High-Fidelity Reprogrammed Human IPSCs Have a High Efficacy of DNA Repair and Resemble hESCs in Their MYC Transcriptional Signature

Human induced pluripotent stem cells (hiPSCs) are reprogrammed from adult or progenitor somatic cells and must make substantial adaptations to ensure genomic stability in order to become “embryonic stem cell- (ESC-) like.” The DNA damage response (DDR) is critical for maintenance of such genomic integrity. Herein, we determined whether cell of origin and reprogramming method influence the DDR of hiPSCs. We demonstrate that hiPSCs derived from cord blood (CB) myeloid progenitors (i.e., CB-iPSC) via an efficient high-fidelity stromal-activated (sa) method closely resembled hESCs in DNA repair gene expression signature and irradiation-induced DDR, relative to hiPSCs generated from CB or fibroblasts via standard methods. Furthermore, sa-CB-iPSCs also more closely resembled hESCs in accuracy of nonhomologous end joining (NHEJ), DNA double-strand break (DSB) repair, and C-MYC transcriptional signatures, relative to standard hiPSCs. Our data suggests that hiPSCs derived via more efficient reprogramming methods possess more hESC-like activated MYC signatures and DDR signaling. Thus, an authentic MYC molecular signature may serve as an important biomarker in characterizing the genomic integrity in hiPSCs.


Introduction
Although human induced pluripotent stem cells (hiPSCs) resemble hESCs in many respects [1,2], the therapeutic utility of hiPSCs is limited by low reprogramming efficiency [3][4][5][6] and poor genomic integrity [7][8][9][10]. A deeper understanding of the mechanisms that control these roadblocks will be vital for the use of hiPSCs in regenerative medicine. Reprogramming efficiency is controlled by intrinsic and extrinsic microenvironmental factors that are determined by the method employed [5]. Standard protocols often utilize inefficient and potentially mutagenic retroviral mediated transgene factor expression (e.g., OSKM: OCT4, SOX2, KLF4, and C-MYC, or OSNL, i.e., OCT4, SOX2, NANOG, and LIN28) [11]. More clinically useful nonviral, nonintegrating methods have also been widely employed (e.g., plasmids, microRNA), albeit with a significantly reduced reprogramming efficiency [11]. The choice of somatic donor and in vitro microenvironmental conditions also significantly influences reprogramming efficiency. For example, we previously demonstrated that bone marrow stromal cell (MSC) activation robustly activated MYC complex-regulated genes of pluripotency that subsequently facilitated high-quality reprogramming of human myeloid progenitors (MP) differentiated from CD34 + hematopoietic stem-progenitor cells [12]. Activation of MYCregulated factors potentially enhanced the rate and efficiency of reprogramming [13]. MYC may also play a key role in regulating promoters and microRNAs associated with core pluripotency-associated genes [14,15]. These findings implicate targets of the MYC network not only in playing a

Materials and Methods
Ethics Statement (Human Embryonic Stem Cell Lines). All hESC lines used in this study were obtained commercially from the WiCell Research Institute (Wisconsin International Stem Cell Bank, WISCB). The use of all WISCB-donated hESC lines in these studies was approved by the Johns Hopkins Embryonic Stem Cell Research Oversight (JHU-ESCRO) Committee and the University of Maryland School of Medicine Embryonic Stem Cell Research Oversight Committee (UMSOM-ESCRO) and conforms strictly to standards of both institutions, including written informed consent. All experiments conducted in these studies also conformed to guidelines outlined for hESC and pluripotent stem cell research by the National Academy of Sciences and the National Institutes of Health (NIH).

