RECQL4 and p53 potentiate the activity of polymerase γ and maintain the integrity of the human mitochondrial genome

Abstract

Germline mutations in RECQL4 and p53 lead to cancer predisposition syndromes, Rothmund–Thomson syndrome (RTS) and Li–Fraumeni syndrome (LFS), respectively. RECQL4 is essential for the transport of p53 to the mitochondria under unstressed conditions. Here, we show that both RECQL4 and p53 interact with mitochondrial polymerase (PolγA/B2) and regulate its binding to the mitochondrial DNA (mtDNA) control region (D-loop). Both RECQL4 and p53 bind to the exonuclease and polymerase domains of PolγA. Kinetic constants for interactions between PolγA–RECQL4, PolγA–p53 and PolγB–p53 indicate that RECQL4 and p53 are accessory factors for PolγA–PolγB and PolγA–DNA interactions. RECQL4 enhances the binding of PolγA to DNA, thereby potentiating the exonuclease and polymerization activities of PolγA/B2. To investigate whether lack of RECQL4 and p53 results in increased mitochondrial genome instability, resequencing of the entire mitochondrial genome was undertaken from multiple RTS and LFS patient fibroblasts. We found multiple somatic mutations and polymorphisms in both RTS and LFS patient cells. A significant number of mutations and polymorphisms were common between RTS and LFS patients. These changes are associated with either aging and/or cancer, thereby indicating that the phenotypes associated with these syndromes may be due to deregulation of mitochondrial genome stability caused by the lack of RECQL4 and p53.

Summary:

The biochemical mechanisms by which RECQL4 and p53 affect mtDNA replication have been elucidated. Resequencing of RTS and LFS patients’ mitochondrial genome reveals common mutations indicating similar mechanisms of regulation by RECQL4 and p53.

Introduction

Rothmund–Thomson syndrome (RTS) is an autosomal recessive disorder caused by germline mutations in RECQL4 that predisposes the affected individuals to cancer (1). Clinical manifestations of RTS include poikilodermal lesions, short stature, sparse scalp hair, juvenile cataract, skeletal abnormalities, radial ray defects, premature aging and predisposition to osteosarcoma and lymphoma. Karyotypic analysis of RTS patient cells show increased level of the chromosomal rearrangements, trisomies, deletions and translocations (2).

RECQL4 mutations include nonsense, missense and splice site mutations with intronic insertions and deletions, which lead to frameshift changes and subsequent termination (2). Multiple RECQL4 knockout mouse models have provided crucial insights into the role of RECQL4 in the development of RTS. The knockout mouse, bearing deletions in exons 9–13 of RECQL4, survived till adulthood and recapitulated some of the clinical symptoms of RTS patients (3). Biochemically it has been recently demonstrated that RECQL4 has ATP-dependent 3′ → 5′ helicase activity and can unwind DNA substrates such as splayed arms, bubbles and blunt-end duplexes. The conserved helicase domain plays a key role in the unwinding activity of RECQL4 (4).

RECQL4, which contains canonical nuclear import and retention domains in its N-terminal region, was initially reported to localize in the nucleus (5). Subsequently, it was shown that RECQL4 was localized both in the nucleus and cytoplasm (6). Recently, we and others have shown that RECQL4 localizes to the mitochondria (7–9) in all phases of the cell cycle except the S phase at which it is found exclusively in the nucleus. The localization of RECQL4 in the mitochondria is due to the presence of a validated N-terminal mitochondrial localization signal (9). The mitochondrial localization has also been verified from cell fractionation studies (7–9). The presence of two putative nuclear export signals (NES2 and NES3) at the C-terminal region of RECQL4 may also abet in its mitochondrial localization (7).

Previous studies have revealed that RECQL4 facilitated the formation of the prereplication complex during nuclear DNA replication in S phase (10). RECQL4 facilitates the binding of DNA polymerase α to chromatin, thereby leading to the formation of the replication fork and subsequent initiation of DNA replication (11). RECQL4 has also been shown to be associated with the essential replication factors MCM10, MCM2–7 helicase, CDC45, GINS complex and SLD5 (12). RECQL4 is recruited to the origins of replication in late G₁ phase and loaded onto the origins in the S phase. Hence, nascent DNA synthesis and the frequency of the firing of the origin of replication are both reduced after RECQL4 depletion (13). Recent evidence indicates that the helicase and the C-terminal domain of RECQL4 may facilitate replication elongation on DNA templates damaged by ionizing radiation (14). The involvement of RECQL4 in the regulation of nuclear replication provided the hint that it might also be involved in mitochondrial DNA (mtDNA) replication. Comparison of the de novo mtDNA replication between the isogenic control and RTS patient cell lines indicated that lack of RECQL4 led to a decrease in the incorporation of bromodeoxyuridine at the sites of nascent DNA synthesis in the mitochondrial nucleoids (9). During the nascent DNA synthesis in the mitochondrial nucleoids, RECQL4 colocalizes with the mtDNA polymerase holoenzyme PolγA/B2 (7,9), thereby indicating its role in mtDNA replication.

The tumor suppressor p53 is a multifunctional protein best known for its transactivation functions in the nucleus (15). Germline mutations (mostly missense and a few nonsense mutations) in p53 lead to the rare autosomal recessive disorder called Li–Fraumeni syndrome (LFS). LFS is characterized by the early onset of multiple forms of tumors (16). The roles of p53 in regulating multiple mitochondrial functions including bioenergetics metabolism, antioxidant effect and mitophagy have gained attention in recent years (17). We have recently provided evidence that in unstressed cells, a fraction of p53 associates with RECQL4. RECQL4 masks the nuclear localization signal of p53, resulting in the accumulation of p53–RECQL4 complex within the mitochondrial nucleoids (9). Apart from mitochondrial entry, the recruitment of p53 to the sites of de novo mtDNA replication is also regulated by RECQL4 (9). Several lines of evidence suggest that p53 interacts with multiple components of the mtDNA replication machinery like mitochondrial transcription factor A (TFAM) and mitochondrial single-stranded DNA-binding proteins (mtSSB) (18,19). Putative interaction between p53 and mitochondrial polymerase (PolγA) has been postulated based on the finding that the loss of p53 resulted in increased mtDNA mutations (20). However, the biochemical mechanisms by which both RECQL4 and p53 affect mtDNA replication remains unclear. Our results reveal how RECQL4 and p53 potentiate the polymerization and proofreading activities of PolγA/B2. Furthermore, the occurrence of a subset of common mutations in patients lacking either RECQL4 or p53 is consistent with the notion that both these factors are essential for the maintenance of mitochondrial genome integrity.

Materials and methods

Antibodies

Anti-RECQL4: mouse monoclonal K6312 (9); goat polyclonal sc-16924 (Santa Cruz). Anti-p53: mouse monoclonal sc-126 (Santa Cruz), rabbit polyclonal SIG-3520-1000 (Covance). Anti-Flag antibody: mouse monoclonal F1804 (Sigma Chemical Company). Anti-Flag antibody beads: A2220 (Sigma Chemical Company). Anti-PolγA: rabbit polyclonal sc-48815 (Santa Cruz Biotechnology). Anti-PolγB: mouse monoclonal SAB1402537 (Sigma Chemical Company). All secondary antibodies were purchased from Jackson ImmunoResearch.

