Generation of Constructs

The mouse Mnda expression construct, pUNO1-mMnda (Invivogen), which expresses Mnda under the control of hEF1α promoter, was linearized with ApaL I and then transfected into ESCs by electroporation. The shRNA expression constructs against mouse Mnda (Qiagen) were linearized with Sca I and used for transfection. To generate the Fgf4 overexpression construct with an HA epitope tag (phEF1α-Fgf4_HA), the ORF of Fgf4 was PCR-amplified from pBluescript-mFgf4 (Addgene) using primer pair 5′-GAATTCTGATG GCGAAACGCGGGCCG-3′ and 5′-GCGGCCGCTCAAGCGTAATCTGGAACATCGTA TGGGTAACCACCCAGTCTAGGAAGGAAGTG-3′ and cloned into the phEF1α expression vector, using the EcoR I and Not I restriction sites.

Cell Culture

The derivation of the mouse maGSC and the respective ESC lines from the 129Sv and Stra8-EGFP backgrounds were described previously [19]. The Fgf4-KO ESC line was a kind gift from Prof. Rizzino [20]. Mouse embryonic fibroblasts (MEFs) were grown in standard fibroblast culture conditions. To generate Mnda and Fgf4 overexpression clones (Mnda-OE and Fgf4-OE, respectively), the constructs were electroporated into ESCs and the cells were selected for neomycin resistance. To generate Mnda downregulation clones (Mnda-DN), the shRNA expression constructs against Mnda were electroporated into ESCs and the cells were selected for neomycin resistance. For differentiation experiments, the cells were cultured in medium containing retinoic acid for the indicated time points without leukemia inhibitory factor (LIF) and feeder cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. List of GO terms associated with upregulated genes in early-apoptotic cells.

https://doi.org/10.1371/journal.pone.0048869.t001

Oligonucleotide Microarray for Detection of Apoptosis-specific Gene Expression

Expression of apoptosis-specific genes in undifferentiated or differentiated maGSCs and ESCs was examined using Mouse Apoptosis OligoGEArray following the manufacturer’s protocol (SABiosciences). Briefly, total RNA extracted from maGSCs and ESCs was used to prepare cDNA followed by biotin 16-dUTP labelling using a GEArray True Amp Labelling kit (SABiosciences). The Mouse Apoptosis OligoGEArray membranes were then hybridized overnight with biotin-labelled cDNA. Signals were detected using a CDP-Star chemiluminescence kit (SABiosciences). Data were analyzed using the GEArray Expression Analysis Suite 2.0 (SABiosciences). For fold differences, an arbitrary threshold of 2 was chosen to signify a substantial fold change in either direction. All experiments were done with two independent biological samples.

Apoptosis Induction, Annexin-V Staining, and Flow Cytometry

To induce apoptosis through the mitochondrial pathway, maGSCs and ESCs were cultured in medium containing citrinin (CTN) (Sigma) or DMSO (Sigma) (as a control) for either 12 or 24 h. To verify and quantify CTN-induced apoptosis in maGSCs and ESCs, annexin-V and 7-amino-actinomycin D (7-AAD) staining was performed using an Annexin V-PE Apoptosis Detection Kit I (BD Biosciences). After staining, flow cytometric measurements were performed on a FACSCalibur flow cytometer (BD Biosciences) and analyzed with CellQuestPro software (BD Biosciences). Early-apoptotic cells were sorted using a FACSCanto II (BD Biosciences). All experiments were done in three or more independent biological replicates.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. List of GO terms associated with downregulated genes in early-apoptotic cells.

https://doi.org/10.1371/journal.pone.0048869.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Identification and characterization of novel pluripotency- and apoptosis-related genes.

