Toxicology 324 (2014) 76–87

Contents lists available at ScienceDirect

Toxicology journal homepage: www.elsevier.com/locate/toxicol

Dynamic changes in energy metabolism upon embryonic stem cell differentiation support developmental toxicant identification Dorien A.M. van Dartel a, *, Sjors H. Schulpen b,c , Peter T. Theunissen b , Annelies Bunschoten a , Aldert H. Piersma b,c , Jaap Keijer a a

Human and Animal Physiology, Wageningen University, P.O. Box 338, 6700 AH Wageningen, The Netherlands Laboratory for Health Protection Research, National Institute for Public Health and the Environment (RIVM), P.O. Box 1, 3720 BA Bilthoven, The Netherlands c Institute for Risk Assessment Sciences, Faculty of Veterinary Medicine, Utrecht University, P.O. Box 80.163, 3508 TD Utrecht, The Netherlands b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 24 June 2014 Received in revised form 24 July 2014 Accepted 24 July 2014 Available online 30 July 2014

Embryonic stem cells (ESC) are widely used to study embryonic development and to identify developmental toxicants. Particularly, the embryonic stem cell test (EST) is well known as in vitro model to identify developmental toxicants. Although it is clear that energy metabolism plays a crucial role in embryonic development, the modulation of energy metabolism in in vitro models, such as the EST, is not yet described. The present study is among the first studies that analyses whole genome expression data to specifically characterize metabolic changes upon ESC early differentiation. Our transcriptomic analyses showed activation of glycolysis, truncated activation of the tricarboxylic acid (TCA) cycle, activation of lipid synthesis, as well as activation of glutaminolysis during the early phase of ESC differentiation. Taken together, this energy metabolism profile points towards energy metabolism reprogramming in the provision of metabolites for biosynthesis of cellular constituents. Next, we defined a gene set that describes this energy metabolism profile. We showed that this gene set could be successfully applied in the EST to identify developmental toxicants known to modulate cellular biosynthesis (5-fluorouracil and methoxyacetic acid), while other developmental toxicants or the negative control did not modulate the expression of this gene set. Our description of dynamic changes in energy metabolism during early ESC differentiation, as well as specific identification of developmental toxicants modulating energy metabolism, is an important step forward in the definition of the applicability domain of the EST. ã 2014 Elsevier Ireland Ltd. All rights reserved.

Keywords: Energy metabolism Embryonic stem cells Embryonic stem cell test (EST) Toxicology Transcriptomics

1. Introduction Embryonic stem cells (ESC) are derived from the inner cell mass of the pre-implementation embryo, and are considered the prototypical stem cells (Evans and Kaufman, 1981). These cells can be cultured for a prolonged period without losing their pluripotent characteristics, and can be induced to differentiate into all foetal and adult cell types. In vitro, it has been shown that ESC can originate a wide variety of cell types, including cardiomyocyte, neural and hepatic cells. These in vitro models can provide a powerful model system to study embryonic development on the level of embryonic cell differentiation. Experiments comparing gene expression profiles of in vivo embryonic development and in vitro ESC differentiation have shown significant overlap in the

* Corresponding author. Tel.: +31 317 484136. E-mail addresses: [email protected], [email protected] (D.A.M. van Dartel). http://dx.doi.org/10.1016/j.tox.2014.07.009 0300-483X/ ã 2014 Elsevier Ireland Ltd. All rights reserved.

regulation of genes annotated to developmental processes (Hettne et al., 2013; Robinson et al., 2011). The good correlation between modulation of processes in vivo and in vitro explains the successful initiatives of ESC differentiation-based in vitro approaches for identification of developmental toxic compounds (Wobus and Loser, 2011). The best known example of these approaches is the embryonic stem cell test (EST) (Spielmann et al., 1997). Initially, identification of developmental toxicants was based on disturbance of ESC differentiation into cardiomyocytes. Later, more objective molecular endpoint measures were introduced (Osman et al., 2010; Seiler et al., 2004; van Dartel et al., 2009a), which improved the efficiency, objectivity, and accuracy of the EST. Gene expression changes upon ESC differentiation generally show enrichment of three processes: cell cycling, development and metabolism (van Dartel et al., 2010b). The first two themes have often been identified to be regulated upon ESC differentiation, whereas altered energy metabolism has been less extensively described. Recently, a de novo analysis of ESC differentiation

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87

transcriptomics data has shown a network of genes that are clearly enriched for energy metabolism (Pennings et al., 2011). This study indicated that altered energy metabolism is a common finding in ESC differentiation studies. Although this study has shown regulation of common genes involved in energy metabolism, it did not provide direct insight into the type of physiological alterations in energy metabolism that take place upon ESC differentiation. Energy metabolism is pivotal during mammalian development. For example, during early mammalian development, embryos are adapted for glycolytic energy production, but acquire the dependence on oxidative metabolism at several times during pre- and post-implantation development (Knudsen and Green, 2004). In in vitro studies, this shift from anaerobic (glycolytic) metabolism to aerobic (oxidative phosphorylation) metabolism has also been observed during differentiation of stem cells (Varum et al., 2011). Moreover, energy metabolism can also be adapted for generation of building blocks for growth, a process characteristic for cells with high proliferative capacity (Lunt and Vander Heiden, 2011). So far, most studies that evaluated alterations of energy metabolism upon ESC differentiation compared mitochondrial activities of stem cells with that of fully differentiated cell types. Although these studies gave important insight into changes of metabolism-related processes upon stem cell differentiation, these studies were limited by the relatively late stages of differentiation that were studied, as well as by their predefined end point measurements. Although it is clear that energy metabolism is one of the major regulated processes during development, a detailed description of the type of changes in energy metabolism upon early ESC differentiation is currently lacking. An improved description of the biological processes that are active during ESC differentiation, including energy metabolism, will contribute to an improved description of the applicability domain of the EST, which is essential for more accurate identification of developmental toxicants. Whole genome gene expression analysis is a technique that does not necessarily require prior mechanistic knowledge and provides an integral platform for identification of effects between groups. It is ideally suited to genome-wide describe developmental changes in detail. In this study, we have used this attractive tool to elucidate the shift in energy metabolism upon ESC differentiation during an early and broad time span. Our results show that energy metabolism is altered during early ESC differentiation to favor cell growth. Moreover, we demonstrate that altered gene expression of an energy metabolism based gene-set can successfully be used to identify developmental toxicants. This study contributes to an improved description of the applicability domain of the EST, and to mechanism-based identification of developmental toxicants using the EST. 2. Materials and methods 2.1. Data set selection We used the largest data repository ArrayExpress (Parkinson et al., 2011), which encompasses over 50.000 MIAME-compliant experiments, to select data sets for our analyses. The search term: ‘stem cell’ AND ‘differentiation’ was used to identify potential usable data sets. Data sets that were not available when we started our analyses (August 7, 2013) were excluded. We reasoned that a study should include at least 12 samples in order to represent sufficient time points to reflect dynamic temporal changes and to obtain sufficient statistical power. Only data derived from human or mouse ESC were included. Finally, the study design of each remaining study was evaluated for inclusion. In this work, gene expression data of our previously published work on neuronal differentiation of ES-ES-D3 ESC was used to