Recombinants

Glutathione S-transferase–p53 (GST–p53) (1–393), GST–p53 (1–75), GST–p53 (76–320), GST–p53 (321–362), GST–p53 (363–393) (21). pcDNA3.1 p53 (1–393) (a kind gift from Ronald Hay, University of Dundee, UK). pcDNA3-Flag-RECQL4 (1–1208), pcDNA3-Flag-RECQL4 (1–459), pcDNA3.1 hygro (+) Flag PolγA have been described earlier (9). pcDNA3-Flag-RECQL4 (460–868) and pcDNA3-Flag-RECQL4 (869–1208) were obtained by cloning RECQL4 fragments between the NheI/XhoI sites of Flag-tagged pcDNA3 construct. pGEX4T-1 mtSSB (17–148) was obtained by cloning the PCR product into EcoRI/XhoI sites of pGEX4T-1 vector. pGEX4T-1 PolγA (53–1239) and pGEX4T-1 PolγB (27–485) were obtained by cloning the respective PCR products into BamH1/NotI sites of pGEX4T-1 vector. pGEX4T-1 PolγA (440–1239) and pGEX4T-1 PolγA (440–815) were obtained by cloning the respective PCR fragments into the EcoR1/Not1 sites of pGEX4T-1. While pGEX4T-1 PolγA (53–439) was obtained by cloning the PCR product into the EcoRI/NotI site of pGEX4T-1, pGEX4T-1 PolγA (816–1239) was obtained by cloning the PCR fragment in the Not1 site of the vector and subsequently checking the orientation. pBacPAK mtSSB, pBacPAK PolγA and pBacPAK PolγB (kind gifts from Maria Falkenberg, Göteborgs Universitet, Sweden) were used as templates for the corresponding clonings. PolγA (53–1239) and PolγB (27–485) polypeptides are referred to as PolγA and PolγB. pcDNA3-Flag-RECQL4 (1–1208) K508A was obtained by site-directed mutagenesis using the site-directed mutagenesis kit by Agilent Technologies.

Cells lines

Immortalized NHF, NHF E6 (22); LFS 041 p53 (−/−), LFS 172 p53 (−/−), LFS 087 p53 (−/−) cells (23). Growth conditions for RTS fibroblasts AG03587; AG05013, AG05139, L9552914-J, B1865425K, D8903644-K, isogenic strains NHF shRECQL4, NHF shControl and Aequorea coerulescens green fluorescent protein (AcGFP)–RECQL4 (1–1208) Clone 1 have been described earlier (9). HCT116 and HCT116 p53^−/− cells were grown in McCoy's 5A medium containing 10% fetal calf serum.

Additional materials and methods

Additional information regarding Materials and methods can be found in Supplementary Materials and mSupplementary Data, available at Carcinogenesis Online. This includes mitochondrial chromatin immunoprecipitation (mtChIP) analysis, expression, purification, interactions of proteins and limited proteolysis, primer extension assay, polymerization assay, exonuclease/proofreading assay, electrophoretic mobility shift assay (EMSA) and mtDNA resequencing.

Results

RECQL4 and p53 regulate the binding of PolγA/B2 to the mitochondrial control region known to form D-loop structures

We have recently demonstrated that RECQL4 and p53 are required for optimal de novo mtDNA replication (9). To further characterize the interaction of RECQL4–p53–PolγA with mitochondrial control region in vivo, we carried out mtDNA ChIP assays with a combination of five primer sets spanning the entire control region (Figure 1A; Supplementary Table S1, available at Carcinogenesis Online). In contrast to NHF shRECQL4 cells, PolγA in NHF shControl cells showed distinct and consistent binding within certain specific regions of the control region. Binding of PolγA was observed only with PCR products I and IV (Figure 1B). PCR product I is adjacent to the cytochrome B coding sequence, whereas product IV, which spans the C-tract or D310 region, is known to be a mutation hot spot in a wide variety of cancers (24) and is involved in mtDNA replication and transcription. This region also serves as one of the origins of mtDNA replication (25). Decreased binding of PolγA to the PCR product IV was observed in multiple RTS patient fibroblasts (Figure 1C), LFS patient fibroblasts (Figure 1D) and in NHF E6 cells lacking p53 expression (Figure 1E). These results indicate that both RECQL4 and p53 are required for optimal binding of mitochondrial polymerase to this specific region. Importantly, both p53 and RECQL4 also bind to the PCR product IV (Figure 1F). To confirm concomitant binding of PolγA–RECQL4 to this region, sequential and reciprocal mtDNA ChIP (re-mtDNA ChIP) was carried out with PolγA and RECQL4 (Figure 1G and H). We observed consistent increase in signal intensity for PCR products I and IV (but not for the PCR product III, which served as the negative control). Altogether, the results indicated that the presence of RECQL4 and p53 were both required for optimal binding of PolγA to specific regions within the control region and thereby regulate mtDNA replication.

Fig. 1.

In vivo interactions between RECQL4 and PolγA/B2 on mitochondrial D-loop. (A) Primer sets used for mtDNA ChIP analysis of the D-loop. Roman numerals (I–V) below the discontinuous lines correspond to the respective PCR products. (B and C) In vivo binding of PolγA to mtDNA is decreased in the absence of RECQL4. mtDNA ChIP analysis with PolγA antibody in (B) NHF shControl and NHF shRECQL4 cells or (C) NHF and RTS fibroblasts. The relative binding of PolγA to IgG has been represented with respect to the five PCR products spanning the D-loop. Bars indicate mean ± SD. (D and E) In vivo binding of PolγA/B2 to mtDNA is dependent on p53. Same as in (C), except the mtDNA ChIP was performed with PolγA antibody in (D) NHF and LFS fibroblasts and (E) NHF and NHF E6 cells. Bars indicate mean ± SD. (F) Both RECQL4 and p53 binds to the D-loop, which contains mtDNA origins of replication. Same as in (B), except the mtDNA ChIP was carried out in NHF cells with either RECQL4 (sc-16924) or p53 (sc-126) antibody. Bars indicate mean ± SD. (G and H) PolγA binds to mtDNA in the same regions as RECQL4. Same as in (B), except that the reciprocal mtDNA ChIP (re-mtDNA ChIP) was carried out in NHF using (G) PolγA antibody followed by RECQL4 antibody (sc-16924) and (H) RECQL4 antibody (sc-16924) followed by PolγA antibody. The relative PolγA–RECQL4 binding has been represented with respect to three PCR products spanning the D-loop. Bars indicate mean ± SD. (I and J) RECQL4 and p53 interacts in vivo with PolγA and PolγB. The experiments were carried out in two isogenic pairs of cell lines (A) AG05013 and AG05013 AcGFP–RECQL4 (1–1208) Clone 1 cells and (B) HCT116 and HCT116 p53^−/−. Left: levels of RECQL4, p53, PolγA and PolγB in the whole cell extracts in the two pairs of isogenic cell lines. Levels of hsp90 were used as the loading control. Right: immunoprecipitations with (A) anti-RECQL4 antibody or (B) anti-PolγA antibody were carried out with lysates from (A) AG05013 and AG05013 AcGFP–RECQL4 (1–1208) Clone 1 cells or (B) HCT116 and HCT116 p53^−/− cells. The respective immunoprecipitates were probed with antibodies against RECQL4 (K6312), p53 (sc-126), PolγA and PolγB.