Gene Expression Microarrays.
Details of the microarray analysis were described before [12]. Human HT-12 Expression BeadChip arrays (Illumina, San Diego, CA) were used for microarray hybridization to examine the global gene expression of hESC, hiPSC, and starting populations (CD34 + progenitors and fibroblasts). The NIH Gene Expression Omnibus has issued the accession numbers GSE44425 (Figure 1,   Figure 1: CB progenitors and CB-derived iPSCs closely resemble hESCs DNA repair gene expression signature. Microarray gene expression of selected DNA repair genes. (a)(i) Shown are hierarchical clustering heatmaps of mRNA from donor fibroblasts, donor CD34 + CB, and CB.iPSC derived with (+MSC) and without (−MSC) bone marrow stromal cell activation. hiPSC lines included sa-CBiPSC derived from stromal-activated CD34 + MP ( = 6; E5C3, E12C5, and E17C6: 6.2, 6.13, and 19.11), standard CB-iPSC, lines derived from CD34 + MP without stromal activation ( = 3, E17C1, E20C2, and E24C1), and standard CB-iPSC lines derived from CB unsorted mononuclear cells ( = 3, iCB9, iCB8, and iCB2.5). hESC lines included ( = 3) H9, H7, and ES03. Signal intensities are from averaged independent biological replicate microarray samples ( as indicated). Expression array data depicts normalized values of the mean transcript levels for a subset of DDR genes in each group of the indicated cell lines.

DNA Damage, Apoptosis, and MYC Inhibition Studies.
For irradiation (IR) studies, pluripotent stem cells were exposed to X-ray radiation using a Pantak HF320 X-Ray machine (250 kV peak, 13 mA; half-value layer, 1.65 mm copper) at a dose rate of 2.4 Gy/min. For experiments involving MYC inhibitor (10058-F4, Sigma-Aldrich, St. Louis, MO), the cells were treated with either the control solvent (DMSO) or the drug at dose of 50 M for 24 h before X-ray IR. Following 24 h treatment, the medium was replaced before exposure to IR. For knockdown studies, siMYC (ON-TARGETplus6, Dharmacon, Thermo Fisher Scientific) was utilized. The cells were transfected with siMYC (2 g) using Lipofectamine6 2000 (Life Technologies), 48 h before exposure to IR.

In Vitro NHEJ Assays (Plasmid Reactivation: PUC18
and EJ5-ISce1). We used the DNA repair fidelity assay (PUC18-based) as described before [30]. For the assay, 2 g of EcoRI linearized PUC18 was incubated with 4 g of nuclear extract. Reactions (in 20 L volume) were carried out in ligation buffer (50 mM triethanolamine-HCl (pH 7.5), 60 mM KOAc, 50 M deoxynucleotide triphosphates, 2 mM ATP, 1 mM DTT, and 100 g/mL BSA). The mixture was incubated for 16 h at 18 ∘ C. Following the incubation, 10 ng of purified plasmid DNA was used to transfect Escherichia coli strain DH5 . Transformed cells were plated on LB agar plates, including 100 g/mL carbenicillin, 20 mg/mL X-gal, and 200 mg/mL isopropyl-1-thio--d-galactopyranoside. To allow for spontaneous rejoining/incomplete EcoRI cutting, assay controls were conducted without nuclear extract. In addition, to correct for bacterial plating numbers and determine whether nuclease activity was affecting plasmid efficacy, cells were plated on Luria-Bertani agar without carbenicillin.
For the EJ5-Isce1 assay, we used a protocol adapted from the one designed by Bennardo and colleagues but modified for in vitro plasmid reactivation analysis [31]. Briefly, the pimEJ5GFP reporter plasmid (Addgene Plasmid 44026) [31] was enzymatically linearized with I-Sce1 (New England Biolabs (NEB), Ipswich, MA) at 37 ∘ C overnight. Linearized plasmid was dephosphorylated using Shrimp Alkaline phosphatase (SAP) (NEB), and column 500 ng DNA was incubated with dialyzed nuclear extracts (2 g) of respective cell lines, and ligation reactions were performed in ligation buffer (10x T4 ligase buffer, 2 mM ATP, and 50 M deoxynucleotide triphosphates). Following in vitro ligation, the plasmid DNA was column-purified and GFP genes were PCR-amplified using the primers p1 (Fwd) 5 -CTGCTAACCATGTTCATGCC-3 and p2 (Rev) 5 -AAGTCGTGCTGCTTCATGTG-3 , as described by Bennardo et al. [31]. Following the PCR, we redigested plasmid with I-Sce1 to differentiate between NHEJ repair that was completed with I-Sce1 restoration (S+) and repair completed with loss of I-Sce1 site (i.e., "S−" with deletions). Undigested and digested PCR products were fractionated on 2% agarose and visualized with the GelStar6 Nucleic Acid Stain (Lonza). S-fragment was excised from the gel and cloned into PCR2.15 TOPO (Life Technologies). Cloned products were transformed into OneShot5 TOP10 chemically competent cells (Life Technologies) and plated on LB plates with kanamycin resistance. DNA from 5 colonies from each experiment was sequenced using the M13 primers. A total of 15 colonies were analyzed from three independent experiments, and TOPO plasmids were sequenced at the UMB TGL/Biopolymer core facility.