77

validate the findings of the de novo analysis. The raw data of this study have been deposited in EBI’s ArrayExpress (http://www.ebi. ac.uk/arrayexpress) and are accessible through ArrayExpress accession number E-MTAB-1108. Methods and data regarding culture conditions and gene expression profiling have been described in full detail by Theunissen et al. (Theunissen et al., 2011). 2.2. Data analysis and statistics All selected studies for our analyses used the Affymetrix platform. Raw data were downloaded from ArrayExpress and the Affymetrix CEL files were normalized using the Robust Multichip Average (RMA) algorithm (Irizarry et al., 2003) using RMAexpress (Bolstad et al., 2003). For probe to gene mapping, a custom Common Data Format (CDF) was performed as we described previously (van Dartel et al., 2010a). In total, 16,346 probe sets were used in our analyses. Statistical analyses were carried out using the R statistical software environment (http://www.r-project.org) using ln-transformed values. Maximal fold ratios (FR) in individual gene expression between the experimental groups were determined by comparing the average normalized signal values per group. Genes that were significantly differentially expressed between any of the experimental groups were identified by a one-way ANOVA on the normalized data, using the cut-off criterium FDR < 0.01. Arrangement of the samples was achieved by hierarchical clustering using GeneMaths XT based on Euclidian clustering in combination with Ward linkage (Applied Maths, Sint-MartensLatem, Belgium). To study similarities in differentiation progression between the selected data sets we compared the gene expression profiles of a previously identified set of genes that have been previously identified to be related to pluripotency (Assou et al., 2007). Additionally, we selected marker genes for to evaluate similarities in proliferation, pluripotency and differentiation. Initially, we selected 3–4 marker genes per process based on scientific literature (all processes) and on our own historic data (differentiation). Marker genes were included in the analysis if these genes were present on all platforms used in our analyses and if these genes were present in the work data set for data analyses. For proliferation we used Ccnd1, Ccne1, and Pcna (Kanehisa and Goto, 2000; Stacey, 2003), for pluripotency we used Pou5f1 and Sox2 (Chambers and Tomlinson, 2009), and for differentiation we used Cyp26a1, Gata4, Vegfa, and Rbp4 (Hescheler et al., 2006). To evaluate enrichment of biological process, molecular function or cellular component of the gene clusters, DAVID EASE was used (http://david.abcc.ncifcrf.gov) (Huang da et al., 2009). This annotation was performed using only terms to which maximal 300 genes are annotated, to exclude terms that are too general for functional interpretation. Terms with an enrichment score >5 were used to describe the enrichment of the gene clusters. Gene set enrichment analysis (GSEA) was performed to discover differential expression of sets of genes that are related to energy metabolism (Subramanian et al., 2005). For the GSEA, we refined the publicly available C5 gene set collection for energy metabolism-related gene sets by only including the child terms of the term ‘metabolic process’. Gene sets were considered being significantly affected if p < 0.05. GSEA was followed by molecular concept analysis, in which the regulated gene sets are visualized within a network based on their overlap in genes, as we described previously (van Dartel et al., 2009b). Analysis of glucose levels in ES-ES-D3 ESC tissue culture medium Glucose levels were analyzed in medium samples of pluripotent ES-D3 ESC and ES-D3 ESC differentiation cultures at day 3, 5 and 7, using three biological replicates. Cells were cultured as we described

78

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87

previously (Theunissen et al., 2010). For glucose analysis, we used an automated enzymatic/colorimetric-based analysis method, which is primarily used to determine whole blood glucose levels (Freestyle, Abbott Diabetes Care, Hoofddorp, The Netherlands). Medium was refreshed 24 h prior to medium sampling, and the difference in glucose concentration was calculated and corrected for protein content as determined using a micro BCA assay (Thermo Fisher Scientific, Breda, The Netherlands). A technical triplicate of each measure was included. To confirm the accuracy of this device on measuring glucose concentration in tissue culture medium samples, medium-based glucose calibration curves were analyzed, which showed to be linear up to 20 mM glucose (data not shown). Significant differences on glucose levels were identified using oneway ANOVA and Dukey’s post-hoc test. Results were considered statistically significant if p-value < 0.05.

Search term: ‘stem cell’ differentiation Data sets: 475

Data should be available Data sets: 357

Total samples in study ≥12 Data sets: 187

Commonly used species Data sets: 177

Relevant study design - first screening Data sets: 25

2.3. Analysis of citrate synthase activity in ES-D3 ESC tissue culture medium Citrate synthase (CS) activity was analyzed in pluripotent ES-D3 ESC and ES-D3 ESC differentiation cultures at day 3, 5 and 7, using three biological replicates. The CS assay kit (Sigma–Aldrich, Zwijndrecht, The Netherlands) was used to determine CS activity according to the manufacturer’s protocol. Total protein concentration was determined using a micro BCA assay, and equal protein concentrations were used for the CS activity determination. A technical triplicate of each measure was included. Significant differences on CS activity were identified using one-way ANOVA and Dukey’s post-hoc test. Results were considered statistically significant if p-value < 0.05. 2.4. Identification of developmental toxicants that modulate energy metabolism To identify developmental toxicants that specifically modulate cellular biosynthesis, a gene set was defined that comprise genes that are annotated to the identified processes of energy metabolism that are altered to favor cellular growth. Next, we used this gene set to visualize ESC differentiation using PCA as we described previously (van Dartel et al., 2010c). Our previously published whole genome gene expression data (deposited in EBI’s ArrayExpress and available via accession number E-TABM-903). This data sets includes control ESC differentiation cultures isolated at 0 h, 24 h, and 48 h after differentiation induction (corresponding with culture day 3, 4, and 5), as well as ESC differentiation cultures exposed for 24 h from the onset of differentiation (embryoid body stage at day 3) to selected (non-) developmental toxicants (0.45 mM 5-fluorouracil, 5 mM lithium chloride, 8.5 mM methoxyacetic acid, 1.4 mM monobutyl

