Mitochondrial DNA (mtDNA) polymerase gamma (Polg) is a heterodimeric enzyme containing a Pol I-like catalytic core (PolgA) and an accessory subunit. Mutations in POLGA, affecting the stability of mtDNA, have been identified in several human pathologies such as progressive external ophthalmoplegia and Alpers' syndrome. Extensive literature shows mitochondrial toxicity effects nucleoside analogue reverse transcriptase inhibitors used in the treatment of HIV and chronic hepatitis B as a consequence of an inhibitory effect on Polg. We have previously shown that mice with an error-prone version of PolgA accumulate higher levels of somatic mtDNA mutations resulting in a premature aging phenotype. In the present paper, we demonstrate PolgA deficiency in mouse embryos causes an early developmental arrest between embryonic days 7.5 and 8.5 associated with severe mtDNA depletion. Heterozygous knockout mice have half the wild-type levels of PolgA transcripts and a slight reduction in mtDNA levels but develop normally. Surprisingly, amounts of PolgA transcripts in heterozygous knockout mice are increased in response to artificially elevated mtDNA copy number, revealing a possible regulatory link between mtDNA maintenance and PolgA expression. Our results show that Polg indeed is the only DNA polymerase capable of maintaining mtDNA in mammalian mitochondria. In addition, presence of Polg is absolutely essential for the organogenesis during mammalian embryonic development.
Fifteen distinct cellular DNA polymerases have been identified in mammalian cells but only four of these are devoted to DNA replication, whereas the rest are devoted to DNA repair and specialized DNA synthetic processes that contribute substantially to the maintenance of genetic integrity (1). Although most of these enzymes are involved in nuclear DNA repair and replication, DNA polymerase gamma (Polg) remains the only DNA polymerase found in mitochondria (reviewed in 2). As such, Polg is proposed to be uniquely responsible for all DNA synthetic reactions including both replication and repair of mitochondrial DNA (mtDNA) (2).
Mitochondria are the sole organelles in animal cells that contain their own DNA. Individual cells have between 1000 and 10 000 copies of the mitochondrial genome (mtDNA) with the more energy consuming tissues, such as heart, having proportionally higher amounts. Replication of mammalian mtDNA is under relaxed control as there is no known mechanism to ensure that each molecule is replicated only once per cell cycle. Polg is not believed to directly regulate mtDNA levels, as overexpression of Polg in cell lines does not result in a corresponding increase in mtDNA copy number (3). The enzyme contains a Pol I-like catalytic core (PolgA), similar to the bacteriophage T7 DNA polymerase, in a heterodimeric complex with an accessory subunit (4–6). This accessory subunit substantially increases both the catalytic activity and the processivity of the enzyme (6). Interaction of the catalytic and accessory subunits is through multiple contact sites and it has been proposed that these contacts may enhance DNA binding and also increase the nucleotide binding of the holoenzyme (7,8). In addition, Polg exhibits a high fidelity in nucleotide selection and incorporation as well as high processivity during replication (9,10). High fidelity of the Polg is partly due to the 3′–5′ exonuclease activity of the catalytic subunit. It has been shown that the exonuclease activity of Polg contributes several hundred-fold to error avoidance in vivo (8). We recently described a knock-in mouse model that expresses an error-prone version of Polg (mtDNA mutator mice) (11). This mouse strain exhibits a dramatic decrease in exonuclease activity whereas Polg polymerase activity is preserved. Abolished exonuclease activity leads to a 3–5-fold increase in somatic mtDNA mutations that, in turn, cause a progressive respiratory chain deficiency and premature aging phenotypes (11). These results offer direct genetic evidence that the collective effect of a variety of somatic mtDNA mutations can cause aging.
