|Home | About | Journals | Submit | Contact Us | Français|
During replication of nuclear ribosomal DNA (rDNA), clashes with the transcription apparatus can cause replication fork collapse and genomic instability. To avoid this problem, a replication fork barrier protein is situated downstream of rDNA, there preventing replication in the direction opposite rDNA transcription. A potential candidate for a similar function in mitochondria is the mitochondrial transcription termination factor 1 (MTERF1, also denoted mTERF), which binds to a sequence just downstream of the ribosomal transcription unit. Previous studies have shown that MTERF1 prevents antisense transcription over the ribosomal RNA genes, a process which we here show to be independent of the transcription elongation factor TEFM. Importantly, we now demonstrate that MTERF1 arrests mitochondrial DNA (mtDNA) replication with distinct polarity. The effect is explained by the ability of MTERF1 to act as a directional contrahelicase, blocking mtDNA unwinding by the mitochondrial helicase TWINKLE. This conclusion is also supported by in vivo evidence that MTERF1 stimulates TWINKLE pausing. We conclude that MTERF1 can direct polar replication fork arrest in mammalian mitochondria.
Human mitochondrial DNA (mtDNA) is a circular, double-stranded molecule. The two strands differ in their buoyant density in alkaline cesium chloride gradients and they are therefore called the light strand (L-strand) and the heavy strand (H-strand), respectively (1). Human mtDNA encodes 13 components of the respiratory chain. The double-stranded genome also produces 22 tRNAs and 2 rRNAs needed for mitochondrial translation. Transcription is initiated from the heavy- and light-strand promoters (HSP and LSP), which generate polycistronic, near full genome-sized transcripts covering the two strands. The primary transcripts are then processed to produce individual RNA molecules (2).
The molecular machineries responsible for transcription and replication of mammalian mtDNA are distinct from those found in the nucleus. For instance, DNA synthesis is performed by the heterotrimeric DNA polymerase γ (POLγ), which consists of a catalytic subunit (POLγA) and two identical accessory subunits (POLγB) (3,4). The replicative mtDNA helicase TWINKLE is a homohexameric protein complex with similarity to the bacteriophage T7 primase/helicase gene 4 protein (T7 gp4), but without the primase activity present in the phage protein (5–8). A mitochondrial single-stranded DNA binding protein (mtSSB) protects the single-stranded parental strand during mtDNA replication and stimulates both POLγ and TWINKLE activities (6,9–12). Mitochondria also contain a specialized transcription machinery, which include a monomeric RNA polymerase (POLRMT) and two accessory factors, mitochondrial transcription factors A and B2 (13–16). There is also a mitochondrial transcription elongation factor (TEFM), which helps POLRMT to transcribe longer stretches of RNA, and to bypass regions generating highly structured RNA (17–19). In addition to producing mRNA, tRNA and rRNA, the mitochondrial transcription machinery also generates primers for initiation of mtDNA replication at the major origins of the two strands (20).
For a long time, it was believed that mitochondrial transcription is regulated at the level of termination. This idea was based on the identification of the MTERF1 (formerly denoted as mTERF), a protein that binds sequence-specifically to a 28-bp region immediately downstream of the 16S rRNA gene (21,22). According to the belief at the time, MTERF1 regulated the levels of heavy-strand transcription that proceeded beyond its binding site, a model that was put forward to explain why rRNA transcripts were almost 50 times more abundant than the mRNA transcripts produced downstream of the MTERF1 binding site (23). More recently, in vivo analysis of cultured human cells (24) and the Mterf1−/− knockout mouse refuted this idea (25). Loss of MTERF1 does not affect the relative steady-state levels of rRNAs and mRNA encoded on the H-strand. Instead, these relative RNA levels are most likely explained by differences in stability, with rRNA having a much longer half-life. Studies of the Mterf1−/− knockout mouse model instead revealed that Mterf1 prevents antisense transcription from entering the rRNA gene region. In the absence of Mterf1 antisense transcription progresses beyond the rRNA genes and may also interfere with light-strand promoter (LSP) function. These in vivo findings agree nicely with the effects of MTERF1 on transcription termination in vitro, which found that the protein completely blocks L-strand transcription, but only has a minor effect on H-strand transcription (26,27).
