as a result of an important mtDNA depletion leading to sterility. This late onset phenotype is related to a high maternal contribution of POLG-1 from heterozygous mother. We first showed that CLO concentrations allowing normal development of polg-1 mutant animals did not that CLO treatment stimulates affect mitochondrial proteostasis. Moreover, the number of laid and hatched eggs as well as the gonadal/intestinal protrusion of homozygous polg-1(ok1548) were significantly improved by CLO treatment. As in wt yeast cells we observed that CLO treatment of wt worms increased the number of replicating mitochondrial nucleoids. Moreover, several data suggested mtDNA replication and we clearly demonstrated a 2-fold increase in mtDNA copy number in polg-1(ok1548) mutant animals as well as wt animals exposed to CLO. Interestingly, the mtDNA copy number of polg-1(ok1548) was only 20% of control after 6 days of CLO treatment, but partially rescued various abnormal phenotypes resulting from polg-1 mutation. This suggests that slight mtDNA increase above a minimal threshold can partly compensate organ dysfunctions.
We, finally, tested CLO on fibroblasts from a patient with cerebrohepatopathy and severe mtDNA depletion in the liver associated with compound heterozygous POLG mutations. Both mutations were located in the polymerase domain of the protein. Cultured skin fibroblasts do not represent a good model for studying mtDNA replication as they are continuously cycling allowing the cytosolic dNTP synthesis to partially supply the mitochondrial dNTP pool, which can rescue the mtDNA replication defect. This has been clearly established as supplementation with four dNMP in POLG-deficient myotubes, presenting low mtDNA content resulted in a significant rescue of mtDNA copy number (34). This phenomenon is not restricted to mammalian cells as in yeast increased dNTP pool availability has also been shown to partially suppress the mtDNA instability due to mitochondrial polymerase deficiencies (17,24,25). We, therefore, used a quiescent condition consisting of serum deprivation that inhibits fibroblasts cycling for studying the effect of CLO on mtDNA copy number. This condition induced a significant decrease of mtDNA copy number in patient cells that was rescued by CLO treatment. Interestingly, as in the budding yeast, treatment of fibroblasts with CLO, leads to an increased level of the POLG protein. This result indicates that the biochemical mechanism whereby clofillum tosylate enhances mtDNA replication is conserved from yeast to human cell.
CLO is a class III anti-arrhythmic agent targeting potassium (K+) efflux. K+ channels belong to a large class of transmembrane proteins that specifically regulate the transport of K+ ions across biological membranes. CLO has been shown to block a large number of potassium channels with different potency (35). Several K+ channels have been identified in human, C. elegans and S. cerevisiae. Whether the CLO effect on mtDNA stability is related to potassium channels is unlikely for several reasons. First, two other class III anti-arrhythmic agent, namely amiodarone and sotalol, were included in the TEBU-BIO and Prestwick libraries that were screened with the yeast mip1 mutant strain and none of them was found as positive hit. Furthermore, a CLO-binding sequence proposed for the human hERG and Slick/Slack channels (35) is also present in one yeast potassium channel, the Tok1 protein. However, it seems unlikely that Tok1 could be a target of CLO since neither TOK1 deletion nor TOK1 overexpression modified the decrease of petite production induced by CLO (Supplementary Material, Fig. S5). Therefore, a yet unknown target of CLO related to mtDNA maintenance may exist that is conserved in yeast, worm and human. It is currently accepted that a drug binds to a multitude of targets in the cells and elicits a number of other off-target effects. All these targets and effects together connect to a phenotypic response. For that reason, identification of the pathway by which CLO can rescue POLG deficiencies would be of great interest, since it may also uncover the underlying mechanism of mtDNA maintenance rescue that may be conserved from yeast to C. elegans and Human. This can be addressed by systematic chemical-genomic profiling using the yeast homozygous genome-wide deletion collection. By this approach, marked sensitivity of homozygous deletion mutants to inhibitory concentrations of a compound indicates the genes and pathways involved in the mechanism of action of the chemical (36–38). Such a strategy has already revealed the sorting of nuclear-encoded mitochondrial proteins as a new intervention point to counteract mitochondrial ATP synthase deficiency (13). Finally, based on the chemical structure of CLO, we showed that two other class III anti-arrhythmic agents, dofetilide and ibutilide (both FDA-approved drugs) also improve the development of POLG-1-deficient C. elegans animals. Therefore, dofetilide and ibutilide could also be considered as promising molecules for the treatment of POLG-related disorders.