Polg is unique among the cellular replicative DNA polymerases as it is highly sensitive to inhibition by nucleoside analogue reverse transcriptase inhibitors (NRTIs) used in the treatment of HIV and chronic hepatitis B and C infections (12,13). Highly active antiviral therapy (HAART) that includes NRTIs has changed AIDS from a lethal illness to a chronic disease. Unfortunately, it was quickly seen that significant mitochondrial toxicity due to HAART was an important clinical entity and has since been shown that it is primarily a consequence of Polg inhibition. Pyrimidine NRTIs, such as AZT, fialuridine and others, cause cardiac dysfunction, hepatic failure, skeletal myopathy, lactic acidosis with defective mtDNA replication, mtDNA depletion and altered mitochondrial ultrastructure (reviewed in 13).
Mutations in POLGA, the gene encoding the catalytic subunit of human mtDNA polymerase has been associated with a number of mitochondrial disorders that affect the stability of mtDNA (14,15). These mutations can lead to both mtDNA deletion and mtDNA depletion syndromes. Most common and best known are the autosomal dominant or autosomal recessive (ar) forms of familial progressive external ophthalmoplegia (15–17). Additional clinical presentations like ar sensory atactic neuropathy with dysarthria and ophthalmoplegia have been associated with POLGA mutations (18). Recently, different POLGA mutations causing the depletion of liver mtDNA have been reported in Alpers' hepatophatic poliodystrophy (19,20). In addition, dominant POLGA mutations have shown to cause a severe multisystem disorder including Parkinsonism and premature menopause, which are not typical of mitochondrial disease (21). Finally, specific polymorphisms in the POLGA gene have been associated with male infertility (22).
The involvement of Polg in a number of different human pathologies, its complex role in mtDNA maintenance and the still ongoing debate about the number of DNA polymerases in mitochondria, makes it a very interesting target for further analysis.
In the present study, we investigated the role of Polg in mammalian development by creating PolgA deficient mice. We generated a transgenic mouse model lacking PolgA through homologous recombination in embryonic stem cells. Loss of Polg in mice leads to developmental arrest at around embryonic day (E) 7.5 associated with severe mtDNA depletion. Mice heterozygous for the PolgA null allele have only half the wild-type levels of PolgA transcripts. However, these heterozygous PolgA knockouts appear completely normal and have close to wild-type mtDNA copy number. Interestingly, mice with artificially elevated mtDNA levels show a corresponding up-regulation of PolgA transcripts, suggesting a link between mtDNA copy number and PolgA expression.
Generation of PolgA knockout mice
The gene encoding PolgA was disrupted in the mouse by utilizing the cre-loxP recombination system. PolgAloxP/loxP mice were mated with animals containing a β-actin cre-transgene that ubiquitously expresses cre-recombinase in the pre-implantation embryo (Fig. 1A). Resulting PolgA+/− offspring were identified by Southern blot analysis detecting the cre-mediated excision of exon III from one allele of these mice (Fig. 1B). Deletion of PolgA exon III results in a frameshift mutation leading to a stop codon early into exon IV. The severely truncated mRNA lacks virtually all the active domains of PolgA, and therefore we do not expect any active protein to be derived from this allele.
PolgA knockout is embryonic lethal
To completely knockout PolgA in the mouse, PolgA+/− mice were intercrossed. The resulting genotype distribution from this cross can be seen in Table 1. No homozygous knockout (PolgA−/−) pups were born, PCR genotyping (Fig. 1C) was, therefore, carried out on embryos at different stages of development. Results showed that PolgA−/− embryos died between E7.5 and E8.5. Wild-type and PolgA+/− embryos (Fig. 2A and data not shown) at E8.5 were morphologically indistinguishable from each other, whereas PolgA−/− embryos (Fig. 2B) were generally smaller and much less developed. Enzyme histochemical staining showed a severe respiratory chain dysfunction. There was no cytochrome c oxidase (COX) activity (catalytic subunit encoded by mtDNA), whereas normal succinate dehydrogenase (SDH) activity was found (all subunits encoded by nuclear genes) (Fig. 2C). Southern blots were performed on total DNA isolated from embryos to confirm that the respiratory chain dysfunction was the result of mtDNA depletion due to a loss of PolgA. Results showed there was no difference in mtDNA levels between wild-type and PolgA+/− embryos, whereas the PolgA−/− embryos contained barely detectable levels of mtDNA (Fig. 2D).