RECQL4 and p53 physically interact with polymerase γ

To elucidate the role of RECQL4 and p53 in mtDNA replication, we first determined whether RECQL4 and p53 were complexed in vivo with the PolγA/B2 holoenzyme. For this purpose, co-immunoprecipitations of endogenous proteins was carried out in two isogenic pairs of cell lines, namely RTS patient fibroblast AG05013 and its isogenic counterpart expressing wild-type RECQL4 namely AcGFP–RECQL4 (1–1208) Clone 1 (Figure 1I) (9). In addition, we also performed co-immunoprecipitations of the endogenous proteins from cell lines HCT116 and HCT116 p53^−/− (Figure 1J). Immunoprecipitation with anti-RECQL4 antibody revealed the presence of a complex consisting of RECQL4, p53, PolγA and PolγB in AG05013 AcGFP–RECQL4 (1–1208) Clone 1 cells. Such a complex was absent in the co-immunoprecipitations of AG05013 cells, which do not express RECQL4, indicating the specificity of the reaction (Figure 1I, right). A similar complex was also observed in HCT116 cells when immunoprecipitated with PolγA antibody. The PolγA immunoprecipitate in HCT116 p53^−/− cells contained only RECQL4 and PolγB (and not p53), thereby again indicating the specificity of the complex formation (Figure 1J, right). The lack of any specific signal when the immunoprecipitations were carried out with IgG beads further confirmed the presence of the RECQL4–p53–PolγA/B2 complex in vivo.

To further elucidate the mechanism by which RECQL4 and p53 interact with members of the mtDNA replication machinery, we performed in vitro pulldown assays with recombinant proteins (Figures 2 and 3). Since p53 is known to interact with mtSSB (19), this served as a positive control for the interaction of p53 with other proteins involved in mtDNA replication. We observed that p53 interacted with both PolγA and PolγB (Figure 3A), whereas RECQL4 interacted with PolγA alone but not with either PolγB or mtSSB (Figure 2B), indicating the specificity of the respective interactions. Treatment with DNase or RNase failed to disrupt the interaction of PolγA with RECQL4 and p53 but instead showed a consistent albeit modest increase in signal intensity (Figures 2C and 3B, left). However, interaction between p53 and PolγB displayed sensitivity to treatment with RNase (Figure 3B, right), implicating that RNA physically assists interaction between these two proteins.

Fig. 2.

RECQL4 N-terminal region binds to the exonuclease and polymerase domains of PolγA. (A) Coomassie blue-stained gels showing purified GST-tagged PolγA (53–1239), PolγB (27–485) and mtSSB (17–148). Pure GST alone is shown as a reference. (B and C) PolγA binds to RECQL4 in vitro. (B) Interaction assays were carried out using in vitro-translated S³⁵ methionine-labeled RECQL4 and GST-tagged PolγA (53–1239), PolγB (27–485) and mtSSB (17–148). The gel was dried and autoradiographed. (C) Same as in (B), except the interaction assays were carried out in presence of DNase or RNase. (D) PolγA binds to the N-terminal region of RECQL4. Interactions were carried out between lysates expressing Flag–RECQL4 fragments and GST or GST-tagged PolγA. The Flag–RECQL4 fragments bound to either GST or GST-tagged PolγA were visualized by western blotting using anti-Flag antibody. The lanes containing only the Flag-tagged RECQL4 fragments represent the respective input lysates (10% of the amount taken for interaction). (E) Coomassie blue-stained gels showing purified GST-tagged fragments: PolγA (53–439), PolγA (440–1239), PolγA (440–815), PolγA (816–1239) and PolγA (53–1239). Pure GST alone is shown as a reference. (F) RECQL4 binds to the exonuclease and polymerization domains of PolγA. Interactions were carried between in vitro-translated S³⁵ methionine-labeled RECQL4 and GST-tagged recombinant wild-type PolγA and its deletion fragments. Bound radioactivity was visualized by autoradiography. (G) Dissociation constant (K_d) for RECQL4–PolγA interaction. Glutathione-beads-bound PolγA (0.166, 0.33, 0.5, 0.66, 1.33, 2, 4.66 and 6.67 µM) was incubated with S³⁵ methionine-labeled in vitro-translated RECQL4 (8.4628 fmol) for 1 h at 4°C. The samples were analyzed on sodium dodecyl sulfate–polyacrylamide gel electrophoresis and the relative amount of bound S³⁵-labeled RECQL4 was quantitated. The rate of product formation was plotted against bound PolγA to obtain the saturation curve from which the K_d and V_max values for RECQL4–PolγA complex were calculated to be 114.7 ± 22.5nM and 0.203 ± 0.006 fmol/h, respectively.

Fig. 3.

p53 binds to PolγA and PolγB. (A and B) PolγA binds to p53. (A) Interactions were carried out using in vitro-translated S³⁵ methionine-labeled p53 and bound GST-tagged PolγA (53–1239), PolγB (27–485) and mtSSB (17–148). Bound radioactivity was visualized by autoradiography. (B) Same as in (A), except that the interaction of S³⁵ methionine-labeled p53 was carried out with purified GST-tagged PolγA (53–1239) or GST-tagged PolγB (27-485) in the presence of DNase or RNase. Asterisk (*) indicates product generated potentially from internal initiation during in vitro transcription cum translation. (C) Coomassie blue-stained gels showing purified GST-tagged fragments: p53 (1–75), p53 (76–320), p53 (321–362), p53 (363–393) and p53 (1–393). Pure GST alone is shown as a reference. (D) Both DNA-binding domain and the extreme C-terminal region of p53 bind to PolγA and PolγB. Interactions were carried out using lysates (produced in 293T cells) overexpressing either PolγA (53–1239) (top) or PolγB (27–485) (bottom) and bound GST-tagged p53 (1–75), p53 (76–320), p53 (321–362), p53 (363–393) and p53 (1–393). The bound PolγA or PolγB were detected by western analysis using antibodies to either PolγA or PolγB. (E) p53 binds to the exonuclease and polymerization domains of PolγA. Interactions were carried using in vitro-translated S³⁵ methionine-labeled p53 and bound GST-tagged recombinant wild-type PolγA and its deletion fragments. Bound radioactivity was visualized by autoradiography. Asterisk (*) indicates product generated potentially from internal initiation during in vitro transcription cum translation. (F) Dissociation constant (K_d) for p53–PolγA interaction. Glutathione-beads-bound PolγA (0.33, 0.66, 2, 3, 4, 5, 6.67, 10 and 13.34 µM) was incubated with S³⁵ methionine-labeled in vitro-translated p53 (9.252 fmol) for 1 h at 4°C. Samples were analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and the relative amount of bound S³⁵-radiolabeled p53 was quantitated. The rate of product formation was plotted against bound PolγA to obtain the saturation curve from which the K_d and V_max values for p53–PolγA complex were calculated to be 2.871 ± 0.62 µM and 1.137 ± 0.6203 fmol/h, respectively. (G) Dissociation constant (K_d) for p53–PolγB interaction. Same as in (F), except the concentrations of PolγB used were 0.63, 1.26, 1.89, 2.53, 3.16, 5.69, 6.32, 9.49, 12.65, 25.31, 31.64 and 37.97 µM. The concentration of radiolabeled in vitro-translated p53 used was 9.252 fmol. K_d and V_max values for p53–PolγB complex were calculated to be 8.349 ± 0.76 µM and 0.614 ± 0.228 fmol/h, respectively.

