Ionizing Radiation Induces Ataxia Telangiectasia Mutated-Dependent Checkpoint Signaling and G2 But Not G1 Cell Cycle Arrest in Pluripotent Human Embryonic Stem Cells

Authors

  • Olga Momčilović,

    1. Department of Human Genetics, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pennsylvania
    2. Pittsburgh Development Center, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Serah Choi,

    1. Medical Scientist Training Program and Molecular Pharmacology Graduate Training Program, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Sandra Varum,

    1. Pittsburgh Development Center, Pittsburgh, Pennsylvania
    2. Centro de Neurociências e Biologia Celular, Departamento de Zoologia, Universidade de Coimbra, Portugal
    Search for more papers by this author
  • Christopher Bakkenist,

    1. Department of Radiation Oncology, University of Pittsburgh School of Medicine, Hillman Cancer Center, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Gerald Schatten,

    1. Pittsburgh Development Center, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Christopher Navara

    Corresponding author
    1. Pittsburgh Development Center, Pittsburgh, Pennsylvania
    2. Department of Biology, University of Texas at San Antonio, San Antonio, Texas
    • University of Texas at San Antonio, Department of Biology, San Antonio, Texas 78249, USA
    Search for more papers by this author
    • Telephone: 210-458-6497


  • Author contributions: O.M.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; S.C.: collection of data; S.V.: provision of study material; C.B.: provision of study material, data analysis and interpretation, manuscript writing; G.S.: financial support, data analysis and interpretation, manuscript writing; C.N.: concept and design, data analysis and interpretation, manuscript writing, final approval of manuscript.

  • First published online in STEM CELLS EXPRESS May 14, 2009.

Abstract

Human embryonic stem (ES) cells are highly sensitive to environmental insults including DNA damaging agents, responding with high levels of apoptosis. To understand the response of human ES cells to DNA damage, we investigated the function of the ataxia telangiectasia mutated (ATM) DNA damage signaling pathway in response to γ-irradiation. Here, we demonstrate for the first time in human ES cells that ATM kinase is phosphorylated and properly localized to the sites of DNA double-strand breaks within 15 minutes of irradiation. Activation of ATM kinase resulted in phosphorylation of its downstream targets: Chk2, p53, and Nbs1. In contrast to murine ES cells, Chk2 and p53 were localized to the nucleus of irradiated human ES cells. We further show that irradiation resulted in a temporary arrest of the cell cycle at the G2, but not G1, phase. Human ES cells resumed cycling approximately 16 hours after irradiation, but had a fourfold higher incidence of aberrant mitotic figures compared to nonirradiated cells. Finally, we demonstrate an essential role of ATM in establishing G2 arrest since inhibition with the ATM-specific inhibitor KU55933 resulted in abolishment of G2 arrest, evidenced by an increase in the number of cycling cells 2 hours after irradiation. In summary, these results indicate that human ES cells activate the DNA damage checkpoint, resulting in an ATM-dependent G2 arrest. However, these cells re-enter the cell cycle with prominent mitotic spindle defects. STEM CELLS 2009;27:1822–1835

INTRODUCTION

Since all organisms are continually exposed to environmental and metabolic factors that cause DNA damage, eukaryotic cells have developed elaborate cell cycle checkpoint control and DNA repair mechanisms that act in an orchestrated manner to arrest the cell cycle until the damage is repaired [1, 2]. Failure to do so can have detrimental consequences—transmission of genetic defects to the daughter cell or cell death. If DNA damage cannot be repaired, cells containing DNA damage may undergo apoptosis, removing their damaged DNA from the pool of cycling cells [1, 3–5].

Ionizing radiation induces a variety of DNA lesions in exposed cells [6–8], of which DNA double-strand breaks (DSB) are particularly toxic because they are more difficult to repair because of the loss of integrity of both DNA strands [9]. In somatic cells, ionizing radiation-induced DNA damage signaling is initiated by the DSB sensor ataxia telangiextasia mutated (ATM) kinase [10]. ATM is a member of the phosphatydilinositol 3′-kinase (PI3K)-related kinase family, but it phosphorylates protein rather than lipid substrates. In the presence of DSB, ATM becomes activated and phosphorylates numerous downstream targets, including Chk2 and p53, which act as signal transducers and effectors that initiate cell cycle arrest and apoptosis in G1, S, and G2 phases of the cell cycle [11].

Whereas DNA damage responses have been extensively studied in somatic cells in vitro, the limiting nature of the available tissues and the ethical concerns in studying early human development in vivo have precluded studying DNA damage responses during very early human development. The successful isolation of human embryonic stem (ES) cells from blastocyst embryos by Thomson et al. [12] enables these pluripotent, immortal cells to be used to study the corresponding developmental stages in vitro.

Both mouse [13, 14] and non-human primate [15] ES cells have previously been reported to lack a functional DNA damage-induced G1/S cell cycle arrest and to be hypersensitive to DNA-damaging agents, responding with high levels of apoptosis [13, 14, 16] and differentiation [16, 17]. Furthermore, mouse ES cells may exhibit uncharacteristic localization and expression of checkpoint control proteins. The DNA damage signaling factor Chk2 has been reported to localize aberrantly to the centrosomes in mouse ES cells and failed to translocate to the nucleus after irradiation [14]. Additionally, conflicting reports of p53 localization and activity have been described in mouse ES cells in response to DNA damage. Aladjem et al. [13] reported that mouse ES cells do not activate p53-dependent DNA damage responses and undergo p53-independent apoptosis in response to ionizing radiation. These authors and others have reported that p53 was inefficiently translocated to the nucleus after DNA damage in these cells [13, 14, 18]. In contrast, others have reported that treatment with DNA-damaging agents results in p53-induced differentiation of mouse [17] and human [16] ES cells by suppressing Nanog expression through direct binding to the nanog promoter, implying that p53 does successfully translocate to the nucleus after DNA damage.