According to the strand-asymmetric model of mtDNA replication, DNA synthesis is continuous on both strands and each strand contains a distinct, major origin of replication, OriH and OriL (1,28). Replication is strand-asynchronous, i.e. synthesis of the H-strand is first initiated at OriH and the reaction proceeds approximately two thirds of the way around the circular genome, displacing the non-template H-strand, which becomes transiently complexed with mtSSB (29). Two-dimensional agarose gel-electrophoresis has also identified the presence of RNA annealed to the displaced, non-template H-strand DNA (30,31). It has been proposed that these RNA species can replace mtSSB and stabilize ssDNA during replication, but this remains a debated issue (29,32). L-strand DNA synthesis is initiated at OriL by a process that is believed to involve activation of the origin when it becomes single-stranded. The main features of this strand-asymmetric replication mode are supported by a number of different observations, including electron microscopy analysis (28), biochemical analysis of replication intermediates in both wild-type cells and tissues, and where expression of key replication factors has been manipulated (33–35), in vivo saturation mutagenesis (36), the in vivo occupancy pattern of mtSSB (29), and the possibility to reconstitute specific steps of this process in vitro (12,20,37).
How DNA replication and transcription are coordinated in human mitochondria is not known. Unregulated clashes between replication and transcription may affect gene transcription and/or impair DNA replication (38). To avoid such problems in nuclear DNA replication, the moving replication fork can be arrested close to the 3′-end of highly transcribed ribosomal DNA (rDNA). A replication fork barrier prevents DNA replication from progressing in a direction opposite that of rDNA transcription, which supposedly prevents head-on collisions between replication and transcription machineries (38). In mouse, the mTTF-1 protein binds to Sal boxes situated just downstream of the ribosomal transcription units and block replication with opposite polarity to that of ribosomal transcription (39). In budding yeast, the Fob1 protein ensures that rDNA replication only occurs in the direction of transcription and that the replication machinery does not disturb transcription of rDNA (40). If a similar replication fork barrier protein also exists in mitochondria is not known, but the binding of MTERF1 just downstream of the 3′-end of rDNA makes it an interesting candidate for such a function. In support of this idea, a previous report demonstrated that overexpression of MTERF1 stimulates replication pausing at MTERF1 binding sites in vivo. The expression levels of MTERF1 correlated with the strength of the pausing and it was hypothesized that MTERF1 acts to coordinate the passage of replication and transcription complexes (41). In this report, we use in vitro biochemistry to demonstrate that MTERF1 acts as a replication fork barrier and delays replication fork progression. The effect is explained by MTERF1's ability to work as a contrahelicase with distinct polarity.
The HSP and LSP transcription templates were cloned as described in (13) and linearized with PstI and EcoRI respectively. The transcription reaction volumes were 25 μl and contained 25 mM Tris-HCl pH 8.0, 10 mM MgCl2, 40 mM NaCl, 100 μg/mL BSA, 10 mM DTT, 400 μM ATP, 150 μM GTP, 150 μM CTP 10 μM UTP, 0,02 μM α-32P UTP (3000 Ci/mmol), 4 U RNase inhibitor Murine (New England Biolabs), 4 nM of indicated plasmid template. The reactions contained POLRMT (20 nM), TFAM (200 nM), and TFB2M (60 nM). TEFM (40 nM) and MTERF1 (2, 8 and 32 nM) were added as indicated. The reactions were stopped and analyzed on 4% denaturating polyacrylamide gels as described previously (13).
We cloned a DNA fragment corresponding to nt 3056–3559 of the mitochondrial human genome containing the MTERF1 binding site (nt 3231–3253) between the HindIII and EcoRI sites in the pBluescript SK(+) and pBluescript SK(−) vectors (Agilent Technologies; La Jolla, CA, USA). The pBluescript SK(−) MTERF1 construct was used as a template for site-directed PCR mutagenesis reactions to generate a mutant variant of the MTERF1 binding site containing a 4-bp deletion (nt 3241–3244). All constructs were confirmed by sequencing and used to isolate ssDNA following the manufacturer's protocol (Stratagene). To produce the rolling-circle DNA replication templates, we annealed a 70-mer oligonucleotide (5′-40[T]- AGT TAC CAA TGC TTA ATC AGT GAG GCA CCT-3′) to the pBluescript SK(+) and (5′-40[T]-TCG CTG AGA TAG GTG CCT CAC TGA TTA AGC-3′) to the pBluescript SK(−) ssDNA (7.5 pmol) and synthesized the second strand as described previously (20).
