Author contributions: M.K.: collection and assembly of data, data analysis and interpretation, and manuscript editing; L.B., R.H., Z.D., L.E., K.M., T.B., P.F. and J.F.: collection and assembly of data and data analysis and interpretation; A.H. and P.D.: conception and design and financial support; V.R.: conception and design, financial support, collection and assembly of data, data analysis and interpretation, and manuscript writing.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS January 12, 2013.
The inevitable accumulation of chromosomal abnormalities in human embryonic stem cells (hESCs) during in vitro expansion represents a considerable obstacle for cell replacement therapies. To determine the source of chromosomal abnormalities, we examined hESCs maintained in culture for over 55 months for defects in telomere maintenance and DNA repair. Although prolonged culture affected neither telomerase activity nor nonhomologous end joining, the efficiency of base excision repair (BER) was significantly decreased and correlated with reduced expression of apurinic/apyrimidinic endonuclease 1 (APE1), the major nuclease required for BER. Interestingly, the expression of other BER enzymes was unchanged. Addition of human recombinant APE1 protein to nuclear extracts from late passage hESCs increased BER efficiency to the level typical of early passage hESCs. The link between BER and double-strand breaks (DSB) was demonstrated by decreased DSB release after downregulation of APE1 in early passage hESCs via siRNA. Correspondingly lower APE1 level in late passage hESC resulted in slower and less intensive but long lasting DSB release upon ionizing radiation (IR). Downregulation of APE1 in early passage hESCs also led to approximately 30% decrease in γ-H2AX signaling following IR, similar to that in late passage hESCs. We suggest that downregulation of APE1 significantly contributes to the failure of BER during long-term culture of hESCs, and further that BER failure is one of the factors affecting the genomic instability of hESCs by altering BER-dependent DSB release and cell cycle/checkpoint signaling. STEM CELLS2013;31:693–702
The majority of human embryonic stem cell (hESC) lines are known to accumulate chromosomal abnormalities [1, 2] that may underlie the tendency of embryonic cells to survive and proliferate under suboptimal in vitro conditions. Supporting this idea, long-term culture of hESCs seems to generate subclones with specific genome rearrangements [1, 3, 4].
Large chromosomal rearrangements, aneuploidy and polyploidy can be generated via various mechanisms. Lack of telomerase activity is associated with such abnormalities in many tumors and immortalized cell lines . Genomic instability can be also triggered by failure of DNA repair or malfunctioning checkpoint mechanisms. During hESC differentiation, several enzymes involved in oxidative stress defense and DNA repair are downregulated . Interestingly, genomic instability in hESCs is typically observed only after a certain period of cultivation, with risk of instability increasing with passage number, even though adaptation to culture does not seem to be dependent on passage number . hESCs accumulate chromosomal abnormalities as a typical feature of culture adaptation. Several DNA repair enzymes are downregulated with differentiation of hESCs into embryonic bodies (EBs), including the BER enzymes UNG, Fen1, Lig3, and NTHL1 .
The increase in mutant frequency (MF) in germ cells with increasing parental age is caused by the failure of BER  and has been attributed to downregulation of apurinic/apyrimidinic endonuclease 1 (APE1, also known as APE, HAP1, Ref-1, and APEX) , although in other tissues, for example, old murine brain, other factors such as decreased activity of glycosylases are implicated in causing BER deficiency (our unpublished data). Downregulation of APE1 is also observed during the senescence of human mesenchymal stem cells . Therefore, we reasoned that investigation of the efficiency of BER and the abundance and activity of APE1 in hESCs at early and later passages would be very informative, because the mechanism might be similar for germ cells and ESCs.
In this study, we focused on examining several mechanisms likely to contribute to the genomic integrity of hESCs, namely telomere maintenance, nonhomologous end joining (NHEJ), and BER. We observed that BER is less efficient during prolonged cultivation/adaptation and that the BER defect is correlated with a decrease in APE1 expression. We hypothesize that the mechanism by which the APE1/BER failure contributes to overall genome instability and adaptation lies in increased MF.