In this study we examined the events in the ATM kinase-dependent checkpoint signaling pathway and cell cycle arrest in human ES cells after exposure to 2 Gy of γ-radiation. We demonstrated phosphorylation and localization of ATM at the sites of DNA DSB and phosphorylation and nuclear localization of ATM downstream targets, Chk2 and p53, similar to that in irradiated human somatic cells. We also showed that irradiated human ES cells arrest in the G2, but not in the G1, phase of the cell cycle. We used the ATM-specific inhibitor KU55933 to evaluate the role of ATM in effecting this G2 arrest. Checkpoint signaling in irradiated human ES cells was inhibited using KU55933, but only at concentrations 10-fold higher than that necessary to inhibit ATM function in somatic cells. Inhibition of ATM function compromised G2 arrest in human ES cells, and irradiated cells proceeded to mitosis in the presence of KU55933.

MATERIALS AND METHODS

Cell Culture and KU55933 Treatment

Human ES cell lines WA07 and WA09 (WiCell Research Institute, Madison, WI, http://www.wicell.org) were cultured in human ES cell medium containing 80% knockout Dulbecco's modified eagle medium (DMEM), 20% knockout serum replacer, 1% nonessential amino acids, 1% penicillin/streptomycin (100 U per 100 μg/ml), 2 mM L-glutamine, and 4 ng/ml basic fibroblast growth factor (bFGF; all from Invitrogen, Carlsbad, CA, http://www.invitrogen.com) on mitomycin C-treated mouse embryonic fibroblasts (MEF; Millipore, Billerica, MA, http://www.millipore.com). Cells were passaged manually every 7 days and medium was changed every 48 hours. For flow cytometry cells were grown on Matrigel (BD Biosciences, Bedford, MA, http://www.bdbiosciences.com) in human ES medium conditioned with feeder cells for 24 hours and supplemented with an additional 4 ng/ml bFGF. For ATM inhibition studies, cells were treated with 10, 50, or 100 μM KU55933 (generously provided by Graeme Smith, Astra Zeneca, Wilmington, DE, http://www.astrazeneca.com) or vehicle dimethyl sulfoxide (DMSO; Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com) 1 hour before γ-radiation treatment. Cells were routinely tested for normal karyotype as previously described [19].

Irradiation

Human ES cells were irradiated 1 week after passaging with 2 Gy of γ-radiation using a Gammacell 1000 Elite cesium137 irradiator (Nordion, Ottawa, Canada, http://www.mds.nordion.com). Immediately after irradiation, cells were returned to the incubator for recovery until the appropriate time point.

Immunocytochemistry and Confocal Microscopy

For immunocyctochemistry human ES cells were grown on Thermanox plastic coverslips (NUNC, Rochester, NY, http://www.nuncbrand.com) on MEF feeder cells, fixed in 2% formaldehyde, and permeabilized in 0.1% Triton X-100 in tris-buffered saline (TBS; both from Sigma). Incubation with the primary antibodies was carried out in a humidified chamber at 37°C for 45 minutes. The following primary antibodies were used: POU5F1 (Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com), Nanog (R&D Systems Inc., Minneapolis, http://www.rndsystems.com), histone H3-serine 10, Chk2-threonine 68, and p53-serine 15 (Cell Signaling Technology, Danvers, MA, http://www.cellsignal.com), ATM-serine 1981 (Epitomics, Burlingame, CA, http://www.epitomics.com), γ-H2A.X-serine 139 (Upstate, Lake Placid, NY, http://www.upstate.com), and β-tubulin and SSEA-4 (Developmental Studies Hybridoma Bank at The University of Iowa, Iowa City, IA, http://dshb.biology.uiowa.edu). Primary antibodies were detected using species-specific fluorescently labeled secondary antibodies (Invitrogen) at 37°C for 45 minutes. DNA was visualized by addition of 1 μM Toto-3 (Invitrogen). Coverslips were mounted onto glass slides with Vectashield antifade mounting medium containing 4′,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). Slides were examined using a Leica TCS-SP2 laser scanning confocal microscope (Leica Microsystems GmbH, Wetzlar, Germany, http://www.leica-microsystems.com). The appropriate species-specific secondary antibody controls were obtained in the same manner, but incubation with primary antibodies was omitted. Co-localization of immunocytochemical probes was determined using ImageJ software [20]. For the analysis of mitotic figure morphology the following criterion was used: mitotic figures that displayed misaligned chromosomes in metaphase, lagging chromosomes in anaphase, or multipolar spindles were regarded as aberrant. Prophase cells were considered as noninformative and were not categorized as either normal or aberrant. Examples of both normal and aberrant mitotic figures are presented in Figure 5D.

RNA Extraction, Reverse Transcription, and TaqMan Low-Density Arrays

Human ES cells grown on MEF feeders were harvested by manual scraping. Total RNA was isolated with TRIzol (Invitrogen) and subjected to DNA cleanup with the DNA-free kit (Ambion, Austin, TX, http://www.ambion.com) following manufacturer's directions. Two micrograms of RNA were used per reverse transcriptase reaction performed with ImProm-II Reverse Transcription System (Promega, Madison, WI, http://www.promega.com). Reverse transcriptase (RT) and no-RT reactions were performed identically, except that in no-RT reactions water replaced reverse transcriptase. The TaqMan Human Stem Cell Pluripotency Array (Applied BioSystems, Foster City, CA, http://www.appliedbiosystems.com) was used following manufacturer's instructions, and data for Nanog, POU5F1, and CD9 expression were analyzed using SDS 2.2.2 software (Applied BioSystems). Expression fold changes were calculated using the −ΔΔCt method and normalized using β-actin as the endogenous control.

Western Blot

Human ES cells were manually scraped, pelleted by centrifugation, and lysed in RIPA buffer supplemented with 1 mM phenylmethylsulphonyl fluoride (both from Sigma) and 2× Halt phosphatase inhibitor cocktail (Pierce, Rockford, IL, http://www.piercenet.com). The protein concentration was determined using bicinchoninic acid assay (Bio-Rad, Hercules, CA, http://www.bio-rad.com) and 5 μg of protein was loaded per well. Proteins were separated by SDS-polyacrylamide gel electrophoresis followed by transfer to BioTrace polyvinylidene difluoride membrane (Pall Life Sciences, East Hills, NY, http://www.pall.com). After transfer, the membrane was blocked in TBS with 5% milk and 0.1% Tween-20 (all from Sigma) for 1 hour at room temperature. Membranes were incubated with the following primary antibodies at 4°C overnight: POU5F1 and p53 (Santa Cruz Biotechnology, Inc.), Nanog (Kamiya Biomedical Company, Seattle, WA, http://www.kamiyabiomedical.com), Chk2-threonine 68, p53-serine 15, p53-serine 20, Nbs1-serine 343, and cleaved caspase-3 (Cell Signaling Technology), ATM-serine 1981 and Nbs1 (Epitomics), ATM and α-tubulin (Sigma), and Chk2 (Thermo Fisher Scientific, Fremont, CA, http://www.thermofisher.com). Horseradish peroxidase-conjugated species-specific secondary antibodies (Invitrogen) were diluted in blocking buffer and incubated for 1 hour at room temperature. Detection of bound antibodies was performed with an ECL Advance Western Blotting Detection kit (Amersham Biosciences, Piscataway, NJ, http://www.gelifesciences.com), according to manufacturer's directions. Chemiluminescent signals were recorded using Hyperfilm (Amersham Biosciences).