Relevant study design - second screening Data sets: 5

Meta-data available Data sets: 5 Fig. 1. Number of data sets that fulfilled the selection criteria for the de novo analysis of the present study.

phthalate, 1.1 mM penicillin G, 12 nM retinoic acid, 0.8 mM valproic acid). Concentrations were calculated to inhibit ESC differentiation with 50% compared to control cultures. For each compound-exposed group, coordinates along the first and second principal component were calculated and deviation of the groups from their timematched control on these coordinates was analysed by a t-test on both the first and second principal component. Cultures were considered to deviate statistically significant from their timematched controls if p-value < 0.05. 3. Results 3.1. Data set selection Five out of 475 data sets related to stem cell differentiation fulfilled all our inclusion criteria (Fig. 1). Of these five data sets, three data sets were published by Hailesellasse-Sene et al. These data sets consist of whole genome gene expression data of three

Table 1 Main characteristics of data sets included in the analyses of the present study. Author

cell line (species)

Hailesellasse- R1 Sene et al. (Mus musculus) Hailesellasse- V6.5 Sene et al. (Mus musculus) Hailesellasse- J1 Sene et al. (Mus musculus) Theunissen ES-D3 et al. (Mus musculus)

Pubmed ID

Data accession number

17394647 E-GEOD2972 17394647 E-GEOD3231 17394647 E-GEOD3749 21613230 E-TABM1108

Platform

n (per n Differentiation protocol group) (total)

Main cell types present at end of differentiation protocol

Sampling (day)

Affymetrix MOE430 A en B Affymetrix MOE430 A en B Affymetrix MOE430 A en B Affymetrix Mouse Genome 430 2.0

3

33

3

33

3

33

8

64

Meso-, ecto-, and endodermal originated cells Meso-, ecto-, and endodermal originated cells Meso-, ecto-, and endodermal originated cells Ectodermal originated cells

0, 0.25, 0.5, 0.75, 1, 1.5, 2, 4, 7, 9,14 0, 0.25, 0.5, 0.75, 1, 1.5, 2, 4, 7, 9,14 0, 0.25, 0.5, 0.75, 1, 1.5, 2, 4, 7, 9,14 0, 3, 4, 5, 7

Embryoid body culture with LIF removal and absence of murine embryonic feeder cells, normoxic Embryoid body culture with LIF removal and absence of murine embryonic feeder cells, normoxic Embryoid body culture with LIF removal and absence of murine embryonic feeder cells, normoxic Embryoid body culture with LIF removal, addition of all-trans retinoic acid, serum deprivation, and addition of neural growth factors, normoxic

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87

4d

48h

5

d1 d0.75 d1

d1.5 d9 d2

0

d0.75 d0.5 d0.5 d1 d0.75 d0.25 d0.5 d0.25 d0 d0 d0

8 4 2 +1 -2 -4 -8 0

d4 d7

-15

10

J1

V6.5 20

PC1 (56.9%)

4

5

6

7

8

9

10 11 12 13 14

8 4 2 +1 -2 -4 -8 1

2

V6.5

3

4

5

6

7

8

9

10 11 12 13 14

Differentiation duration (days) Gene expression (average fold change)

J1

t=d0 t=d0.25 t=d0.5 t=d0.75 t=d1 t=d1.5 t=2d t=d4 t=d7 t=d9 t=d14 t=d0 t=d0.25 t=d0.5 t=d0.75 t=d1 t=d1.5 t=2d t=d4 t=d7 t=d9 t=d14 t=d0 t=d0.25 t=d0.5 t=d0.75 t=d1 t=d1.5 t=2d t=d4 t=d7 t=d9 t=d14

R1

3

J1

16

0

B

2

Differentiation duration (days)

d9 d7 d14 d9 d4 d14

R1 0

1

d2

-10

d0.25

-10

R1

16

d1.5

Gene expression (average fold change)

10

36h

-5

PC2 (14.7%)

14d

7d

Gene expression (average fold change)

20

C

15

A

79

V6.5

16 8 4 2 +1 -2 -4 -8 0

1

2

3

4

5

6

7

8

9

10 11 12 13 14

Differentiation duration (days) Proliferation-related

Differentiation-related

Pcna

Cyp26a1

Ccne1

Gata4

Ccnd1

Vegfa Rbp4

Pluripotency-related Pou5f1 Sox2

-4

-2

_ +1

2

4

Fig. 2. Identification of similarities and differences of basic cell differentiation characteristics between the R1, J1 and V6.5 data sets. PCA-based differentiation tracks based on 2834 significantly regulated genes (A). Gene expression profiles of pluripotency-related genes (B). Gene expression regulation of marker genes for proliferation, pluripotency, and differentiation (C).

cell lines, namely mouse R1, J1, and V6.5 ESC, each representing a unique data set (Hailesellasse Sene et al., 2007). The fourth data set described transcription profiling of mouse CGR8 ESC, which was obtained by Hescheler and co-workers and was published as part of the FunGenES database study (Schulz et al., 2009). The fifth data set is one of our own, which we included as validation data set a priori, and describes neuronal differentiation of ES-D3 ESC (Theunissen et al., 2011a). Initially, we aimed to include one data set for our main analyses. However, since we identified that three data sets obtained by the same research group were available, we selected these three data sets for our main analyses. These three datasets are from three distinct cell lines, which provides the necessary biological robustness to the analysis. As decided in advance, the data set of Theunissen et al. was used to validate our findings. The main characteristics of the selected data sets are summarized in Table 1.