Deletion analysis indicated that the N-terminal 1–459 amino acids of RECQL4 interact with PolγA (Figure 2D). In contrast, using recombinant p53 fragments (Figure 3C), we showed that that the central-DNA-binding domain of p53 and the extreme C-terminal regulatory domain interacted with both PolγA and PolγB (Figure 3D). Pulldown assays carried out with recombinant PolγA deletion mutants demonstrated that both RECQL4 and p53 interacted separately with the exonuclease and polymerase domains. The interaction of both these proteins diminished in the presence of the spacer domain and the two thumbs of PolγA (Figures 2F and 3E, respectively).

The reported K_d values for PolγA–PolγB interaction is in the range of 27–35 nM (26,27), whereas the K_d values for PolγA–DNA interaction is 39 nM (28). To understand the RECQL4–PolγA, p53–PolγA and p53–PolγB interactions in quantitative terms, we determined the K_d values for the above interactions. Binding assays were performed using radiolabeled in vitro-translated RECQL4 or p53 and GST-tagged PolγA. The results yielded a K_d value of 114 ± 22.5nM for RECQL4–PolγA interaction (Figure 2G), indicating that PolγA–PolγB or PolγA–DNA interactions are almost 3- to 5-fold stronger than PolγA–RECQL4 interaction. The K_d values for p53–PolγA and p53–PolγB binding were 2.871 ± 0.62 and 8.349 ± 0.76 µM, respectively (Figure 3F and G), indicating that the affinity of these two interactions was weaker than RECQL4–PolγA binding. Hence, RECQL4 and p53 do not interfere with either the binding of PolγB to the spacer domain of PolγA or the binding of PolγA to mtDNA. Instead, RECQL4 and p53 subsequently bind to the polymerase and exonuclease domains of PolγA, thereby allowing RECQL4 and p53 to carry out their regulatory functions. Taking into account the relative K_d values, p53–PolγA and p53–PolγB associations occur after the PolγA–PolγB, PolγA–DNA and RECQL4–PolγA binding events.

RECQL4 potentiates the binding of PolγA to mtDNA

Based on the above results, we speculate that RECQL4 may potentiate binding of PolγA to the template DNA, thereby increasing its activity. To this end, EMSA was performed with a radiolabeled 469bp fragment (PCR product IV; Supplementary Table S1, available at Carcinogenesis Online), encompassing a stretch of double-stranded DNA present within the control region known to contain the mtDNA replication origins (25). At lower protein concentrations (i.e. 30nM for PolγA/B2 or 5nM for RECQL4), neither PolγA/B2 nor RECQL4 exhibited measurable binding to the labeled DNA. However, at higher protein concentrations, both PolγA/B2 holoenzyme and full-length RECQL4 were able to bind double-stranded DNA (Figure 4A). The extremely fine titration by which both RECQL4 and PolγA binds DNA is interesting and warrants further investigation. Quantitatively, 15% and 25% of the radiolabeled template was retarded in the presence of 40nM PolγA/B2 or 10nM RECQL4, respectively. Supershift assays with PolγA or RECQL4 antibodies corroborated the binding of the two proteins to the template DNA. The similar mass of wild-type RECQL4 (1–1208) and PolγA (53–1239) could be the basis for similar mobility of the supershifted complexes. Interestingly, addition of 5nM of full-length RECQL4 (which alone did not bind to the DNA) to 30nM of PolγA/B2 (which alone did not bind to the DNA) to the radiolabeled template led to the binding of the holoenzyme to the DNA, as revealed by the supershift with PolγA antibody (Figure 4A). Similar enhancement of binding of PolγA/B2 to the radiolabeled template was observed for the N-terminal region of RECQL4, amino acid fragment 1–459 that binds to PolγA, but not with the non-binding C-terminal region of the helicase, RECQL4 (868–1208) (Figure 4B). The presence of negative regulatory domain(s) within the full-length protein is likely to be the reason why a higher concentration, i.e. 5 nM RECQL4 (1–1208) (Figure 4A), was required compared with 0.75 nM of RECQL4 (1–459) (Figure 4B) for the enhancement of PolγA/B2 binding to the radiolabeled substrate. We also note that in both Figure 4A and B, the mobility differences between the trimeric complex of RECQL4–PolγA–DNA and the binary complexes (PolγA–DNA and RECQL4–DNA), either in the absence or presence of anti-RECQL4 antibody, is small, probably because of the limitation of the gel system.

Fig. 4.

Binding of PolγA with RECQL4 enhances the DNA binding activity of polymerase. (A) RECQL4 enhances the binding of PolγA/B2 to mtDNA in vitro. EMSAs were carried out with the indicated concentrations of RECQL4 (1–1208) and PolγA/B2. Stimulation of PolγA/B2 binding to the control region of mtDNA by RECQL4 (1-1208) was verified by supershift using antibodies against PolγA. (B) RECQL4 (1–459) but not RECQL4 (868–1208) stimulated the binding of PolγA/B2 to the control region of mtDNA. EMSAs were carried out with the indicated concentrations of RECQL4 fragments [RECQL4 (1–459) and RECQL4 (868–1208)] and PolγA/B2. Stimulation of PolγA/B2 binding to the control region of mtDNA by RECQL4 (1–459) was verified by supershift using antibodies against PolγA. (C) Limited proteolysis of PolγA indicates alteration in its conformational changes after binding to RECQL4. The products were obtained after limited proteolysis of PolγA alone or after pre-incubating PolγA with either RECQL4 (1–459) or RECQL4 (868–1208). The products were analyzed by western analysis with PolγA antibody.

To investigate whether binding of RECQL4 actually induces conformational changes in the PolγA, we subjected PolγA to limited proteolysis with trypsin alone or in the presence of either RECQL4 (1–459) or RECQL4 (868–1208). Western blot analysis with PolγA antibody indicated that the accessibility of PolγA to trypsin was enhanced in presence of RECQL4 (1–459) (arrows in the right-hand side of Figure 4C). Similar enhancement was not observed when PolγA was incubated with RECQL4 (868–1208). Based on these results, we speculate that RECQL4 induces subtle conformational changes in PolγA by allowing the latter to bind more effectively with mtDNA, thereby potentially enhancing its activities.