Flow Cytometric Analysis

Analysis of Cell Cycle Distribution. Cell cycle analysis by flow cytometry was performed using propidium iodide (PI; BD Biosciences Pharmingen, San Diego, CA, http://www.bdbiosciences.com). Human ES cells grown on Matrigel were harvested using Accutase (Chemicon International, Temecula, CA, http://www.chemicon.com), pelleted, and washed in phosphate-buffered saline (PBS; Invitrogen). Five hundred thousand cells were resuspended in 1 ml of PBS, fixed by addition of ice-cold 70% ethanol dropwise, and placed at −20°C overnight. After fixation, cells were washed with PBS and centrifuged at 200g for 5 minutes. The cell pellet was resuspended in PBS solution containing PI (5 μg per 106 cells) and RNase (10 mg/ml) and incubated for 30 minutes at room temperature in the dark. Cells were examined using a Becton Dickinson FACSVantage DiVa (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com), and DNA-PI-Area histograms were analyzed using ModFit (Verity Software House, Topsham, ME, http://www.vsh.com). Results were gated to exclude cellular debris, sub G0 population, and doublets.

EdU Pulse-Chase Incorporation. Human ES cells grown on Matrigel-coated plates were labeled with 10 μM 5-ethynyl-2′-deoxyuridine (EdU; Invitrogen) for 30 minutes (pulse) before replacement with fresh medium. Human ES cells were immediately collected (nonirradiated control cells), or irradiated with 2 Gy, and left to recover for 4, 8, 16, and 24 hours after irradiation (chase). Cell fixation, permeabilization, and EdU detection were performed following manufacturer's instructions for the Click-iT EdU flow cytometry kit (Invitrogen). DNA content was measured using PI. Cells were analyzed using a Becton Dickinson FACSVantage DiVa. Dot plots representing EdU-647 fluorescence against DNA content were analyzed using BD FACSDiva software 6.1.1 (BD Biosciences, San Jose, CA, http://www.bdbiosciences.com).

Annexin V Labeling. Human ES cells were grown on Matrigel and harvested by manual scraping in PBS 8 hours after exposure to 2 Gy of γ-radiation. Staining with Annexin V and PI was performed according to manufacturer's instructions for Annexin V-FITC Apoptosis Detection Kit I (BD Pharmingen). Dot plots representing PI fluorescence against Annexin V-FITC fluorescence were analyzed using BD FACSDiva software 6.1.1.

Histone H3-Serine 10 Flow Cytometry. Human ES cells were grown on Matrigel and collected with Accutase (Chemicon) as previously described. Staining was performed using histone H3-serine 10 Alexa Fluor 488 conjugate (Cell Signaling Technology) following manufacturer's directions. Appropriate species-specific isotype control was used to estimate the nonspecific binding of the conjugated antibody. Cells were analyzed using a Becton Dickinson FACSVantage DiVa, and data were analyzed using BD FACSDiva software 6.1.1.

Statistical Analysis. Means and SEMs were calculated, and statistically significant differences for categorical data were determined by χ2 test. Significance was determined at p|<|.05.

RESULTS

Pluripotency and Radiosensitivity in Human ES Cells

Both mouse and human ES cells have been reported to reduce the expression of Nanog and differentiate in response to DNA damage [16, 17]. To confirm the populations of cells that we were studying were pluripotent human ES cells, we followed the expression of the pluripotency transcription factors Nanog and POU5F1 by quantitative polymerase chain reaction (Fig. 1A), Western blot (Fig. 1B), and immunocytochemistry (Fig. 1C, 1D) for 24 hours after 2 Gy of irradiation. Additionally, we examined the cell surface markers CD9 (Fig. 1A) and SSEA-4 (Fig. 1E). We observed that the mRNA levels for CD9, Nanog, and POU5F1 decreased relative to nonirradiated cells 6 hours after irradiation, similar to what was previously described in mouse and human ES cells [16, 17] (Fig. 1A). However, when we continued to follow human ES cells at later time points, the mRNA levels returned to those values observed in nonirradiated cells by 24 hours after irradiation. We did not detect a similar decrease in the protein level of Nanog and POU5F1 (Fig. 1B). In accordance with the Western blot results, POU5F1 (Fig. 1C), Nanog (Fig. 1D), and SSEA-4 (Fig. 1E) were detected in human ES cells before and after irradiation by immunocytochemistry.

Figure 1.

Pluripotency and radiosensitivity of human embryonic stem (ES) cells. (A): Human ES cells were irradiated, or left untreated, and allowed to recover for the indicated time periods before collection. Total RNA was isolated and the expression of pluripotency markers CD9, Nanog, and POU5F1 analyzed by TaqMan human stem cell pluripotency array. The mRNA fold changes were calculated using the −ΔΔCt method and normalized using β-actin as endogenous control. The results in nonirradiated cells were normalized to 1, and the value in irradiated cells was calculated with respect to this. (B): Western blot analysis of Nanog and POU5F1 protein levels after the irradiation of human ES cells. α-Tubulin served as the loading control. Confocal microscopy for (C) POU5F1, (D) Nanog, and (E) SSEA-4 in human ES cells at indicated time points after irradiation. (F): Western blot analysis for cleaved caspase-3 after irradiation. α-Tubulin served as the loading control. Bar = 100 μm (C); 50 μm (D,E)

After exposure to 5 Gy of ionizing radiation, large holes began appearing in the colonies within 6 hours (Fig. 1C, supporting information Fig. 1). To determine whether this was cell death, as previously reported for ES cells [13, 14, 16], or a detachment of cells from the substrate, we performed a Western blot for cleaved caspase-3 (Fig. 1F) and flow cytometry for Annexin V and PI (supporting information Fig. 2) after exposure to 2 Gy of γ-radiation. Cleaved caspase-3 began appearing in the samples 4 hours after irradiation and continued to increase for at least 24 hours. A marked increase in percentage of cells positive for Annexin V and doubly positive for PI was observed 8 hours after irradiation. This finding confirms that the observed loss of cells was due to cell death, most likely via apoptosis. We also noted a change in the size of the cells. We believe this is not due to differentiation, as demonstrated by maintenance of pluripotency markers, but rather the result of human ES cells expanding into space freed by the cell death.

