|Home | About | Journals | Submit | Contact Us | Français|
Genomic integrity depends on DNA replication, recombination and repair, particularly in S phase. We demonstrate that a human homologue of yeast Elg1 plays an important role in S phase to preserve genomic stability. The level of ELG1 is induced during recovery from a variety of DNA damage. In response to DNA damage, ELG1 forms distinct foci at stalled DNA replication forks that are different from DNA double strand break foci. Targeted gene knockdown of ELG1 resulted in spontaneous foci formation of γ-H2AX, 53BP1 and phosphorylated-ATM that mark chromosomal breaks. Abnormal chromosomes including fusions, inversions and hypersensitivity to DNA damaging agents were also observed in cells expressing low level of ELG1 by targeted gene knockdown. Knockdown of ELG1 by siRNA reduced homologous recombination frequency in the I-SceI induced double strand break-dependent assay. In contrast, spontaneous homologous recombination frequency and sister chromatin exchange rate were upregulated when ELG1 was silenced by shRNA. Taken together, we propose that ELG1 would be a new member of proteins involved in maintenance of genomic integrity.
Genomic DNA is constantly challenged by environmental as well as intracellular stressors such as radiation, chemicals and reactive oxygen species. Multiple DNA repair mechanisms protect the genome from DNA damage caused by such stressors.1 Mutations in genes participating in DNA repair have been identified in many cancer-prone syndromes. These mutations cause cells to become sensitive to DNA damaging agents and increase incidences of gross chromosomal rearrangements such as translocations, deletions, inversions, amplifications, chromosome end-to-end fusions and aneuploidy.2–4
Recently, the ELG1 gene was identified as a suppressor of genomic instability and a protein for telomere maintenance in yeast.5–8 Yeast Elg1p was suggested to form an alternative replication factor C complex and function in stabilization of the DNA replication fork.9 Eukaryotic replication factor C (RFC) is composed of one large RFC1 subunit (95–140 kD) and four small subunits (RFC2-5; 36–40 kD). RFC loads proliferating cell nuclear antigen (PCNA) onto DNA during DNA replication.9,10 There are three alternative RFC like complexs (RLC) that have been identified: RAD17-RLC for the loading of the 9-1-1 complex composed of Rad9, Rad1 and Hus1 for checkpoint activation (Rad24-RLC in S. cerevisiae),11 and CTF18-RLC which participates in sister chromatin cohesin.12 Recently, yeast studies added a new RLC called Elg1-RLC.5–8 Yeast Elg1-RLC was suggested to stabilize or restart the DNA replication fork and participates in lagging strand DNA synthesis. Recently, a putative mouse homologue of yeast Elg1p called Frag1 was reported.13 Mouse Frag1 was suggested to function in the RAD9-dependent DNA damage checkpoint and in apoptosis. However, the actual function of ELG1 in DNA damage repair, especially in human cells is completely unknown.
In the present study, we demonstrate that human ELG1 protein is stabilized and localized to form foci at sites of stalled DNA replications that are different from DNA double strand breaks (DSBs) foci. Knockdown of ELG1 expression caused sensitivity to DNA damaging agents and severe genomic instability. Collectively, these results support a model that ELG1 functions to faithfully complete the repair of DNA damage.
We previously demonstrated that a Saccharomyces cerevisiae Elg1p is important for the suppression of gross chromosomal rearrangement and maintenance of telomere homeostasis.5,14 We performed a database search and identified FLJ12735 (named ELG1, also known as ATAD5) (Accession no: NM_024857), located on chromosome 17q11.2, as a putative human homologue of yeast Elg1p similar to previously reported.8 To determine whether ELG1 responds to DNA damage, the level of ELG1 protein was investigated in response to DNA damage. The ELG1 protein level increased within two hours post 10 Gy γ-ray irradiation and continued to increase in a time dependent manner (Fig. 1A). In addition, when HEK 293T cells were treated either with 0.01% of methylmethane sulfonate (MMS), an alkylating agent for one hour, 4 mM hydroxyurea, which depletes the dNTP pool, for three hours, or 0.4 nM aphidicolin, an inhibitor of DNA polymerases for two hours, washed and allowed to recover in fresh media, the level of ELG1 protein increased two- to three-fold two hours post-treatment and further increased in a time dependent manner (Fig. 1A). Therefore, the level of ELG1 increases following treatment of DNA damaging agents.