MATERIALS AND METHODS
hESC Lines and Cell Culture
Experiments were performed using the hESC lines CCTL10, CCTL14, and CCTL12, which have been characterized in detail [10, 11]. Cells were divided into three groups according to the passage number: early (passages 1–50), medium (passages 50–100), and late (passages above 100). The hESCs were propagated on mitotically inactivated mouse embryonic fibroblasts (MEF) as described previously . Human foreskin fibroblasts (SCRC 1041) were obtained from the American Type Culture Collection (Manassas, VA) and used at passages 11–13.
hESC-derived fibroblasts were produced as follows. hESCs growing in colonies on MEF feeder cells were treated with collagenase IV (Invitrogen, Carlsbad, CA, http://www.invitrogen.com, 1 mg/ml in Dulbecco's modified Eagle's medium-F12) and released from the feeder layer by gentle pipetting. Whole colonies were then transferred to MEF medium supplemented with 20% Fetal Bovine Serum and plated onto nonadhesive bacteriological plates, where they were maintained for 5–15 days with a change of medium every other day. When large EBs developed, they were transferred onto tissue culture plates precoated with porcine gelatin (Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com). Cells were kept in culture without passaging until fibroblast-like cells started to emerge around the EBs. EBs were then removed. The fibroblast-like cells were then trypsinized (trypsin-EDTA, Invitrogen) and plated on a tissue culture plate without gelatin. The resulting culture was designated as passage 0, and cells were further passaged with trypsin-EDTA every 2 days after reaching confluence. Cells become morphologically homogenous and began to resemble primary fibroblasts between passage 3 and 4. Cells from passage 5–7 were used for the experiments described in this article.
We developed a database of cultured hESC lines for monitoring cell line cultivation, named DabLin. The web-based system was developed using PHP scripting language and MySQL open source database. Data from DabLin were used to measure the passaging frequency of hESC lines.
Analysis was always performed simultaneously for early, medium, and late passage cells to enable quantitative comparison of APE1 expression. Briefly, approximately 1 × 105 cells were plated on Ibidi μ-Dish plates (Ibidi, Munich, Germany, http://ibidi.com). Cells were allowed to create colonies for 48 hours. Cells were then fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) and blocked with 1% BSA (Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com) and 0.1% Triton X-100 (USB, Cleveland, OH) in PBS. Rabbit polyclonal anti-APE1/Ref1 (Trevigen, Gaithersburg, MD) or rabbit anti-Pol β, rabbit anti-Ogg1, goat anti-Oct3/4 (Santa Cruz Biotechnology, Santa Cruz, CA, http://www.scbt.com) and rabbit anti γ-H2AX antibody (BioLegend, San Diego, CA) were used at a dilution of 1:1,000 or 1:200, respectively. Anti-rabbit or anti-goat Alexa 488 and Alexa 568 conjugates (Invitrogen) were used as a secondary antibody according to manufacturer recommendations. Cell nuclei were counterstained with 4',6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich). The cells were observed either with a confocal microscope (Olympus FluoView 500) or an inverted fluorescent microscope (Olympus IX71).
Cells were extracted with CHAPS lysis solution from the TRAPezeXL Telomerase Detection Kit (Chemicon International, Temecula, CA, http://www.chemicon.com) following the manufacturer's recommendations. Total protein concentration was determined using the Bio-Rad protein assay kit (Hercules, CA, http://www.bio-rad.com).
Telomerase activity assays were performed in dual-color real-time mode using the Rotorgene 3000 (Corbett Research, Sydney, Australia) as described previously . For the determination of telomerase activity, absolute quantification was used. A standard curve was obtained using dilutions (0.02–0.4 amol) of the TSR8 control template (a substrate oligonucleotide extended by eight telomeric repeats) included in the TRAPezeXL Telomerase Detection Kit. One microliter of sample extract containing 250 ng of total protein was added to each reaction. Telomerase activity was expressed in relative telomerase units (T.U.; 100 T.U. corresponds to 0.1 amol of TSR8 control template).