ATM Activation in Human ES Cells

ATM kinase activation is one of the earliest and most sensitive responses to DNA damage in irradiated cells [21]. After γ-irradiation, ATM is recruited to the sites of DSB, where it becomes activated through autophosphorylation at serine 1981 [22]. We investigated ATM phosphorylation at this residue by Western blot (Fig. 2A). Phosphorylation of ATM at serine 1981 was detected in human ES cells 1 hour after exposure to 2 Gy of γ-radiation. High levels of this phosphorylation were maintained until 4 hours after irradiation, at which point the levels declined, but remained above steady state for at least 24 hours. During the same time frame, the level of total ATM protein did not change. Localization of ATM to the sites of DNA DSB was investigated by co-localization with the DNA DSB marker, γ-H2AX (Fig. 2B). Phosphorylated ATM was detected only weakly in nonirradiated cells, but ATM-serine 1981 co-localized with γ-H2AX foci 15 minutes after irradiation. The number of γ-H2AX foci increased immediately following irradiation and returned close to basal levels within 24 hours. Furthermore, we double-labeled cells for ATM-serine 1981 and Nanog and analyzed them by confocal microscopy (Fig. 3A), confirming that the ATM response occurs in pluripotent (Nanog-positive) human ES cells.

Figure 2.

ATM autophosphorylation and localization in human embryonic stem cells. (A): Western blot of ATM-serine 1981 and total ATM at the indicated time points after γ-irradiation. α-Tubulin served as the loading control. (B): ATM-serine 1981 localizes to the sites of DNA double-strand breaks (DSB) after irradiation, demonstrated by co-localization with a marker of DSB, γ-H2AX. Abbreviation: ATM, ataxia telangiectasia mutated. Bar = 10 μm. Blue, DNA; green, ATM-serine 1981; red, γ-H2AX; white, co-localization.

Figure 3.

ATM activation in pluripotent human embryonic stem (ES) cells. (A): Double labeling of human ES cells with Nanog (green) and ATM-serine 1981 (red). (B): Human ES cells are actively proliferating on day 7 after passaging. Human ES cells were labeled with EdU 30 minutes before harvesting on day 5 (left) and day 7 (right) after passaging. Cells were collected, stained for EdU and PI, and analyzed by flow cytometry. Dot plots (top) represent EdU fluorescence versus DNA content determined with PI. After gating, the G1, S, and G2/M phase cells were plated on the DNA histograms (bottom). The table represents the percentage of cells in each stage of the cell cycle on day 5 and day 7 after passaging. Abbreviations: ATM, ataxia telangiectasia mutated; EdU, 5-ethynyl-2′-deoxyuridine; PI, propidium iodide. Bar = 10 μm. Blue, DNA; green, Nanog; red, ATM-serine 1981.

Human ES cells are known to cycle rapidly through the growth phases after passaging and to decrease proliferation as they approach the next passage. We performed immunocytochemistry for ATM-Ser1981 and Nanog on day 4 and day 7 after passaging to ensure that the kinetics of ATM signaling was not dependent on the cells' growth phase. The kinetics of ATM activation was identical in both cases, as is observed in Figure 3A. In addition, we compared the proliferation of the human ES cells on day 5 and day 7 after passaging, and observed that the cells remain actively proliferating on the later day of the cell culture (Fig. 3B).

Activation of ATM Downstream Targets, Signal Transducers

Activated ATM phosphorylates numerous substrates at the sites of DSB, activating downstream signal transducers [23–26]. Phosphorylation of Chk2 and p53 is essential for induction of checkpoint arrest, whereas phosphorylation of Nbs1 facilitates formation of Mre11/Rad50/Nbs1 foci that are implicated in DNA repair [27]. We monitored the phosphorylation of these ATM targets by Western blot analysis using phospho-specific antibodies (Fig. 4). Phosphorylation of Chk2 at threonine 68 (Fig. 4A) was maximal 1 hour after irradiation and rapidly declined thereafter, so only a small fraction was still phosphorylated at 6 hours. The level of total Chk2 did not change over the same time course. Nbs1 phosphorylation at serine 343 (Fig. 4A) followed a similar time course, peaking 1 hour after irradiation and returning to control levels by 24 hours. Again, there was no change in total protein level during this time frame.

Figure 4.

Phosphorylation of ataxia telangiectasia mutated targets in human embryonic stem cells. (A): Western blot of Chk2-threonine 68, total Chk2, Nbs1-serine 343, and total Nbs1. (B): Confocal microscopy of Chk2-threonine 68. (C): Western blot of p53-serine 15, p53-serine 20, and total p53. (D): Confocal microscopy of p53-serine 15 after radiation treatment. α-Tubulin served as the loading control for all Western blots. Bar = 10 μm. Blue, DNA; green, Chk2-threonine 68 (B), p53-serine 15 (D).

Canonically, Chk2 is activated at DNA DSB by a mechanism that requires ATM kinase-dependent phosphorylation on threonine 68 [25] and localizes diffusely throughout the nucleus of the irradiated somatic cells [28]. In contrast to observations in somatic cells, phosphorylated Chk2 has been described to localize to the centrosomes of mouse ES cells [14]. We have also observed localization of Chk2-threonine 68 to the mitotic spindle poles in nonirradiated human ES cells (supporting information Fig. 3). However, phosphorylated Chk2 was detected in the nucleus 20 minutes after irradiation of human ES cells (Fig. 4B), unlike published reports in mouse ES cells [14]. Levels detected by immunocytochemistry followed a similar time course to those observed by Western blot analysis, with nuclear levels decreasing rapidly after 2 hours.

