An Important Role for CDK2 in G1 to S Checkpoint Activation and DNA Damage Response in Human Embryonic Stem Cells§

Authors


  • Disclosure of potential conflicts of interest is found at the end of this article.

  • Author contributions: I.N.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing and final approval of manuscript; F.V.: collection and assembly of data, data analysis and interpretation, manuscript writing and final approval of manuscript; S.P.A.: collection and assembly of data and interpretation and final approval of manuscript; M.L.: collection of data and final approval of manuscript; J.F.P.: collection and assembly of data, data analysis and interpretation and final approval of manuscript; T.V.Z.: data analysis and interpretation, manuscript writing and final approval of manuscript; J.-E.C: collection and assembly of data and final approval of manuscript; D.B.: data analysis and interpretation, manuscript writing, and final approval of manuscript; R.J.: collection and assembly of data, data analysis and interpretation, manuscript writing and final approval of manuscript; L.A.: conception and design, financial support, manuscript writing and final approval of manuscript; M.L.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript and fund raising.

  • §

    First published online in STEM CELLSEXPRESS February 11, 2011.

Abstract

A precise understanding of mechanisms used by human embryonic stem cells (hESCs) to maintain genomic integrity is very important for their potential clinical applications. The G1 checkpoint serves to protect genomic integrity and prevents cells with damaged DNA from entering S-phase. Previously, we have shown that downregulation of cyclin-dependent kinase 2 (CDK2) in hESC causes G1 arrest, loss of pluripotency, upregulation of cell cycle inhibitors p21 and p27 and differentiation toward extraembryonic lineages. In this study, we investigate in detail the role of CDK2 in cellular processes, which are crucial to the maintenance of genomic stability in hESC such as G1 checkpoint activation, DNA repair, and apoptosis. Our results suggest that downregulation of CDK2 triggers the G1 checkpoint through the activation of the ATM-CHK2-p53-p21 pathway. Downregulation of CDK2 is able to induce sustained DNA damage and to elicit the DNA damage response (DDR) as evidenced by the formation of distinct γ-H2.AX and RAD52-BRCA1 foci in hESC nuclei. CDK2 downregulation causes high apoptosis at the early time points; however, this is gradually decreased overtime as the DDR is initiated. Our mass spectrometry analysis suggest that CDK2 does interact with a large number of proteins that are involved in key cellular processes such as DNA replication, cell cycle progression, DNA repair, chromatin modeling, thus, suggesting a crucial role for CDK2 in orchestrating a fine balance between cellular proliferation, cell death, and DNA repair in hESC. STEM Cells 2011;29:651–659

INTRODUCTION

The maintenance of genomic stability is essential for survival of all cell types. Consequently, eukaryotic cell have developed multiple mechanisms to cope with DNA damage and this involves activation of checkpoint pathways operating at the G1/S, intra S-phase and G2/M boundary. The DNA damage checkpoint network consists of DNA damage sensors, signal transducers, and various effector pathways, which mediate cell cycle arrest at G1, S, or G2 phases [1] to facilitate DNA repair and prevent transmissions of DNA damage to daughter cells. However, when the DNA damage is too severe, checkpoint activation can also lead to apoptosis, thereby eliminating cells with damaged DNA. In murine embryonic stem cells, DNA damage induced by ultraviolet light or doxorubicin treatment often induces differentiation through p53-mediated suppression of Nanog expression.

Human embryonic stem cells (hESCs) are pluripotent cells derived from the inner cell mass of embryos and are characterized by high rate of proliferation [2], short G1 phase, and an extended S-phase [3–5]. Although the short G1 and extended S-phase are also observed in murine, nonprimate, and primate ESC [6, 7], the G1 to S transition seems to be subject to species-specific regulation [5].