(A) Heatmap representation of qPCR data of selected candidate genes in ESCs (ES RI) and adult mouse tissues. (B) The qPCR analysis confirming the downregulation of Fgf4 in early apoptotic ESCs and maGSCs. (C) The qPCR analysis confirming the downregulation of Mnda in early apoptotic ESCs and maGSCs. (D–H) Stacked bar graphs showing the percentage of viable, early-apoptotic, and late-apoptotic cells in either DMSO- or CTN-treated cells. Induction of apoptosis and analysis by annexin-V/7-AAD staining after 12 h and 24 h of CTN treatment in wild-type ESCs (D), Fgf4-OE (Fgf4-overexpressing) cells (E), Fgf4-KO (Fgf4-knock-out) cells (F), Mnda-OE (Mnda-overexpressing) cells (G), Mnda-DN (Mnda-downregulated) cells (H). The flow cytometry data of three or more independent biological replicates were calculated and represented as a mean ±SD. The data was analyzed for statistical significance and found no significant differences in D–H.

https://doi.org/10.1371/journal.pone.0048869.g004

DNA Damage Induction and Cell Cycle Analysis

For induction of DNA damage, the cells were treated with 50 ng/ml neocarzinostatin (NCS) diluted in PBS. After 30 min of treatment, the cells were washed with PBS and supplemented with culture medium. For cell cycle analysis, the cells were trypsinized after indicated time points and were washed with PBS followed by ethanol fixation at −20°C for a minimum of 2 h. After fixation, the cells were washed with PBS, resuspended in PBS containing 10 µg/ml propidium iodide (PI) and 1 mg/ml RNase A, and incubated at 37°C for 30 min. After the incubation, cells were measured on a FACSCalibur flow cytometer and analyzed after exclusion of cell doublets. All experiments were performed in three or more independent biological replicates.

RNA Extraction and qPCR

Total RNA was isolated with Nucleospin miRNA kit (Macherey Nagel) and 5 µg of DNase-I treated total RNA was used to prepare cDNA using the Superscript-II Reverse Transcriptase kit (Invitrogen). Real-time qPCR was performed in triplicate using 1 µl cDNA (1∶20 dilution) with SYBR green (Invitrogen). Primer sequences used for qPCR are listed in Table S1. The qPCR data was first normalized to two housekeeping genes, Sdha and Gapdh, and represented as relative to one of the cell type. All experiments were performed in two or more independent biological replicates consisting of three technical replicates each.

Whole Genome Microarray Analysis and Statistical Interpretation

Whole genome microarray analysis was performed essentially as described earlier [16]. Briefly, total RNA isolated from cells undergoing apoptosis that had been sorted by FACS was used to prepare cDNA. The Cy3- or Cy5-labeled cDNA was hybridized to Agilent Technologies 44K Mouse Whole Genome Microarrays (G4122A). Data intensities were extracted using the software ‘Feature Extraction 9.1′ (Agilent Technologies, Germany), and the raw microarray data were normalized with a non-linear loess regression method [21]. Differentially regulated genes were identified by ANOVA, and the resulting data were adjusted with the Benjamini-Hochberg method to control the False-Discovery-Rate [22]. All experiments were performed in three independent biological replicates.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Role of Fgf4 and Mnda during genotoxic stress response of ESCs.

(A) Stacked bar graphs showing the percentage of cells at various stages of cell cycle (SubG1/Apoptotic, G1, S and G2/M) in untreated, Control ESCs – wildtype ESCs; Fgf4-OE – Fgf4 overexpression cells; Fgf4-KO – Fgf4 knockout cells; Mnda-OE – Mnda overexpression cells; Mnda-DN – Mnda downregulation cells. (B–F) Genotoxic stress was induced by treatment with NCS for 30 min followed by recovery for indicated time points and analyzed for cell cycle parameters in Control ESCs – wildtype ESCs; Fgf4-OE – Fgf4 overexpression cells; Fgf4-KO – Fgf4 knockout cells; Mnda-OE – Mnda overexpression cells; Mnda-DN – Mnda downregulation cells. The cell cycle data of three or more independent biological replicates were calculated and represented as a mean ±SD. The values which are statistically significant are indicated with asterisks (∗p<0.05).

https://doi.org/10.1371/journal.pone.0048869.g005

Teratoma Formation Assay

Teratoma formation assay was performed as previously described [23], [24]. Briefly, the indicated cell line (1×10⁶ cells) were injected subcutaneously into 8 to 10 weeks old female immunodeficient RAG2^−/−cγc^−/− mice. Tumor growth was monitored weekly by palpation and size was recorded using linear calipers. Animals were sacrificed when a tumor diameter of 1 cm was reached. Autopsies were performed and tumor tissue was placed in phosphate-buffered 4% formalin for 16 h and then embedded in paraffin. For histological analysis, the specimens were stained with haematoxylin and eosin (HE).