RECQL4 and p53 potentiate the functions of PolγA/B2

Similar to the classical DNA polymerases, PolγA/B2 possess both 3′ → 5′ exonuclease and DNA polymerase activity. It has been earlier demonstrated that PolγA/B2 carries out the exonuclease/proofreading function by an intramolecular strand-transfer mechanism without dissociation of the DNA from the enzyme. The authors of this study conclusively demonstrated that PolγA/B2-dependent excision rate was enhanced by the presence of 4–7 mismatches (29). To determine whether the 3′ → 5′ exonuclease or proofreading activity of PolγA was regulated by RECQL4 and p53, we used limiting amounts of PolγA in the exonuclease/proofreading assay (Figure 5A), so that the effect of the trans-acting proteins (Figure 5B) can be deciphered. Both RECQL4 and p53 were able to individually enhance the 3′ → 5′ exonuclease/proofreading activity of PolγA on the mutant template (Figure 5C and E). The N-terminal 459 amino acid fragment of RECQL4, which interacts with PolγA was sufficient to enhance this enzymatic activity (Figure 5D and E). Interestingly, the helicase-dead mutant of RECQL4 (K508A) was unable to enhance the PolγA proofreading activity to the same extent as that of wild-type RECQL4, indicating that the 3′ → 5′ unwinding activity of RECQL4 (4) may have an effect on the stimulation of PolγA proofreading activity (Figure 5D and E). With the wild-type template, PolγA failed to show any exonuclease/proofreading activity, possibly due to the absence of mismatches. Neither RECQL4 n (RECQL4 nor p53) or p53 were able to potentiate PolγA to act on the wild-type template (Figure 5E), thereby indicating that these accessory factors can only enhance the known properties of PolγA.

Fig. 5.

RECQL4 and p53 stimulate PolγA-dependent exonuclease/proofreading and polymerization activities. (A) Schematic diagram depicting the principle of the exonuclease/proofreading assay. (B) Coomassie blue-stained gel showing the proteins used in the assays for exonuclease/proofreading and polymerization activities. The proteins are Flag–RECQL4 (1–1208), Flag–RECQL4 (1–1208) K508A, GST–PolγA (53–1239), GST–PolγB (27–485) and GST–p53 (1–393). (C) RECQL4 and p53 enhance the exonuclease/proofreading activity of PolγA. Exonuclease/proofreading activity was carried out on radiolabeled mutant template using 50nM PolγA without or with RECQL4 (1–1208) (1.14 µM) or p53 (1–393) (1.14 µM). Denatured products were separated by electrophoresis on an acrylamide gel. Numbers on right denote the length of the original primer (25mer) and the successive cleaved products. (D) N-terminal region of RECQL4 and its helicase activity were essential for the enhancement of the exonuclease/proofreading activity mediated by PolγA. Exonuclease/proofreading activity was carried out as in (C). The effects of either RECQL4 (1–459), RECQL4 (869–1208), RECQL4 (1–1208) or RECQL4 (1–1208) K508A (all 1.14 µM) were evaluated on the PolγA-mediated exonuclease/proofreading activity. (E) Quantitation of the exonuclease/proofreading assays carried out with PolγA alone or in combination with RECQL4 (1–1208), RECQL4 (1–1208) K508A and p53 (1–393) on either the mutant or wild-type template. All the experiments were done at least four times. Bars indicate mean ± SD. (F) Schematic diagram depicting the principle of the polymerization assay. (G) Wild-type RECQL4 enhances the polymerization activity of PolγA/B2. Time-dependent polymerization assays were carried out with 150nM PolγA/B2 without and with RECQL4 (1–1208) (650nM). Denatured products were separated by electrophoresis on an acrylamide gel. ‘nt’ indicates nucleotides present in λHindIII marker. (H) N-terminal region of RECQL4 is sufficient to enhance PolγA/B2-dependent polymerization assay. Same as in (G), except along with RECQL4 (1–1208), RECQL4 (1–459) and RECQL4 (869–1208) were also used in the assay. All RECQL4 fragments were used at the concentration of 650nM. (I) Wild-type p53 (1–393) (650nM) enhances the polymerization activity of PolγA/B2. Same as in (G), except along with RECQL4 (1–1208), p53 (1–393) (both 650nM) was also used. (J) Enhancement of the polymerization activity of PolγA/B2 by RECQL4 did not involve its helicase function. Same as in (G), except along with RECQL4 (1–1208), ATPase-dead mutant RECQL4 (1–1208) K508A was also used. Both RECQL4 derivatives were used at the concentration of 650nM.

To define the role of RECQL4 and p53 in mtDNA replication, we used two assays: (i) in vitro primer extension assay using a primer-template pair containing the first 40 nucleotides (nt) of the mtDNA replication origin sequence within the control region (20) (Supplementary Figure S1A, available at Carcinogenesis Online) and (ii) in vitro polymerization assay involving single-stranded M13mp18 DNA annealed to a 26 nt primer (30) (Figure 5F). Compared to NHF shControl, decrease in the polymerization activity was observed while using NHF shRECQL4 and NHF E6 mitochondrial extracts (Supplementary Figure S1B, available at Carcinogenesis Online), indicating that both RECQL4 and p53 individually regulated mtDNA polymerization. Mitochondrial extracts from RTS patient fibroblasts also exhibited decreased DNA polymerization activity (Supplementary Figure S1C, available at Carcinogenesis Online). To elucidate the specific role of RECQL4 and p53 during the polymerization process, purified proteins (Figure 5B) were added to the polymerization or primer extension assay. The concentration chosen for PolγA/B2 holoenzyme was such that the basal rate of polymerization was minimal. RECQL4 alone did not show any polymerization activity (Figure 5G; Supplementary Figure S1D, available at Carcinogenesis Online). DNA polymerization activity of both types of holoenzymes (i.e. PolγA/B2 purified either from baculovirus or Escherichia coli) was equivalent (Supplementary Figure S1D, available at Carcinogenesis Online). PolγA/B2-dependent polymerization was stimulated by wild-type RECQL4 in a dose- and time-dependent manner on both the shorter 40 nt partial duplex (Supplementary Figure S1E, available at Carcinogenesis Online) and also the longer M13mp18 DNA template (Figure 5G). The N-terminal 459 amino acid fragment of RECQL4, which interacted with PolγA (Figure 2D) but not with the non-interacting C-terminal domain (amino acids 860–1208), stimulated PolγA/B2-dependent polymerization (Figure 5H; Supplementary Figure S1F, available at Carcinogenesis Online).

Next, we compared the effect of RECQL4 and p53 in the polymerization assays. We observed that p53 alone was able to enhance the PolγA/B2-holoenzyme-mediated polymerization during in vitro polymerization and primer extension assays (Figure 5I; Supplementary Figure S1G, available at Carcinogenesis Online). However, the effect of RECQL4 on the polymerization reaction was greater than p53 (Figure 5I). Interestingly, addition of both RECQL4 and p53 caused negligible additive effect on the rate of polymerization (Supplementary Figure S1G, available at Carcinogenesis Online and data not shown). The effect of RECQL4 and p53 together was always same as RECQL4 alone. This maybe due to the stronger PolγA–RECQL4 interaction (compared with p53–PolγA binding) (Figures 2G and 3F), thereby preventing p53 to further bind to the same interacting regions on PolγA.