Preparation of Nuclear Extracts
Nuclear extracts were prepared from hESCs, 10- to 15-day old EBs, human feeder fibroblasts, and hESC-derived fibroblast cells as described previously  with minor modifications. Briefly, complete Mini protease inhibitor cocktail (Roche, Mannheim, Germany, http://www.roche-applied-science.com) was added to the homogenization and lysis buffers according to the manufacturer's recommendations. An ice-cold loose-fitting glass homogenizer was used to homogenize hESCs as well as to lyse the nuclei. For each sample, the final ammonium sulfate-precipitated protein pellet was resuspended in 30 μl buffer B and dialyzed against 1 l of the same buffer. Extracts were aliquoted, quickly frozen in liquid N2, and kept at −80°C until use. Protein content in the extracts was quantified using Bradford reagent (Bio-Rad, Hercules, CA, http://www.bio-rad.com) with gamma globulin (Bio-Rad) as a standard. Luciferase (Sigma-Aldrich) was added to nuclear extract as internal standard. Luciferase activity was measured using a plate luminometer (Beckman Coulter DTX 880 Multimode detector) as described previously .
Nuclear extracts were tested for NHEJ efficiency using a previously described protocol  with minor modifications. Briefly, nuclear extracts were incubated for 30 minutes at 37°C with EcoRV-linearized pBluescript SK-plasmid that had been end labeled with gamma-32P ATP using polynucleotide kinase (Roche Diagnostics GmbH, Roche Applied Science, Penzberg, Germany, http://www.roche-applied-science.com). The reaction was stopped by addition of proteinase K (final concentration of 1 mg/ml) in buffer PK (0.1 M Tris-Cl, pH 8, 5 mM EDTA, 0.2% SDS, 0.2 M NaCl) and placed on ice. Samples were incubated at 65°C for 1 hour and subsequently analyzed using gel electrophoresis on a 2% agarose gel in TAE buffer (40 mM Tris-Acetate, 1 mM EDTA, pH 8,3). After electrophoresis, the gel was fixed with 4% trichloroacetic acid (Sigma-Aldrich), dried, and analyzed by autoradiography on a Storm phosphoimager (GE Healthcare, Piscataway, NJ). Bands were quantified using ImageQuant software (GE Healthcare).
Base Excision Repair Assay
Short-patch BER (spBER) assays were carried out as described previously [14, 16] with a few modifications. Three picomoles of fluorescein-labeled duplex oligonucleotide were incubated with nuclear extract for 10 minutes at 37°C. Luciferase was added to the nuclear extract for use in normalization. Buffered solution without nuclear extract was used as negative control. SpBER efficiency was expressed as incorporation of radioactively labeled dCTP into the 51-mer. The oligonucleotide signals were normalized to the untreated fluorescein-labeled oligonucleotide signal (detected using a Bio-Rad Gel Dock transilluminator). SpBER efficiency was normalized for recovery of oligonucleotide fluorescein signal and recovery of luciferase activity in nuclear extracts. The 51-mer oligonucleotide labeled with 32P was used as a size standard to determine the correct size of the BER reaction product.