CB Progenitors and CB-Derived iPSCs Closely Resemble hESCs in DNA Repair Gene Expression Signature.
Previous studies indicated that progenitor donor cells were more amenable than differentiated cells in reprogramming to a pluripotent state [32,33]. We performed microarray-based analysis to determine the DDR gene expression profile of hiP-SCs (Table S1) derived via different methods (Figure 1(a)(i)). We found that donor CD34 + CB progenitors cluster more closely with hESCs than adult fibroblasts (Ad.Fib) donors in baseline expression of DNA repair genes, including poly (ADP-ribose) polymerase 1-PARP1 (involved in single-strand break repair and DSB repair), XRCC5 (a.k.a. Ku80), and XRCC6 (a.k.a. Ku70) (involved in NHEJ DSB repair). Of note, expression of MYC and XRCC6 in CB progenitors was even higher than that for hESCs (Figure 1(a)(i)) [20]. Additionally, PARP1 and XRCC5 were expressed at higher baseline levels in sa-CB-iPSCs than in standard CB-iPSC lines (Figure 1(a)(ii)).

Stem Cells International 5
To determine whether the levels of expression of these repair gene transcripts translated into functional differences in protein levels, we performed immunoblot analyses on hiPSCs from these representative categories. Although steady state protein levels of ATM, Ku80, and PARP1 in sa-CB-iPSCs were similar to standard CB-iPSCs and hESCs, donor CD34 + CB progenitor baseline expression of these DNA repair proteins more closely resembled hESCs ( * < 0.05), compared to Ad.Fib ( * * < 0.01 difference) ( Figures  1(b)(i) and 1(b)(ii)). These results suggested that CD34 + CB progenitors may already possess hESC-like expression of DDR pathway components, even prior to initiation of reprogramming.

Sa-CB-iPSCs Resemble hESCs in Their DDR Response to
Radiation. Irradiation (IR) elicits several posttranslational modifications of the components of DDR pathway. Irradiated hESCs and hiPSCs rapidly activate the ataxia telangiectasia and Rad3-related (ATR) and ataxia telangiectasia mutated (ATM) kinase-dependent DDR signaling [34], phosphorylating targets, such as p53 and H2AX [17,35]. While ATR responds mainly to single strand breaks (SSBs) and stalled replication forks, ATM is activated in response to DSBs. Moreover, ATM deficiency confers hypersensitivity to IR [36].
We next examined the activity of another ATM target, the tumor suppressor p53, whose expression is stabilized upon DNA damage, thus activating the DNA binding function of p53. Posttranslational modification of p53 via phosphorylation at Ser15 has been proposed to be an important mechanism by which p53 is stabilized and its functions are regulated [37]. However, phosphorylation is not an absolute necessity for DNA damage-induced stabilization of p53 [37]. Our results show that P53 activation, measured by monitoring total p53 protein and phosphorylation at Ser15, occurred with similar kinetics in all the hiPSCs and hESCs, with levels increasing between 0 and 4 h after IR (Figures 3(a)-3(c)).
Moreover, standard hiPSC lines (e.g., iCB9 and iHuF3) consistently displayed higher baseline levels of total p53 protein in untreated controls, in comparison to hESC (H9), ESO3, and sa-CB-iPSC (CB6.2) (Figures 3(a)-3(d)). In our observation, activation of p53 in cells following IR is mostly contributed by the stabilization of total p53 protein, as the relative changes in levels of phosphorylated protein were insignificant when its expression is normalized to total p53 (except for CB6.2 (2 h), < 0.05) ( Figure S2).
We next investigated apoptotic responses of hiPSC cell lines to IR-induced damage. All pluripotent stem cells have been reported to exhibit hypersensitivity to radiation, with substantial cell death observed within 24 h after exposure to a low dose of ionizing radiation (1-2 Gy IR) [17,20,34]. We therefore reasoned that cells with higher levels of cytotoxic DSBs may induce apoptosis to avoid genotoxic stress. Using PARP1 cleavage as an apoptotic marker, IRexposed cells were examined by immunoblotting. Notably, there were only subtle differences observed in the kinetics of PARP1 cleavage among hESCs and all hiPSCs. PARP1 was observed predominantly in the cleaved form 4 h after IR in all examined cell lines (Figures 3(a), 3(b), and 3(d)). These results indicated that despite the subtle differences in levels of DNA damage, reprogramming renders all hiPSCs equally hypersensitive to ionizing radiation-induced apoptosis.