To prepare the helicase substrates indicated in the figure legends, three different oligonucleotides; Helicase (WT) forward; TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT AAG ATG GCA GAG CCC GGT AAT, Helicase (WT) reverse; TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT ATT ACC GGG CTC TGC CAT CTT A, or Helicase (del) reverse; TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TTT TCG ATT ACC GGG GCC ATC TTA AC were labeled with 32P at the 5′-termini using T4 polynucleotide kinase (Stratagene) and annealed to ssDNA (described above) containing the corresponding DNA sequence creating the helicase substrates indicated in the figure legends. After annealing, the samples named SK+Terf1, SK-Terf1, and SK- TERF1 del were run through Centricon 100 columns (Millipore Company) to remove unannealed oligonucleotide as previously described (6).
The reaction mixtures (20 μl) contained 10 fmol of the indicated dsDNA template and were performed as described previously but in the presence of MTERF1 when indicated in the figure legends. The reactions were incubated at 37°C and stopped after 40 min (or at the times indicated) by adding 200 μl of transcription stop buffer, 0.5% SDS and 2 μl of proteinase K. After another 45 min incubation at 42°C, the samples were precipitated and analyzed as described before (12).
DNA templates for TWINKLE helicase assays were prepared as described above. The reaction mixture (25 μl) contained 20 fmol of the indicated DNA substrate, 20 mM Tris–Cl (pH 7.6), 4.5 mM MgCl2, 10 mM DTT, 3 mM UTP, 0.1 mg/ml BSA, 40 mM NaCl, 600 fmol TWINKLE or 100 fmol T7 gp4 protein and 500 fmol of MTERF1 if nothing else is indicated in the figure legends. The reactions were incubated at 32°C for 20 min (or as indicated) and stopped by the addition of 2 μl stop buffer (90 mM EDTA, 6% SDS, 30% glycerol, and 0.25% bromphenol blue), and analyzed on a 6% non-denaturing polyacrylamide gel.
MTERF1-MycHis expressing Flp-In™ T-RexTM-293 cells (41) were cultured in DMEM medium with 5% fetal bovine serum, hygromycin B (50 μg/ml) and blasticidin-HCl (15 μg/ml) in 10 × 75 cm2 tissue culture flasks to 80% confluence. Of these, 5 × 75 cm2 flasks cells were induced to express the Myc-tagged mTERF transgene by addition of 100 ng/ml doxycycline, whereas the other 5 × 75 cm2 flasks were kept as a non-induced control. After 24 h induction, the cells were harvested and washed once with ice-cold phosphate buffered saline (PBS). Isolation of mitochondria was carried out following a protocol described in (43). The mitochondria were washed once with ice-cold PBS, and next incubated with 1% formaldehyde in PBS for 10 min at rt. The crosslinking reaction was quenched by addition of glycine (final conc. 125 mM) and then incubated for five additional minutes. After washing twice in ice-cold PBS, mitochondria were lysed in 25 mM HEPES-KOH (pH 7.6), 10% glycerol, 5 mM MgCl2, 0.5 mM EDTA, 0.5% tween-20, 0.15 M KCl, 1 mM phenylmethlsulfonylfluoride, 2 mM pepstatin A, 0.6 mM leupeptin, and 2 mM benzamidine. The mitochondrial lysates were sonicated in a Bioruptor UCD 200TM (Diagenode) for 10 min at high output, with intervals of 30 s on and 30 s off, and then centrifuged for 5 min at 14 000 × g. Aliquots of 100 μl wt and induced supernatant were incubated with 1.5 μl of human TWINKLE polyclonal antibody (Agrisera, Sweden) or 1.5 μl rabbit IgG (ab37415, Abcam) overnight in a rotator at 4°C. 50 μl of protein A beads (GE healthcare) were added to supernatants and the indicated antibody for 1 h at 4°C. After wash and elution, samples were incubated overnight at 65°C to reverse crosslinking. RNA contaminations were removed by incubation with 100 ng/ml RNaseA for 15 min at 37°C, and proteins were removed by addition of 20 μg proteinase K and incubation for 2 h at 56°C. DNA was purified by phenol/chloroform extraction, followed by ethanol precipitation. The purified DNA was used for real-time PCR analysis (Bio-Rad) with the primer pairs listed in Supplementary Table S1. Quantifications were performed using real time PCR Software (Bio-Rad) and Excel (Microsoft); ratios of IP/input are depicted in the figures after subtracting ratios obtained from the rabbit IgG control. Three independent biological replicates were carried out for ChiP analysis. Averages and standard deviations were calculated and plotted in Microsoft Excel.
