1. Introduction
The genetic information encoded in the genome is constantly under endogenous and exogenous threat [
1]. During evolution, cells evolved efficient DNA repair processes to rapidly correct these predominant forms of DNA damage. One of the most common types of DNA injuries is nucleobase damage, comprising oxidation, deamination and alkylation. Many base lesions are pro-mutagenic (i.e., 7,8-dihydro-8-oxoguanine, 8-oxoG), as their chemical structures favour non-canonical base-pairing, thus giving rise to transition or transversion mutations [
2]. Likewise, cytosine and 5-methylcytosine (5mC) modifications—either by Ten-Eleven Translocation (TET)-mediated catalytic oxidation or by hydrolytic deamination of the bases during the cytosine demethylation process [
3]—give rise to a plethora of cytidine and uridine analogues, whose effects on the mutational status and their physiological consequences could be potentially devastating for cell survival. These small nucleobase modifications are mainly detected and repaired by the base excision repair (BER) system [
4,
5]. BER is initiated by one of eleven DNA glycosylases, which efficiently function on one hand as sensors of the type of modification and on the other hand as executors of the non-distorting damaged DNA base by performing the glycosyl bond cleavage step [
6]. The type of damaged nucleobase and the DNA glycosylase employed dictate the chemical structure of the single strand break (SSB) intermediate created during the fixation process. Upon incision of the sugar-phosphate backbone by the AP endonuclease 1 (APE1), the XRCC1 scaffold protein serves as a platform for the coordinated recruitment of polymerase β (POLβ), which fills the one-nucleotide or larger gap, by either Ligase IIIα (LIG3), the enzyme of choice to carry the ligation step during short-patch BER, or Ligase 1 (LIG1) during the long-patch subpathway [
4,
6,
7]. Owing to the fact that BER intermediates are passed on from one enzyme to the next, enabling the so-called “handover model”, which has been compared to “passing the baton” in a relay race, BER is considered a highly coordinated and efficient pathway of repair base damage [
8]. Despite the “handover” model, under some conditions of “unscheduled” SSBs or stochastic core BER components disassembly during intermediate processes, PARP1 and PARP2 are recruited to these incomplete intermediates to enhance the recruitment of the XRCC1 preligation complex in order to accelerate the repair reaction [
9]. XRCC1 is a scaffold protein comprised of an N-terminal domain that mainly interacts with POLβ, a central BRCT domain that binds DNA and PARP1, and a C-terminal BRCT domain that interacts with LIG3 [
10,
11]. Upon cell exposure to the broad range cell cycle-independent alkylating agent MMS, PARP1 is activated and required to recruit XRCC1 at SSBs to promote its accumulation [
12].
In addition to its well-established role during SSB repair and BER, XRCC1 also carries out other functions associated to the maintenance of the replisome stability [
13,
14]. At the replication fork, XRCC1 has been found to physically interact with several bona fide replication factors (i.e., PCNA, DNA polymerase α-primase, FEN1) [
15,
16,
17]. During ligation of Okazaki fragments processing (OFP), PARP1 promotes recruitment of XRCC1 to accelerate nick ligation, whereas the use of PARP inhibitors (PARPi) disrupts OFP ligation, leaving unrepaired SSBs, which are further converted into double strand breaks (DSBs) during DNA replication [
17,
18]. In agreement with this notion, mutations in the XRCC1 gene mainly associate with neurological disorders occurring by excessive accumulation of unrepaired SSB in postreplicative tissues [
5,
10], and cancer predisposition, presumably as a consequence of increased replication-associated DSBs formation [
13,
14]. XRCC1 has also been reported to play crucial roles during microhomology-mediated end joining (MMEJ) and replication fork restart in the absence of the fork protector BRCA2, by promoting nascent DNA degradation mediated by MRE11 and mutagenic repair [
14]. Since the discovery that PARPi sensitize homologous recombination (HR)-deficient tumours, the primary sensitizing lesion has been attributed to the formation of toxic DSBs during replication fork collapse [
19,
20] and, more recently, to lagging strand-associated single stranded gaps [
21]. Olaparib traps PARP1 on DNA, interferes with DNA replication and promotes fork collapse, necessitating a proficient HR mechanism for full restoration of an intact fork structure [
22,
23,
24].