3.2. Gene expression analyses – impression of selected data sets Our first analyses focused on identifying similarities and differences of cell differentiation routes and associated basic characteristics between the J1, R1 and V6.5 data sets, which is essential for interpretation of our subsequent data analyses. Principal component analysis (PCA)-based differentiation track analysis has proven to be a useful strategy for this (van Dartel et al., 2010c). Fig. 2A shows the differentiation tracks of the three data sets based on 2834 genes that were significantly regulated (FDR < 0.01) in at least two out of the three data sets. The PCA showed that 71.6% of all variance between experimental groups could be described using only the first and second principal components of the PCA analysis. The remaining principal components had minor contributions to total gene expression variance and produced no significant shifts between experimental groups. The PCA-based differentiation tracks can be regarded as continuous representation of ESC differentiation, since

80

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87

B

t=d14

t=d9

t=d7

t=d4

t=2d

t=d1.5

t=d1

t=d0.75

t=d0.5

t=d0

t=d0.25

A

GLUTAMINE METABOLIC PROCESS

TRICARBOXYLIC ACID CYCLE_LITERATURE CERAMIDE METABOLIC PROCESS ORGANIC ACID METABOLIC PROCESS LIPID METABOLIC PROCESS MEMBRANE LIPID METABOLIC PROCESS CARBOHYDRATE METABOLIC PROCESS PHOSPHOLIPID METABOLIC PROCESS

GLUCOSE METABOLIC PROCESS

ALCOHOL METABOLIC PROCESS GLUCOSE CATABOLIC PROCESS GLYCOLYSIS_LITERATURE

Lipid metabolism

Glucose metabolism

Other metabolism

Fig. 3. Hierarchical clustering of all experimental groups of the R1 data sets based on 6688 significantly regulated genes (A). Molecular concept analysis of regulated gene sets after 24 h of differentiation versus oh. The size of the gene set corresponds with the size of the node, and the width of the connecting lines indicates the degree of overlap between the respective gene sets (B).

for each cell line samples appear mostly in chronological order in the PCA. This differentiation track analysis for each cell line follows its own unique route of differentiation, but overall the routes are relatively similar, since most deviation is seen on PC2 that only describes 14.7% of all variance. Genes previously identified to be related to pluripotency (Assou et al., 2007) show temporal changes for all three data sets (Fig. 2B). Moreover, these patterns of gene expression alterations are highly identical for these data sets. Next, we evaluated temporal changes in marker genes of proliferation (Ccnd1, Ccne1, Pcna), pluripotency (Pou5f1, Sox2), and differentiation (Cyp26a1, Gata4, Vegfa, Rbp4) (Fig. 2C). As expected, expression levels of the proliferation and pluripotency markers decreased with time for all three data sets. Little variation in the onset of decreased expression of proliferation and pluripotency markers between the three datasets was observed. The R1 and J1 cell line showed decreased expression of pluripotency and proliferation markers 24 h after differentiation induction, and this was 48 h for the V6.5 cell line. In line with this finding, differentiation markers increased with time for all three data sets. The onset of increased expression of differentiation markers was 24 h after differentiation induction for the R1 and J1 cell line, and 48 h after differentiation induction for the V6.5 cell line. The variation in this data was limited with a median relative standard deviation (95% CI) of 4.8% (0.2–23.8%), 4.9% (0.2–23.3%), 5.6% (0.2– 30.3%), and 2.0% (0.3–7.2%) for respectively R1, J1, V6.5 and ES-D3. 3.3. Gene expression analyses – identification regulation energy metabolism To study changes in energy metabolism upon ESC differentiation, we first identified which time interval should be used to study

early alterations in energy metabolism. As described above, the gene expression alterations of the three data sets were very similar, therefore we continued our detailed analyses with only the R1 data set. Hierarchical clustering on all significantly regulated genes (FDR < 0.01, 6688 genes) showed a clear grouping of early (0–24 h), mid (36–48 h) and late sampled cultures (4–14 days) (Fig. 3A). Because both data as shown in Fig. 2C and Fig. 3A point towards induction of differentiation after 24 h of ESC differentiation culturing, and first changes in metabolism are expected to occur previously or simultaneously to development-related changes (Rehman, 2010), our further analyses focused on gene expression changes in the first 24 h of differentiation. To verify if this time interval was correctly selected, process enrichment of significantly regulated genes after 24 h and 48 h was studied. Indeed, these enrichment analyses showed clear induction of differentiationrelated processes and proliferation after 48 h of differentiation, while this was much less pronounced after 24 h differentiation induction (data not shown). Molecular concepts analysis of GSEA on metabolic processrelated terms showed the interactions among the significantly regulated gene sets at 24 h compared to 0 h. Clearly, two themes appeared to be enriched: glucose metabolism, and lipid metabolism (Fig. 3B). These GSEA results were subsequently used to further study energy metabolism-related processes in detail. The processes glycolysis and lipid synthesis, as well as the linking processes tricarboxylic acid (TCA) cycle, and glutamine metabolism were selected for detailed analyses on the individual gene expression level (Fig. 4). 3.4. Gene expression regulation of selected metabolic pathways Detailed analysis of all genes annotated to glycolysis revealed high gene expression values of every step of the glycolysis process, which suggests active glycolysis and pyruvate production during the complete differentiation procedure. Nevertheless, three gene expression profiles could be discriminated (Fig. 5A). In the initial phase, 0, and 6 h after differentiation induction, gene expression was relatively low as compared to later time points. At 12 and 18 h after differentiation induction, overall gene expression of glycolysis-annotated genes was increased, indicating an increased glycolytic flux. From day 1 after differentiation induction onwards, the expression of the latter steps of glycolysis, namely Phosphoglycerate 1/2 (Pgam1/2), Enolase 1/2/3 (Eno1/2/3), and Pyruvate kinase (muscle, and liver and red blood cell; Pkm, Pklr) became relatively downregulated. Pyruvate dehydrogenase (Pdh) is rate limiting in the conversion of pyruvate to acetylCoA (Fig. 4), and was high expressed up to day 1 (Pdha1, Pdha2, Pdhb; Fig. 5B). Accordingly, most pyruvate dehydrogenase kinases (Pdk1/2/4), inhibitors of Pdh activity, showed low expression at these early stages (Fig. 5B), although Pdk3 behaved differently. Acetyl-CoA can be further metabolized in the tricarboxylic acid (TCA) cycle (Fig. 4). Genes annotated to the TCA cycle were upregulated up to 24 h after differentiation induction, with exception of the probes annotated to aconitase (Aco2) and isocitrate dehydrogenase (Idh1, Idh2, Idh3a) (Fig. 5C). This TCA gene expression profile suggests that up to 24 h after differentiation induction, the TCA cycle is truncated after the formation of citrate up to and including the formation of a-ketoglutarate. From day 2 onwards, this truncated profile disappeared and overall the expression of TCA-annotated genes become downregulated. Oxidative glutamine metabolism can function as substrate to generate a-ketoglutarate for further metabolisation in the TCA cycle (Fig. 4). Indeed, expression of genes that catalyze the reaction from glutamine to a-ketoglutarate were upregulated, albeit in a subtle way. Also, the expression of the glutamine importer solute

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87

Hk

CYTOSOL

glucose

glucose 6-phosphate

Gpi

fructose 6-phosphate

Pfk

81

fructose 1-6-biphosphate

Aldo

Aldo

GLYCOLYSIS

Glyeraldehyde 3-phosphate

(Figure 5A)