Downregulation of APE1 Expression
The expression of APE1 was downregulated using siRNA against APE1. The APE1 small interfering RNA (siRNA) sequence (5′-UAUGCCGUAAGAAACUUUG 3′ and 5′ CAAAGUUUCUUACGGCAUA 3′) was obtained from prof. Jozef Dulak (Jagiellonian University, Krakow, Poland) and synthesized by Eurofins MWC Operon (Ebersberg, Germany). siRNA was reconstituted in water to final concentration of 100 μM and stored at −20°C. X-tremeGENE siRNA transfection reagent (Roche) was used for transfections according to the manufacturer's recommendations. Briefly, early passage hESCs seeded 1-day earlier to the media without antibiotics were transfected with 50 nmol APE1 siRNA or scrambled siRNA in a final volume of 200 μl [including transfection reagent, DNA and Opti-MEM I reduced serum media (Invitrogen)] in a 1.5-cm tissue culture plate. After 4 hours of treatment, 300 μl Opti-MEM was added. Cells were cultivated 2–3 more days in standard hESC medium, until APE1 expression reached a minimum (see supporting information Fig. 6A–6C).
Analysis of Ionizing Radiation-Induced DSBs
Cells were irradiated with a single dose (3 Gy) of 137Cs γ-ray (OGL-1 VF a.s. Cerna Hora, Czech Republic, 2 Gy/minute). Comets were analyzed following single cell electrophoresis according to Singh  with several modifications. Cells were harvested either immediately after ionizing radiation (IR) treatment, or each 15 minutes within next 90 minutes in the incubator, by centrifugation at 2,000 × g at 4°C. Cells were then mixed with 0.625% low melting point agarose (Invitrogen) in PBS that had been preheated to 37°C and spread on microscope slide. Slides were treated as described . Neutral electrophoresis (TAE, 20 V, 4°C, 20 minutes) was carried out after equilibration of the slides in TAE buffer for 20 minutes. DNA was visualized with propidium iodide (2 μg/ml, Sigma-Aldrich), and comets were photographed using a fluorescent microscope. Double-strand break (DSB) damage was then quantified using CometScore software (TriTek, Sumerduck, VA). The DSB burden was then expressed as Olive tail moment.
Karyotypic Abnormalities Acquired During Long-Term Culture
We systematically analyzed the karyotypes of CCTL12 and CCTL14 hESC lines in culture and observed karyotypic changes in both lines. Apparent karyotype instability, defined as a mixture of normal and aberrant mitoses, was first observed between passages 13 and 64, and stable karyotypic change, defined as overgrowth of hESC subclones, was typically observed between passages 42 and 76 (see example in supporting information Fig. 1A). Typical stable clonal chromosomal changes included extra copies of chromosomes 1, 7, 10, and 12, and translocations of chromosome 12 and 17 supporting information Fig. 1A, 1B). These changes are associated with accelerated growth rate, as demonstrated by the shortened time between two subsequent passages (supporting information Fig. 1C).
Telomerase Activity in hESCs Remains High Regardless of Time in Culture
We measured telomerase activity in hESCs in early, medium, and late passage as well as in differentiated human cells, such as human foreskin fibroblasts, hESC-derived fibroblasts, and 10- to 15-day old EBs. Mouse embryonic fibroblasts were used as a control. We observed that telomerase activity was up to 20 times higher in CCTL14 hESCs through passages 32–366 than in various differentiated cell types, including EBs and human foreskin fibroblasts as well as autologous fibroblasts derived from the CCTL14 line (supporting information Fig. 2). EBs, an example of early nonspecific differentiation, showed telomerase activity more than two times higher than that in human foreskin fibroblasts or hESC-derived fibroblasts, and similar to the activity observed in mouse embryonic fibroblasts (supporting information Fig. 2). The telomerase activity in hESCs remained high even at late passages, and the differences in the telomerase activity between nondifferentiated and differentiated cells remained similar throughout the tested range of passage numbers.
NHEJ Efficiency Shows a Stable Pattern in Normal and Culture-Adapted hESCs
We next determined the NHEJ efficiency in early, medium, and late passage hESCs and compared it with NHEJ efficiency in differentiated cells. We observed efficient NHEJ in hESCs in all samples tested, and importantly, found no significant differences in the efficiency of NHEJ in cells from passages 33 to 290 (supporting information Fig. 3). Furthermore, the NHEJ efficiency in undifferentiated hESCs was comparable with that of differentiated hESC-derived fibroblasts (supporting information Fig. 3).