Sa-CB-iPSCs More Closely Resemble hESCs in Nonhomologous End Joining (NHEJ) Repair.
Differences in baseline levels of DNA damage markers between hiPSCs noted above may also be accounted for by differences in DSB repair [17]. For example, increased DSB formation could result from decreased efficiency of repair, which can lead to increased error-prone repair or misrepair. Thus, we next determined whether the CB.iPSCs derived with the same factors but using distinct episomal reprogramming methods demonstrated different DSB repair efficiencies. We employed an established end-joining plasmid-reactivation repair assay and observed that hESC H9 and sa-CB-iPSC CB6.2 displayed the lowest percentage of misrepair (approximately 3%). In contrast, standard hiPSCs iHuF3 and iCB9 possessed a significantly higher percentage of misrepair (approximately 8-12%; * < 0.05), when either compatible DSB ends or noncompatible DSB ends (which require additional processing steps in end joining) were used (Figures 4(a)(i) and 4(a)(ii)). To further confirm these results, we utilized an additional modified endjoining assay designed by Gunn and Stark [38] that measures DSB repair junctions representing repair of complementary or noncomplementary ends ( Figure S1). We incubated I-Sce1-linearized pimEJ5GFP plasmid with nuclear extracts of pluripotent cell lines for measurement of in vitro plasmid reactivation (Figure 4(b)(i)), and the I-Sce1 resistant fraction ("S−" products) was further analyzed for quantification and characterization of DNA deletions (Figure 4(b)(ii)). Sequencing of approximately 10-15 "S−" DNA clones recovered from end-joining experiments using H9 and CB6.2 extracts indicated that deletions in the DSB junctions were mainly in I-Sce1 overhangs and were restricted to 1-5 nucleotides (nt) (33% and 54%, resp.). In contrast, only 1 out of 11 (9%)   junctions that were recovered from iCB9 extracts showed 1-5 nt deletions ( * < 0.05, compared to H9, CB6.2), 64% of the junctions had 6-9 nt deletions, and ∼27% of the junctions had >20 nt deletions. Strikingly, we observed that 70% of junctions (7 out of 10) recovered from iHuF3 had deletions >20 nt ( * < 0.05, compared to H9, CB6.2, and iCB9) (Figure 4(b)(ii)). This confirms that DNA end joining in sa-CB-iPSC CB6.2 more closely resembles that of hESCs and is less error-prone, compared to end-joining in the fibroblastderived standard hiPSCs.