The G1 to S checkpoint prevents replication of damaged DNA. Essentially, this process is able to arrest the cell cycle through the functions of ataxia telangiectasia mutated (ATM) and ataxia telangiectasia and Rad3 related, which phosphorylate a number of substrate proteins in response to DNA damage. Typically, phosphorylation substrates include p53 and/or checkpoint kinases CHK1 (CHK1 checkpoint homolog, S. pombe) or CHK2 (CHK2 checkpoint homolog, S. pombe, [8]) although the choice of substrate is largely dependent on the type of DNA damage encountered. Phosphorylation of p53 on serine 20 reduces its ability to bind MDM2 (Mdm2 p53 binding protein homolog, mouse), which normally assigns p53 to proteasome-mediated degradation. The resulting accumulation and stabilization of p53 [9] leads to transcriptional activation of p21 (also known as p21CIP1 or cyclin-dependent kinase N1 [CDKN1], [8, 10, 11]) that in turn binds to CDK2/cyclin E complexes, triggering cell cycle arrest at the G1/S transition. The phosphorylated forms of CHK1 and CHK2 antagonize the function of CDC25 (cell division cycle 25 homolog C, S.pombe) phosphatases, thus, constraining CDK activity and arresting the cell cycle [12]. Although CDK2 is a known effector of the G1 to S DNA damage checkpoint in mammalian cells, the role of this molecule in human ESC with regard to G1 checkpoint activation has not been studied previously [13–15]. hESC do use the CHK1/CHK2-mediated downregulation of CDC25A and subsequently CDK2 [16] to activate the G1/S checkpoint in response to UV irradiation and our own data suggest that siRNA-mediated knockdown of CDK2 causes cell stalling in G1 and upregulation of two key cell cycle inhibitors, p21 and p27 [5]. However, the role of CDK2 in G1 checkpoint activation in hESC has not been studies beforehand.

Published data suggests that a minimal level of CDK activity is needed to promote DNA damage response (DDR) and ensure that efficient DNA repair can occur [17, 18]. It has also been suggested that CDK2 may function in DNA repair in a cell cycle-independent manner [13, 19] but the mechanism by which this may occur is largely unknown. In this study, we report that downregulation of CDK2 in hESC triggers activation of the G1/S checkpoint, increases apoptosis, and delays clearance of DNA damage. Our proteomic analysis suggests that CDK2 interacts with a large number of key proteins that are involved in cellular processes such as DNA replication, chromosome segregation, DNA repair, and chromatin modeling in hESC in addition to the G1 to S cell cycle regulation function already reported by our group in a previous publication [5].

MATERIALS AND METHODS

Cell Culture, Transfection Experiments, and ATM Inhibition

Human H1 and H9 embryonic stem cell lines (WiCell Research Institute, Madison, MI) lines were grown on mitotically inactivated mouse embryonic fibroblasts and passaged as described in [5]. All Western blot images are representatives of experiments that have shown similar results in both cell lines.

A few passages prior to use in experiments, hESCs were transferred to Matrigel-coated plates with feeder conditioned media as previously described in [5]. Downregulation of CDK2 was achieved using small interfering RNAs (Invitrogen Ltd, Paisley, U.K., www.invitrogen.com). The sequence of control and CDK2 siRNAs and the transfection procedure is described in our previous publication [5]. The cells were analyzed at 48, 72, and 96 hours after transfection. For the ATM inhibition studies, hESCs were treated with 100 μM CGK733 ATM kinase inhibitor (Calbiochem, San Diego, CA, www.merck-chemicals.co.uk/life-science-research/calbiochem) or 0.01% dimethyl sulfoxide (Sigma, St. Louis, MO, www.sigmaaldrich.com), which was used for dissolving the inhibitor for 16 hours starting from 24 hours post-transfection with siRNAs.

Western Immunoblotting

Protein extraction and Western blotting were performed as published before in [5]. The details of the antibodies used in this work can be found in the Supporting Information Annex A. The antibody to β-actin was used after membrane stripping to confirm uniform protein loading. Antibody/antigen complexes were detected using ECL (Amersham Biosciences, www.gelifesciences.com) and images were acquired using a luminescent image analyzer FUJIFILM and LAS-3000 software (FUJI, Abingdon, U.K., www.rndsystems.com).

CDK2 Immunoprecipitation Studies and Systematic Analysis of Proteins by Liquid Chromatography–Mass Spectrometry with Peptide Mass Fingerprinting Analysis of Samples from One-Dimensional Gels

Immunoprecipitations were performed with 1 μg of monoclonal anti-CDK2 (D-12; Santa Cruz Biotechnology, Heidelberg, Germany, www.scbt.com). hESCs were lysed in RIPA lysis buffer and 400 μg of extract was immunoprecipitated for at least two hours at 4°C with mixing. A total of 100 μl of swollen protein A/G Sepharose beads (Invitrogen, Barcelona, Spain) were added for an additional 3 hours, washed three times with complete RIPA lysis buffer, and separated on a Nu-PAGE Bis-Tris 4%–12% gel Invitrogen. The gel was stained with SimplyBlue Safestain (Invitrogen). Individual gel lanes were cut manually into 25 × 1.0 mm2 bands with a razor blade, and each band was transferred to a 96-well plate (PRO10003, Genomic Solutions). Gel bands were processed robotically using a ProGest instrument. For additional details, refer to Supporting Information Annex A. Confirmatory coimmunoprecipitation details are also given in Supporting Information Annex A.