The Efficiency of spBER Decreases with Prolonged Time in Culture
We then tested nuclear extracts from hESCs at early, medium, and late passages to determine the efficiency of spBER on G:U lesions, a typical lesion that occurs often in vivo and has been used in many in vitro studies . We observed a significant decrease in BER efficiency in late passage cells (>150; p < .05) compared with early (∼30) and medium passage (∼70) cells in CCTL14 line (Fig. 1A). Some individual experiments on CCTL14 medium passage cells also showed a decrease in BER compared to early passage cells; however, when average efficiency was calculated, the overall decrease was not significant. Although we do not possess such a wide range of passages for other hESC lines, we were able to demonstrate that the CCTL12 line shows a similar significant decrease (p < 0.05) in spBER at passages between 59 and 168 (Fig. 1B). The spBER activity was approximately 10-fold higher in both tested hESC lines compared with differentiated cells (Fig. 1C).
Decreasing Expression of APE1 in Culture-Adapted hESCs Correlates with Decreasing Efficiency of BER
We observed a decrease in APE1 protein level in both medium and late passage hESCs in the CCTL14 line (Fig. 2A, 2C) and also in the CCTL12 line between passages 59 and 168 (Fig. 1B). APE1 was strictly localized in the nuclei of hESCs, indicating that its localization was unaffected by prolonged passaging. Interestingly, analysis of APE1 expression in late passage hESCs revealed an occasional “spotty” pattern (Fig. 2A, 2B), indicating considerable heterogeneity in APE1 expression between colonies as well as within the same colony. APE1 decrease and/or heterogeneity is not linked to localization to cytoplasm or mitochondria, as we detected APE1 only in nuclei (supporting information Fig. 4). We then examined the levels and expression of two other enzymes involved in BER: Pol-β (Fig. 3A, 3C, 3D) and 8-oxoguanine DNA glycosylase Ogg1 (Fig. 3B–3D) and found that their protein levels remained unchanged (Fig. 3B, 3C) and their expression levels did not change either (Fig. 3D) during prolonged cultivation. To exclude the possibility that the lower amount of APE1 in late hESCs might be caused by a gradual loss of pluripotency, an Oct3/4 (Fig. 3A, 3B), NANOG, and SSEA3 (supporting information Fig. 5) embryonic cell markers were also examined. No decrease or heterogeneous expression of either marker was observed in medium or late passage hESCs.
Restoration of BER Efficiency by Addition of Recombinant APE1/Ref-1
To confirm that the decreased levels of APE1 are causally linked to the decrease in BER efficiency, we added recombinant human APE1 endonuclease to nuclear extracts from late passage hESCs (passage 290) and assayed their BER efficiency (Fig. 4). Addition of recombinant APE1 yielded a significant increase in BER activity in late passage CCTL14 hESCs, with higher levels of exogenous APE1 leading to BER efficiency similar to that of early passage hESCs (Fig. 4, passage number 34).
Slower Processing of IR-Induced Generation of DSBs in Late Passage hESCs
We induced DNA damage by treatment with IR and followed the course of DSB repair in early and late passage hESCs using the comet assay. Treatment of hESCs with IR induced significantly higher (p < .05) numbers of DSBs in early passage cells than in late passage cells immediately after IR (0 minute), whereas no difference was detected between early and late passage cells 15 minutes after IR (Fig. 5). The level of background (uninduced) damage was similar in early and late passage cells (Fig. 5). By 45 and 60 minutes after IR, the early hESCs returned the level of damage to the uninduced background level, but a significant (p < .01) amount of residual damage was still present in late passage hESCs (Fig. 5). It took 90 minutes to fully ameliorate the DSBs in late passage hESCs, suggesting a significant shift in the dynamics of DSB release after IR treatment in late passage hESCs relative to early passage hESCs.