Tpi

Dihydroxyacetone phoshate

Gapdh Pk

pyruvate

pyruvate

phosphoenol pyruvate

Pdh

Pgam1/2

Eno

2-phosphoglycerate

3-phosphoglycerate

Pgk

1,3-biphospho glycerate

glycerol-3-phosphate

Pdk

acetyl-CoA Cs

OAA

Mdh2

Acly

mal

citrate

Fh1

oxaloacetate

isocitrate

(Figure 5B )

(to amino acid metabolism)

Idh1/2 Sdh

a-KG

succ Sucl

succ-CoA

Mdh1

fatty acids

lipids

LIPID SYNTHESIS (Figure 5D)

malate Me1/2

Ogdh Glud1/Gpt2/Got2

MITOCHONDRION

Fasn

acetyl-CoA

Aco2

TCA CYCLE

fum

citrate

pyruvate (to lactate/to TCA cycle)

glutamate Got1/Gpt1 Gls

GLUTAMINE METABOLISM

a-KG Glul

MYC

glutamine

glutamate

(Figure 5C) Slc1a5/Slc38a5

glutamine Fig. 4. Schematic overview of the selected cellular metabolic pathways glycolysis, TCA cycle, lipid synthesis, and glutamine metabolism that are subject of detailed analysis in the present study.

carrier family 5 member 1 (Slc1a5) was higher expressed in the first phase after induction of differentiation (Fig. 5D). Expression of the transcription factor Myc was clearly upregulated from day 0 until day 1. Evaluation of regulation of genes involved in lipid synthesis showed upregulation of both ATP citrate lyase (Acly) and fatty acid synthase (Fasn) up to 24 h (Fig. 5E). Also, expression of malate dehydrogenase 1 (Mdh1) was high expressed during this time interval. The subsequent enzymatic step by malic enzyme 1 and 2 (Me1 and Me2) was not increased on gene expression level up to 24 h, although it was upregulated in a later stage of the differentiation process. 3.5. Validation of energy metabolism regulation in differentiating ESC In order to link our detailed gene expression analyses with physiological experimentation (CS activity and glucose usage measurements) performed in the ES-D3 ESC line, we anchored gene expression profiles identified in R1 to ES-D3 gene expression profiles (Fig. 6A versus Fig. 2C). Markers of proliferation, pluripotency, and diffentiation showed similar responses, although the fold ratios in the ES-D3 cell line were significantly higher. Moreover, differentiation progression in the ES-D3 cell line was faster as compared to the R1 ESC differentiation. Anchoring of the ES-D3 data set to all three other data sets under study using genes that were significantly regulated in at least one of the four data sets (FDR 0.01; 9940 genes) revealed that differentiation

responses were similar, as shown by co-clustering, and that progression of the ES-D3 ES differentiation cultures progressed faster compared to the data sets of Haileselasse-Sene (Fig. 6B). Comparison of the gene expression profiles of glycolysis, pyruvate to acetyl-CoA conversion, TCA cycling, lipid synthesis, and glutamine metabolism (genes used in Fig. 5), revealed similar patterns in ES-D3 differentiation as compared to R1 ESC (Fig. 6C). 3.6. Glucose usage and citrate synthase activity during ES-D3 differentiation Determination of glucose usage at multiple stages of ES-D3 differentiation revealed that glucose usage was high when cells were pluripotent (Fig. 7A). At later stages during differentiation the relative use of glucose was significantly lowered. Glucose usage at day 3 could not be evaluated due to too low cell numbers relative to the medium volume. CS activity data revealed that also during the pluripotent phase of ES-D3 cells CS activity was high (Fig. 7B). This was significantly lowered from day 5 onwards, but increased again at day 7 of ES-D3 differentiation. 3.7. Identification of energy metabolism-deregulating developmental toxicants We used the differentiation track approach based on 55 genes annotated to glycolysis, to pyruvate to acetyl-CoA conversion, to

82

t=d0 t=d0.25 t=d0.5 t=d0.75 t=d1 t=d1.5 t=2d t=d4 t=d7 t=d9 t=d14

t=d0 t=d0.25 t=d0.5 t=d0.75 t=d1 t=d1.5 t=2d t=d4 t=d7 t=d9 t=d14

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87

t=d0 t=d0.25 t=d0.5 t=d0.75 t=d1 t=d1.5 t=2d t=d4 t=d7 t=d9 t=d14

Cs Aco2 Idh1 Idh2 Idh3a ogdh Sucla2 Suclg1 Sdha Sdhaf1 Sdhaf2 Sdhb Sdhc Sdhd Fh1 Mdh2

C

t=d0 t=d0.25 t=d0.5 t=d0.75 t=d1 t=d1.5 t=2d t=d4 t=d7 t=d9 t=d14

Hk1 Hk2 Gpi1 Pfkfb3 Aldoa Aldoc Aldob Tpi1 Gapdh Pgk1 Pgk2 Pgam1 Pgam2 Eno1 Eno2 Eno3 Pkm Pklr

A

Glud1 Gpt1 Gpt2 Got1 Got2 Glul Slc1a5 Slc38a5 Myc

D Pdha1 Pdha2 Pdhb Pdk1 Pdk2 Pdk3 Pdk4 t=d0 t=d0.25 t=d0.5 t=d0.75 t=d1 t=d1.5 t=2d t=d4 t=d7 t=d9 t=d14

B

-4

-2

_ +1

2

Acly Fasn Mdh1 Me1 Me2

E

4

Fig. 5. Gene expression profiles of the R1 data sets of genes annotated to glycolysis (A), to pyruvate to acetyl-CoA conversion (B), to TCA cycling (C), to glutamine metabolism (D), and to lipid synthesis (E).