The transcription elongation factor TEFM was just recently identified (17–19). Before proceeding to studies of MTERF1 and DNA replication, we decided to investigate if the presence of TEFM affected MTERF1 activities in vitro and conclusions drawn from these previous studies (27). In mitochondria, transcription can approach the MTERF1 binding site from two different directions. To mimic each of these situations, we used linearized DNA templates with either the HSP or LSP promoter, followed by the MTERF1 binding site oriented in the same direction relative the promoter as that observed in vivo (Figure (Figure1A).1A). In our analysis, we added increasing concentrations of MTERF1 to transcription reactions containing POLRMT, TFAM, TFB2M, and the DNA template. Using the LSP template in the absence of MTERF1 we observed synthesis of the expected full-length run-off transcripts (Figure (Figure1B,1B, lane 1, LSP RO) and also a previously described pre-terminated transcript at CSBII (CSBII PT). When we added increasing amounts of MTERF1, we observed the appearance of an additional transcript with the expected size for an RNA transcript prematurely terminated at the MTERF1-binding site (Figure (Figure1B,1B, lanes 2–4, MTERF1 PT). In agreement with previous reports, addition of TEFM stimulated transcription and abolished transcription termination at CSBII. However, it did not affect MTERF1-dependent transcription termination (Figure (Figure1B,1B, lanes 5–8).
We next analyzed effects on the HSP template. In the absence of MTERF1 we observed synthesis of the expected full-length run-off transcripts (Figure (Figure1B,1B, lane 9, HSP RO). Addition of increasing concentrations of MTERF1 was much less effective in terminating transcription in this orientation and we only observed the appearance of a very weak band at the expected size for an RNA transcript terminated at the MTERF1-binding site (Figure (Figure1B,1B, lanes 10–12, MTERF1 PT). Addition of TEFM abolished this weak band (Figure (Figure1B,1B, lanes 13–16). From our observations, we conclude that MTERF1 terminates transcription initiated from LSP also in the presence of TEFM. In contrast, TEFM does decrease the already very weak MTERF1 termination activity on transcription initiated from HSP.
We next addressed the effects of MTERF1 on mtDNA replication. The purified POLγ, mtSSB, and TWINKLE proteins together form a minimal mitochondrial replisome, which in rolling circle replication can support synthesis of single-stranded DNA stretches of more than 20 000 nt (12). We constructed a DNA template that contained a replication fork on which the replication machinery could be loaded. The template also contained an approximately 3.5 kb long dsDNA region and a free 3′-end that can prime leading-strand DNA synthesis. On this type of circular template, leading-strand DNA synthesis coupled to continuous unwinding of the double-stranded template can, in principle, progress indefinitely. To monitor the effects of MTERF1, we introduced a 38-bp sequence covering the MTERF1 region, in two different orientations (Figure (Figure2A2A).
First, we incubated the circular templates with the components of the mtDNA replication machinery and monitored DNA synthesis using radiolabeled dNTPs in a rolling-circle experiment (Figure (Figure2B).2B). To test if MTERF1 has a replication fork-barrier activity, we added increasing amounts of MTERF1 and monitored effects on DNA synthesis. When the MTERF1 site was positioned in the forward orientation (replication progressing in the same orientation as rRNA transcription initiated at HSP), we could not detect a significant effect on rolling-circle replication (Figure (Figure2B,2B, lanes 6–10). In contrast, when the MTERF1 site was in the reverse orientation (replication progressing in the opposite orientation to rRNA transcription, i.e. antisense transcription initiated at LSP), we observed a dramatic decrease in DNA synthesis (Figure (Figure2B,2B, lanes 1–5). These results indicated that MTERF1 blocks rolling-circle DNA replication in a direction-dependent manner.