Statistical Analysis

All qPCR data for RNA expression analysis (two or more biological replicates) were calculated using delta-delta-Ct method. For statistical significance calculations, 2way ANOVA or Student’s t-test were used (GraphPad Prism 4.0).

Apoptosis-related Gene Expression Profiling of ESCs and maGSCs

Genotoxic stress leads to the activation of p53-dependent mechanisms to either repair DNA damage or undergo apoptosis. The high sensitivity of ESCs to genotoxic stress is presumed to be due to their particular expression of pro-apoptotic and anti-apoptotic genes. To compare the apoptosis-related gene expression profiles of ESCs and maGSCs, we used OligoGEArray consisting of 112 genes implicated in the process of apoptosis (Fig. 1A). The comparison between undifferentiated ESCs and maGSCs of the 129Sv genetic background revealed similar gene expression patterns without any cell type-specific gene expression (Fig. 1B). The majority of the expressed genes are involved in the mitochondrial apoptotic pathway (intrinsic) and included no genes, such as Fas and Fadd, which are implicated in the extrinsic apoptotic pathway. Among all the analyzed caspases, only Casp-2 was expressed at detectable level in both cell types. Quantification of the signal intensities using the OligoGEArray suite and scatterplot analysis revealed that, although many genes are equally expressed (below the fold difference threshold of 2) in ESCs and maGSCs, the pro-apoptotic gene Bok and three anti-apoptotic/tumor suppressor genes (p53, Birc2, and Birc5) showed significant upregulation in maGSCs (Fig. 1C). We verified the expression of apoptosis and anti-apoptosis genes in our previous global transcriptome comparison of undifferentiated ESCs and maGSCs [16], and found the upregulation of Birc5, but not other genes (Fig. S1A, E). Next, we compared the apoptosis-related gene expression profile of ESCs and maGSCs generated from Stra8-EGFP transgenic mice, hereafter referred as ESCs-Stra8 and maGSCs-Stra8, respectively. The gene expression profile of ESCs-Stra8 and maGSCs-Stra8 is identical to each other (Fig. 1D) and is also similar to those of ESCs and maGSCs of the 129Sv background (Fig. 1B). The quantification and subsequent scatterplot analysis showed the slight downregulation of pro-apoptotic genes, Bnip3, Tnfrsf12a, and p53 in maGSCs-Stra8 (Fig. 1E). The opposing expression pattern of p53 between two different genetic backgrounds of maGSCs (maGSCs-129Sv and maGSCs-Stra8) prompted us to verify these differences by more sensitive method, qRT-PCR. The qPCR analysis of p53 as well as two other differentially expressed genes (Bax and Nfkb1), and two equally expressed control genes (Atf5 and Bok) revealed no significant differences between maGSCs and ESCs of different backgrounds (Fig. S2).