Kinase Activity Assays

Kinase activity assays were carried out using the PKLight Assay Kit (LT07-500) (Cambrex Bio Science Rockland, Inc., Rockland, ME, www.cambrex.com) following manufacturer's instructions. The PKLight Assay exploits the intrinsic ATPase activity of kinases, resulting in the cleavage of the γ-phosphate moiety of ATP and its subsequent insertion into the target substrate. This results in the phosphorylation of the substrate and the conversion of ATP to ADP. The PKLight Assay measures the consumption of ATP and is based on the bioluminescent measurement of the remaining ATP present in the wells after the kinase reaction. Bioluminescent signal of PKLight Assay is inversely proportional to kinase activity. Phosphorylation of RB or H1 was measured by incubating for 10 minutes at room temperature 20 μl of immunoprecipitation product for the kinase of interest (see above) with 1 mM ATP, kinase buffer (50 mM Tris, pH 7.5, 5 mM MgCl2) and RB or H1 (5 mg/ml) as substrate. A total of 10 μl of kinase stop solution was added to each sample at room temperature for 10 minutes. Finally, 20 μl of ATP detection reagent was added to each sample at room temperature for 10 minutes, and the readings were taken using a luminometer. The difference in luminometer reading between the no antibody control and IP product containing the antibody was calculated. This figure which is indicative of remaining ATP in the solution was inversely correlated to the kinase activity.

Reverse Transcription Polymerase Chain Reaction

Total RNA was extracted using TRIzol reagent (Invitrogen, Paisley UK) according to manufacturer's instructions. Following DNaseI treatment using RQ1 DNaseI (Promega, Southampton, U.K., www.promega.com), cDNA was synthesized using SuperScript Reverse Transcriptase (Invitrogen) from 1 μg of total RNA. Quantitative reverse transcription polymerase chain reaction (RT-PCR) analysis was carried out using SYBR Green PCR master mix (Sigma) and the primers are listed in Supporting Information Table 1. All samples were analyzed using an AB7900HT real-time analyzer and were normalized to multiple housekeeping genes expression (RPL13a, SDHA, GAPDH, TBP, and G6PD).

Immunocytochemistry and Confocal Microscopy

hESCs were cultured on Matrigel-covered glass slide flasks (SlideFlask, NUNC, Roskilde, Denmark, www.nuncbrand.com). Cells quickly washed (thrice) with phosphate-buffered saline (PBS), prior to being fixed with 2% formaldehyde for 10 minutes and permeabilized with 0.1% Triton X-100 in PBS for 15 minutes at room temperature. Unspecific binding was blocked by incubation of samples in PBS containing 5% normal goat serum for 40 minutes. Staining with mouse monoclonal anti-phospho histone H2A.X (Ser 139; Millipore, Watford, U.K., www.millipore.com) was carried out as described before [20]. Slides were examined using a Zeiss confocal microscope (Carl Zeiss, Jena, Germany, http://www.zeiss.com). Quantification was performed by counting γ-H2A.X-positive foci in 150–200 nuclei per experiment. Details of the colocalization of RAD52 with BRCA1 are given in Supporting Information Annex A.

The images were acquired with a Leica TCS SP2 AOBS (Leica Microsystems Heidelberg GmbH, Mannheim, Germany) inverted laser scanning confocal microscope using a ×63 Plan-Apochromat-Lambda Blue 1.4 N.A. oil objective. The excitation wavelengths for fluorochromes were as follows: 488 nm (Argon laser) for fluorescein isothiocyanate (FITC), 594 nm (HeNe laser) for Texas Red, and 405 nm (Blue Diode) for 4′,6-diamidino-2-phenylindole (DAPI). The emission apertures for fluorescence detection were 500–565 nm for FITC, 605–675 nm for Texas Red, and 415–475 nm for DAPI. Two-dimensional pseudocolor images (255 color levels) were gathered with a size of 1,024 × 1,024 pixels. All confocal images were acquired through sequential scan mode and using the same settings. The distribution of fluorescence was analyzed using the Leica Confocal Software “Leica Lite” version 2.61.

Measurement of DNA Damage by the Comet Assay

The DNA damage was determined using the comet assay at 48 hours post-transfection of siRNAs using TREVIGEN Comet assay kit (Trevigen, Gaithersburg, MD, www.trevigen.com). The detailed protocol is shown in Supporting Information Annex A. Slides were viewed using the ×20 objective of a Zeiss Axioskop microscope equipped with epifluorescence optics. For each sample 100 nuclei were analyzed within at least five different zones with same surface area.