C-MYC Maintains the DDR and NHEJ in hESCs and Is
Required for Less Error-Prone Repair in sa-CB-iPSCs. MYC modules, along with Core and Polycomb group genes, represent key gene circuits that contribute to the ES cell expression signature [39]. C-MYC depletion from the reprogramming   cocktail significantly reduces the efficiency of reprogramming [40]. Interestingly, sa-CB-iPSCs were characterized by hESC-like MYC-regulated expression module and robustly expressed MYC complex genes [12]. In a different context, C-MYC has also been shown to regulate the transcription of several key DSB repair genes including Ku70 and BRCA1 in somatic cells [26]. Therefore, we questioned whether C-MYC contributes to enhanced efficacy and efficiency of repair in hESCs. As a proof of principle, C-MYC was depleted in hESC H9, using chemical inhibitor 10058-F4, which prevents MYC/MAX association and downstream signaling [41]. Following C-MYC inhibition (50 M, 24 h), the control and drug-treated cells were exposed to IR (1 Gy) and cells were examined at 0, 1, 2, and 4 hours after IR for expression of DDR proteins by immunoblotting. Notably, compared with untreated cells, C-MYC inhibition resulted in increased levels of H2AX 1 h after IR and persists until 4 h after IR (Figures 5(a)(i) and 5(a)(ii)). Whereas pATM expression changes after IR in untreated cells are more subtle, C-MYC inhibition results in persistence of pATM that decreases by 4 h. While Ku80 expression decreases with C-MYC inhibition, it is not significant compared with controls. These data suggest that C-MYC is involved in the radiation-induced DSB repair response in hESCs, facilitating repair. We therefore next determined whether MYC inhibition reduced quality and efficiency of DNA end-joining in H9 cells. Remarkably, siRNA-mediated MYC knockdown (KD) in H9 resulted in a significant decrease in total NHEJ efficiency of these cells, as measured by counting total number of colonies (blue plus white) in an in vitro PUC18 assay ( Figure 5(b)(i)). Moreover, we also observed a significant increase in the proportion of misrepaired colonies from hESC H9 cells treated with MYC siRNA (Figure 5(b)(ii)).
Since MSC activation of CB donors during reprogramming robustly activated MYC complex genes of pluripotency and facilitated high-capacity reprogramming of human MP differentiated from CD34 + cells [12], we sought to evaluate the MYC module expression networks in sa-CB-iPSC versus other hiPSC lines. Interestingly, microarray expression of MYC-regulated circuit genes in sa-CB-iPSC was more hESClike relative to standard CB-iPSC ( Figure 5(c)). We next determined whether inhibition of C-MYC affected the quality of end-joining in these categories of hiPSCs. For these experiments, we utilized I-Sce1-based assays ( Figure S1) and Stem Cells International 11 measured DSB repair in these cells in vitro (see Section 2). As shown in Figure 5(d)(i), the majority of the GFP genes amplified from the PCR reaction were S+ (I-Sce1 sensitive), indicating that these extracts mostly produced distal-end joining products that are error-free. However, to determine the character of the errors from plasmid reactivation, PCR products resistant to ISce1 restriction digestion (S−) were cloned into PCR2.1. Remarkably, similar to C-MYC depletion end-joining results in H9 ( Figure 5(d)(ii)), analysis of DSB repair junctions indicated that the efficacy of DNA endjoining significantly deteriorated and became more errorprone when C-MYC was depleted in sa-CB-iPSC (CB6.2) ( * < 0.05) ( Figure 5(d)(ii)). Specifically, while none of the 13 clones from WT CB6.2 had deletions of >20 nt, ∼ 33% of clones (5 out of 15) showed deletions of >20 nt when C-MYC was depleted ( * < 0.05). Interestingly, in WT iCB9, 20% of clones had deletions of >20 nt nucleotides that further increased following C-MYC KD (38% versus 27% in WT) ( Figure 5(d)(ii)). These results imply that C-MYC gene expression signature is linked to efficacious NHEJ DSB repair in pluripotent cells. Moreover, these data indicate that expression of C-MYC gene expression circuits in hiPSCs could be an important indicator of not only overall efficiency of reprogramming, but also overall DDR signaling and, in particular, repair of DSBs.