Here, we describe a protective role for the BER factor XRCC1 during misincorporation of cytidine analogues. Both 5hmC and 5-formyl-2′-deoxycytosine (5fC), but not 5mC or 5-carboxy-2′-deoxycytosine (5caC), activated the BER pathway during their misincorporation. Moreover, Xrcc1 deficient cells displayed marked sensitivity upon 5hmC or 5hmU exposure, a 5hmC deamination product. 5hmC or 5hmU exposure led to increased chromatin retention of PARP1 and replication fork instability, that were exacerbated by treatment with the PARP1-trapping drug olaparib. Remarkably, replication fork instability by 5hmC occurred during the ongoing S-phase, thus suggesting a novel role of XRCC1 at coordinating BER during replication fork dynamics. Collectively, these data suggest that persistent PARP1 at BER intermediate structures during removal of 5hmC-derived lesions are a novel source of replication fork insults that challenge genome integrity.
3. Discussion
This work describes a cytotoxic role for 5hmC and its deamination product 5hmU upon misincorporation during replication. Whereas epigenetically modified DNA bases present in gene promoters and bodies are essential factors required for the implementation of transcriptional programmes controlling cell fate decisions and differentiation [
2,
28], their harmful potential to alter gene transcription upon misincorporation in the genome is not well understood. Here, we identify three epigenetically modified nucleotides as novel sources of genomic instability associated to replication fork defects and widen the scenario in which BER actively removes cytosine analogues during misincorporation by the replicative polymerases. Our results indicate that in
wild type cells, 5hmC and its deamination product 5hmU, and 5fC to a lesser extent, activates a DNA damage response featured by increased PARylation and γ-H2AX signals, consistent with a BER mechanism activation at replication forks. Consistent with this notion, the loss of XRCC1, an essential scaffold protein functioning at MMEJ [
14] at the replication fork [
13] in addition to SSB repair and BER [
7,
26], shows a marked sensitivity to 5hmC, 5hmU and to 5fC to a lesser extent. Our data shows that the mechanism behind the aforementioned cytotoxicity is related to increased PARP1 chromatin retention at 5hmC/5hmU removal intermediates associated to replication fork collisions. In agreement with our data, a recently published work shows that contamination of the nucleotide pool by 5hmU-derived from 5hmC leads to 5hmU misincorporation in the genome, requiring some components of the BER pathway (SMUG1, PARP1, PARG) to promote cell survival [
27]. The mechanism for this cytotoxicity seems to be related to olaparib trapped-PARP1 at 5hmU-processing sites persisting long enough as to be head-on collided by the replisome during the following cell cycle. However, our work uncovers that 5hmU processing at replication forks directly impact in fork dynamics, and diverges from the model proposed by Fugger et al. [
27]. Our model would imply that BER processing of 5hmC or 5hmU substrates occurred straight after misincorporation, probably in the context of reversed replication forks. In such scenario, 5hmU in the nascent DNA would be processed by BER, which, in the absence of XRCC1, would lead to persistent PARP1 retention and formation of ssDNA gaps at nascent DNA, necessitating HR for replication fork restart. Several reports also describe that XRCC1 interacts with numerous bona-fide replication factors such as MRE11, PCNA, α-primase, the DNA glycosylase UNG2 or the alternative OFP pathway at replication forks [
11]. Moreover, it has also been proposed that the end protection factor DNA-PKcs/KU70/KU80 regulates XRCC1 function at a subset of difficult-to-resect reversed replication forks to promote their restart [
13]. We could envisage a model whereby, upon fork reversal of 5hmU-containing nascent DNA, BER would initiate at the chicken foot in nascent DNA where XRCC1 could play dual functions by coordinating PARP1 removal and preligation steps to fork resumption, regulated by PARP1 and DNA-PKcs throughout its interaction with PCNA, thus ensuring SSB fixation. Whereas PARP1 trapping seems to be the major determinant associated to fork instability during olaparib treatment, other researchers have proposed SSBs arising from BER intermediates rather than trapped PARP1 as the main cause of lethality in HR deficient cells [
21]. However, it has also been reported that the marked sensitivity of
XRCC1−/− cells to the cell cycle independent alkylating agent MMS is almost completely suppressed by
PARP1/2 deletion [
25], suggesting that trapped PARP1 rather than SSBs arising from BER intermediate lesions during damaged base removal accounted for the loss of viability, implying SSBs fixation perhaps throughout postreplicative repair mechanisms. Whether these differences are due to the use of particular agents such as MMS, which engages BER across all stages of the cell cycle, or 5-hydroxymethylated nucleosides which mainly induce BER activation during replication, needs further research.