Both ATM and Chk2 kinases phosphorylate and thereby stabilize and activate p53. ATM phosphorylates p53 at serine 15 [23, 24], whereas Chk2 phosphorylates p53 at serine 20 in response to activation by ATM [29, 30]. We examined phosphorylation of these serine residues after irradiation using Western blot analysis (Fig. 4C). Nonirradiated human ES cells had no detectable p53 phosphorylated at serine 15 or serine 20 (Fig. 4C), but within 1 hour of irradiation, phosphorylation of serine 15 and serine 20 was detected, reached maximal levels by 2 hours, and declined thereafter, remaining elevated above baseline for 24 hours. The level of total p53 also increased after irradiation (Fig. 4C), most likely because of p53 stabilization after its phosphorylation [31, 32]. We also analyzed the phosphorylation of Chk2 and p53 in floating, presumably dying, human ES cells 6 and 24 hours after irradiation (supporting information Fig. 4). We observed the response in floating cells is in accordance with the response of attached cells at these time points after irradiation.

Conflicting results regarding p53 subcellular localization have been reported in mouse and human ES cells [13, 14, 16–18]. Our data show that p53-serine 15, detected using two different p53-serine 15 antibodies, is nuclear in two human ES cell lines (WA07 and WA09, not shown) after irradiation, by both immunocytochemistry (Fig. 4D) and immunohistochemistry (supporting information Fig. 5). Our results demonstrate that nonirradiated human ES cells have no detectable nuclear staining for phospho-p53. Twenty minutes after irradiation, both the number of p53-positive nuclei and the intensity of staining increased, peaked at 2 hours, and declined thereafter, similar to Western blot results (Fig. 4C). Localization of p53 to the nucleus followed a dose dependency, as higher levels of radiation induced a stronger response when examined at the same time point (3 hours; supporting information Fig. 6).

Cell Cycle Arrest After Irradiation

We next investigated whether human ES cells halt progression through the cell cycle in response to γ-irradiation. To assess the cell cycle profile of irradiated human ES cells, we performed flow cytometric analysis of DNA content using PI (Fig. 5A). Similar to non-human primate [15] and mouse [14] ES cells, irradiated human ES cells arrested in the G2/M phase of the cell cycle. The percentage of cells in G1 was significantly reduced 4 hours after irradiation, and essentially no cells were detected in G1 8 hours after irradiation. Human ES cells returned to the cell cycle 16 hours after irradiation, indicated by a decrease in the percentage of cells in G2/M between 16 and 20 hours (Fig. 5A). By 48 hours the cell cycle distribution closely resembled nonirradiated cells.

Figure 5.

Cell cycle analysis of irradiated human embryonic stem (ES) cells. (A): Analysis of DNA content of human ES cells after irradiation by flow cytometry using PI. Left panel: cell cycle profiles, as measured 0–48 hours after irradiation of human ES cells. Right panel: percentages of cells in G1, S, and G2/M as a function of time after irradiation. Percentages were calculated using ModFit software (Verity Software House). Results were gated to exclude cellular debris, sub G0 population, and doublets. Data presented are means ± SEM calculated from three independent experiments. The table represents the percentage of cells in each stage of the cell cycle after irradiation. (B): Histone H3-serine10 time course immunocytochemistry in irradiated human ES cells. (C): Left panel: Quantification of mitotic indexes in nonirradiated cells and 24 hours after irradiation. Human ES cells were irradiated, or left untreated, and fixed 24 hours later. Coverslips were stained with histone H3-serine10 antibody and the percentage of histone H3-serine10 positive (mitotic) cells was determined. The data represent mean ± SEM from three independent experiments. At least 1,000 cells were analyzed per condition in each experiment. Statistical significance was determined by χ2 test. Right panel: Quantification of percentage of aberrant mitotic figures in nonirradiated and irradiated human ES cells. Cells were treated as in the left panel and assayed for the presence of aberrant mitotic figures. Three independent experiments were performed, and at least 100 mitotic figures were analyzed per condition in each experiment. The result in nonirradiated cells was normalized to 100%, and the value in irradiated cells was calculated with respect to this. Statistical significance was determined by χ2 test. (D): Examples of normal (top row) and aberrant (bottom row) mitotic figures visualized by confocal microscopy. Abbreviations: A, anaphase; B, anaphase bridge; M, metaphase; n.s., not significant; P, prophase; PI, propidium iodide; PM, prometaphase; T, telophase. Bar = 10 μm. Asterisks, poles of mitotic spindle; arrowheads, misaligned chromosomes; blue, DNA; green, histone H3-serine10; red, β-tubulin. ∗∗∗, p|<|.001.

To further understand the events that occurred after release from cell cycle arrest, we performed immunostaining with the mitosis-specific marker histone H3-serine 10 (Fig. 5B). Similar to somatic cells, histone H3-serine 10 only labeled human ES cells in late G2/M (supporting information Fig. 7). Twenty minutes after irradiation, cells continued to divide and these mitotic figures were indistinguishable from controls (Fig. 5B). Two hours after irradiation, no histone H3-serine 10 positive cells were detected, suggesting that cells were arrested in G2 phase. Similarly, 6 hours after irradiation, no dividing cells were observed and there was a considerable amount of cell death. Mitotic figures began to be detected 20 hours after irradiation, indicating that cells returned to the cell cycle. We determined the mitotic indexes before and 24 hours after irradiation (Fig. 5C, right). There was no significant difference in quantified mitotic index between nonirradiated (3.91 ± 0.50%) and irradiated (3.57 ± 0.37%, 0.5|<|p|<|.7, n = 3) cells, but a high percentage of cells entering mitosis 24 hours after irradiation displayed aberrant mitotic spindles (Fig. 5D, Materials and Methods). The presence of aberrant mitotic figures, including anaphase bridges, multipolar spindles, and lagging and misaligned chromosomes, was elevated to 414.51 ± 45.98%, in comparison to that of nonirradiated cells (100 ± 38.86%, p|<|.001, n = 3; Fig. 5C, left).