The increase in ELG1 protein level could result from either an increase in transcription or suppression of proteolysis of ELG1. The ELG1 mRNA level did not change significantly during the recovery after treatment of ionizing radiation (IR) (Fig. 1B). DNA damage could make cells accumulate at specific cell cycle phases. ELG1 protein level increase by DNA damage could be due to differential mRNA expression during cell cycle. However, we did not observe any significant change of mRNA expression level during the cell cycle (Fig. 1C). In contrast, when cells were treated with a proteosome inhibitor, MG132, the basal level of endogenous ELG1 protein increased and there was a further increase upon treatment with DNA damaging agents (Fig. 1D). Consistently, the ectopically expressed FLAG-ELG1 was also stabilized after treatment with 0.01% MMS for one hour or 30 J/m2 UV (Fig. 1E). Therefore, the induction of ELG1 protein in response to DNA damage was due at least in part to the inhibition of proteolysis.
Many DNA damage response proteins are recruited to sites of DNA damage and form discrete nuclear foci.15–18 The order and timing of the recruitment of these proteins are critical for DNA repair.19,20 To investigate whether ELG1 forms damage response foci, ELG1 protein was monitored after treatment with DNA damaging agents in RPE or U2OS cell lines stably expressing CFP-ELG1 or GFP-ELG1, respectively. Cells were treated either with 15 J/m2 UV irradiation, 0.4 nM aphidicolin for one hour, 0.01% MMS for one hour, or 1 mM hydroxyurea overnight. ELG1 nuclear foci formation was observed in every case (Fig. 2A). In contrast to γ-H2AX and 53BP1, which form foci rapidly in less than an hour responding to DNA damage, the ELG1 foci were detected later than six hours post-treatment. The number of ELG1 foci increased in a time dependent manner, reaching a maximum at twelve hours and persisting for up to twenty-four hours post-treatment (Fig. 2B). We conclude that ELG1 functions at the late stage of DNA repair in response to a broad range of DNA damaging agents. Interestingly, when we stained for the double strand break (DSB) marker γ-H2AX after MMS treatment, there was no significant colocalization between γ-H2AX and ELG1 foci suggesting ELG1 foci do not mark DSBs (Fig. 2B).
Because ELG1 knockdown cells persisted at S phase (data not shown), we hypothesized that the observed increase of ELG1 protein level and foci formation would occur specifically during DNA replication. To test this hypothesis, cells after MMS treatment were stained with an antibody detecting cyclin A that is expressed at S and M phases of cell cycle. ELG1 foci upon MMS treatment were only observed in cells stained with an antibody detecting cyclin A (Fig. 2C). Because a majority of cells expressing cyclin A in cultures are at S phase, ELG1 foci formation after DNA damage appears to be specific at S phase. S phase specific foci formation by ELG1 suggests that ELG1 could accumulate at stalled replication forks. To investigate whether ELG1 accumulates at stalled replication forks, we looked for the presence of ELG1 at the incorporation sites of pulse labeled BrdU following treatment of DNA damaging agents. Because the pulse treatment of BrdU labels the stalled replication forks, the co-localization of ELG1 with BrdU following 0.01% MMS treatment suggesting ELG1 accumulates at the stalled replication forks (Fig. 2D).
The assembly of ELG1 into DNA damage-induced nuclear foci suggested that ELG1 might function in genomic stability and DNA repair. Therefore, we next examined HCT116 cell lines expressing low levels of ELG1 by shRNA for genomic instability. We confirmed the reduction of ELG1 mRNA or protein levels by quantitative RT-PCR and western blot hybridization, respectively (Fig. 3A and B). The reduction of ELG1 expression increased the level of p53 protein and the phosphorylation of several DNA damage response proteins including ATM, H2AX and SMC1, suggesting that inactivation of ELG1 caused spontaneous DNA damage (Fig. 3B).