Slower Dynamics of DSB Generation After IR of hESCs with Downregulated APE1 Expression
To determine whether there is a causal link between BER and altered dynamics of DSB after IR treatment, we examined the effect of downregulating APE1 expression on IR-induced generation of DSB in hESCs. APE1 expression was downregulated using siRNA (as demonstrated in supporting information Fig. 6A–6C), and DSBs were examined immediately, after, and 45 minutes after IR treatment, time points at which significant differences were observed between early and late passage hESC DSB release (see Fig. 5). Downregulation of APE1 expression in early passage hESCs caused a significant decrease in DSB release after IR both immediately and 45 minutes after the treatment in all three tested independent cell lines (CCTL10, CCTL12, and CCTL14, Fig. 6A–6C). The recapitulation of the behavior of late passage hESCs in early passage hESCs with downregulated APE1 clearly demonstrates the effect of BER downregulation on metabolism of DSBs (Fig. 5).
Slower Dynamics of DSBs Signaling Via Phosporylated Histone H2A After IR of hESCs with Downregulated APE1
To determine whether the BER-mediated increase in DSB generation was associated with checkpoint-mediated cell cycle regulation, we monitored the level of phosphorylated histone γ-H2AX foci. We did not observe any significant difference in levels of γ-H2AX foci formation between late passage and early passage hESCs with downregulated APE1 immediately after IR in case of APE1 siRNA (supporting information Fig. 8A, 8B), while we detected less foci containing cells in sample treated with APE1 inhibitor methoxamine, which was not irradiated (supporting information Fig. 10 Panel A). 60 minutes after IR, the total number of foci was significantly lower (p < .05) in APE1 siRNA-treated early passage hESCs and in late passage hESCs than in early passage hESCs (Fig. 7A, supporting information Figs. 8B and 10). Stable γ-H2AX levels on Western blot analysis (Fig. 7B) suggest that foci appear by migration of phosphorylated H2A histones rather than by de novo phosphorylation of histone H2A. Downregulation of APE1 also did not lead to altered proliferation (supporting information Fig. 7A) or altered cell cycle distribution (supporting information Fig. 7B), which ruled out the change in cell cycle properties as a cause of altered γ-H2AX signaling.
Chromosomal abnormalities, karyotypic changes, meiotic aberrations, and a plethora of mutations have been identified in long-term hESCs adapted cultures . We have observed karyotypic changes in all examined hESC lines. We need to ask what molecular mechanisms trigger these adaptations to in vitro conditions [20, 21].
Under normal conditions, normal tissues can accumulate up to 500,000 apurinic sites per cell per day , and a significant number of these apurinic lesions may derive from clustered oxidative damage [recently reviewed in [23–29, 30]]. hESCs appear to be very proficient in the repair of certain types of DNA damage , especially DNA damage after exposure to IR . BER efficiency is also higher in testis compared to somatic tissues under normal conditions ( and our unpublished data). Decreased BER efficiency and increased MF  in germ cells (but not other tissues) from older animals is related to decreased levels of APE1 in the germ cells ( and our unpublished data) and it leads to chromosomal abnormalities . We also observed a significant decrease in the level of APE1 in late passage hESC (Fig. 2A, 2C). APE1 remained localized in the nuclei in early, medium, and late passage hESCs (Fig. 2A; supporting information Fig. 4). Thus, the lower BER activity measured in nuclear extracts from late hESC passages could not be caused by APE1 localization in the cytoplasm or mitochondria, in contrast to some other cell types, such as HeLa cells, in which APE1 is typically localized in the cytoplasm and translocated to the nucleus after DNA damage (e.g., IR) . Such strict pattern of APE1 localization observed in hESCs together with higher elevated level of APE1 in hESC compared to differentiated cells (Fig. 1C) indicate the need of stem cells to safeguard the integrity of their genome. Despite the fact, that Western blot analysis reveals only about 30% average decrease in APE1 (Fig. 2C) while BER activity decreased 2- to 10-fold (Fig. 1A, 1B), addition of recombinant APE1 not only restores BER efficiency in late passage hESC nuclear extracts to a normal level but also can increase the efficiency above that of nuclear extracts from early passage hESCs (Fig. 4). Disparity between APE1 level and BER activity suggests an existence of other APE1 activity regulation mechanisms, such as protein–protein interactions and protein modifications, which may be of interest when finding ways to prevent these changes in long-term cultures of hESC. These data also suggest that other enzymes of the BER machinery are present in excess in the extracts, which is supported by stable protein levels of Pol-β and Ogg1 during prolonged cultivation (Fig. 3A–3C) as well as their stable expression levels (Fig. 3D).