In conclusion, this study allowed the identification of CLO as a potential therapeutic molecule for POLG-related diseases. Indeed, using three different model organisms deficient in POLG, we found that CLO was able to increase mtDNA replication and POLG steady-state level. Therefore, the mechanism underlying these effects seems conserved. These results lead, for the first time, to the identification of a potential pharmacological molecule against mtDNA instability associated with mitochondrial polymerase defects.
Analyses in yeast
Yeast strains and growth conditions
The S. cerevisiae strains used and their genotypes are listed in Supplementary Material, Table S3. Strains W303-1B deleted for CCE1 or PIF1 genes were obtained by a one-step gene disruption with a KanMX4 cassette amplified from the corresponding BY4742 deleted strains, using primers CCE1DFw and CCE1Drv, or PIF1DFw and PIF1DRv, respectively (Supplementary Material, Table S4). The Δmir1 mutant strain has been described previously (39).
Strains were grown in either synthetic complete (SC) media (0.19% YNB without amino acids and 0.5% NH4SO4) supplemented with 1 g/l of dropout mix with amino acids or bases necessary to complement the auxotrophies or in the YP medium (0.5% yeast extract, 1% peptone). Media were supplemented with indicated carbon sources at 2%, in liquid phase or after solidification with 20 g/l agar. The YPAEG-Ery medium contained 1% yeast extract (Difco), 2% peptone (Difco), 100 mg/l adenine, 3% (v/v) glycerol, 3% (v/v) ethanol and 3 g/l of erythromycin (Sigma).
Plasmids, cloning and transformation
MIP1 and mip1G651T alleles were tagged at their 3′-end with the HA-tag sequence using overlapping PCR primers MIP1PreAvrGI, MIP1HAFw, MIP1HaRv and MIP1PostBsrRv (Supplementary Material, Table S3) and PCR fragments were cloned in pFL39MIP1 (17) at BsrGI and SalI transformation was performed according to (40). Plasmid shuffling, necessary to obtain rho+ mip1 mutant strains, was performed according to Ref. (41).
Determination of petite frequency and of EryR
Petite frequency was measured as previously reported (41). Briefly, strains transformed with wt or mutated mip1 alleles were grown overnight in the YP medium supplemented with 2% ethanol and then inoculated at a final concentration of 5 × 105 cells/ml in the same medium supplemented with 2% glucose with and without CLO. After 24 h of growth, cells were re-inoculated in the same medium at a final concentration of 5 × 105 cells/ml. After 24 h, cells were plated on SC agar plates supplemented with 2% ethanol and 0.3% glucose at a dilution that gave ∼200 cells/plate. Petite frequencies were defined as the percentage of colonies showing the petite phenotype after 5 days of incubation. Statistical analysis was performed through an unpaired two-tailed t-test. EryR mutant frequency was measured as previously reported (16).
Determination of the minimal inhibition concentration and of the optimal working concentration of CLO
The minimal inhibition concentration (MIC) of CLO (Sigma-Aldrich, C2365) for yeast mip1G651S cells was determined by inoculating yeast at a final concentration of 0.05 OD600/ml in the liquid YP medium supplemented with 1 to 512 μm of CLO solved in DMSO. OD600 was measured after 24 h. MIC was defined as the concentration at which no growth was visible.