Normal mtDNA levels in PolgA+/− mice
PolgA+/− mice were born at the expected Mendelian ratio, developed normally and displayed no obvious phenotype even at 18 months of age. TaqMan PCR was performed on cDNA generated from kidney total RNA to confirm that expression from the remaining PolgA allele was not up-regulated in PolgA+/− mice. Relative levels of PolgA mRNA were reduced by approximately one half in PolgA+/− mice compared with age-matched controls (Fig. 3A). These results are in accordance with previous reports that PolgA is constitutively expressed at similar levels in different tissues (3). We failed to quantify PolgA protein levels in both total cell and purified mitochondrial extracts by western blot analysis (data not shown). Available antibodies were raised against human POLGA and although they show cross-reaction with the mouse recombinant protein, they were not sensitive enough for quantification by western blot.
As a total loss of PolgA resulted in an almost complete depletion of mtDNA in embryos, we looked at mtDNA levels in adult PolgA+/− mice. Total DNA isolated from heart, muscle and liver was analysed by Southern blot and the mtDNA levels were found to be comparable in PolgA+/− and wild-type mice (Fig. 3B). As Southern blots showed a tendency towards reduced mtDNA levels in PolgA+/− mice, the analysis was repeated on the same DNA samples using real-time PCR. Again, we observed the same trend of slightly lower mtDNA levels, a finding that was now significant in heart tissue (Fig. 3C).
TFAM overexpression increases mtDNA levels in PolgA+/− mice
Recently, it was demonstrated that overexpression of the human TFAM protein in mice leads to an increase of mtDNA copy number of up to 60% (23). As mtDNA levels were normal in PolgA+/− mice we decided to cross them with TFAM overexpressors (TFAMOE) and study if PolgA expression was sufficient to allow for a similar increase in mtDNA. Total DNA was isolated from mice of three genotypes obtained from this cross (PolgA+/+; PolgA+/+, TFAMOE and PolgA+/−, TFAMOE) and analysed by both Southern blot and real-time PCR to determine levels of mtDNA. Both methods confirmed that the overexpression of human TFAM indeed increases mtDNA levels (∼70% above wild-type) and also showed that PolgA+/−, TFAMOE mice were just as capable of raising mtDNA levels as PolgA+/+, TFAMOE mice (Fig. 4A and B). PolgA mRNA levels were determined by TaqMan PCR analysis of cDNA generated from total RNA isolated from mice of the three genotypes mentioned earlier. Surprisingly, we detected an increase in PolgA transcripts in both PolgA+/+,TFAMOE and PolgA+/−, TFAMOE mice (Fig. 4C). The values observed in PolgA+/−, TFAMOE animals are equivalent to PolgA steady-state transcript levels in wild-type animals and are almost double those seen in the PolgA+/− animals (Fig. 4C).
We have shown that PolgA is absolutely essential for mammalian embryonic development and mtDNA maintenance. Homozygous disruption of the mouse PolgA gene leads to embryonic death at the time of late gastrulation and before early organogenesis. Loss of PolgA coincides with a dramatic decrease in mtDNA levels, with PolgA null embryos having almost no detectable mtDNA at E8.5. PolgA null embryos at E8.5 are much smaller than wild-type embryos and have a severe respiratory chain deficiency, as shown by an absence of COX activity, whose catalytic subunits are encoded by mtDNA.