We also analyzed the expression profile of pro-apoptotic and anti-apoptotic genes in ESCs and maGSCs that have been induced to differentiate for 20 days with retinoic acid (RA). The expression pattern of apoptosis-related genes is quite similar in both differentiated cell types (Fig. 2A) and is markedly distinct from the expression pattern of undifferentiated cells (Fig. 1B–E). The scatterplot analysis revealed the upregulation of Nfkb1 and the downregulation of Pycard and Bnip3 genes in differentiated maGSCs in comparison to differentiated ESCs (Fig. 2B). To verify these results, we analysed the global transcriptome data of differentiated ESCs and maGSCs [16], and found no obvious differences in the expression of apoptosis and anti-apoptosis genes (Fig. S1B, F). To highlight the differences between undifferentiated and differentiated cells and also to identify the genes that are specifically expressed in undifferentiated ESCs and maGSCs; we generated a heatmap of all OligoArray data sets (Fig. 2C). All undifferentiated cells possessed similar gene expression patterns and are clustered together, whereas differentiated cells possessed an expression pattern that is quite distinct and are clustered separately (Fig. 2C). Further analysis of the heatmap revealed strong and specific expression of Mcl1, Bax, Hells, and Nfkb1 genes in undifferentiated ESCs and maGSCs, whereas the expression of these genes was highly downregulated or absent in their differentiated counterparts (Fig. 2C′). The expression of p53, Api5, Birc2, Cradd, and Bid genes was moderately high in undifferentiated pluripotent cells compared to RA-differentiated cells (Fig. 2C′). In agreement with these results, we found a significant downregulation of Hells, Nfkb1, and p53 in our previous transcriptome data of differentiated ESCs and maGSCs [16] (Fig. S1C, D and G, H). In contrast to OligoGEArray results, we found few apoptotic as well as anti-apoptotic genes (Dapk1, Ltbr, Xiap, and Bcl2l1) to be upregulated in the transcriptome data of differentiated ESCs and maGSCs (Fig. S1C, D and G, H). However, these upregulated genes were not detectable in our OligoGEArray studies, suggesting their possible low expression levels which are difficult to detect using OligoGEArray.

Citrinin-induced Apoptosis and Transcriptome Analysis of Early-apoptotic ESCs and maGSCs

The particular expression pattern of intrinsic apoptotic pathway-related genes led us to evaluate how both ESCs-129Sv and maGSCs-129Sv (hereafter referred as ESCs and maGSCs, respectively) react to the induction of apoptosis through the intrinsic pathway. Citrinin (CTN), a mycotoxin, is known to induce apoptosis through the intrinsic pathway in ESCs and to generate reactive oxygen species (ROS), which cause DNA damage and subsequent p53- induced apoptosis [25]. Because of these facts, CTN is a suitable compound to study mitochondrial as well as p53-mediated apoptosis. In this context, we induced apoptosis in ESCs and maGSCs with CTN (200 µM) for 24 h and analyzed the percentage of apoptotic cells by annexin-V and 7-AAD staining (Fig. 3A′, B′). After 24 h of CTN treatment, both ESCs and maGSCs had lost the characteristic colony morphology and appeared as typical apoptotic cells (Fig. 3A, B). Flow cytometric analysis after annexin-V and 7-AAD staining revealed a substantial presence of early-apoptotic cells (∼35% of the total cell population, annexin-V^+ve/7-AAD^−ve) in CTN-treated ESCs and maGSCs, but only ∼8–15% of un-treated and DMSO control cells displayed an early-apoptotic phenotype (Fig. 3A′, B′). The percentage of late apoptotic (annexin-V^+ve/7-AAD^+ve) and viable (annexin-V^−ve/7-AAD^−ve) cells were also significantly similar between CTN-treated ESCs and maGSCs. Overall, these results indicate that maGSCs, like ESCs, are reactive to CTN and undergo apoptosis at a similar rate.