To determine the optimal CLO concentration to be used in yeast experiments, i.e. the higher sub-MIC concentration without negative effect on the division time and on viability of petite mutants, we first analyzed the effect of the drug on the petite fitness, by growing equal quantity of isogenic rho+ and rho0MIP1 cells in the same culture in the presence of increasing concentrations of CLO (0–128 μm). Aliquots of the cultures were plated at T = 0, 24 and 48 h, on solid media supplemented with 0.3% glucose and 2% ethanol to determine the ratio of respiratory sufficient and petite (respiratory deficient) colonies. The molecular concentration at which (after 24 and 48 h of treatment) the ratio of rho0 to rho+ was not statistically different (two-tailed unpaired t-test, P < 0.05) from culture grown without CLO was considered as optimal to use in the following experiments. In fact, in this condition the reduction of petite mutants should be ascribed only to the diminished onset of novel petite cells exerted by CLO and not to a negative effect on the fitness of the petite cells. The molecular concentration at which the ratio of rho0 and rho+ during their competitive growth was not statistically different (two-tailed unpaired t-test, P < 0.05) compared with culture without CLO was considered as optimal to use in the following experiments. In fact, in this condition the reduction of petite mutants should be ascribed only to the diminished onset of novel petite cells exerted by CLO and not to a negative effect on the fitness of the petite cells.
O2 consumption was continuously measured at 28°C in a thermostatically controlled chamber equipped with a Clark oxygen electrode (Oxygraf, Hansatech Instruments). Cells were grown in Yeast extract 0,5%, bacto peptone 1%, ethanol 2% (YPE) supplemented or not with CLO 32 µm (OD600 = 3) and a cell suspension containing 107 cells in 250 µl of YPE was used for each measurement. The O2 uptake rate was measured as a tangent at the initial part of the progress curve and expressed as nmoles of O2 consumed/2 × 106 cells/min. CCCP (Sigma-Aldrich) was dissolved in DMSO. Measurements were repeated at least three times.
Western blot analyses
Yeast mitochondria were prepared from cells grown for 7–8 generations in YPE at 28°C in the presence of 32 µm CLO or equivalent quantity of DMSO (compound vehicle) as described in Ref. (42) without sucrose step gradient. Mitochondrial proteins were precipitated by trichloroacetic acid. Equivalent amounts of mitochondrial protein extracts were loaded on 6% sodium dodecyl sulphate–polyacrylamide gel electrophoresis gels and transferred to polyvinylidene difluoride membranes by wet transfer. Membranes were probed with monoclonal anti-HA antibody (Roche) and anti-porine antibody (Invitrogen). Immunodetection was performed with horseradish peroxidase-conjugated anti-mouse antibodies using chemoluminescence western blotting reagents (Pierce). For fibroblasts, cells were lysed in a NP40 based lysis buffer (50 mm Tris–HCl, pH 8.4, 300 mm NaCl, 10 mm MgCl2, 1 mm ethylenediaminetetraacetic acid, 0.5% NP40, protease inhibitor mixture). The samples were analyzed by western blot analysis.
Analyses in C. elegans
Strains and culture conditions
Caenorhabditis elegans strains were grown at 20°C on 6 cm Petri dishes containing nematode growth medium NGM agar and a lawn of Escherichia coli OP50. The wt N2 strain was obtained from the Caenorhabditis Genetics Center (CGC; Minneapolis, MN, USA). The strain containing the mitochondrial stress reporter hsp-6::gfp(zcIs13) V has been previously described (28). The strains hsp-6::gfp(zcIs13) V and TB2143 (polg-1(ok1548) /+ II; +/mln [dpy-10(e128) mls14] II) were a generous gift from Aleksandra Trifunovic (Cologne, Germany). Heterozygous animals hereafter named polg-1(ok1548/+) are GFP positive while homozygous animals named polg-1(ok1548) do not express GFP.
Pharmacological treatment, brood size measurement and determination of the mtDNA copy number
All pharmacological compounds were obtained from Sigma-Aldrich. Concentrated solutions of each drug were prepared by dissolving compounds in DMSO. Caenorhabditis elegans tolerates DMSO up to a final concentration of 0.6%. Compounds were added to liquid NGM previously autoclaved and cooled to 55°C and the medium was immediately mixed and dispensed into Petri dishes. It is admitted that drugs penetrate the animals both by diffusion through the cuticle and ingestion.