In addition, the data presented directly shows the necessity of mtDNA replication during early mouse embryogenesis. Previous mouse models have shown that mtDNA maintenance is critical during embryogenesis; however, here we have described the first mouse model where only mtDNA replication has been disrupted (24,25). The importance of replication and mtDNA transcription and maintenance in early mammalian embryogenesis was somewhat underestimated as it is believed that during embryogenesis mtDNA transcription starts as early as the two-cell stage, whereas mtDNA replication could be dispensable until organogenesis occurs (26–28). This idea was further supported by the finding that in Xenopus laevis embryos, mtDNA replication begins at the heartbeat stage, emphasizing the need for mature mitochondria at the time the circulatory system becomes available (29,30). In addition, mice lacking mitochondrial transcription factor A (Tfam−/−) or ribonuclease H (Rnaseh−/−) die around E8.5 showing signs of organogenesis, although impaired and unfinished (24,25). Common problems in Tfam−/− and Rnaseh−/− embryos include cardiovascular malformations and neuronal tube defects (24,25). Here, we show that knockout of PolgA has a lethal phenotype comparable to those previous mouse models. In addition, PolgA−/− embryos seem to be more severely affected, as they are stalled at the egg-cylinder stage and there is a lack of any signs of organogenesis occurring. Therefore, our results show that early mouse organogenesis critically depends on active mtDNA replication and not just maintenance of mtDNA. Although, we cannot rule out that the more severe phenotype may partly be a consequence of the different genetic backgrounds between these mouse models or perhaps the higher maternal contribution of both Tfam and RNase H in comparison to Polg, the difference in developmental arrest between embryos of these three knockout models makes it unlikely.
Our results show that partial PolgA deficiency in heterozygous animals (PolgA+/−) is not compensated for by an increased transcription of the PolgA gene. This fits with the idea that PolgA expression is not tightly regulated as it is constitutively expressed in tissues even in the absence of mtDNA (31). Steady-state levels of PolgA transcripts have previously been shown not to correlate with mtDNA content in different tissues and indeed, overexpression of the PolgA gene did not change mtDNA content in cell lines (3).
Our findings show that PolgA+/− animals have a slight reduction in mtDNA levels. A study of animals heterozygous for the Tfam null mutation has demonstrated that a 35% decrease in mtDNA levels does not have any effect on mitochondrial function (24). Hence, it is no surprise that the 15% reduction in mtDNA levels in PolgA+/− mice does not cause any phenotype. Increased transcription and mitochondrial RNA stability are believed to have the potential to compensate for decreased mtDNA copy number and preserve the overall mitochondrial protein demands (24,32).
In this study, we also show that increased mtDNA copy number leads to increased expression of PolgA. We chose to increase mtDNA copy number in PolgA+/− mice by overexpressing the TFAM protein. Overexpression of TFAM leads to an overall increase in mtDNA copy number by ∼70% in both wild-type and PolgA+/− animals. Our results demonstrate that the up-regulation of mtDNA levels leads to an increase in steady-state levels of PolgA transcripts. Although it is generally believed that PolgA expression is not regulated by the rate of mitochondrial biogenesis, there are some known cases where an increase in PolgA is required. First, levels of PolgA transcripts coincide with cellular mtDNA copy number changes in Xenopus laevis during vitellogenesis (4). Secondly, a simultaneous increase in both Polg and cytochrome c activities was observed immediately before an increase in ATP levels, as a response to increased energy demands, in the early S phase of the cell cycle (33). Finally, a consistent and direct relationship between mtDNA copy number, mtRNA content, respiratory chain capacity and Polg activity was observed in murine, bovine and rabbit hearts (34).
Our results suggest that there is an active feedback mechanism that increases PolgA expression as a result of an overall increase in TFAM and mtDNA levels. This further supports the idea that TFAM is a major regulatory factor in determining mtDNA levels within mitochondria as it has been shown that mtDNA copy number mirrors the amounts of TFAM within cells (23,24). Others have failed to detect changes in PolgA expression levels coinciding with increasing amounts of mtDNA, following chronic motor nerve stimulation in rabbit muscle. However, the authors found that the 3.5-fold increase in mtDNA levels was paralleled by a comparable increase in Polg activity (3,34). One possible explanation for the feedback mechanism could be activation of PolgA expression via NRF1, a transcription factor driving Tfam expression as well as a number of other nuclear-encoded mitochondrial genes (35). It has been shown that the PolgA promoter region has a DNA element potentially recognized by NRF1 (36). The importance of this NRF1-like element in PolgA transcription regulation is further supported by the finding that mutations in this site significantly reduce activity of the PolgA promoter in Schneider cells (37).