Apoptosis Assay

Apoptosis was detected using the Annexin-V-PE apoptosis detection kit (BD Bioscience, Oxford, U.K., www.bdbiosciences.com). Cells were harvested using Accutase, washed twice with ice-cold phosphate-buffered saline and counted. A total of 1 × 105 cells were suspended in 100 μl of 1× binding buffer then 5 μl of annexin V–FITC, and 5 μl of propidium iodide solution were added. The mixture was vortexed gently and incubated for 15 minutes at room temperature in the dark. A total of 400 μl of 1× binding buffer was added and the cells analyzed by flow cytometry (FACS Calibur, Becton Dickinson, Oxford, U.K., www.bd.com).

Cell Cycle Analysis

hESCs were collected using Accutase (Chemicon, Temecula, CA, www.millipore.com). Cell cycle analysis was performed using the CycleTest Plus DNA reagent kit (Becton Dickinson) with FACS Calibur (BD Biosciences). The data were analyzed using FlowJo software to generate percentages of cells in G1, S, and G2/M phases.

Measurement of Cell Proliferation Using EdU Incorporation Method

hESC proliferation was measured by incorporation of EdU (Invitrogen) into the genomic DNA during the S-phase of the cell cycle. hESCs were incubated and processed with a Click-iT EdU Pacific Blue Flow Cytometry assay Kits (Invitrogen A10034) according to the manufacturer's protocol. Flow cytometry analysis was carried out using LSRII (BD) and data processed by FacsDiva software.

Statistical Analysis

Two-tailed pairwise Student's t test was used to analyze the results obtained from two samples with one time point. The results were considered significant if p < .05.

RESULTS

Downregulation of CDK2 Causes Activation of the G1/S Checkpoint in ESC

We achieved 77%, 69%, and 63% knockdown of CDK2 expression using the protocol detailed in [5] (Supporting Information Fig. 1A) and confirmed a significant decrease in CDK2 activity by kinase activity assays (Supporting Information Fig. 1B). This led to cessation of DNA replication (Supporting Information Fig. 1C) coupled to increased levels of phosphorylated ATM (serine 1981) and phosphorylated CHK2 in addition to total CHK2 protein (Fig. 1A) consistent with the activation of ATM/CHK2 checkpoint signaling. The total level of ATM did not change and only the phosphorylated form showed an increase as early as 2 days after CDK2 downregulation (Fig. 1A) although this is consistent with published reports showing activation of ATM through intermolecular autophosphorylation [21]. Conversely, total CHK1 protein decreased, whereas the serine 345 phosphorylated form showed a very slight upregulation within 48 hours of CDK2 knockdown (Fig. 1A). Levels of p53 and both its serine 15 and serine 20 phosphorylated forms increased and total CDC25A decreased (Fig. 1B) in accordance with activation of ATM and CHK2. Similarly, p21 (Fig. 1B), p27 (Fig. 1B, [5]), and p38 mitogen-activated protein kinase (MAPK) (Thr180/Tyr182, data not shown) all showed increased levels. The p38 MAPK stress activated kinase induces p21 mRNA stabilization in response to DNA damage at the G1/S checkpoint [22] and the p27/pRb pathway is downstream of p38 MAPK signaling independently of the canonical ATM/p53/p21 pathway where it is believed to assist the maintenance of genomic integrity under prolonged exposure to DNA breaks [23]. In view of this, the upregulation of p27 and p38 MAPK (Thr180/Tyr182) may suggest a later event (it occurs only 4 days post siRNA transfection) that consolidates the maintenance of the p53/p21 cellular response. It must be noted that transcript levels of proteins active in the G1/S checkpoint did not always correlate with the level of proteins present after CDK2 knockdown (Supporting Information Fig. 2).

Figure 1.

Knockdown of CDK2 in human embryonic stem cell (hESC) triggers activation of ATM checkpoint signaling. (A, B): Western blot analysis showing activation of ATM checkpoint signaling and phosphorylation of its substrates (p53, CHK1, CHK2) and upregulation of cell cycle regulators (p27, p21, and CDC25A) at days 2, 4, and 6 post-transfection of CDK2 and control siRNAs. β-actin served as a loading control for all Western blots. The data shown are representative of at least three independent Western blots carried out in both hESC lines. Abbreviation: ATM, ataxia telangiectasia mutated.