Evidence that decreased AP endonuclease activity results in less efficient BER (Figs. 1, 2 and 4), decreased ability to cope with induced damage (Figs. 5, 6, and 7A), and increased MF (as suggested by the study on murine germ cells ) may fit the model of oxidative damage-induced telomere shortening. Telomerase is associated with two mechanisms of telomere shortening. The first is the direct absence of telomerase, which leads to failure to renew the telomeres, which is ruled out in the case of hESCs by the steady telomerase activity during prolonged cultivation (supporting information Fig. 2). Also telomerase activity did not fluctuate with prolonged hESC cultivation (up to passage 130) in a different study . Conversely, and in agreement with our findings, high telomerase activity has been shown to be indispensable for self-renewal of hESCs , and telomerase activity was downregulated during hESC differentiation to EBs ( and supporting information Fig. 2). The second mechanism is oxidative damage-induced telomerase-mediated shortening . This model is also supported by finding, that IR-induced mostly oxidative DNA damage  to telomeres appears to cause more genome instability in human fibroblasts expressing human telomerase reverse transcriptase (hTERT) than does damage induced by Cr(VI) , processing of which is dependent on MSH2-MSH3 . The increase in the DSB burden after time (not limited to the telomeres) in late passage hESC in our comet analysis (Fig. 5), together with the reports described above, suggests that increased oxidative damage to the telomeres in hESCs might be key to chromosomal rearrangements in hESCs.
An interesting phenomenon of spotty staining phenotype was often observed when analyzing APE1 abundance and localization in medium passage hESCs. There was a variable level of APE1 fluorescence in neighboring colonies on the same plate as well as randomly distributed cells with varying APE1 fluorescence within a single colony (Fig. 2A, 2B). This observation supports the recent supposition that hESC populations are highly heterogeneous . APE1 heterogeneity may be linked to fluctuation in activity of BER in medium passage hESCs as demonstrated by large error bars compared to early and late passage hESCs (Fig. 1). Limited heterogeneity in NHEJ activity on the other hand suggests that BER variability is linked to stem cell specific mechanisms as NHEJ activity is comparable in hESC and differentiated cells (supporting information Fig. 3), while BER activity is approximately 10-fold higher in hESC (Fig. 1C). Our data support the hypothesis that hESC cultures are highly heterogeneous and decreased APE1 might be either the result of some heterogeneity or a driving force for further processes of adaptation including overgrowth of BER-inefficient cells. This hypothesis is further supported by the variability of spBER efficiency in medium passage cells, which is probably also caused by fluctuations in APE1 content on a colony-to-colony and cell-to-cell basis (Fig. 1A, medium).