The yeast elg1 mutation increased the sensitivity of cells to DNA damaging agents.5,8,14 Cells with reduced expression of ELG1 were treated with different doses of MMS and their sensitivity was compared to cells expressing an empty shRNA control vector. The reduction of ELG1 expression caused a higher sensitivity of HCT116 to MMS (Fig. 3C). Consistently, HeLa and HEK 293T cells expressing low level of ELG1 by shRNA showed sensitivity to the DNA crosslinking agent mitomycin C (data not shown).
Proteins responding to spontaneous DNA damage also accumulate at DNA damage, especially DNA double stranded breaks sites and form foci.15–18 We hypothesized that ELG1 knockdown would trigger foci formation of DNA damage response proteins, γ-H2AX and phosphorylated-ATM foci formation. There was high level of induction in spontaneous γ-H2AX and phosphorylated-ATM foci formation in HCT116 cells expressing low level of ELG1 by targeted knock-down (Fig. 3D). Similarly, silencing ELG1 expression in RPE cells by siRNA induced another DSB marker, 53BP1 foci formation (Fig. 3E).
To investigate whether reduced expression of human ELG1 increases gross chromosomal rearrangement, metaphase spreads from both ELG1 knockdown cell lines (derived from HCT116 and RPE) and control cells were examined for chromosomal aberrations with DAPI staining or spectral karyotyping analysis. Chromosomes from the knockdown cells showed a higher incidence of end-to-end fusions, inversions and aneuploidy compared to wild-type (Fig. 3F and G). Approximately 10% of cells examined had defects in chromosomes compared to no defects detected in control cells.
Defects in DNA repair enhance chromosomal abnormalities when cells are treated with DNA damaging agents. Compared to control cells, the two ELG1 knockdown cell lines increased DNA DSBs when treated with 0.01% MMS or 4 mM hydroxyurea (Fig. 4A and B and data not shown). In addition, siRNA knockdown in HEK 293T cells similarly caused high level of chromosome breaks in response to 0.01% MMS treatment (Fig. 4C and D). Collectively, these results indicate ELG1 has a clear and important role in suppressing genomic instability.
The elg1 mutation in yeast reduced the frequencies of MMS- or phleomycin-induced HR.21 To know whether the ELG1 functions in HR, we measured the HR frequencies when ELG1 was silenced by siRNA in the I-SceI DSB induced HR assay (Fig. 5A). ELG1 knockdown caused a reduction in the HR frequency. In contrast, BLM knockdown increased I-SceI induced HR frequency as suggested role of BLM helicase in HR.22 Therefore, similar to yeast Elg1p, human ELG1 functions in DSB induced HR.
In contrast to DSB induced HR, the yeast elg1 mutation increased the frequency of spontaneous HR21 which is mechanistically distinct from DSB induced HR. To investigate whether the inactivation of ELG1 similarly affects spontaneous HR, we measured the spontaneous intrachromosomal recombination between two direct repeats.23 The reduced expression of ELG1 enhanced the spontaneous intrachromosomal recombination frequency (Fig. 5B). Similarly, ELG1 knockdown enhanced sister chromatid exchange rate (Fig. 5C). Taken together, ELG1 has similar role in HR repair to yeast Elg1p.