To confirm the involvement of ATM activity in checkpoint activation, we used a specific ATM kinase inhibitor, CGK733 (Fig. 2A) which caused a great reduction in levels of phosphorylated forms of ATM, CHK2, and p53 (Fig. 2B). Application of this inhibitor in control siRNA-treated cells did not result in any changes in cell cycle regulation (Fig. 2C, 2E); however, application of this inhibitor after CDK2 knockdown caused the abrogation of cell stalling in G1 (Fig. 2D, 2F), arguing that the observed CDK2-G1-related cell accumulation is mediated by ATM activation. Together, these data suggest activation of the ATM checkpoint signaling cascade on downregulation of CDK2 in hESC, which corroborates data obtained in cancer cell lines [18, 19, 24].

Figure 2.

CDK2-induced G1 checkpoint activation is achieved via ATM-CHK2-p53-p21 pathway. (A): Schematic representation of the experimental strategy used to investigate the role of ATM inhibitor, CGK733 in CDK2-induced checkpoint activation in human embryonic stem cell (hESC). (B): Western blot analysis showing inhibition of ATM-CHK2-p53 pathway. The data represent at least three independent experiments. β-actin served as a loading control for all Western blots. (C): Impacts of ATM inhibitor on CDK2-induced G1 stalling in hESC demonstrated by flow cytometry analysis. ModFit analysis of the cell cycle distribution (propidium iodide staining) of hESC transfected with control and CDK2 siRNA in hESC in absence (B, C) and presence of ATM inhibitor (D, E). Percentage of cells in each phase of the cell cycle is indicated on the top right hand corner of each flow histogram. The data represent at least three independent experiments. Abbreviations: ATM, ataxia telangiectasia mutated; DMSO, dimethyl sulfoxide.

Downregulation of CDK2 in Human ESC Induces DNA Damage and Activates DDR

If CDK2 plays a role in DNA repair, it might be inferred that CDK2 knockdown could lead to increased accumulation of DNA damage especially in the light of studies showing that irradiated cells lacking CDK2 show prolonged cell cycle arrest and are slow to resume DNA repair [25, 26]. DNA damage levels were measured by single-cell gel electrophoresis (comet) assays in control and CDK2 siRNA transfected cells. At day 2 post-transfection, 61.4% of the hESC transfected with CDK2 siRNAs showed comet-like tails compared with cells treated with control siRNAs (Fig. 3A, 3B), suggesting higher DNA damage on CDK2 knockdown. Further analysis at day 4 post-transfection again indicated higher DNA damage in CDK2 siRNA-treated cells, albeit the level of damage was lower in both control and CDK2 siRNA-treated cells (Fig. 3C, 3D) most probably due to ongoing DNA repair.

Figure 3.

Downregulation of CDK2 by RNAi induces double-strand breaks (DSBs) in human embryonic stem cell (hESC) as revealed by the comet assay. hESC transfected with either control or CDK2 siRNAs were subjected to single-cell gel DNA electrophoresis assay (comet assay) on day 2 (A, B) and day 4 post-transfection (C, D). The percentage of comet tails as a marker of DSBs is shown on the top right-hand corner of each panel (magnification, ×200).

Activation of ATM by double-strand break (DSB) formation induces the histone variant H2A.X (a critical determinant of the DNA damage response, [27]) as well as complexes involved in resolving replication such as the Mre11/Rad50/NBS1 complex [8]. Downregulation of CDK2 by RNA interference resulted in higher levels of total H2A.X at 48 hours post-transfection of CDK2 siRNAs as well as the phosphorylated form of H2A.X (Serine 139: γ-H2A.X phosphorylated form) at all time points examined (Fig. 4A). DSBs are normally visualized by the formation of γ-H2AX foci as these two events show close (1:1) correlation [28]. Greater numbers of γ-H2A.X foci per nucleus were present in CDK2 siRNA-treated cells compared with controls, which showed a background level of around 1 foci per nucleus, similar to human fibroblasts (Fig. 4B, 4C). In addition, the DNA damage is sustained over the time course of this investigation as in hESC treated with CDK2 siRNA, the average number of γ-H2A.X foci per nucleus is higher at day 10 than at day 2 post-transfection.

Figure 4.