The helicase activity of RECQL4 is important for its role during initiation of nuclear DNA replication (10) and also during the enhancement of the exonuclease/proofreading activity of PolγA (Figure 5D and E). Hence, we wanted to determine whether the helicase activity of RECQL4 was also essential for its functions during the PolγA/B2-dependent polymerization of mtDNA. We found that the helicase-dead mutant RECQL4 K508A had equivalent activity as the wild-type RECQL4 (Figure 5J; Supplementary Figure S1H, available at Carcinogenesis Online), thereby indicating that the helicase activity of RECQL4 was not essential for its effect on PolγA/B2 during latter’s polymerization function. In all the polymerization assays (Figure 5G–J), major bands are produced at intermediate range, suggesting pausing or transient arrest during the polymerization reaction. It is interesting to note that the presence of RECQL4 and p53 helps PolγA/B2 to replicate across these pause sites during the polymerization process. Moreover, it should be noted that the effects of RECQL4 on PolγA/B2-mediated proofreading and polymerization activities have been studied in vitro using recombinant enzymes, which are much in excess compared to in vivo physiological conditions.

mtDNA from RTS and LFS fibroblasts exhibit increased somatic mutations and polymorphisms

Based on the above results, we speculate that the lack of RECQL4 and/or p53 may lead to suboptimal binding of PolγA/B2 to mtDNA, which may be subsequently reflected in increased error rate during the replication process (as also observed for class I and II mutants of PolγA (30)). The diminished fidelity of PolγA/B2 may lead to accumulation of mtDNA mutations in the patient cells in which either RECQL4 or p53 are absent. Hence, we carried out complete mitochondrial genome resequencing of NHF, five RTS [AG03587, AG05139, AG05013, B1865425K (in all of which RECQL4 is mutated, p53 is wild-type) and L9552914-J (both RECQL4 and p53 are mutated)] and three LFS (LFS172 p53^−/−, LFS041 p53^−/− and LFS087 p53^−/−) fibroblasts. The sequences were annotated and compared with the revised Cambridge Reference Sequence (‘rCRS’) (GenBank: NC_012920). The rare polymorphisms (263A, 311C–315C, 750A, 1438A, 4769A, 8860A, and 15326A), which are considered part of the reference sequence, were all found in NHF and in the RTS and LFS patient fibroblasts. Table I represents the summary of the additional nucleotide changes (both somatic mutations and polymorphisms) observed in the RTS and LFS patient fibroblasts when compared with revised Cambridge Reference Sequence (rCRS). The complete list of somatic mutations and polymorphisms in the mtDNA from RTS and the LFS fibroblasts are presented in Supplementary Tables S2 and Supplementary Data, available at Carcinogenesis Online, respectively. The results have also been summarized in Table I.

Table I.

Summary of somatic mutations and polymorphisms observed in the mitochondrial genomes of RTS and LFS patient fibroblasts

Type of analysis	Number of items	Table identification
A. RTS patient fibroblasts whose mitochondrial genomes were entirely resequenced	5	Supplementary Table S2, available at Carcinogenesis Online
B. Total somatic mutations and polymorphisms detected in the mitochondrial genomes of RTS patient fibroblasts	66	Supplementary Table S3, available at Carcinogenesis Online
C. Unique somatic mutations and polymorphisms detected in the mtDNA D-loop of the RTS patient fibroblasts	24	Supplementary Table S3, available at Carcinogenesis Online
D. New somatic mutations and polymorphisms being reported for the first time after resequencing of the mitochondrial genomes in RTS patient fibroblasts	12	Supplementary Table S4, available at Carcinogenesis Online
E. Common somatic mutations in the mitochondrial genomes of multiple RTS patient fibroblasts	9	Supplementary Table S5, available at Carcinogenesis Online
F. Common polymorphisms in the mitochondrial genomes of multiple RTS patient fibroblasts	7	Supplementary Table S6, available at Carcinogenesis Online
G. LFS patient fibroblasts whose mitochondrial genomes were entirely resequenced	3	Supplementary Table S7, available at Carcinogenesis Online
H. Total somatic mutations and polymorphisms in the mitochondrial genomes of LFS patient fibroblasts	72	Supplementary Table S8, available at Carcinogenesis Online
I. Unique somatic mutations and polymorphisms in the mtDNA D-loop of the LFS patient fibroblasts	32	Supplementary Table S8, available at Carcinogenesis Online
J. New somatic mutations and polymorphisms being reported for the first time after resequencing of the mitochondrial genomes in LFS patient fibroblasts	7	Supplementary Table S9, available at Carcinogenesis Online
K. Common somatic mutations in the mitochondrial genomes of LFS patient fibroblasts	6	Supplementary Table S10, available at Carcinogenesis Online
L. Common polymorphisms in the mitochondrial genomes of multiple LFS patient fibroblasts	5	Supplementary Table S11, available at Carcinogenesis Online
M. Somatic mutations which are common in the mitochondrial genomes obtained from the fibroblasts of RTS and LFS patients	14	Supplementary Table S12, available at Carcinogenesis Online
N. Polymorphisms which are common in the mitochondrial genomes obtained from the fibroblasts of RTS and LFS patients	15	Supplementary Table S13, available at Carcinogenesis Online

Among the five RTS fibroblasts, 66 unique nt changes (both somatic mutations and polymorphisms) have been found (Table I; Supplementary Table S3, available at Carcinogenesis Online). We found that 24 out of the 66 nt changes (36.3%) are in the control region known to form D-loop (Supplementary Table S3, available at Carcinogenesis Online), and the rest 42 nt changes being in the genes encoding the oxidative phosphorylation enzymes and the t-RNAs and 12/16S RNAs. Among these, 12 out of the 66 changes (18.1%) are being reported for the first time (Table I; Supplementary Table S4, available at Carcinogenesis Online). Further analysis was carried out to determine whether the same somatic mutations and polymorphisms were present in more that one RTS patient. Indeed among the 66 unique nt changes being reported, 9 somatic mutations (Table I; Supplementary Table S5, available at Carcinogenesis Online) and 7 polymorphisms (Table I; Supplementary Table S6, available at Carcinogenesis Online) (cumulatively 16/66, i.e. 24.2%) are present in more than one RTS patient. Interestingly, while three of the nine somatic mutations in the RTS patients are being reported for the first time, the rest six are found in patients with either different types of cancers (310T>C and 16519T>C) or are known to be associated with aging phenotypes (73 A>G, 146T>C, 152T>C, 195T>C) (Supplementary Table S5, available at Carcinogenesis Online).