B

D3

16 8 4 2 +1 -2 -4 -8 0

1

2

3

4

5

6

7

Differentiation duration (days) Proliferation-related

Differentiation-related

Pcna

Cyp26a1

Ccne1

Gata4

Ccnd1

Vegfa

D3_d0 D3_d3 R1_d0 J1_d0.25 R1_d0.25 R1_d0.5 J1_d0.5 J1_d0 V6.5_d0 V6.5_d0.25 V6.5_d0.5 V6.5_d0.75 V6.5_d1 R1_d0.75 R1_d1 V6.5_d1.5 J1_d0.75 J1_d1 D3_d4 D3_d5 R1_d1.5 V6.5_d2 R1_d2 J1_d1.5 V.5_d4 R1_d4 J1_d7 J1_4 J1_d14 V6.5_d9 J1_d2 J1_d9 R1_d7 R1_d9 R1_d14 V6.5_d7 V6.5_d14 D3_d6 D3_d7

Gene expression (average fold change)

A

C

D3_d7

D3_d6

R1_14

R1_9

R1_7

D3_d5

R1_4

R1_2

R1_1.5

D3_d4

R1_1

R1_0.75

R1_0.5

R1_0.25

R1_0

Pou5f1 Sox2

D3_d3

Pluripotency-related

D3_d0

Rbp4

Fig. 6. Gene expression profile anchoring of ESC differentiation ES-D3 data. Gene expression regulation of marker genes for proliferation, pluripotency, and differentiation in the ES-D3 data set (A). Hierarchical clustering of all experimental groups of the R1, J1, V6.5 and ES-D3 data sets based on 9940 significantly regulated genes (B). Clustering of gene expression profiles of the experimental groups of the R1 and ES-D3 data sets of genes annotated to glycolysis, pyruvate to acetyl-CoA conversion, TCA cycling, lipid synthesis, and glutamine metabolism (genes used in Fig. 5) (C).

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87

A Glucose usage (μmole glucose /ug protein/24h)

0.350

**

0.300

***

0.250

0.200

0.150

0.100

0.050

0.000 day 5

ESC

B Citrate synthase activity (μmole/ml/min/ug)

0.020

**

**

**

*

day 7

0.016

*

*

0.012 0.008 0.004 0.000 ESC

day 3

day 5

day 6

day 7

Fig. 7. Glucose usage (A). and citrate synthase activity (B) during ES-D3 ESC differentiation. Error bars indicate sd.

TCA cycling, to glutamine metabolism, and to lipid synthesis (summarized in Supplementary Table 1) to identify developmental toxicants that disturb early energy metabolism dynamics. The developmental toxicants that are known to disturb this process, namely 5-fluorouracil and methoxyacetic acid, were identified because these samples deviated significantly from their timematched control in the PCA (Fig. 8; Table 2), with a p-value of 1.57E-7, and 4.17E-11, respectively (Table 2). As expected, the samples of other compounds, including the non-developmental toxicant penicillin G, and the developmental toxicants lithium chloride, monobutylphthalate, retinoic acid and valproic acid, did not significantly deviate from their time-matched controls. 4. Discussion Over the past several decades, embryonic stem cell differentiation has been increasingly used as an in vitro model to study embryonic development and cancer cell reprogramming. In toxicology, ESC differentiation has been studied as a tool to predict developmental toxicity (van Dartel and Piersma, 2011b; Wobus and Loser, 2011). A myriad of studies have investigated molecular regulation of differentiation and pluripotency (Fathi et al., 2014; Niwa et al., 2000; Pardo et al., 2010), processes that are well reflected in toxicological ESC-based assays (van Dartel and Piersma, 2011a). This is of great importance to define the applicability domain of these assays, and to further improve their accuracy (Hartung et al., 2004). Recently, advanced progress in studies on metabolic reprogramming in ESC differentiation has

83

deepened our understanding on the crucial role of energy metabolism in directing cellular differentiation. It appears that specific metabolic requirements are needed to maintain stem cell self-renewal, and that modulation of metabolic pathways can modulate self-renewal and lineage specification (Shyh-Chang et al., 2013). This opened a new avenue for ESC-based identification of developmental toxicants. Here, this was explored and we defined an energy metabolism-focused gene set that can faithfully be used to identify compounds that may display specific developmental toxicity. Our study is among the first studies that analyze whole genome expression data to specifically characterize metabolic changes upon ESC early differentiation. So far, most studies compared differentiated gene signatures with pluripotent signatures to identify metabolic alterations. Very recently, Gaspar et al. used transcriptome analysis to identify metabolic changes during the ESC into cardiomyocyte differentiation process in order to further elucidate the role of metabolic signatures in pluripotency and differentiation (Gaspar et al., 2014). Using various data sets of their lab, mostly using whole gene expression analysis of samples taken after lineage selection, they suggested active anaplerosis via amino acid metabolism in order to resupply the TCA cycle with intermediates for biosynthetic pathways, regretfully the authors did not specifically elaborate on this finding. Our data analyses specifically focused on dynamic alterations of early stage ESC differentiation, since it has been shown that metabolic alterations occur simultaneously or even precede the induction of differentiation (Folmes et al., 2012). For our initial analyses, three data sets were selected that were obtained by the same research group. These data provided a solid basis for our study, since we reasoned that regardless of the ESC line that was used, similar alterations in key physiological processes that accompany ESC differentiation can be expected. Our results confirmed that the dynamic gene expression changes of the three data sets were significantly comparable. Our detailed analyses using the R1 data set clearly revealed enrichment of glucose and lipid metabolism in the early stage of ESC differentiation. The simultaneous regulation of both these processes pointed towards the function of energy metabolism in the provision of metabolites for biosynthesis of cellular constituents, as was indicated by Gaspar et al. In this respect, upregulation of glycolysis is to be expected, since this allows for ATP production and for funneling into the pentose phosphate pathway to generate NADPH for lipid synthesis and ribose-6-phosphate for nucleotide biosynthesis (Lunt and Vander Heiden, 2011). Furthermore, pyruvate, the penultimate product of glycolysis, can enter mitochondria and feed into the TCA cycle, after conversion into acetyl-CoA. The TCA cycle plays a key role in providing proliferating cells with biosynthetic precursors, by functioning as a hub for amino acid and, especially, lipid synthesis (Keijer and van Dartel, 2013; Lunt and Vander Heiden, 2011). Activation of this process results in cataplerosis, the continuous efflux of intermediates. This is different from the role of the TCA in non-proliferating cells, where the main function is full oxidation of acetyl-CoA to generate maximal ATP production. Our detailed data analyses on the R1 data set indeed revealed that glycolysis was highly active during the complete differentiation process as indicated by high expression values, particularly in the early stages of ESC differentiation. Hierarchical clustering of the expression of these genes of the R1 data set with the ES-D3 data set revealed that a similar pattern was observed for ES-D3 ESC differentiation. Here, overall high expression of glycolysis genes was shown in cells at a pluripotent stage, namely at day 0 and 3. At later stages, only increased expression of genes that function in the first part of glycolysis was identified, suggesting reduced glycolysis. Additional physiological