Our results demonstrate that ELG1 is a new mammalian DNA damage response protein involved in DNA repair following exposure to various genotoxic stresses during DNA replication. Higher levels of spontaneous DNA damage (Fig. 3B and D–G), higher sensitivity to DNA damaging agents (Fig. 3C) and higher persistent DSBs even twenty-four hours following exposure to DNA damaging agent (Fig. 4) when the expression of ELG1 was silenced strongly argue that ELG1 plays a critical role in repairing DNA damage. Protein stabilization and foci formation by DNA damage in S phase (Figs. 1 and and2)2) suggest that ELG1 functions in DNA repair during DNA replication. Because ELG1 accumulates at DNA replication stall sites (Fig. 2D), we propose that ELG1 functions at stalled sites of DNA replication. Because ELG1 forms an RLC complex,8 the ELG1 RLC might load or unload PCNA or ubiquitinated PCNA from chromatin to facilitate DNA repair. However, recent in vitro biochemical analysis of yeast Elg1p failed to detect PCNA loading or unloading activity.24
The absence of ELG1 leads to the accumulation of spontaneous DNA damage (Fig. 3). ELG1 therefore have a physiological role in repairing spontaneous DNA damage despite the fact that the level of ELG1 without DNA damage is relatively low. Constant oxidative stress and DNA sequences that could stall DNA replication such as repetitive sequences could be sources for spontaneous DNA damage requiring ELG1 for repair. In addition, DNA damage from abnormal Okazaki fragment maturation in the lagging strand synthesis could be another source for spontaneous DNA damage requiring ELG1 for repair because yeast Elg1p interacts with Rad27p, a yeast homologue of human FEN1 flap endonuclease which specifically removes RNA primers during lagging strand synthesis.8 Consistently, ELG1 makes spontaneous as well as damage-induced S phase specific foci (Fig. 2 and data not shown).
Recently, Ishii et al. reported that the level of the mouse Elg1 (Frag1) mRNA is rapidly decreased upon chronic replication stress.13 We also observed that the chronic exposure of human cells to replication stress caused reduction in ELG1 mRNA and caused apoptosis (Fig. 6 and data not shown). In contrast, following treatment with and recovery from various DNA damaging agents, ELG1 protein levels were stabilized (Fig. 1A) and ELG1 formed nuclear foci (Fig. 2A) within six hours. Therefore, ELG1 responds to various forms of DNA damage that block DNA replication and facilitates DNA repair. However, if DNA damage is too high for cells to recover, ELG1 mRNA appears to be rapidly degraded and cell death occurs.
I-SceI DSB induced HR was reduced by siRNA knockdown of ELG1 (Fig. 5A). Similarly, the elg1 mutation in yeast generated reduced HR upon MMS or phleomycin treatment.21 In stark contrast to DSB-induced HR, increases in spontaneous HR and sister chromatid exchange were observed in ELG1 knockdown cells (Fig. 5B and C). These results suggest that the lack of ELG1 enhances spontaneous recombinational repair. The higher HR frequency in the ELG1 knockdown cells indicates that low expression of ELG1 could increase more replication associated DSBs that initiate HR. Consistently, there was high increase of γ-H2AX and 53BP1 foci in the ELG1 knockdown cells (Fig. 3D and E). Similarly, the yeast elg1 mutant exhibited a thirty-fold increase in the level of long terminal repeat recombination, a five-fold increase in ectopic recombination between Ty elements and a five-fold elevation of direct repeat recombination.7 These phenotypes of the yeast elg1 mutant appeared to be caused by the increase of DNA damage during DNA replication and may indicate the same in human cells. ELG1 has a clear, but complex role in HR, which needs to be investigated further due to its complex role in multiple pathways affecting HR such as DNA replication, chromatin cohesion and DNA damage sensing.6–9,21,25–27
Collectively, our results illustrate a critical role for human ELG1 in the maintenance of genomic stability and suggest that ELG1 may be an important tumor suppressor. To date, no mutation in ELG1 has been directly linked to any cancer-prone syndrome. Intriguingly, high incidence of cancers in approximately 5–10% of Neurofibromatosis type I patients who have a micro-deletion of the NF1 gene together with several other genes, including ELG1,28 suggests a putative role of ELG1 in the suppression of carcino-genesis. Therefore, ELG1 could be a candidate as a new tumor suppressor gene.
The human colon cancer cell line HCT116 was grown in McCoys, immortalized normal retinal pigment epithelial cell line RPE in DMEM-F12, and human embryonic kidney HEK 293T cells were grown in DMEM media (Invitrogen) with 10% (v/v) fetal bovine serum and 1% (v/v) Pen-strep (Invitrogen). Transfections of plasmids were performed with either Lipofectamine 2000 (Invitrogen) or Fugene 6 (Roche) by following the manufacturers’ protocols.