Interestingly, 30 out of 66 unique nt changes present in the RTS patients (45.4%) were observed in L9552914-J, the only RTS patient in which both RECQL4 and p53 were both non-functional. The high number of nucleotide changes in L9552914-J is in contrast to three (AG03587, AG0539 and B1865425K) out of the four sequenced RTS fibroblasts in which only RECQL4 is mutated. These findings indicate an independent role of p53 in the regulation of mtDNA replication. To test this hypothesis, the resequencing of the entire mitochondrial genome was carried out in three LFS patient (LFS172 p53^−/−, LFS041 p53^−/− and LF087 p53^−/−) fibroblasts. We found 72 unique somatic mutations and polymorphisms were detected in the mitochondrial genomes of three LFS patients, of which 32 changes (i.e. 44.4%) were present in the control region (Table I; Supplementary Table S8, available at Carcinogenesis Online). Importantly, 7 out of the 72 somatic mutations and polymorphisms (9.7%) in the LFS patients are being reported for the first time (Table I; Supplementary Table S9, available at Carcinogenesis Online). In addition, six somatic mutations and five polymorphisms (cumulatively 11/72, i.e. 15.2%) are found in more than one LFS patients (Table I; Supplementary Tables S10 and Supplementary Data, available at Carcinogenesis Online). Among the six somatic mutations in multiple LFS patients, two are associated with aging phenotypes (73A>G and 195T>C), whereas the rest four (310T>C, 4216T>c, 16126T>C, 16519T>C) are found in different forms of cancers (Table I; Supplementary Table S10, available at Carcinogenesis Online).

Using a combination of biochemical (Figures 1–5) and de novo mtDNA replication assays (9), we have established that both RECQL4 and p53 regulate mtDNA replication via regulating the activities of PolγA alone (for RECQL4) or both PolγA and PolγB (for p53). The lack of functional PolγA or PolγB leads to mutations in mtDNA as seen in progressive external ophthalmoplegia patients (31). Since both RECQL4 and p53 regulate PolγA, a subset of mtDNA nucleotide changes may be shared between RTS and LFS patients. Indeed, comparison of the somatic mutations and polymorphisms revealed that cumulatively 29 nt changes (14 somatic mutations and 15 polymorphisms) were common between RTS and LFS fibroblasts (Table I; Supplementary Tables S12 and Supplementary Data, available at Carcinogenesis Online). Out of the 14 common somatic mutations between RTS and LFS patient fibroblasts, one mutation is being reported for the first time (309_310insCT), one mutation is exclusively associated with aging (73A>G), three mutations associated are with both aging and cancer (146T>C, 152T>C, 195T>C), whereas nine mutations are associated exclusively with cancer development (310T>C, 4216T>C, 8697G>A, 10463T>C, 12705C>T, 16126T>C, 16189T>C, 16223C>T, 16519T>C). Altogether, these results provide evidence that deregulation of the mitochondrial functions of RECQL4 and p53 may play an important role in the development of the clinical phenotypes associated with RTS and LFS patients.

Discussion

We have shown previously that both RECQL4 and p53 localize to the mitochondria under unstressed conditions and modulate de novo mtDNA replication (9). We now reveal the mechanistic basis underlying the roles of RECQL4 and p53 in mtDNA replication. Using mtDNA ChIP, we demonstrate that both RECQL4 and p53 bind to the origins of replication present within the control region of mtDNA along with PolγA (Figure 1A–H). Using co-immunoprecipitations involving endogenous proteins, we provide conclusive evidence that RECQL4 and p53 are present in a complex with the PolγA/B2 holoenzyme in vivo (Figure 1I and J). Furthermore, we elucidated the determinants involved in PolγA–RECQL4, PolγA–p53 and PolγB–p53 interactions (Figures 2 and 3). The fact that the K_d values for the interactions were either in high nanomolar or subnanomolar range indicated that both RECQL4 and p53 were accessory proteins, which potentiate the PolγA–PolγB and PolγA–DNA interactions. Interestingly, conserved mechanisms of interactions exist between RECQL4, p53 and the subunits of the Polγ. Physical interaction studies indicated that both p53 and RECQL4 associate with the exonuclease and polymerase domains of PolγA (Figures 2F and 3E) and not with the spacer region of PolγA, which contains the accessory-interacting determinant subdomain (residues 511–570) that provides an interface for the binding of the processivity factor PolγB (30). Thus, the interaction of RECQL4 with PolγA is unlikely to affect PolγA–PolγB interaction. However, binding of RECQL4 and p53 to the exonuclease and polymerase domains of PolγA enhances the interaction between PolγA/B2–DNA, resulting in the enhancement of the proofreading and polymerization activities of the mitochondrial polymerase (Figure 5). Further, both PolγA and PolγB interact with the DNA-binding domain and the extreme C-terminal region of p53 (Figure 3D), indicating another level of conservation of the interacting modules. We also note that RNA seems to play a role in interaction, especially between p53 and PolγB (Figure 3B, right), supporting the idea that p53 might regulate the initiation of the mtDNA replication via short RNA molecules that may serve as a primer for DNA synthesis (32). Indeed PolγB is required for mtDNA replisome function (33) and has been postulated to play a role in the recognition of RNA primers during mtDNA replication (34).

It is known that both the catalytic and accessory subunits of PolγA make specific contacts with each other and also with mtDNA (30). RECQL4 induces subtle conformation changes in PolγA (Figure 4C), thereby enhancing the binding of PolγA/B2 with DNA (Figure 4A and B). RECQL4 (and maybe p53) possibly achieves this effect by maximizing contact points between the holoenzyme and DNA. The enhancement of binding of the holoenzyme with mtDNA is likely to be the reason why RECQL4 and p53 increase the proofreading and polymerization activities of PolγA/B2 (Figure 5), thus culminating in the enhanced fidelity of mtDNA replication (Supplementary Figure S2, available at Carcinogenesis Online). Hence, lack of either RECQL4 or p53 (as in RTS and LFS patients) compromises the fidelity of mtDNA replication, resulting in somatic mutations and polymorphisms throughout the mitochondrial genome (summarized in Table I).

The forgoing results suggest that RECQL4 helicase activity was required for enhancing the proofreading activity of PolγA (Figure 5D and E) consistent with the recent reports that it possesses 3′ → 5′ helicase activity (4). However, the loss of helicase activity in the RECQL4 K508A mutant did not affect the polymerization activity of PolγA/B2 (Figure 5J), thereby indicating that the unwinding activity of RECQL4 was dispensable during the polymerization step. RECQL4 has been known to participate during the initiation of nuclear DNA replication where it is involved in chromatin binding of DNA polymerase α (10,11). In contrast to its effect on mtDNA polymerization, the role of RECQL4 during initiation of nuclear DNA replication was helicase dependent (10). This difference may occur due to the different mechanisms by which RECQL4 affects the functioning of the nuclear versus mtDNA polymerases. Reconstitution of mtDNA replication in vitro indicates a role for the other known mitochondrial helicase Twinkle at the mtDNA replication fork (35). Twinkle can function without a specialized helicase loader and hence can assemble on a closed circular template and support the initiation of DNA replication (36). We have observed that RECQL4 and Twinkle interact in vitro and colocalize in vivo within the mitochondrial nucleoids (data not shown). It maybe postulated that the two helicases, RECQL4 and Twinkle, form a part of a multi-helicase complex that is required for the efficient and optimal initiation, polymerization and proofreading during mtDNA replication. An analogy for such a possibility can be drawn from nuclear DNA replication where it has been shown that RECQL4 is a component of MCM replicative helicase complex (12).