In conclusion, this study emphasizes the role of PolgA as a key player in mtDNA maintenance that is absolutely necessary for mtDNA replication from an early stage in embryogenesis. Our work also provides evidence that PolgA expression can be induced as a consequence of increased demand for the replication of mtDNA. However, the minor reduction in mtDNA levels observed in PolgA+/− mice appears not to be significant enough to induce a compensatory induction of PolgA expression. Finally, our results put an end to the long-standing debate of whether mtDNA Polg is the sole polymerase capable of replicating and maintaining mtDNA in the cell.
MATERIALS AND METHODS
Matings and genotyping of transgenic animals
The generation of mice with a loxP-flanked PolgA-allele (Fig. 1) have previously been described (11). PolgA knockout animals are genotyped with both Southern blot and PCR analysis. For Southern blot analyses, ∼5 µg of genomic DNA was digested with BglI and blotted to Hybond-C+(Amersham) membrane using standard procedures. The membranes were hybridized with a 2.7 kb probe spanning exons 11–14 (Fig. 1A and B) labelled with [α-32P]dCTP (3000 Ci/mmol; Amersham) and visualized by autoradiography with standard procedures.
PCR genotyping was performed by multiplex PCR with primers F1 (5′-ttt gcc aag ttc taa ttc ca-3′), F2 (5′-gga tgg gca gga aca gtt ag-3′) and R1 (5′-ctg cca ttc acc ttc ccc-3′). Cycling conditions were 94°C for 5 min, followed by 35 cycles of 94°C for 30 s, 57°C for 45 s, 72°C for 30 s, with final extension at 72°C for 7 min. The PCR reaction designed in this way can give two differently sized products corresponding to the wild-type fragment of 658 bp and the knockout fragment of 471 bp.
Heterozygous PolgA knockout animals (PolgA+/−) were crossed with transgenic animals overexpressing TFAM (TFAMOE; previously called PAC19) (23) generating PolgA+/−,TFAMOE. From the same cross, we generated two control groups of animals used in our experiments: PolgA+/+ and PolgA+/+, TFAMOE. Double transgenic animals were genotyped to detect the PAC transgene (23) and PolgA+/− allele as described earlier. Animal studies were approved by the animal welfare ethics committee and performed in compliance with Swedish law.
Enzyme histochemical analyses of SDH and COX activities were performed on 14 µm cryostat sections of fresh frozen wild-type and PolgA+/− embryos as described previously (24).
Southern and western blot analyses
DNA was prepared by SDS and proteinase K treatment as previously described (38). The DNA was purified by phenol:chloroform extraction, precipitated with ethanol and dissolved in TE pH 8.0. Total DNA (5 µg) from heart, liver and skeletal muscle of adult mice (>10 weeks of age) and from whole E8.5–E9.5 embryos was digested overnight with the SacI restriction enzyme, precipitated with 0.2 m NaCl and ethanol and separated on 0.8% agarose gels. Separated DNA was transferred to nitrocellulose membranes by southern blotting. The membranes were probed with [α-32P]-dCTP random-labelled murine total mtDNA or a CoxI DNA fragment. Phosphorimager analyses were used to assess total mtDNA levels normalized to the nuclear 18S rRNA gene. Two-sided, unpaired t-tests were used to assess statistical significance.