CDK2 downregulation triggers activation of H2.AX and formation of RAD52/BRCA1 foci in human embryonic stem cell (hESC). (A): Western blot analysis showing induction of γ-H2A.X in hESC at days 2, 4, and 6 post-transfection of CDK2 siRNAs. (B): Confocal microscopy analysis showing the presence of γ-H2A.X foci in hESC at 2, 4, 6, 8, and 10 days post-transfection of control and CDK2 siRNAs. Phosphorylated histone H2A.X (γH2A.X foci) is shown as white dots. Chromatin is stained with DAPI (blue). Scale bar = 10 μm. (C): Graphic representation of the analysis of frequencies of the number of γ-H2A.X foci per nucleus in transfected hESC during a 10-day time course post-transfection of CDK2 and control siRNAs. Data is shown as average ± SEM, n = 3. (D): Impacts of CDK2 inhibition on the regulation of key factors involved in DNA damage response analyzed by Western blotting. β-actin served as a loading control for all Western blots. The data represent at least three independent experiments. (E): Formation and colocalization of RAD52-BRCA1 foci revealed by confocal microscopy on days 2 and 4 post-transfection of CDK2 siRNA in hESC. RAD52, green; BRCA1, red; DNA, blue. Scale bar = 10 μm. The data represent at least three independent experiments. Abbreviation: NBS, Nijmegen breakage syndrome.

Increased levels of several proteins involved in DSB repair were observed in CDK2 siRNA-transfected hESC. We detected increases in the levels of total as well as Ser343 phosphorylated form of Nijmegen breakage syndrome 1 and in the phosphorylated form of BRCA1 (both proteins are involved in DSB repair by homologous recombination) on CDK2 knockdown (Fig. 4D). Conversely, a decrease in the level of Ku86 (involved in nonhomologous end joining [NHEJ]) was observed. However, we also found a strong correlation between downregulation of CDK2 and formation of BRCA1 and RAD52 foci, which was not observed in hESC treated with control siRNAs only (Fig. 4E). These data suggest that although the CDK2 siRNA-treated cells may upregulate some components of the DSB repair system, this damage is not fully repaired after 10 days post-transfection. This could be due to either continuous generation of DNA damage or low repair efficiency.

Downregulation of CDK2 Leads to Increased Apoptosis in Human ESC

Our flow cytometry analysis showed that the percentage of both early apoptotic (Annexin V+) and late apoptotic cells (Annexin V+ 7-aminoactinomycin D+) was higher in cells transfected with CDK2 siRNAs compared with control siRNA-treated cells (Fig. 5A). Repeat of this analysis at day 6 post-transfection showed lower apoptosis in both control and CDK2 siRNA-transfected cells, albeit the hESC with reduced CDK2 showed higher levels of apoptosis when compared with control siRNA-transfected cells (Fig. 5B).

Figure 5.

Downregulation of CDK2 leads to increased apoptosis in human embryonic stem cell (hESC). (A): Flow cytometry analysis of the early apoptotic (Annexin V+) and late apoptotic (Annexin V + and 7-AAD+) cells on day 2 (A) and 6 (B) after transfection of CDK2 and control siRNAs in hESC. Percentage of early and late apoptotic cells is shown on the respective quadrants. The data represent at least three independent experiments. (C): Quantitative real-time reverse transcription polymerase chain reaction for expression of key genes involved in cell apoptosis. Results are presented as average ± SEM, n = 3. The value for the hESC transfected with control siRNA was set to 1 and all other values were calculated with respect to that. Abbreviation: 7 AAD, 7-aminoactinomycin D.

Apoptosis is a recognized response of cells with excessive DNA damage especially those lacking CDK2 expression [29–33]. It has been shown that one of the mechanisms by which CDK2 may affect cell death in the presence of extensive DNA damage is by direct interaction of FOXO1 with CDK2, which results in nuclear export of FOXO1 and further activation of apoptosis-related genes such as Fas, Bclx, and Bim [33]. In accordance with these published data, we also observed increased expression of BIM1 and BAD1 on CDK2 knockdown (Fig. 5C). However, other apoptotic genes such as BCL2, CASP4, CASP9, NOXA, and FOXO1 were downregulated (Fig. 5C), suggesting that CDK2 may play a role in modulation of apoptosis in human ESC but downstream of the main caspases.

Identification of CDK2 Interacting Proteins in hESC

We performed immunoprecipitation assays using CDK2 antibody and whole hESC extract followed by mass spectrometry analysis (LS/MS/MS) with the aim of identifying CDK2 interacting proteins (Fig. 6A). Using a high stringency analysis and a threshold of 1.5% false discovery rate, we identified 423 proteins that physically interact with CDK2 (see Supporting Information Table 2). CDK2 itself was identified through mass spectrometry analysis, and this was also confirmed by our Western blot analysis on the immunoprecipitation sample (data not shown). As an additional control, we performed the same analysis on whole cell extracts without antibody to identify unspecific interactions that may arise from sticking of proteins to the agarose beads that are used in this process. None of the 423 proteins was identified in the no-antibody control analysis (data not shown), suggesting that they are highly unlikely to be unspecific interactions. To further prove this, we performed direct co-immunoprecipitation studies between CDK2 and two of the proteins identified among 423 interactors, single-stranded DNA binding protein 1 (SSBP1) and high-mobility group box 2 (HMGB2). In both cases, a direct interaction was confirmed between CDK2 and SSBP1 and HMGB2 (data not shown), suggesting true physical interactions.