The effect of RECQL4 and p53 on the proofreading and exonuclease activities of PolγA/B2 holoenzyme led to the hypothesis that lack of these proteins in RTS and LFS patients may affect the mitochondrial genome integrity. Resequencing of mitochondrial genomes from five RTS patient fibroblasts and three LFS patient fibroblasts revealed that somatic mutations and polymorphism are widespread over the entire mitochondrial genome, including the control region (Table I; Supplementary Tables S2, S3, S7 and Supplementary Data, available at Carcinogenesis Online). In both RTS and LFS patient fibroblasts, we detected the presence of new somatic mutations and polymorphisms (Table I; Supplementary Tables S4 and Supplementary Data, available at Carcinogenesis Online). Interestingly, a significant number of somatic mutations and polymorphisms were common, not only within the RTS and LFS patients (Supplementary Tables S5, S6, S10 and Supplementary Data, available at Carcinogenesis Online) but also between the RTS and LFS patients (Supplementary Tables S12 and Supplementary Data, available at Carcinogenesis Online). The presence of somatic mtDNA mutations underscores some of the clinical features of both RTS and LFS patients. Like the mice expressing the error prone version of the catalytic subunit of mtDNA polymerase PolγA (37), RECQL4 patients accumulate specific mtDNA mutations, which are associated with premature aging phenotype. A vast majority of the mutations in RTS cells were homoplasmic in nature, consistent with high frequency of homoplasmic mutations in human tumors (38). Evolution of homoplasmic mutations may occur due to the rapid emergence of mitochondria whose mutant mtDNA confers it a selective advantage or disadvantage, thereby leading to its clonal selection. Finally, like known mitochondrial diseases, RTS also conforms to the concept of ‘phenocopy’ and ‘genocopy’. RTS patients have a spectrum of mtDNA mutations, which may account for the generation of a spectrum of clinical features in RTS patients. In fact, none of the patients had all the features used for characterizing RTS patients in the clinics (2). Some of the clinical features of RTS (predisposition to aging and formation of cataracts and evidences of mitochondrial replication) (ref. 9 and this study) overlap with those of progressive external ophthalmoplegia patients, which primarily occur due to mutations in Pol G gene. These observations argue that the clinical features associated with RTS maybe at least partially due to mitochondrial dysfunction.

Over the years, a vast amount of data has accumulated supporting a role for p53 in mitochondria (17). Transgenic mice expressing mutant p53 (39) and also mice with altered p53 alleles display increased incidence of many cancers and aging-associated phenotypes. Indeed, p53 is a key regulator of both tumor suppression and aging (40). Analysis of p53 status has shown that mtDNA mutations correlated positively with p53 mutations, thereby suggesting a role for mtDNA during tumor formation due to the lack of functional p53 (41). Interaction of p53 with different components of mtDNA replication machinery has been earlier reported (18–20). Our previous results (ref. 9 and the present study) provide a mechanistic framework regarding the contribution of the mitochondrial p53 during mtDNA replication. The presence of multiple somatic mutations in the mtDNA genome of LFS patients and their link to diverse forms of cancer (Supplementary Tables S8 and Supplementary Data, available at Carcinogenesis Online), indicates a need to reevaluate the contribution of the fidelity of mtDNA replication to the tumor-suppressive functions of p53. In both LFS and RTS patients, mtDNA mutations accumulate in the D-loop, t-RNAs, 12/16S RNAs and within the polypeptides of the oxidative phosphorylation system encoded by the mitochondrial genome, thereby potentially altering their functions (Supplementary Tables S2 and Supplementary Data, available at Carcinogenesis Online). Hence, loss of RECQL4 and p53 may lead to persistent oxidative stress resulting in a feed-forward loop, which causes more changes in mitochondrial biogenesis and further oxidative stress, leading to the activation of oncogenes, inactivation of tumor suppressors like p53 itself, deregulated mitochondrial apoptotic pathways, all of which ultimately may result in neoplastic transformation. Indeed, damaging somatic mutations in mtDNA have been proven to confer a selective advantage in oncogenesis for multiple forms of tumors (42). We propose that the mtDNA mutations seen in both RTS and LFS patients may reprogram the nuclear genome by retrograde signaling (43), thereby together may contribute to neoplastic transformation.

Mitochondria possesses components of both short-patch and long-patch base excision repair (BER) pathways (44). One of the most important enzymes in BER is hOgg1, a BER glycosylase that specifically incises 8-oxoG opposite cytosine. Mitochondrial hOgg1 protects cells from DNA-damage-induced death (45). RECQL4 is known to interact with multiple key enzymes of the nuclear BER pathway and modulate the process. Indeed, the levels of many of the BER enzymes are upregulated in RTS patient fibroblasts (46). Hence, it will be interesting to determine whether RECQL4 also modulates the functions of mitochondrial hOgg1. The p53 pathway has been demonstrated to promote efficient mitochondrial BER in colorectal cancer cells (47) and thereby enhance the accuracy of the mtDNA synthesis (48). Thus, the functions of both RECQL4 and p53 in mtDNA replication may be further complemented by their respective roles in mtDNA repair.

In this study, we have further characterized the functions of two nuclear-encoded tumor suppressors, RECQL4 and p53, in mitochondria. Resequencing of the entire mitochondrial genome of RTS and LFS patient fibroblasts has revealed the presence of common somatic mtDNA mutations, which can be correlated with some of the phenotypes associated with these autosomal recessive disorders. It is interesting to note that ataxia telangiectasia mutated gene product (ATM), mutated in autosomal recessive disorder ataxia telangiectasia (A-T), and Fanconi anemia group G protein (FANCG), mutated in heterogeneous recessive disorder Fanconi anemia, also localize to the mitochondria (49,50). Similar to RECQL4 and p53, both ATM and FANCG have defined roles in nuclear DNA metabolism. Further studies aimed at understanding the interplay among proteins with proven roles in neoplastic transformation, which shuttle between mitochondria and nucleus, will provide important insight into the relative roles of the nucleus and mitochondria during cancer progression.

Supplementary material

Supplementary Materials and methods can be found at Supplementary Data

Funding

National Institute of Immunology core funds, Department of Biotechnology (DBT), India (BT/PR3148/AGR/36/706/2011); Department of Science and Technology (DST), India (SR/SO/BB-08/2010); Indo-French Centre for the Promotion of Advanced Research (IFCPAR) (IFC/4603-A/2011/1250); Council of Scientific and Industrial Research (CSIR), India [37(1541)/12/EMR-II].

Abbreviations:

Abbreviations:
• AcGFP

  Aequorea coerulescens green fluorescent protein

• BER

  base excision repair

• EMSA

  electrophoretic mobility shift assay

• GST

  glutathione S-transferase

• LFS

  Li–Fraumeni syndrome

• mtDNA ChIP

  mitochondrial DNA chromatin immunoprecipitation

• mtDNA

  mitochondrial DNA

• mtSSB

  mitochondrial single-stranded DNA-binding proteins

• nt

  nucleotides

• RTS

  Rothmund–Thomson syndrome.

Acknowledgements

The authors would like to acknowledge Maria Falkenberg and Ronald Hay for plasmids and Whitney Yin for recombinant proteins. Shweta Tikoo for proofreading the manuscript and Vinoth Madhavan for help in protein purification. K.M. is the recipient of J.C. Bose National Fellowship from the Department of Science and Technology, New Delhi, India.

Conflict of Interest Statement: None declared.