The appearance of mitotic cells 16-20 hours after irradiation can be explained (a) by arrested cells re-entering the cell cycle or (b) by cells that were in the very early stages of G1 reaching mitosis 16-20 hours after irradiation and never arresting. We performed a pulse-chase experiment in which we labeled human ES cells with a thymidine analogue, 5-ethynyl-2′-deoxyuridine (EdU) for 30 minutes just before irradiation to distinguish between these two possibilities. EdU was washed out after the 30-minute pulse, and cells were collected at 4, 8, 16, and 24 hours after irradiation and analyzed by flow cytometry (Fig. 6). The pulse only labeled the human ES cells in S-phase at the time of irradiation. Eight hours after irradiation, these cells clearly moved into the G2/M phase of the cell cycle. At 16 hours, the first G1 cells that stain for EdU are observed, indicating that arrested cells have undergone mitosis. Additionally, in the dot plot it is observed that the G1 population has a lower EdU signal, as would be expected after mitosis. EdU labeling combined with immunocytochemistry also clearly labels mitotic figures 24 hours after irradiation (supporting information Fig. 8).

Figure 6.

Human embryonic stem (ES) cells resume the cell cycle after irradiation-induced G2/M arrest. Human ES cells were pulse-labeled with EdU before irradiation, exposed to 2 Gy of γ-radiation, and left to recover for 4, 8, 16, and 24 hours, without addition of EdU. At indicated time points after irradiation, human ES cells were collected, stained for EdU and PI, and analyzed by flow cytometry. Dot plots (left) represent EdU fluorescence versus DNA content determined with PI. After gating, the EdU-positive cells (blue) were plotted together with the single cells (green) on DNA histograms (on the right). Note that EdU-positive cells are cells that were in the S-phase at the time of irradiation and do not necessarily represent S-phase cells at any of the time points after irradiation. Abbreviations: EdU, 5-ethynyl-2′-deoxyuridine; PI, propidium iodide.

ATM Is Required for the G2 Arrest

To test whether ATM signaling is necessary for G2 arrest, we inhibited ATM kinase using the ATM-specific inhibitor KU55933 [32]. To the best of our knowledge, KU55933 inhibits ATM kinase activity in all somatic cells examined to date, when added to the cell culture medium at a concentration of 10 μM [33, 34]. However, our titration experiments revealed that this dose is only partially effective in human ES cells and that 100 μM is needed to inhibit ATM signaling (Fig. 7A). ATM phosphorylation and activity are maximal 1 hour after irradiation; thus, we titrated the concentration of KU55933 needed to inhibit the ATM function in human ES cells at this time point. We treated cells starting 1 hour before irradiation (minus 1 hour) until 1 hour after irradiation (plus 1 hour) with vehicle (DMSO), 10 μM, 50 μM and 100 μM KU55933. A dose-dependent decrease in the level of ATM-serine 1981, Chk2-threonine 68, p53-serine 15, and Nbs1-serine 343 after the addition of increasing concentration of KU55933 confirmed inhibition of ATM signaling in irradiated human ES cells. Under the same treatment, the level of total ATM, Chk2 and Nbs1 did not change. Similarly, treatment with KU55933 dramatically reduced p53 stabilization after irradiation.

Figure 7.

Role of ATM signaling in induction of G2 arrest in human embryonic stem (ES) cells. (A): Inhibition of ATM signaling with KU55933. Western blot for ATM-serine 1981, ATM, Chk2-threonine 68, Chk2, p53-serine 15, p53, Nbs1-serine 343, and Nbs1 in human ES cells. Human ES cells were pretreated with vehicle (DMSO) or 10 μM, 50 μM, and 100 μM KU55933 for 1 hour (minus 1 hour), irradiated, or left untreated, and harvested 1 hour later (plus 1 hour). α-Tubulin served as the loading control. (B): ATM is required for functional G2 arrest after irradiation of human ES cells. Human ES cells were treated with vehicle (DMSO) or 100 μM KU55933 starting 1 hour before irradiation (minus 1 hour), irradiated, or left untreated, and fixed 2 hours later (plus 2 hours), stained for histone H3-serine10, and analyzed by confocal microscopy. (C): Top panel: Quantification of mitotic indexes of KU55933- or vehicle-treated nonirradiated and irradiated cells. Human ES cells were treated as in (B), and the percentage of mitotic cells was quantified 2 hours after irradiation. Three independent experiments were performed, and at least 1,500 cells were analyzed per condition in each experiment. The data represents means ± SEM. Statistical significance was determined by χ2 test. Bottom panel: Quantification of percentage of aberrant mitotic figures in KU55933-treated cells. Human ES cells were treated as in (B), and assayed for the presence of aberrant mitotic figures. The percentage of aberrant mitotic figures in nonirradiated KU55933-treated cells was normalized to 100%, and the value in irradiated KU55933-treated cells was calculated in respect to this. Three independent experiments were performed, and at least 100 mitotic figures were analyzed per condition in each experiment. Statistical significance was determined by χ2 test. Abbreviations: ATM, ataxia telangiectasia mutated; DMSO, dimethyl sulfoxide; n.s., not significant. Bar = 50 μm. Blue, DNA; green, histone H3-serine10. ∗∗∗, p|<|.001.

To determine whether ATM activation is necessary for G2 arrest, we performed histone H3-serine 10 immunocytochemistry and determined mitotic index 2 hours after irradiation in cells in which ATM kinase was inhibited. Human ES cells were treated starting 1 hour before irradiation (minus 1 hour) until 2 hours after irradiation (plus 2 hours) with vehicle (DMSO) or 100 μM KU55933, fixed and stained with histone H3-serine 10 antibody (Fig. 7B). No significant difference in mitotic index was observed between nonirradiated cells treated with DMSO (2.85 ± 0.53%) and KU55933 (2.60 ± 0.21%, .3|<p|<.5, n = 3; Fig. 7C, top panel). However, 2 hours after irradiation, KU55933-treated cells had higher mitotic index (1.35 ± 0.34%) than DMSO-treated cells (0.16 ± 0.02%, p|<|.001; Fig. 7C), suggesting that ATM function is required for functional G2 arrest in human ES cells. However, even this high dose (100 μM) of ATM inhibitor did not completely restore the mitotic index in irradiated cells (1.35 ± 0.34%) to the one detected in nonirradiated cells (2.60 ± 0.21%).