Addition of MTERF1 not only led to inhibition of DNA replication, but also to the formation of replication products of defined lengths (e.g. Figure Figure2B,2B, lane 5). To further elucidate this effect, we performed a time-course experiment. We used the rolling-circle template with the MTERF1 site oriented in the reverse direction. We incubated this template with our reconstituted in vitro DNA replication system in the presence of constant amounts of MTERF1 (Figure (Figure2C).2C). In the absence of MTERF1, the replication products formed a continuous smear (Figure (Figure2C,2C, lanes 1–4). In the presence of MTERF1 protein, we observed the formation of specific-sized replication products (Figure (Figure2C,2C, lanes 6–9) The size intervals between neighbouring replication products were constant, about 3.5 kb, which corresponded to the size of the DNA template molecule, suggesting that the replication fork arrested in the same region each time the circle was traversed. In addition, the first band was at about 5.5 kb, corresponding to the size expected for replication pausing at the MTERF1 binding site during the first such cycle. We next repeated the experiment with an identical template, but with the MTERF1 binding site oriented in the forward direction. With this template, we failed to observe any effects of the MTERF1 protein on DNA synthesis (Figure (Figure2D).2D). Our results support the conclusion that MTERF1 arrests the mtDNA replication machinery when approaching the 3′-end of the rRNA transcription unit, with the same directionality as transcription from LSP. The time-course experiment also demonstrated that mtDNA replication was not definitively terminated at the MTERF1 binding site, but that MTERF1 stimulated pausing at this site, since shorter replication products were chased into longer products over time (Figure (Figure2C,2C, compare 5.5 kb product in lanes 7–9).
To verify the relevance of MTERF1 binding for the observed effects on mtDNA replication, we prepared a rolling circle replication template with 4-bp deletion in the MTERF1 binding site (Figure (Figure33 and Supplementary Figure S1). The effects of MTERF1 on mtDNA replication were monitored in a time-course experiment. The binding site mutation completely abolished the fork-pausing effect of MTERF1.
POLγ is strictly dependent on TWINKLE for replication using a dsDNA template. We decided to monitor if MTERF1 could function by blocking TWINKLE-dependent DNA unwinding. To this end, we created DNA helicase substrates by annealing oligonucleotides 32P-labeled in the 5′-end (for sequences, see Materials and Methods) to the complementary region of pBluescript II SK single-stranded DNA (SK+ or SK−) to form helicase substrates with a short double-stranded region and a 40-nt 5′-single-stranded tail. The double-stranded region contained the MTERF1-binding site in the forward or reverse orientation relative to the 40-nt 5′-single-stranded tail. In the absence of MTERF1, the TWINKLE helicase was able to unwind the DNA substrates. Addition of increasing amounts of MTERF1 had no effect on DNA unwinding of the template with the MTERF1 binding site in the ‘forward’ orientation (Figure (Figure4A,4A, lanes 1–6). In contrast, increasing amounts of MTERF1 dose-dependently blocked DNA unwinding of the template with the MTERF1 binding site in the reversed orientation (Figure (Figure4A,4A, lanes 7–12). We concluded that the ability of MTERF1 to block the TWINKLE DNA helicase is dependent of the relative orientation of the MTERF-binding site, since we observed a contrahelicase activity when unwinding progressed in an orientation opposite that of rRNA transcription, but when we reversed the orientation of the binding site, the contrahelicase activity was lost.