Discussion
Generating hiPSCs from adult cells represents one of the most exciting developments in regenerative medicine. However, potential clinical applications of hiPSCs are severely hampered by low efficiency of production and suboptimal genomic integrity. One study estimated that ∼13% of hESC and hiPSC cultures demonstrated aberrant aneuploid karyotypes [42]. Comparative genomic analyses have revealed a high frequency of DNA copy-number variations (CNVs) in hiPSCs when compared to either hESCs or somatic cells of origin [9,43]. DNA damage and inaccurate "followup" repair mechanisms likely present a significant source of genomic aberrations [44]. For example, reprogramming methods may introduce DNA lesions in the form of lethal DSBs [44]. DSB lesions are introduced by ectopic expression of reprogramming factors and appear to develop irrespective of the reprogramming methodology (i.e., integrative or nonintegrative) [44,45]. DSB repair components also play an important role in controlling the efficiency of reprogramming [44,[46][47][48][49][50]. Cells that are impaired in HR genes, such as BRCA1/BRCA2 or NHEJ factor DNA ligase IV (LIG4), show significantly decreased capacity for reprogramming [44,49]. However, it is not well understood whether the features that promote reprogramming further translate into hiPSCs with more robust and efficacious DSB repair properties.
Our study demonstrates that CB-iPSCs generated with high efficiency (sa-CB-iPSC) possess an hESC-like C-MYC transcript signature and have a DDR that more closely resembles hESCs, relative to hiPSCs derived via standard methods. Moreover, sa-CB-iPSCs also performed end-joining DSB repair with less errors, compared with standard CB.iPSCs. Notably, depletion of C-MYC led to increased end-joining errors, suggesting for the first time that MYC-regulated circuits may be required for maintaining genomic integrity in hiPSCs.
Cell differentiation leads to a decline in DNA repair capacity, which can further lead to accumulation of DNA damage and mutations [32,33]. In contrast, stem progenitors possess greater overall capacity for efficient DNA repair. Stem-progenitor cells may also be more amenable to cellular reprogramming, compared with differentiated somatic cells [12,51,52]. However, sa-CB-iPSCs derived from human myeloid progenitors through MSC activation signals are generated even more efficiently (1-4%) and possess minimal interline variability when differentiated to vascular progenitors, compared with hiPSCs derived from CB mononuclear cells generated without MSC activation (0.2-0.3%) [52]. While no significant differences in baseline expression of mRNA transcripts and translated proteins for DDR genes were observed between CB-iPSCs derived via different methods, most significant differences emerged when these cells were analyzed for their DSB repair activities. sa-CB-iPSCs exhibited end-joining repair which was less error-prone and more closely resembles DSB repair properties in hESCs.
Repair of nonligatable ends by NHEJ requires an endprocessing step for ligation and thus is prone to errors resulting in deletions of a few nucleotides at DSB repair junctions. IR damage induces NHEJ-mediated DNA misrepair events in late G2 cell cycle stage [53]. Interestingly, ATM suppresses genomic aberrations and incorrect end utilization during NHEJ, known as "distal-end joining," formed as a consequence of multiple DSBs due to genotoxic stress [54,55]. Although hESCs can uniquely employ high-fidelity NHEJ that can operate independently of ATM [56], hiPSCs perform error-prone DSB repair in particular when exposed to genotoxic stress [53]. Our studies indicate that despite similarities in levels of total ATM and ATM phosphorylation kinetics after IR, sa-CB-iPSCs and standard CB-iPSC have differences in NHEJ responses. In particular, standard fibroblast and CB-iPSCs demonstrated a higher percentage of large deletions (≥20 nt) in DSB junctions, compared to sa-CB-iPSCs and hESCs. Remarkably, "error-proneness" of NHEJ significantly escalates when pluripotent cells are subjected to IR stress under conditions of MYC inhibition.
MYC is an important regulator of transcription in hESCs and is one of the key factors employed in the generation of hiPSCs. Indeed, ectopic MYC is necessary for efficiently generating iPSCs [57,58]. MYC interacts with the NuA4 complex, a regulator of ESC identity. and is the master regulator of a key ESC transcription program [14,59,60]. MYC also activates high telomerase activity during reprogramming via regulation of TERT [61]. Hematopoietic growth factor (GF) stimulation of myeloid progenitors differentiated from CD34 + CB cells activates C-MYC-regulated modules to hESC-like levels and facilitates their pluripotency induction [12]. These GF-activated progenitors robustly overexpress MYC complex genes, which have been found to be vital for pluripotency and facilitation of somatic reprogramming [12]. Interestingly, the C-MYC module signature in ESCs highly resembles the C-MYC module that is found in cancer cells [39]. Our data reveals that hESCs and sa-CB-iPSCs have a similar C-MYC module signature. Moreover, MYC inhibition results in more repair errors in hESC and hiPSCs. Thus, while NHEJ in somatic cells is considered error-prone [19], in normal pluripotent cells, C-MYC appears to be required for maintaining a more error-free NHEJ repair. Notably, putative C-MYC binding sites have been identified in the regulatory regions of several NHEJ genes, suggesting a potential mechanism through which C-MYC may maintain error-free NHEJ in hESCs and hiPSCs [26,62].
In conclusion, our studies show that the various methods for generating hiPSCs may affect the pathways that regulate genomic integrity. Further characterization is required to determine how these pathways are interconnected and will enable improvement of the genomic integrity of hiPSCs. Knowing that C-MYC is also a master regulator of chromatin modifications [13,60], its role in facilitating repair might be not only transcriptionally regulated but also epigenetically controlled. Thus, further elucidation of the role of C-MYC in maintenance of genomic integrity, regulating the balance between "good repair" and "bad repair" in pluripotent cells, is required.