The hypothesis that apurinic repair intermediates are converted to DSBs via BER machinery and clustered damage, as suggested by Yang , in hESCs is supported by the altered dynamics of DSB release after IR in hESCs treated with APE1 siRNA compared with untreated hESCs (Fig. 6A–6C). The level of DSBs in untreated hESCs is significantly higher immediately after IR treatment, as well as 45 minutes later, than in cells with downregulated APE1 (where it does not significantly differ from the level in unirradiated cells), suggesting that all oxidative damage has been quickly detected and processed by BER in untreated cells, while downregulated cells were left with their load of abasic lesions without further processing by APE1 to release DSBs. Level of APE1 in late passage hESC (lower than early passage hESC and higher than siRNA depleted) presents intermediate phenotype, where DSBs are released during prolonged period of time (60 minutes, Fig. 5) changing the short and high peak of DSB release in early passage cells into slow and long release detected in later passage. We suggest that the low expression of APE1 is one of the factors leaving the cells struggling with a heavy burden of abasic lesions, which cannot be converted to DSBs, cannot trigger early checkpoint response (Fig. 7A), and cannot be repaired by NHEJ within minutes (Fig. 5).
In cultured mammalian cells, NHEJ is a major tool for DSB repair, with homologous recombination (HR) playing a minor role . Antiapoptotic effect of Ku80 and Ku70 knockout in p53−/− mice double mutant suggests their possible role in altering cell cycle checkpoint response to DNA damage [28, 29, 37]. Ku80 modulates also chromosome rearrangements in certain Ku80−/− mouse tissues . We found strong NHEJ activity in hESCs, comparable to that in differentiated cells, such as hESCs-derived fibroblast cells. The NHEJ activity did not change with passage number (supporting information Fig. 3). These findings suggest that a normal level of NHEJ in hESCs in the early, medium, and late passage cells might render checkpoint mechanisms less effective in response to DNA damage which in combination with less BER induced-DSB release can overload HR.
Thus decrease in BER-mediated DSB damage may also lead to increase in MF via HR. Clonal expansion and persistence of cells after HR between misaligned sequences has been shown to contribute to differences in MF between tissues and at different ages, while the frequency of HR events and the ability to initiate them appears to be unchanged . Cellular detection of DSBs signals through ataxia talangescia mutated (ATM) and ATM and Rad3 related (ATR) kinases to the G2/M checkpoint mechanism with one of its targets being histone H2A . γ-H2AX is associated with apoptosis of hESCs 3 days after DNA damage . While the dynamics of the foci formation (supporting information Fig. 8C, 8D) does not seem to be different from published data , we observed a significant shift in dynamics of γ-H2AX foci formation upon IR in late passage hESCs (Fig. 7A). Also hESCs with downregulated BER activity due to APE1 siRNA treatment exhibited lower phosphorylation of γ-H2AX foci, suggesting the decreased number of DSBs released immediately after IR treatment in late passage or siRNA-treated and methoxamine-treated cells (supporting information Fig. 10, Panel B) leads to the formation of fewer foci. Although the APE1 depletion presents only about 30% decrease in γ-H2AX foci signaling and majority is mediated by alternative mechanisms (Fig. 7A; supporting information Fig. 10A, 10B), such significant drop can present an increased risk of acquiring mutation. Similar effect of APE1 siRNA and methoxamine demonstrates, that it is the DNA repair function of APE1, mediating the effect. Such affected proapoptotic signaling at early stage after DNA damaging event might reflect escape the cellular checkpoint mechanisms during G2M transition, which seems to be important with respect to the later apoptosis of DNA damaged cells [41, 43]. These results are not affected by changed cell cycle distribution (supporting information Fig. 7A, 7B) or apoptosis detected 6 hours after irradiation with 3 Gy of IR (supporting information Fig. 9A, 9B).
We thank Martina Vodinská, Klára Koudelková, and Iveta Peterková for their technical support. We also thank students Jan Raška and Monika Šeneklová for their help with revision of this manuscript. This study was supported by the Ministry of Health of the Czech Republic (grant NS10237-3), the Ministry of Education, Youth and Sport of the Czech Republic (MSM0021622430), and European Commission FP6 funding (contract LSHG-CT-2006-018739) and Grant Agency of the Czech Republic (Grant No. P302/12/G157) and by the project FNUSA-ICRC (no. CZ.1.05/1.1.00/02.0123) from the European Regional Development Fund.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.