The intrinsic-pathway related gene expression pattern seen in our OligoArray suggests that this pathway is constitutively active in pluripotent cells and might play a role in resistance to genotoxicity. Because, CTN is also known to induce DNA damage through the generation of ROS and the subsequent activation of the intrinsic apoptotic pathway, we performed transcriptome analysis using early-apoptotic cells as outlined (Fig. 3C) to identify the factors that are responsible for maintaining genomic integrity. Transcriptome analysis of early-apoptotic ESCs and maGSCs revealed that the majority of genes were equally expressed in both cell types (∼94% similarity), but 1257 (∼6%) genes were differentially expressed (Fig. 3D). Several of the pluripotency marker genes expression was slightly downregulated in early-apoptotic cells transcriptome compared to our previously described transcriptome of undifferentiated ESCs and maGSCs [16] (Table S2). Similar analysis for differentiation marker genes of three germ layers revealed no overt changes in their expression, except for downregulation of Vimentin, a mesodermal marker gene (Table S3). Comparison of early-apoptotic ESCs transcriptome with transcriptome of undifferentiated ESCs [16] showed ∼80% similarity, indicating that only ∼20% genes are differentially regulated in early-apoptotic cells (Fig. S3A). Similarly, early-apoptotic maGSCs showed ∼67% similarity to their undifferentiated counterparts and ∼33% differentially regulated genes (Fig. S3B). Further comparison of differentially regulated genes in early-apoptotic ESCs and maGSCs showed ∼44% of commonly upregulated and ∼55% commonly downregulated genes (Fig. S3C, D, and Tables S4 & S5). The functional gene ontology (GO) analysis revealed that genes associated with several biological processes such as signal transduction, cell cycle processes, response to stress, DNA damage, and DNA repair are upregulated in early-apoptotic cells compared to wild-type cells (Table 1). Interestingly, genes associated with biological processes such as regulation of transcription, oxidation reduction, intracellular signalling cascade, cell proliferation, cell death, cell adhesion are downregulated in early-apoptotic cells (Table 2). The principle component analysis (PCA) of the transcriptome of early-apoptotic cells and previously described transcriptomes of undifferentiated and differentiated cells [16] clearly showed the clustering of early-apoptotic ESCs and maGSCs, which are quite distant from their undifferentiated and differentiated counterparts (Fig. 3E). Similar results were obtained when we analyzed the transcriptome data using a hierarchical clustering model (Fig. S4).

Identification of New Pluripotent Cell-specific and Apoptosis-related Candidate Genes

The transcriptome analysis of early-apoptotic ESCs and maGSCs facilitated the identification of new candidate genes that might be specifically expressed in pluripotent cells and might play a role in genotoxicity-triggered cell death. We focused on genes with either direct or indirect role in apoptosis and are highly modulated in early-apoptotic cells (Tables S4 and S5), to characterize their role in response to genotoxicity. Further, we performed quantitative real-time PCR (qPCR) for the selected candidate genes and found downregulated candidate genes, Matrix metalloproteinase-10 (Mmp10) and Fibroblast growth factor 4 (Fgf4) to be specifically expressed in pluripotent cells but not in any of the analyzed adult mouse tissues (Fig. 4A). Another downregulated candidate gene that we selected is Myeloid cell nuclear differentiating agent (Mnda) which bears a death domain with an unknown function. Although Mnda is expressed in heart and liver, apart from pluripotent cells, it is expressed in no other mouse tissue (Fig. 4A). Expression analysis of other downregulated candidate genes (Adm, Fbln2, and Hist1h1b) revealed ubiquitous expression (data not shown), hence omitted from further analysis. Similarly, we have validated the upregulation of three upregulated candidate genes (Hist2h2be, Hspa1, and Mia2) using qRT-PCR (Fig. S5), however, these genes showed ubiquitous tissue expression (data not shown). Hence, these genes were not considered for further analysis. Based on the above results, we choose Fgf4 and Mnda for further analysis, because of their known direct or indirect association with apoptosis [26], [27], and performed qPCR to validate the downregulation seen in the transcriptome. The qPCR data further confirmed the downregulation of Fgf4 and Mnda in early-apoptotic cells (Fig. 4B, C).