In humans, biallelic mutations in the
XRCC1 gene are associated with neurodegenerative disorders, whose symptoms encompass ocular motor apraxia, axonal neuropathy, and progressive cerebellar ataxia [
5,
10].
XRCC1-deficient patient cells show elevated PARylation and reduced efficiency of single-strand break repair. Moreover, the loss of cerebellar neurons and the ataxic phenotypes observed in
Xrcc1−/− mice were completely rescued by deletion of
Parp1, suggesting that excessive PARP1-DNA complexes trapped at sites of endogenous DNA damage associated to BER intermediates seen in rodents may account for the neuropathological symptoms observed in these patients. Our work also uncovers that misincorporation of 5′-hydroxymethylated DNA bases at the nascent DNA is a key contributor to replication fork instability in the absence of XRCC1 and predicts that tissues containing highly enriched 5hmC-derivatives genomic DNA could be more susceptible to replication fork defects associated to collision with the BER machinery. Whether misincorporation of the epigenetically modified DNA bases 5hmC, 5fC or 5hmU during DNA repair process or during DNA synthesis in early neural progenitor cells stems from the base of these neurodegenerative disorders, paves the way for further investigations.
4. Materials and Methods
Transformed wild type and Xrcc1 deficient murine embryonic fibroblasts (MEFs) and the pRFP-HsXRCC1 plasmid were kindly provided by Dr. Keith W. Caldecott (U. Sussex, UK). U2OS cells were transfected with HsXRCC1-RFP plasmid using TurboFect Transfection Reagent (Thermo Scientific, Waltham, MA, USA) according to the manufacturer’s protocol, and selected on G418-supplemented (1 mg/mL) growth medium. MEFs and U2OS were grown in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Waltham, MA, USA) supplemented with 10% FBS and penicillin/streptomycin (1%, Gibco 10,000 U/mL). Cells were grown at 37 °C in a 5% CO2 incubator.
2′-deoxycytidine (dC) (D3897; Sigma-Aldrich, St. Louis, MI, USA) and its derivatives 5mC (PY 7635; B&A, Dexter, MI, USA), 5hmC (PY 7588; B&A, Dexter, MI, USA), 5fC (PY 7589; B&A, Dexter, St. Louis, MI, USA), 5caC (PY 7593; B&A, Dexter, St. Louis, MI, USA), 5-chloro-2′-deoxyuridine (CldU) (Sigma-Aldrich, St. Louis, MI, USA, C6891) and 5-iodo-2′-deoxyuridine (IdU) (Sigma-Aldrich, St. Louis, MI, USA, I7125) were dissolved in PBS (10 mM stock solution) and added to cell cultures as indicated. Olaparib and 5-ethynyl-2′-deoxyuridine (EdU) were purchased from Selleckchem (Houston, TX, USA, S1060) and Baseclick (Neuried, Germany, BCN-001-100), respectively, and dissolved in DMSO (10mM stock solution).
The antibodies used were against phospho ser139-H2AX (Millipore, Waltham, MA, USA, 05-636), PAR (Millipore, Waltham, MA, USA, MABE1016), BrdU (AbDSerotec, San Diego, CA, USA, OBT0030), BrdU (BD Bioscience, Newark, NJ, USA, 347580), Histone H3 (Invitrogen, Waltham, MA, USA, PA5-16183), PARP1 (Cell Signalling, Waltham, MA, USA, 46D11), alpha-tubulin (GeneTex, San Diego, CA, USA, GTX628802).
4.1. Cytotoxicity Assays
A quantity of 2.5 × 103 cells were seeded in a 96-well dish and incubated at 37 °C for 3 days in the presence of the indicated 5′-modified 2′-deoxynucleoside or olaparib at the indicated doses. Cell Proliferation Kit I (MTT) (Roche, Basel, Switzerland) assay was performed according to the manufacturer’s instructions.