Total liver homogenate and mitochondrial preparations were suspended in protein buffer [equal parts of suspension buffer (100 mm NaCl, 10 mm Tris–HCl pH 7.6, 1 mm EDTA pH 8.0, 1% Aprotinin and 1% PMSF) and loading buffer (100 mm Tris–HCl pH 6.8, 4% SDS, 20% glycerol and 200 mm DTT)]. Each sample was boiled for 10 min, sonicated and the supernatants containing the proteins were collected after centrifugation for 10 min at 10 000g. We used 1:100–1:2000 dilutions of the Ab-4 cocktail antibody consisting of antibodies raised against the C- and N-terminal domains of POLGA (Lab Vision). In addition, we used 1:2000–1:8000 dilution of the CD-7 antibody, raised against the C-terminal POLGA domain, as recommended by William C. Copeland. Western blot analyses to determine protein levels of mouse Polg were performed as described (24).
Real-time PCR to quantify mtDNA
DNA primers to detect CoxI and β-actin were designed to give a maximum amplicon length of 150 bp. β-actin forward primer 5′-TGTTCCCTTCCACAGGGTGT-3′ and reverse primer 5′-TCCCAGTTGGTAACAATGCCA-3′. CoxI forward primer 5′-GCCCCAGATATAGCATTCCC-3′ and reverse primer 5′-GTTCATCCTGTTCCTGCTCC-3′. The 25 µl PCR reaction mixture contained 1× Platinum SYBR Green qPCR SuperMix UDG (Invitrogen), 500 nm each primer and ∼1 ng of total genomic DNA. Real-time PCR conditions were 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 60 s at 60°C. Fluorescent signal intensities were recorded and analyzed during PCR in an ABI Prism 7700 Sequence Detector System (Applied Biosystems) using the SDS (version 1.91) software. The threshold cycle (CT) is the cycle at which a significant increase in the reaction product is first detected from the increasing fluorescent signal. Thus, the CT values within the linear exponential phase were used to measure the original DNA template copy numbers from a standard curve generated from five 10-fold dilutions of either pure mtDNA (CoxI) or pure nuclear DNA (β-actin). Relative values for CoxI and β-actin were compared within each sample to generate a ratio which is the relative level of mtDNA in that sample.
RNA isolation and TaqMan PCR
Isolation of total RNA from mouse kidney was carried out following the RNAeasy midi-kit protocol (Qiagen). Reverse transcription was performed using 5 µg of total RNA and Superscript reverse transcriptase (Invitrogen). Amplification was performed with the ABI PRISM 7700 Sequence Detector (Applied Biosystems) with a two-step PCR protocol (pre-incubation for 10 min at 95°C followed by 40 cycles at 95°C for 15 s and at 60°C for 1 min). All primers and probes were designed using Primer Express software (Applied Biosystems) avoiding contamination of the genomic DNA amplification by positioning one of the primers over the exon/intron boundary (Table 2). The PolgA probe was labelled with FAM at the 5′ end as reporter dye and TAMRA at the 3′ end as quencher dye. The 18S probe was labelled with VIC at the 5′ end as reporter dye and TAMRA at the 3′ end as quencher dye. Relative quantification was performed using the standard curve method where curves were constructed using five 10-fold dilutions of the standard sample. Standards were run in triplicates with 18S being the endogenous control. Relative amounts of PolgA and endogenous control were calculated for one sample, and the results were used to calculate the final amount of PolgA in that sample as a ratio between the two relative amounts.
The authors wish to acknowledge Maja Jagodic for excellent technical support, Maria Falkenberg and Claes Gustafsson for helpful discussions and William C. Copeland for donation of the human PolgA antibodies. A.T. is supported by Funds of Karolinska Institutet and Gun and Bertil Sthones Stiftelse and a fellowship from Loo and Hans Ostermans Stiftelse. N.H. is supported by a grant from The Royal Swedish Academy of Sciences.
Conflict of Interest statement. None declared.
|Stage||Total (tot. lit.)||+/+||+/−||−/−|
|Stage||Total (tot. lit.)||+/+||+/−||−/−|
*resorbed embryos; **1/4 of embryos resorbed.
|Gene||5′ primer||3′ primer||Probe|
|Gene||5′ primer||3′ primer||Probe|