Figure 6.

Identification of CDK2 interacting proteins in human embryonic stem cell (hESC) using liquid chromatography–mass spectrometry with peptide mass fingerprinting (LC/MS/MS) analysis. (A): Representative image of the silver-stained gel that was used to cut the 24 bands subjected to LC/MS/MS. Lane 1, marker; lane 2, CDK2 immunoprecipitation product, and lane 3, whole cell extract of hESC. The two dense bands shown in lane 3 represent the heavy and low chain of the antibody used for the inmunoprecipitation. (B): Diagram of the protein ontology classification of all CDK2 interacting proteins shown in Supporting Information Table 1. (C): Schematic presentation of CDK2 interacting proteins that are important for the purpose of this study.

Bioinformatic analysis of all 423 interactors using curated information from NCBI suggested that a large fraction of those cluster under the RNA processing and stability, protein metabolism/signal transduction, and cell metabolism (Fig. 6B). Of great interest, for this article, are the candidates that cluster within the cell cycle progression, DNA replication, DNA repair, cell motility and cytoskeleton, and transcriptional and chromatin modulation and are summarized in Figure 6C and Supporting Information Table 2. These interactions of course remain to be confirmed experimentally in hESC. Notwithstanding this, the mass spectrometry data taken together suggest multiple roles for CDK2 in addition to the well-known functions in cell cycle regulation.

DISCUSSION

The principal question addressed in this study is whether CDK2 depletion activates G1 to S checkpoint signaling. Recent publications have suggested that hESC possess functional G1 and G2 checkpoints that are activated by DNA damage [16, 34] and that may operate by downregulation of CDK2, but further study of this mechanism was needed. We have investigated this question in more detail and made the observations summarized as follows: (a) the G1 checkpoint is activated in hESC in response to CDK2 downregulation; (b) activation of the G1 checkpoint under these conditions results in phosphorylation of ATM and CHK2 and induces both the rapid (degradation of CDC25A) and slow response (p53 and p21 activation) as well as the activation of p38 MAPK and p27; (c) G1 checkpoint activation is an upstream event that forces hESC stalling in G1 in response to CDK2 downregulation; (d) activation of the p53 and p21 axis is not perturbed in the absence of CDK2; (e) CDK2 downregulation results in almost complete block of replication; (f) downregulation of CDK2 results in more extensive DNA damage although it does not stop the formation of DNA repair foci and activation of key factors involved in DNA repair; (g) downregulation of CDK2 results in increased apoptosis at the early time points after siRNA transfection, possibly to protect genomic stability whereas the checkpoint activation is taking place and enabling activation of the DNA repair machinery, although not sufficiently enough to clear the damage (these data are summarized in Fig. 7).

Figure 7.

Schematic figure representing the role of CDK2 in G1 to S checkpoint activation and downstream effects on cell replication, apoptosis, and DDR. Abbreviations: ATM, ataxia telangiectasia mutated; DDR, DNA damage response; p38MAPK, p38 mitogen-activated protein kinase.

The siRNA knockdown methodology employed to reduce CDK2 expression results in activation of the ATM-CHK2-p53-p21 pathway, but this system remains active for some time after the siRNA oligonucleotides would be expected to be cleared from the cell. Activation of this pathway is noticeable even 6 days post-transfection with CDK2 siRNAs, although as shown in our previous publication the cell cycle reverts to normal by this time [5]. It is possible that this effect results from the long-term activation of p21 and mitochondrial dysfunction as shown recently [35] and/or activation of p38 MAPK, which has been suggested to consolidate the p53/p21 cellular response.

In this article, we indicate a complete block in DNA replication as result of CDK2 downregulation and one of the more interesting questions arising from this is whether the block in DNA replication itself a consequence of cell stalling in G1 or an indicator of direct involvement of CDK2 in DNA replication process? Although we cannot currently discriminate between the two possibilities, our experimental evidence places checkpoint activation upstream of cell stalling in G1. Whether this is caused by a DNA replication defect is currently unclear, but evidence from work carried out in human cancer cells supports a direct role for CDK2 in the DNA replication process and further checkpoint activation. For example, it has been shown that both CDK2 and CDK1 phosphorylate several minichromosome maintenance (MCM) factors, which are core components of a helicase complex and in their hypophosphorylated state help to unwind the DNA and establish a replication fork [36]. Loss of CDK2 function in G1 phase of the cell cycle results in reduced and suboptimal MCM4 phosphorylation in human ovarian cancer cell lines, thus, causing increased loading of the MCM complex onto chromatin and activation of checkpoint response [37]. Our proteomic analyses suggest that CDK2 interacts directly with several MCM core factors, including MCM2, MCM6, MCM7 as well as other helicases such as DHX9, SMARCA5, and replication factors including proliferating cell nuclear antigen. Together, these data suggest a likely and direct role for CDK2 in the DNA replication process in human ESC and further work within our group is directed at understanding the role of CDK2 in establishment of the replication fork, activation of helicases, and subsequent effects on checkpoint activation.