Role of Fgf4 in Response to Apoptosis and Genotoxicity Induction

To study the role of Fgf4 in response to induced genotoxicity and apoptosis, we generated an ESC line with stable Fgf4 overexpression (Fgf4-OE) that expressed Fgf4 at levels approximately 15-fold higher than those found in control cells (Fig. S6A) and also used a previously described Fgf4^−/− ESC line (Fgf4-KO) [20]. Both Fgf4-OE and Fgf4-KO cell lines showed normal colony morphology and proliferation under standard ESC culture conditions (data not shown). Moreover, when induced to differentiate for 12 days with RA treatment, both cell types demonstrated differentiation potential comparable to control cells, as assessed by qPCR of pluripotency and differentiation marker genes (Fig. S7A and B). Previously, Fgf4-KO cells were shown to differentiate into all three germ layers in teratoma formation assay [20]. We tested the teratoma formation potential of Fgf4-OE cells and found cell types from all three germ layers (Fig. S7C), suggesting that the alteration of Fgf4 expression has no overt effect on their differentiation potential. We challenged both Fgf4-KO and Fgf4-OE cells with CTN for 12 h and 24 h, and assayed for survival with annexin-V and 7-AAD staining. Analysis of these results indicated that both Fgf4-OE and Fgf4-KO cells responded to CTN in a similar manner with no noticeable differences from the wild-type cells response (Fig. 4D–F).

To analyze the DNA damage response in Fgf4-OE and Fgf4-KO, we induced DNA damage with neocarzinostatin (NCS), a known, potent inducer of DSBs, similar to γ-irradiation, in various cell types. Currently, there are no reports of NCS treatment on ESCs or other pluripotent cell lines. To verify the DSBs induction in ESCs, we stained the control and NCS treated ESCs with γ-H2A.X, a marker for DSBs (Fig. S8). Analysis of these results indicated the presence of a weak but significant amount of γ-H2A.X staining in control cells and higher signal intensities in NCS-treated cells (Fig. S8). The γ-H2A.X foci are gradually decreased by 24 h of recovery (Fig. S8) indicating DNA damage induction by NCS and subsequent activation of DNA damage repair machinery. Treatment of maGSCs with NCS revealed DNA damage induction and repair with similar pattern like ESCs (Fig. S8), suggesting the existence of similar DNA damage response mechanisms in both cell types. Thereafter, we treated Fgf4-OE, Fgf4-KO, and wild-type control ESCs with NCS and analyzed cell cycle parameters at various time points of recovery (Fig. 5A–F and Fig. S9). We found that control ESCs displayed a G2/M arrest, as previously shown with γ-irradiation [11] (Fig. 5A–F). Similarly, Fgf4-KO and Fgf4-OE cells also displayed a G2/M arrest (Fig. 5A–F). Approximately 35–40% of control cells were subG1/apoptotic by 48 h (Fig. 5F). Strikingly, fewer Fgf4-KO cells were apoptotic at 48 h (∼20%), whereas approximately 35–40% of both control and Fgf4-OE cells were apoptotic at this time point (Fig. 5F). The representative images of flow cytometry measurements at various time points are provided in Fig. S9.

Role of Mnda in Response to Apoptosis and Genotoxicity Induction

To study the role of Mnda in response to genotoxicity- and apoptosis-inducing stimuli, we generated a stable Mnda-overexpressing cell line (Mnda-OE), in which Mnda is expressed at levels ∼3-fold higher than in wild-type cells (Fig. S6B). We also generated a cell line (Mnda-DN), in which Mnda is downregulated through stable expression of shRNA against Mnda, resulting in approximately 90% downregulation (Fig. S6B). These two cell lines displayed normal colony morphology and proliferation under standard ESC culture conditions (data not shown) and also differentiated to all three germ layers when treated with either RA or assayed for teratoma formation (Fig. S7A–C). We challenged Mnda-DN and Mnda-OE cells with CTN for 12 h and 24 h, and analyzed for the survival rate. No overt differences in cell survival rates were observed between control, Mnda-OE, and Mnda-DN cells (Fig. 4D, G, H).

Additionally, we treated Mnda-OE and Mnda-DN cells with NCS and analyzed the cell cycle parameters at various time points (Fig. 5A–F and Fig. S9). Cell cycle analysis indicated that a majority (∼80%) of both Mnda-OE and Mnda-DN cells undergo G2/M arrest by 24 h and that this arrest persisted even after 36 h of recovery (Fig. 5D, E). Finally, we did not observe any differences in the percentage of apoptotic cells between Mnda-OE, Mnda-DN, and control cells (Fig. 5F). The representative images of flow cytometry measurements at various time points are provided in Fig. S9.