Cells were plated in LabTek chamber slides (Nunc) or 60 cm dishes one day prior to transfection. Cells were incubated for 24 hours in growth media after transfection and treated with different DNA damaging agents. To prepare cells for Immuno-fluorescence microscopy, cells were fixed either with fixation solution (50% (v/v) methanol, 50% (v/v) acetone) at −20°C for fifteen minutes or with 3.7% (v/v) formaldehyde at room temperature for 15 minutes and stained with different antibodies according to the manufacturer’s recommendation. Fluorescence conjugated anti-mouse IgG and anti-rabbit IgG antibodies (Molecular Probe) were used as a secondary antibody to detect proteins, followed by washing with PBS with 0.01% Tween 20. Cells were visualized in a Zeiss Axioskop2 epi-fluorescence microscope equipped with a Hamamatsu Orca CCD camera using IPLab software v3.9. Each experiment was repeated at least twice.
Proteins from cells were isolated and detected by western blot analysis as previously described.29 Antibodies used to detect proteins were α-phospho-Histone H2AX (Ser139) (Upstate), α-FLAG (Sigma), α-p53 (DO-1), α-p53-phospho-Ser15, α-1981-phospho-ATM (Cell Signaling) and α-HA-IgG-HRP from Santa Cruz Biotechnology, α-phospho-SMC1 (Cell Signaling), α-Rad51 (Novagen), α-FITC-BrdU (Roche), 53BP1 (Abcam).
To mark stalled DNA replication forks, cells were incubated with 50 μM BrdU (Sigma) for 45 minutes before fixation with 2% formaldehyde in PBS for 10 min. Followed with treatment with 2 M HCl at 37°C for 40 min and neutralization in 0.1 M borate buffer (pH 8.5), cells were incubated for one hour at room temperature with the FITC conjugated anti BrdU mAb (ROCHE) for BrdU and anti GFP antibody for CFP (Invitrogen). Cells were then washed with 0.05% tritonX-100 in PBS three times and mounted in ProLong Gold mounting media (invitrogen) containing DAPI. 200 nuclei were analyzed for each cell line and treatment. Photographed of cells were collected with a Zeiss LSM 510 NLO Meta system mounted on a Zeiss Axiovert 200M microscope with an oil immersion Plan-Apochromat 63x/1.4 DIC objective lens.
We silenced the expression of ELG1 with shRNAs in two different cell lines: HCT116, a colon cancer cell line, and RPE, a retinal pigment epithelial cell line immortalized by telomerase overexpression. Two plasmids expressing different shRNAs from Open Biosystems, one targeting sequences upstream of the P loop (Access no: NM_024857 (v2HS_158190)) and the other targeting the N terminal end (Access no: NM_024857 (v2HS_158187)) of ELG1 were transfected and ten stable clones from each cell line showing puromycin resistance were selected. Several stable clones were selected with 5 μg/ml puromycin (Sigma) from single colonies. An empty plasmid was transfected into HCT116 or RPA cells and puromycin-resistant clones were selected to use as a negative control. We chose two clones from each cell line, one silenced with the N-terminus targeting shRNA (MKJM49, MKJM61) and the other silenced with shRNA that targeted upstream of the P loop (MKJM56, MKJM63) from HCT116 and RPE cell lines, respectively. siRNA SMART pools from Dharmacon was used to knockdown the expression of genes. The reduction of ELG1 expression was confirmed with quantitative RT-PCR and western blot analysis with a polyclonal ELG1 antibody raised in our laboratory (N-EIE PCK KRK KDD DTS TCK TIT K-C peptide was used to produce the ELG1 antibodies). At least two clones from HCT116 and RPE cells were analyzed for all experiments.
Full-length ELG1 cDNA was constructed by ligating a partial 1.8 kb N-terminus ELG1 cDNA (Open Biosystems) to a 3.7 kb ELG1 cDNA amplified by 5′-RACE and double stranded cDNAs synthesis using one step cDNA synthesis kit (Invitrogen) with total RNAs from A459 lung adenocarcinoma cell line as a template. The 5.6 kb full-length ELG1 cDNA was cloned into p3XFLAG CMV10 expression vector and named as pKJM470.