4.2. Immunofluorescence Microscopy
For immunofluorescence analysis, 5 × 104 cells were seeded in coverslips (VWR, Monroeville, PA, USA) and incubated with the chemicals for the indicated times. Cells were permeabilized with 0.25% Triton X-100 in PBS at 4 °C for 2 min and fixed with 4% formaldehyde in PBS at 4 °C for 15 min. Then, cells were blocked with 0.3% Tween20 + 3%BSA in PBS and incubated with the primary antibody of interest overnight at 4 °C. Next day, cells were incubated with alexa fluor-labelled secondary antibody and 4′6-diamidine-2-phenylindole dihydrochloride (DAPI) was added at 0.1 mg/mL for 1 h at RT in dark conditions. Finally, coverslips were mounted in ProLong® Gold Antifade Reagent (Invitrogen, MA, USA). Immunofluorescence images were taken using ×40 magnification lens and quantified using the ImageJ software. Nuclear DAPI staining was used to outline cell nuclei, and a mask corresponding to nuclear DAPI area was used over the red or green channels to quantify PAR or γ-H2AX nuclear staining. Background levels were substracted from each image. For EdU immunofluorescence, cells were incubated with 10 µM EdU for 2 h. Cells were permeabilized, fixed and blocked as previously described, and click-it reaction (100 mM Tris-HCl 1.5M pH 7.5, 1 µM 488 fluorescence azide, 1 mM CuSO4, 100 mM ascorbic acid) was carried out for 30 min at RT in dark conditions.
4.3. Chromosome Aberrations Assay
To assess for chromosome aberrations, cells were grown on media containing 5hmC or 5hmU during two cell cycles. KaryoMAX colcemid solution (Gibco, MA, USA) was added at a final concentration of 0.1 μg/mL 1.5 h before collecting the cells. Cells were collected and incubated in a hypotonic solution (75 mM KCl) for 50 min at 37 °C. Fixative solution (3:1 methanol/acetic acid) was added dropwise, and cells were washed three times with this fixative solution. After washing, cells were resuspended in 500 μL of fixative solution and spread on cleaned slides dropwise. The slides were stained with Giemsa 1:10 (Sigma-Aldrich, St. Louis, MI, USA) for 30 min. To score chromosome aberrations, pictures of metaphases were scored with a DFC390 camera attached to a Leica DM6000 optical microscope using ×100 magnification lens and analyzed using LAS AF Leica software. The pictures were scored blinded for chromosome aberrations.
4.4. DNA Fiber Technique
For replication fork blockage purpose, 7.5 × 104 cells were seeded in a 6-well dish. Cells were pulsed with 25 μM CldU for 30 min. Then, the growth medium was removed and replaced by fresh medium containing 50 μM IdU in the presence of the indicated concentration of 5hmC or 5hmU for 30 min. Cells were placed on ice, washed with cold PBS and harvested by scrapper. The cells were centrifuged at 1200 rpm 5 min 4 °C, the cell pellet resuspended in cold PBS and kept on ice. A 2 μL-drop of cells was placed on the slides, incubated for 6 min until the drop dried out. Then, 7 μL of spreading buffer was added on, mixed and incubated for 30 min at RT. DNA fibers were spread, fixed for 10 min in 3:1 methanol/acetic acid and stored at 4 °C. DNA fibers were denatured with 2.5 M HCl for 1 h, washed with PBS + 1%BSA + 0.1%Tween20, and incubated with rat-anti-BrdU antibody (1:500) for 1 h to detect CldU. The slides were washed and stained with alexa-488 anti-rat antibody for 1.5 h. Then, the slides were washed, fixed with 4% paraformaldehyde for 10 min, and stained with a mouse anti-BrdU antibody (1:500). Then, the slides were rinsed and stained with Cy3 anti-mouse antibody for 2 h. Finally, the slides were mounted in Fluoromount-G (eBioscience, San Diego, CA, USA). Pictures of fibers were scored using ×100 magnification lens.
4.5. Chromatin Retention Assay and Immunoblotting
PARP1 chromatin retention assay was performed essentially as described by Demin and colleagues [
25].
4.6. Statistical Analysis
The number of independent technical repeats (n) are indicated in figure legends. Data are shown as mean ± standard error of the mean (s.e.m.) or median. The nonparametric Mann–Whitney test or Student’s t-test were employed to determine statistical significance (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). Data analyses were performed in GraphPad Prism.