Finally, we characterized mitotic cells that were observed after irradiation of KU55933-treated cells. The percentage of aberrant mitotic figures after irradiation of KU55933-treated cells was elevated (229.01 ± 36.15%, p|<|.001, n = 3; Fig. 7C) compared to that of KU55933-treated nonirradiated cells (100 ± 4.56%). The most prevalent type of mitotic error was anaphase bridges. We did not observe a statistically significant difference in the frequency of aberrant mitotic figures between vehicle-treated (78.03 ± 22.44%) and KU55933-treated (100 ± 4.56%, p|>|.1, n = 3) nonirradiated cells.

DISCUSSION

Several phenotypes associated with response of embryonic stem cells to DNA damage are as follows: (a) they reduce expression of pluripotency factors [16, 17], (b) they lack a G1 checkpoint [14, 15], and (c) they are extremely sensitive to DNA-damaging agents, resulting in cell death within several hours of exposure to DNA-damaging agents [13, 14, 16]. Additionally, a number of cell cycle regulatory proteins have been demonstrated to have aberrant functions in ES cells with sometimes conflicting results reported between laboratories. In this study, we sought to investigate the functioning of the ATM-signaling cascade in human ES cells, including analysis of the cell cycle distribution after ionizing radiation. We found that ATM signaling is intact and that human ES cells activate and properly localize the checkpoint signaling proteins ATM, Chk2, and p53, resulting in a temporary G2 arrest before the cells resume mitosis. Furthermore, our results describe an essential role of ATM in induction of G2 arrest, as an ATM-specific inhibitor, KU55933, inhibits this G2 arrest.

First, we investigated the expression of pluripotency markers Nanog, POU5F1, CD9, and SSEA-4 after irradiation. Surprisingly, we found that the protein levels do not change over the 24-hour period after irradiation, suggesting that human ES cells remained pluripotent. This observation is in contradiction with reports of DNA damage-induced differentiation of mouse [17] and human [16] ES cells. However, in these studies authors investigated the mRNA levels within 6 hours of irradiation. When we examined mRNA levels, we also observed a drop at 6 hours after irradiation. However, at 24 hours the levels returned to near that of controls.

Embryonic stem cells of different species show tremendous sensitivity to DNA damage and undergo extensive cell death within hours of DNA damage. In this paper, we demonstrated cell death within hours after exposure of human ES cells to 2 Gy of γ-radiation. Cleavage of caspase-3 was detected 4 hours after irradiation, and cell loss was visualized 6 hours after irradiation. Interestingly, human embryonal carcinoma (EC) cells display a higher survival rate after ionizing radiation, when compared to their differentiated counterparts [35]. After confirming the radiosensitivity of human ES cells, we investigated checkpoint signaling and function in irradiated human ES cells.

ATM is rapidly activated in cells exposed to agents that induce DNA double-strand breaks [21]. Cells deficient in ATM exhibit hypersensitivity to radiation and radiomimetic drugs, defective cell cycle arrest, increased chromosome breakage, and reduced p53 response after irradiation [36–39]. Cell cycle checkpoint defects include a diminished arrest in G1, radioresistant DNA synthesis, and reduced arrest in G2 [36, 38, 39]. In this study, we monitored the kinetics of ATM phosphorylation and localization to the sites of DNA DSB. Confocal microscopy revealed that ATM was phosphorylated and recruited to the DNA DSB in human ES cells within 15 minutes of irradiation. Phosphorylation of ATM and its target proteins Chk2, p53, and Nbs1 reached maximum within the first hour of irradiation, suggesting activation of checkpoint signaling in human ES cells.

Hong and Stambrook [14] reported that Chk2 is hyperphosphorylated and localized to the centrosomes in mouse ES cells rather than diffusely distributed in the nucleus, making it unavailable to act as a motile signal transducer. We also observed phosphorylated Chk2 at the poles of mitotic spindle in nonirradiated human ES cells (supporting information Fig. 3). However, in contrast to mouse ES cells, phospho-Chk2 was mobilized to the nucleus of human ES cells within 20 minutes of irradiation. Centrosomal Chk2 was detected in nonirradiated somatic cells as well and mobilized to the nucleus in response to DNA damage [40], in agreement with its canonical role as a mobile signal transducer. In addition, several proteins involved in DNA damage response, such as ATM [40, 41], ATR [40], ATRIP [40], Chk1 [42], p53 [43], and BRCA1 [44] have been detected to associate with centrosomes during mitosis. On the basis of these emerging data, it has been suggested that centrosomes might have a functional role in DNA damage responses, serving either as “command centers” [45], where DNA damage response proteins come in close proximity and/or are sequestered from unfavorable interactions, or as a subject of DNA damage response [46].

Conflicting results regarding p53 localization have been reported in ES cells. Several groups have reported that p53 is not translocated into the nucleus of mouse ES cells after DNA damage [13, 14, 18]. In contrast, other data suggest that p53 induces differentiation in mouse [17] and human [16] ES cells after DNA damage by binding to the nanog promoter, and suppressing Nanog expression, implying that p53 does translocate into the nucleus of ES cells after DNA damage. Our data demonstrate that p53 is stabilized in irradiated human ES cells. The observation that both ATM kinase-dependent p53-serine 15 and ATM- and Chk2-dependent p53-serine 20 phosphorylation are maximal 2 hours after irradiation, before maximal p53 protein level 4 hours after irradiation, is consistent with a function of these two kinases in p53 stabilization and activation. Furthermore, our results conclusively demonstrate that p53 is nuclear in irradiated human ES cells.

Mouse [14] and non-human primate [15] ES cells, as well as human EC cells [35], have been shown to lack DNA damage-induced G1 cell cycle arrest. Here, we extend this finding to human ES cells. Flow cytometry revealed that irradiated human ES cells do not accumulate in G1 but rather at the G2/M stage of the cell cycle. However, this cell cycle arrest is temporary, and human ES cells re-enter the cell cycle approximately 16 hours after irradiation. Interestingly, in spite of the numerous similarities between ES and EC cells, human ES cells do not exhibit S-phase delay, as do human EC cells, and undergo G2 arrest much earlier in comparison to human EC cells.