There is evidence to suggest that CDK2 plays an indirect role in DNA repair by activating several of the proteins involved in the repair pathway, and this is supported by a number of studies in cancer cell lines showing that CDK knockdown or inhibition of their activity, delays the normal activation of DNA damage signaling, slows the rate of DNA repair, and prolongs the persistence of DSBs [26]. This response is enhanced when Chk1 function is compromised, especially under conditions of replication stress [38]. CDK2/Cyclin A can phosphorylate BRCA2 in vitro [39] and our data indicate that CDK2 directly interacts with XRRC5 and the catalytic subunit of DNA-PKC, which together with XRCC5 participates in DNA repair via NHEJ as well as HMGB1 and HMGB2 that are known to be involved in the final steps of the DNA end joining process. Reduced levels of CDK2 may impair the activation of these proteins, however, CDK2 is also known to function in the chromatin decondensation steps required to allow the DNA repair complexes to access the site of damage [40]. One of the CDK2 interacting proteins highlighted by the proteomic study, SMARCA 5 is involved in remodeling chromatin structures and in epigenetic regulation of transcriptional activity [41], however, further studies are needed to confirm any functional interaction with CDK2.

At first glance, there is some disparity between our observed increase in γ-H2A.X expression following CDK2 knockdown and the current understanding of histone gene expression. Human histone gene expression including that of H2A.X is controlled at the level of transcription initiation and is dependent on the G1/S-phase transition by the Cyclin E/CDK2-mediated induction of p220(NPAT)/HiNF-P complexes [42–44]. Moreover, increased γ-H2A.X expression is usually observed only in cells that are undergoing DNA replication [45]. Given that CDK2 downregulation causes cell stalling in G1, it is hard to reconcile the increased expression of γ-H2A.X with the absence of S-phase in these cells. Our hypothesis is that histone gene transcription is deregulated in the absence of CDK2 or that CDK2 may directly phosphorylate H2A.X. Our proteomic analysis (see next section) indicates direct interaction between CDK2 and some histone clusters such as H2AH and H2BE. However, these hypotheses need to be addressed experimentally, and this work is currently ongoing in our groups.

Our observations of enhanced apoptosis soon after CDK2 knockdown and a possible interaction between CDK2 and caspase 3 leads us to speculate that CDK2 may have a possible role in the execution of apoptosis. Given that apoptosis is reduced at later time points after CDK2 siRNA transductions when DDR is activated, there is also the possibility that role of CDK2 in apoptosis may also be linked to DNA repair. Such a role for CDK2 is speculative with the data available but is worthy of further investigation.

CONCLUSION

In summary, we like to conclude that CDK2 plays a number of important roles in hESC ranging from cell cycle regulation to checkpoint activation, DNA repair, and DNA replication. Although current knowledge does not explain the mechanisms by which CDK2 is involved in this multiplex process, we have started to make some inroads toward answering some of these questions by using a proteomic approach that has identified CDK2 interacting partners involved in each of these cellular events. We believe that further understanding of these cellular events will be important for understanding how pluripotent stem cells coordinate the maintenance of self-renewal, pluripotency, and genomic stability in culture for prolonged periods of time.

Acknowledgements

We are grateful to Conselleria de Sanidad (Generalitat Valenciana) and the Instituto de Salud Carlos III (Ministry of Science and Innovation). We thank Alberto Hernandez Cano for help with confocal microscopy analysis, Ian Dimmick, Rebecca Stewart, Joseph Collin, and Alicia Martínez-Romero for their help with flow cytometry, and Dennis Kirk for general technical support. We also thank Ms A. Khnykina for help with figure preparation. This study was supported by awards from BBSRC/EPSCR(CISBAN) and funds from Newcastle University.

Disclosure of Potential Conflicts of Interest

R.J. is a cofounder of MS Bioworks. The other authors indicate no potential conflicts of interest.

Ancillary