84

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87

Lithiumchloride

0.0

PC2

0.0 -1.0

PC2

1.0

1.0

5-Fluorouracil

-2.0

-1.0

*** -1

0

1

-1

2

2

Monobutylphthalate

PC2

-1.0

0.0

*** 0.0

0.0

1.0

1.0

Methoxyacetic acid

-1.0

PC2

1

PC1

-1.0

-1

1.0

PC1

0

1

2

PC1 Retinoic acid

PC2

0.0

-1.0

-1.0

0.0

1.0

1.0

Penicillin G

PC2

0

PC1

-1

0

1

2

PC1

-1

0

1

2

PC1

Valproic acid 1.0

Differentiation cultures

0.0

Control; t=24h Control; t=48h

-1.0

PC2

Control; t=0h

Compound-exposed; t=24h -1

0

1

2

PC1 Fig. 8. Principal component analysis of non-exposed ESC differentiation cultures (CON; n = 12 per group) at 0 h (red), 24 h (yellow), and 48 h (green) after differentiation induction, and compound-exposed cultures (black, n = 6–8 per group) 24 h after differentiation induction, using the 55 genes that describe energy metabolism dynamics during early ESC differentiation. *** indicate significant deviation of compund-exposed cultures from time-matched control on the diffenetiation track with p-value < 0.001. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87 Table 2 Significance values of deviation from the PCA-based differentiation track (Fig. 8) of compound-exposed differentiation cultures using the 55 genes that describe energy metabolism dynamics during early ESC differentiation. Compound

p-value

5-Fluorouracil Lithium chloride Methoxyacetic acid Monobutyl phthalate Penicillin G Retinoic acid Valproic acid

1.57E-07 0.08 4.17E-11 0.08 0.47 0.90 0.06

experimentation in ES-D3 ESC confirmed that glucose usage was high in undifferentiated ESC, and was significantly lower at day 5 and 7 after differentiation induction. These data confirm our hypothesis that glucose is used as a substrate for generating biosynthetic precursors when cells are in a pluripotent state. For ES-D3 cells, one of the most commonly used ESC line in toxicological research, a metabolic profile that favors cellular biosynthesis was particularly active during the first three days of the differentiation protocol. This data corresponds with our previous data in which we discuss that at the first phase of ESD3 differentiation cells predominantly proliferate, and that differentiation is induced from day 3 onwards (van Dartel et al., 2009b). A truncated TCA cycle as indicated by up-regulation of part of the TCA cycle genes is characteristic for TCA-based cellular biosynthesis, and this profile was identified during early stages of differentiation in our gene expression analyses. The TCA-gene expression profile between the R1 and ES-D3 cell line showed high similarity, with a truncated TCA cluster during early stages of differentiation, and relatively low TCA-related gene expression in the mid-phase of differentiation. Since the differentiation progression of ES-D3 cells showed to run significantly faster, also a third cluster was identified in the ES-D3 data set, which showed relative up-regulation of the first part of the TCA cycle, and an overall high expression of the TCA-annotated genes. Additional CS activity measurements confirmed our gene expression findings. A plausible explanation for the reactivation of CS at the late differentiation stage (day 7) in ES-D3 differentiation cultures is to activate the function of the TCA cycle for ATP generation, in agreement with return to a post-glycolytic state. Furthermore, it has previously been shown for ES-D3 differentiation to cardiomyocytes that differentiated cells start contracting at day 7 (van Dartel et al., 2009b), a process that definitely requires substantial amounts of ATP. In line with the increased expression of glycolysis genes that support cellular biosynthesis, expression of genes that function in the conversion of pyruvate into acetyl-CoA (Pdh) and its inhibitors (Pdk) were regulated. We studied the expression levels of all four Pdk genes (Pdk1–4) in our analyses, although recent research showed that specifically Pdk2 and Pdk4 functions in regulating Pdh activity in stem cells (Takubo et al., 2013). This data most likely explains the, at first unexpected, upregulation of Pdk3 from day 0– 2 in the R1 data set. Our analyses also showed strong up regulation of the transcription factor Myc. This transcription factor facilitates reprogramming of metabolism for growth, similarly as we describe in the present study: boost of glycolytic metabolism, and induced glutaminolysis (Dang 2013). This result thus further supports our conclusion of reprogrammed energy metabolism to support cell growth and proliferation. Our finding that ESC before differentiation induction and in the early stage of differentiation have a predominantly glycolytic and truncated TCA cycle metabolic profile fits with the observation that

85

ESC at these stages contain ‘immature’ mitochondria (Cho et al., 2006; Ramalho-Santos et al., 2009; Varum et al., 2011). It has been described that upon differentiation, when metabolism changes to mitochondrial-dependent ATP production using oxidative phosphorylation, mitochondrial density concomitant with the appearance of cristae dense morphology and increased mitochondrial membrane potential. Our observation that pluripotent ESC have a glycolytic signature, even under conditions of sufficient oxygen supply, has been considered characteristic for ESC as well as other rapidly growing cells, including cancer cells (DeBerardinis et al., 2008; Zhou et al., 2012). In the data sets used in the present study, ESC were cultured under normoxic conditions. It has been discussed, however, that maintenance of pluripotency and induction of differentiation is favored when pluripotent ESC are cultured under hypoxic conditions. This will contribute to stabilization of the transcription factor hypoxia inducible factor alpha (HIF1A) and the subsequent expression of stemness factors (Mandal et al., 2011; Yanes et al., 2010; Yoshida et al., 2009). HIF1A has been implicated in regulating many genes to support aerobic glycolysis (Denko, 2008), as also observed in pluripotent ESC. Although we used data of ESC differentiation cultured under normoxic conditions, our analyses showed increased expression of HIF1a as well as genes downstream of HIF1a during early differentiation in our data sets. Most likely, a hypoxic environment was created under normoxic conditions via the 3D-culturing procedure of cells in embryoid bodies. However, to improve standardization of ESC differentiation-based assays it might be useful to adapt oxygen levels during different stages of the differentiation process. Our previous research has shown that specific gene sets can be used within the same assay to identify process- or compound class-specific dysregulation (van Dartel et al., 2011a; van Dartel et al., 2011b). Also, we showed that the accuracy of the gene sets to identify developmental toxicant was optimal when the gene expression modulation was evaluated within the most dynamic part of ES differentiation, which appeared to be at day 3–5 of the EST protocol (0–48 h after differentiation induction) (van Dartel et al., 2010a). Here, we revealed that in this same timeframe also energy metabolism-related changes are dynamically changing. Indeed, a first test showed that a gene set that describes energy metabolism alterations in the early stage of ESC differentiation was successful in identification of developmental toxicants that are known to affect this process. Further optimization of such a gene set could further improve identification of developmental toxicants that affect ESC differentiation-specific metabolic requirements. 5. Conclusion In conclusion, the present study shows metabolic reprogramming from a cellular biosynthesis profile to an ATP generation profile in differentiating embryonic stem cells, a phenomenon that was identified in multiple cell lines. These energy metabolism features are also activated in processes such as embryonic development and cancer cell growth, which justifies the use of ESC and ESC differentiation as an in vitro model in these areas of research. Particularly ES-D3 ESC appear to be useful as in vitro model, because of their relative homogeneous response as compared to the other ESC lines evaluated in the present study. Moreover, we showed that a gene set that comprises genes annotated to the metabolic processes regulated during early differentiation could be used to identify developmental toxicants that specifically modulate this process. This is an important step forward in defining the applicability domain of the EST.