Studies with histone H3-serine 10 immunocytochemistry demonstrated that human ES cells promptly cease mitosis, further indicating that cells are arresting in G2. Twenty-four hours after irradiation, a fourfold higher proportion of mitotic figures appear aberrant in comparison to that of nonirradiated cells. It is unclear on the basis of these results why aberrant mitoses are observed. It is possible that human ES cells resume cycling before all DNA damage is removed. However, this speculation is not supported by our γ-H2AX data because the number of positive foci returns to near baseline levels by 24 hours. In addition, defective DNA DSB repair would increase DNA defects at mitosis such as chromosomal cross bridges. We do not observe this phenotype, and instead see numerous types of defects including spindle pole abnormalities, suggesting perhaps other cell cycle errors. Additional experiments will be required to discern the cause of these abnormal mitotic structures.

To determine whether the observed accumulation of irradiated human ES cells in G2 was ATM-dependent, we employed a selective small molecule competitive ATM inhibitor, KU55933 [32]. KU55933 inhibits ATM kinase activity in all somatic cells examined to date when included in a culture medium at a concentration of 10 μM [33, 34]. Interestingly, we observed that a concentration of 100 μM KU55933 is necessary in human ES cells to inhibit ATM function, as determined by ATM-serine 1981, Chk2-threonine 68, and Nbs1-serine 343 phosphorylation and by p53 stabilization. The partial inhibition of ATM kinase with 10 μM KU55933 is not due to insufficient time for inhibition because White at al [34] reported that ATM can be inhibited within 5 minutes of drug addition to the medium. It is of importance to note that even at 100 μM concentration KU55933 specifically targets ATM, and not other members of PI3K-related kinase family [33]. One possible explanation for this difference between human ES and somatic cells is high expression level of multiple drug resistant proteins in ES cells, which could efficiently remove the drug from human ES cells. This is of particular importance for the potential use of KU55933 as a radiosensitizing drug in anti-cancer therapy since an implication may be that cancer stem cells may not be radiosensitized with 10 μM KU55933.

We demonstrated an essential role of ATM in induction of G2 arrest in irradiated human ES cells. After inhibition of ATM, we observed an escape of irradiated cells from G2 arrest. However, abrogation of the G2 checkpoint was not 100% effective, as the mitotic index 2 hours after irradiation of KU55933-treated human ES cells was not restored to the levels observed in nonirradiated KU55933-treated cells. Two explanations are possible for this observation: (a) ATM kinase activity and signaling are not completely inhibited with 100 μM KU55933 or (b) there are two independent checkpoint mechanisms in human ES cells. The observation that ATM kinase-dependent phosphorylation of Chk2, Nbs1, and p53 was entirely abrogated in 100 μM KU55933-treated irradiated human ES cells suggests that a second, ATM-independent G2 checkpoint is operating in human ES cells. However, additional experiments will be required to identify this mechanism.

Finally, we characterized the cells that escape G2 arrest after KU55933 treatment and enter mitosis 2 hours after irradiation. We detected twofold higher frequency of mitotic errors, in particular anaphase bridges, in comparison to that of KU55933-treated control cells. At this time point γ-H2AX foci are still numerous and DNA DSB are not completely removed; the presence of free chromosome ends provides conditions in which chromosome cross-links and anaphase bridges can occur, explaining the high rate of mitotic errors in these conditions.

It is interesting that although ATM-dependent signaling in human ES cells appears indistinguishable from that in human somatic cells, ATM-dependent cell cycle arrest in G1 phase does not occur in ES cells. Canonically, maintenance of G1 arrest is dependent upon upregulation of p21 by p53. Mouse ES cells have been shown to lack this p53-p21 axis [14], which explains the absence of G1 arrest in these cells. However, it is not clear whether human ES cells induce p21 in response to DNA damage. Qin et al. reported the absence of p21 gene upregulation after UV irradiation, but noticed a twofold increase in protein level [16]. In contrast, Becker et al. demonstrated 250- to 300-fold induction of p21 mRNA levels after irradiation with 5 Gy of ionizing radiation [47]. In our preliminary data, p21 protein was undetectable in nonirradiated human ES cells, and we observed only weak induction of protein after 2 Gy of γ-irradiation. Taken together, it is possible that human ES cells have extremely low basal levels of p21, and after DNA damage induce p21, but at levels that are insufficient to inhibit Cdk2 activity in ES cells.

It has been suggested that the G1 phase of the cell cycle is a time when ES cells are sensitive to differentiating cues from the environment, and that shortening of G1 can protect ES cells from differentiation [48, 49]. Therefore, by escaping G1 arrest after DNA damage, ES cells might be reducing the risk of differentiation. Indeed, Maimets et al. recently demonstrated that activation of p53 by nutlin induces rapid differentiation of human ES cells by promoting accumulation of cells in G0/G1 phase in a p21-dependent manner [50]. Another possibility is that ES cells prefer repairing DNA damage during G2 when the sister chromatid is present, allowing for error-free DNA repair by homologous recombination, rather than error-prone nonhomologous end joining. Under this hypothesis, cells that are in G0/G1 at the time when DNA damage is inflicted may undergo differentiation or apoptosis, and those cells that are in G2 phase of the cell cycle would arrest and attempt repair of the damage. Further experiments need to be performed to test this hypothesis, as the kinetics and efficiency of DSB repair in human ES cells are unknown.

CONCLUSIONS

Collectively, our results demonstrate that the ATM checkpoint signaling cascade is intact in pluripotent human ES cells and ATM, Chk2, and p53 are phosphorylated and properly localized in response to induction of DNA DSB. Human ES cells temporarily arrest progression through the cell cycle at the G2 stage and re-enter the cell cycle approximately 16 hours after irradiation. However, after radiation exposure a fourfold higher proportion of mitotic figures appears aberrant. Finally, we have shown that ATM function is essential for induction of G2 arrest in irradiated human ES cells, but there may be additional cell cycle regulatory mechanisms, as ATM inhibition does not completely abrogate cell cycle arrest.

Acknowledgements

This research was supported by a grant from the National Institute of Child Health and Human Development, 1PO1HD047675. We thank Graeme Smith for generously providing KU55933 (AstraZeneca). For providing critical assistance, we thank Lynda Guzik for performing FACS analysis, Stacie Oliver for karyotyping human ES cell lines, Carrie Redinger, Jocelyn Mich-Basso, and David McFarland for help with human ES cell culture, Dan Constantinescu, Charles Easley, and Ahmi Ben-Yehudah for critical reading of this manuscript, and Robert Ferrell, Laura Niedernhofer, and Susanne Gollin for support and guidance.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

The authors indicate no potential conflicts of interest.

Ancillary