Total RNA isolated with TRIZOL reagent (Invitrogen) was used as templates for RT-PCR analysis. One step RT-PCR kit (Invitrogen) was used with 500 ng of total RNA. The expression of β-actin was used as an internal control.
The frequency of homologous recombination was determined by using an extrach-romosomal HR assay that measures the frequency of HR between two GFP sequences in a plasmid DNA as described previously.23 The recombination frequency was calculated as RF = [(EBFP+ and GFP+) × (GFP+)]/[(EBFP+ and GFP+) × (GFP+) × (EBFP+)], where (EBFP) and (GFP) represent the number of blue and green fluorescent cells, respectively.
To determine DSB induced HR frequency, we used a U2OS cell line carrying a DR-GFP reporter in genome.30 A DSB was introduced in the reporter in genome by expressing I-SceI endonuclease and HR frequency was determined by comparing GFP positive cells upon DSB with no DSB.
Confocal images, exclusively live cell microscopy, were collected with a Zeiss LSM 510 NLO Meta system mounted on a Zeiss Axiovert 200M microscope with an oil immersion Plan-Apochromat 63x/1.4 DIC objective lens. Sequential excitations at 458 nm and 514 nm were provided by an argon gas laser. Confocal fluorescent images were collected as described above. Confocal Z sectioning was performed to obtain all foci throughout the entire nucleus. Optical sections were set at 1.0 mm with a Z-interval of 0.48 mm for Z-sectioning of the nuclei. Excitation at 458 nm was used to excite the CFP positive foci as well as simultaneous excitation of a DIC image using a filter BP480-520 and a light transmitted channel, respectively. Each image stack was post-processed using Bitplane’s Imaris software package v5.7.2. The image was first cropped so that only the nucleus of interest was left in the processing field. Then a volume rendering (VR) was performed to best fit the original raw data in x, y and z. Once the VR was completed a surface rendering (SR) was constructed using the best fit of the VR again in x, y and z. Each SR focus was then split into individual foci with no less than 100 triangles making up the individual focus. Counts were then averaged and sorted in pivot tables generated in MS Excel.
SCE was carried out as previously described.31 Briefly, cells were cultured in the presence of 3 μg/ml BrdU for forty-eight hours and treated with 0.05 μg of colcemid for three hours. Metaphase spreads on slides were prepared by standard hypotonic treatment and fixation with methanol and glacial acetic acid and incubated at 45°C for eighteen hours before staining with 2 mg/ml Hoechst 33258 in 2x SSC solution. The slides, mounted with Sorensen’s phosphate buffer pH 7, were then treated by UV light for thirty minutes and subsequently incubated at 65°C for one hour in 2x SSC. Chromosomes on slides were stained with 4% Giemsa in Sorenson’s buffer pH 6.8 for thirty minutes and monitored under microscope. Images of twenty metaphase spreads for each cell line were captured with two independent times.
Cells were treated with indicated doses of DNA damaging agents and plated onto 100 mm plates at defined cell densities. Cells were cultured for two weeks and cells surviving to form colonies were counted.
We thank D. Bodine (NHGRI), A. D’Andrea (DFCI), P. Liu (NHGRI), M. Lichten (NCI), P. Meltzer (NCI) and Y. Shiloh (Tel Aviv U., Israel) for helpful discussions; J. Taylor (NIEHS) for plasmids; T. Wolfsberg (NHGRI) for helping sequence analysis, S. Anderson (NHGRI) and M. Kirby (NHGRI) for FACS analysis; and D. Bodine, S. Burgess (NHGRI), A. D’Andrea, D. Lim (KAIST, Korea), P. Liu, A. Nussenzweig (NCI), Y. Shiloh, and members in Myung laboratory for comments on the manuscript; J. Fekecs (NHGRI) for figure preparation. K.M. especially thanks E. Cho. This research was supported by an NIH grant (RO154688 to M.J.), and the intramural research program of the NHGRI, NIH (HG012003-06 to K.M.).