We next performed a time-course experiment with constant amounts of MTERF1. We used the fork-like helicase substrates with the double-stranded region containing a binding site for MTERF1 in the reverse direction (Figure (Figure4B,4B, lanes 1–10). We also included a substrate in which the MTERF1 binding site had been mutated by deleting 4 bp, AGAG, in the middle of the MTERF1 binding site (Figure (Figure4B,4B, lanes 11–20 and Supplementary Figure S1). As expected, MTERF1 was able to block TWINKLE DNA unwinding on the wild-type binding site template (Figure (Figure4B,4B, lanes 7–10). The effect was abolished when the MTERF1 binding site was mutated (Figure (Figure4B,4B, lanes 17–20). To facilitate further analysis and to investigate if the MTERF1 contrahelicase activity was specific to TWINKLE, we also investigated MTERF1 effects on a related DNA helicase, the bacteriophage T7 gene 4 protein (gp4) (Figure (Figure4C).4C). The MTERF1 and gp4 protein concentrations were kept constant and we again monitored DNA unwinding of the template in a time course experiment. In the absence of MTERF1, the gp4 helicase could unwind both templates (Figure (Figure4C,4C, lanes 3–6 and 13–16). Addition of MTERF1 blocked gp4-dependent unwinding of the substrate containing the MTERF1 binding site in the reverse orientation (Figure (Figure4C,4C, lanes 7–10). MTERF1 did not affect the gp4-driven unwinding reaction, when the binding site was mutated (Figure (Figure4B,4B, lanes 17–20). MTERF1 can thus efficiently block DNA unwinding by gp4.
We observed that the MTERF1-dependent block of DNA unwinding was not complete, since there was a weak, but noticeable increase in DNA unwinding at later time points, even in the presence of MTERF1 (Figure (Figure4B,4B, compare lanes 6 and 10). To investigate the possibility that MTERF1 stimulates helicase pausing rather than termination, we performed a more detailed time-course experiment using the DNA helicase substrate with the MTERF1 binding site in both forward (Figure (Figure5A)5A) and reverse direction (Figure (Figure5B).5B). In the forward direction, the addition of MTERF1 did not affect the DNA unwinding rate (Figure (Figure5A5A and C). In the reverse direction, i.e. that corresponding to TWINKLE approaching MTERF1 in a direction opposite that of rDNA transcription, MTERF1 inhibited DNA unwinding. Interestingly, MTERF1 did not completely block TWINKLE-dependent unwinding, instead slowed it down significantly (Figure (Figure5B5B and D). We repeated the experiment with the gp4 helicase and obtained similar results (data not shown). Based on these observations we conclude that MTERF1 is a potent contrahelicase that delays TWINKLE-dependent DNA unwinding in a direction-dependent way.
A previous study has demonstrated that overexpression of MTERF1 can increase replication pausing at MTERF1 binding sites in vivo (41). To investigate if the observed effect correlates with increased TWINKLE pausing in vivo, we performed chromatin immunoprecipitation (ChIP) using seven different primer pairs that covered the MTERF1 binding site and surrounding regions. In cells over-expressing a C-terminally Myc epitope-tagged MTERF1, we noted a peak of TWINKLE protein adjacent to the MTERF1 binding site (Figure (Figure6).6). The peak was barely noticeable in the non-induced control cells. The results confirm that MTERF1 can increase TWINKLE occupancy levels at the MTERF1 binding site, hence supporting the idea that MTERF1 acts as a contrahelicase, inducing TWINKLE pausing also in vivo.
DNA replication and transcription are fundamental processes that must be performed with high fidelity. The enzymes responsible, DNA polymerase and RNA polymerase, act in coordination with a number of additional factors and travel for long distances on the DNA template. To the best of our knowledge, DNA replication and transcription in mitochondria are not separated in time, which means that these enzymatic machineries may use the same DNA template and therefore have the potential to interfere with each other. Studies in many different systems have demonstrated that head-on collision between transcription and replication can cause decreased replication fork stability, leading to replication stress (38). This helps to explain why highly transcribed genes are associated with increased levels of DNA mutations (transcription associated mutations, TAM) and recombination (transcription associated recombination, TAR) (44–47). Head-on collision may also affect the many proteins associated with co-transcriptional RNA processing, which in mitochondria include the precise endonucleolytic cleavage of the polycistronic transcripts. Bacterial genomes seem to have evolved to minimize the problem of replication - transcription collision. For instance, in Escherichia coli, the seven highly transcribed ribosomal genes are mainly present in regions near the origin of replication (48), oriented in a way so that replication and transcription take place in the same direction (49). Another means by which collisions between replication and transcription are prevented is via specific fork-progression barriers found at highly transcribed rDNA in eukaryotic nuclei (40).