EtBr (Fluka) was dissolved in water and added to NGM agar plates to the indicated final concentration. Measurements of worm sensitivity toward EtBr have been performed according to (29).
Gravid animals were bleached to obtain a synchronous population that was used in CLO treatments. For brood size measurement, singled L4-staged polg-1(ok1548) hermaphrodites (n = 12) were placed on NGM plates supplemented or not with 50 µm CLO seeded with E. coli OP50 bacteria and worms were transferred every day on new plates. The number of laid eggs and hatched embryos were counted over the next 5 days. The mtDNA copy number was measured by qPCR. L4-staged N2 and polg-1(ok1548) worms were grown in the presence or absence of 50 µm CLO until Day 6 of adulthood. Worms were lysed and total DNA was extracted according to (29). Primers for NADH dehydrogenase subunit 1 (nd1, mitochondrial reference) and adenine nucleotide translocase ANT (ant-1.3, nuclear reference) were used for determination of mtDNA copy number as described in Ref. (29). For total worm respiration, oxygen consumption rates were measured using an Oxytherm oxygen electrode (Hansatech, Norfolk, UK). Synchronized worms were grown on NGM seeded with OP50 bacteria. Worms were washed three times with M9 buffer before being introduced in the measurement chamber maintained at 20°C. To reach sufficient respiration rates, at least 500 worms were used in each measure, the respiration value was then standardized to 1000 animals.
Analyses in human cells
Informed consent was obtained for all subjects and approved by local National Ethic Committee.
Primary human skin fibroblast cultures from a patient with POLG mutations and a healthy control of nearly the same age were used. The patient was compound heterozygous for R807P and D1184N mutations. Fibroblasts were grown in Dulbecco's modified Eagle's medium with 4.5 g/l glucose supplemented with 110 mg/l sodium pyruvate, 100 U/ml penicillin and streptomycin and 10% dialyzed FBS in a humidified incubator at 37°C and 5% CO2. After reaching confluence, FBS was reduced to 0.1% (vol/vol) to induce quiescence. Four days later (Day 0), CLO was added to the cell medium. All experiments were performed in triplicates. During the treatment period, cell media were replaced every 3 days. At the times required according to the experimental designs, media were removed and cells harvested by trypsinization, washed with PBS, pelleted and stored at −20°C until DNA isolation.
Quantification of mtDNA copy number
Total DNA was isolated from cell pellets by the phenol/chloroform procedure and the copy number of mtDNA per cell was determined by real time quantitative PCR (RT-PCR) using an ABIPRISM 7300 sequence detector (Applied Biosciences). Determination of mtDNA and nuclear DNA content was performed as a multiplex reaction. TaqMan Universal PCR Master Mix and designed sets of TaqMan probes and primers were used for relative mtDNA (12S rRNA) versus nuclear DNA (RNAseP) copy number quantification. A standard curve with 12SrRNA and RNAse P amplicons (from Roche) containing the 12SrRNA and RNAse P genes was used for absolute quantification of mtDNA and nuclear DNA in the samples. Triplicates were assessed for each experimental condition.
This work was supported by the Indonesian government (Beasiswa Unggulan Kemendiknas RI), the University Paris-Sud (to L.P.) and by grants from Fondation pour la Recherche Medicale (DRM20101220457), Asociation Française contre les Myopathies (MitoScreen 17122) and Telethon (Italia, GGP11011).
We thank Aleksandra Trifunovic for her generous gift of the C. elegans polg-1(ok1548/+) and hsp-6::gfp(zcIs13) V strains. We are grateful to Marc Blondel and Deborah Tribouillard for their warm welcome, advices and help for the yeast-base screening of drug libraries. We also thank Jean-Paul diRago, Marc Blondel, Vincent Procaccio, Pascal Reynier and Geneviève Dujardin for scientific discussions; Sylvie Hermann le Denmat and Monique Bolotin-Fukuhara for proofreading the manuscript.
Conflict of Interest statement. None declared.