Similar to fork-barrier proteins in many other systems, MTERF1 binds immediately downstream of the rDNA transcription unit (21). Previous reports have demonstrated that MTERF1 blocks antisense transcription across the rDNA region in both human and mouse mtDNA (24,25,27). As demonstrated here, MTERF1 can also arrest the moving replication with the same directionality as that observed for transcription. MTERF1 thus acts as a replication fork barrier and causes a delay in the progression of DNA replication in the direction opposite that of rDNA transcription. As a consequence, MTERF1 can act to avoid or manage head-on collisions between replication and transcription machineries in the rDNA region, thereby ensuring that POLRMT has completed synthesis of highly structured rRNA molecules before collision with the mtDNA synthesis machinery. Our findings also provide a biochemical explanation for the observation that overexpression of MTERF1 causes replication pausing in vivo (41). The effect is explained by the ability of MTERF1 to function as a contrahelicase with distinct polarity. In support of this notion, we employed ‘chromatin’ immunoprecipitation to demonstrate that MTERF1 stimulates TWINKLE pausing in vivo.
It is tempting to speculate that MTERF1 supports a process whereby POLRMT completes rRNA transcription initiated from HSP before H-strand DNA replication continues over the rDNA region. According to this model, the mitochondrial replisome pauses at the MTERF1 binding site (Figure (Figure7,7, step 1). Once POLRMT has passed the rDNA region, transcription would facilitate the displacement MTERF1 and the block of mtDNA replication in the opposite direction is removed (Figure (Figure7,7, step 2). This model predicts that POLRMT encounters the paused TWINKLE helicase, which encircles the transcribed strand. The most plausible outcome of this collision is that POLRMT terminates transcription and that the H-strand DNA synthesis is resumed (Figure (Figure7,7, step 2A). This finding would be in line with findings in many other systems, where transcription normally yields to a moving replisome (38). A similar concept has been proposed for sea urchin (50,51) and Drosophila (52) mtDNA. In fact, the sea urchin mtDBP protein functions as a contrahelicase, inasmuch as it blocks the helicase activity of the heterologous simian virus 40 large T antigen (50).
Potentially, the interaction with MTERF1 weakens that of POLRMT and the template, which allows TWINKLE to displace POLRMT from the template strand. Termination of transcription in this manner would be a rare event, unlikely to significantly affect the relative levels of rRNA and mRNA molecules produced downstream of the MTERF1 binding site. The alternative outcome (Figure (Figure7,7, step 2B) appears less likely. If POLRMT displaces TWINKLE it would entail replisome reloading after passage of the transcription machinery, a process that could lead to mispriming and increased levels of mutations. In addition, POLRMT has very low processivity on ssDNA and, if it managed to displace TWINKLE, the polymerase would only continue transcribing for another 25–75 nt before termination (42).
What could be the consequences of unregulated, head-on collisions in the mitochondrial rDNA region? One possibility is that active replication can impair rRNA processing and maturation. Arguing against this possibility, characterization of the Mterf1 knockout mice has not demonstrated any major effects on rRNA biogenesis (25). We instead favour the idea that MTERF1 blockage of DNA replication prevents transcription associated mutations or genetic alterations in mtDNA. Given the presence of thousands of mtDNA molecules in each mammalian cell, functional consequences of increased mutation levels and/or genetic stability may be difficult to detect in a short time span, which could explain the complete lack of observable phenotypes in the Mterf1 knockout mice model. Nevertheless, the maintenance of mitochondrial genomic fidelity over evolutionary time must be sufficiently important for species survival, at least within the mammalian lineage where MTERF1, as well as its binding site, are conserved (53). In future experiments, we will develop tools to specifically identify effects on low levels genetic alterations in these mice, to see if we can detect increased levels of e.g. mtDNA mutations, which may corroborate this hypothesis.
We thank the Mammalian Protein Expression (MPE) Core Facility at the University of Gothenburg for cell production.
Supplementary Data are available at NAR Online.
Swedish Research Council grants [2013-3621 to M.F., 2012-2583 to C.M.G.]; Swedish Cancer Foundation (to M.F. and C.M.G.); European Research Council [261248 to M.F., 268897 to C.G.]; IngaBritt and Arne Lundberg foundation (to M.F.); Knut and Alice Wallenbergs foundation (to M.F. and C.M.G.); Academy of Finland (FinMIT Centre of Excellence 118654 and Professorship 256615 to H.T.J.). Funding for open access charge: European Research Council.
Conflict of interest statement. None declared.