86

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87

Conflict of interests The authors declare that there are no conflicts of interest. Transparency document The Transparency document associated with this article can be found in the online version. Acknowledgements The authors gratefully acknowledge Kagnew Hailesellasse Sene and colleagues for making their transcriptomic data sets publicly available. In addition, we thank our colleagues Jeroen Pennings and Evert van Schothorst for their helpful suggestions and discussions. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.tox.2014.07.009. References Assou, S., Le Carrour, T., Tondeur, S., Strom, S., Gabelle, A., Marty, S., Nadal, L., Pantesco, V., Reme, T., Hugnot, J.P., Gasca, S., Hovatta, O., Hamamah, S., Klein, B., De Vos, J., 2007. A meta-analysis of human embryonic stem cells transcriptome integrated into a web-based expression atlas. Stem Cells 25, 961–973. Bolstad, B.M., Irizarry, R.A., Astrand, M., Speed, T.P., 2003. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19, 185–193. Chambers, I., Tomlinson, S.R., 2009. The transcriptional foundation of pluripotency. Development 136, 2311–2322. Cho, Y.M., Kwon, S., Pak, Y.K., Seol, H.W., Choi, Y.M., Park do, J., Park, K.S., Lee, H.K., 2006. Dynamic changes in mitochondrial biogenesis and antioxidant enzymes during the spontaneous differentiation of human embryonic stem cells. Biochem. Biophys. Res. Commun. 348, 1472–1478. Dang, C.V., 2013. MYC, metabolism, cell growth, and tumorigenesis. Cold Spring Harb. Perspect. Med. 3, . DeBerardinis, R.J., Lum, J.J., Hatzivassiliou, G., Thompson, C.B., 2008. The biology of cancer: metabolic reprogramming fuels cell growth and proliferation. Cell Metab. 7, 11–20. Denko, N.C., 2008. Hypoxia, HIF1 and glucose metabolism in the solid tumour. Nat. Rev. Cancer 8, 705–713. Evans, M.J., Kaufman, M.H., 1981. Establishment in culture of pluripotential cells from mouse embryos. Nature 292, 154–156. Fathi, A., Hatami, M., Vakilian, H., Han, C.L., Chen, Y.J., Baharvand, H., Salekdeh, G.H., 2014. Quantitative proteomics analysis highlights the role of redox hemostasis and energy metabolism in human embryonic stem cell differentiation to neural cells. J. Proteomics 101, 1–16. Folmes, C.D., Dzeja, P.P., Nelson, T.J., Terzic, A., 2012. Metabolic plasticity in stem cell homeostasis and differentiation. Cell Stem Cell 11, 596–606. Gaspar, J.A., Doss, M.X., Hengstler, J.G., Cadenas, C., Hescheler, J., Sachinidis, A., 2014. Unique metabolic features of stem cells, cardiomyocytes, and their progenitors. Circ. Res. 114, 1346–1360. Hailesellasse Sene, K., Porter, C.J., Palidwor, G., Perez-Iratxeta, C., Muro, E.M., Campbell, P.A., Rudnicki, M.A., Andrade-Navarro, M.A., 2007. Gene function in early mouse embryonic stem cell differentiation. BMC Genomics 8, 85. Hartung, T., Bremer, S., Casati, S., Coecke, S., Corvi, R., Fortaner, S., Gribaldo, L., Halder, M., Hoffmann, S., Roi, A.J., Prieto, P., Sabbioni, E., Scott, L., Worth, A., Zuang, V., 2004. A modular approach to the ECVAM principles on test validity. Altern. Lab. Anim. 32, 467–472. Hescheler, J., Sachinidis, A., Chen, S., Winkler, J., 2006. The FunGenES consortium: functional genomics in engineered embryonic stem cells. Stem Cell Rev. 2, 1–4. Hettne, K.M., Boorsma, A., van Dartel, D.A., Goeman, J.J., de Jong, E., Piersma, A.H., Stierum, R.H., Kleinjans, J.C., Kors, J.A., 2013. Next-generation text-mining mediated generation of chemical response-specific gene sets for interpretation of gene expression data. BMC Med. Genomics 6, 2. Huang da, W., Sherman, B.T., Lempicki, R.A., 2009. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 37, 1–13. Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D., Antonellis, K.J., Scherf, U., Speed, T.P., 2003. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4, 249–264. Kanehisa, M., Goto, S., 2000. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 28, 27–30. Keijer, J., van Dartel, D.A., 2013. Reprogrammed Metabolism of Cancer Cells as a Potential Therapeutic Target. Current Pharmaceutical Design.

Knudsen, T.B., Green, M.L., 2004. Response characteristics of the mitochondrial DNA genome in developmental health and disease. Birth Defects Res. C Embryo Today 72, 313–329. Lunt, S.Y., Vander Heiden, M.G., 2011. Aerobic glycolysis: meeting the metabolic requirements of cell proliferation. Ann. Rev. Cell Dev. Biol. 27, 441–464. Mandal, S., Lindgren, A.G., Srivastava, A.S., Clark, A.T., Banerjee, U., 2011. Mitochondrial function controls proliferation and early differentiation potential of embryonic stem cells. Stem Cells 29, 486–495. Niwa, H., Miyazaki, J., Smith, A.G., 2000. Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat. Genet. 24, 372–376. Osman, A.M., van Dartel, D.A.M., Zwart, E., Blokland, M., Pennings, J.L.A., Piersma, A. H., 2010. Proteome profiling of mouse embryonic stem cells to define markers for cell differentiation and embryotoxicity. Reprod. Toxicol. 30, 322–332. Pardo, M., Lang, B., Yu, L., Prosser, H., Bradley, A., Babu, M.M., Choudhary, J., 2010. An expanded Oct4 interaction network: implications for stem cell biology, development, and disease. Cell Stem Cell 6, 382–395. Parkinson, H., Sarkans, U., Kolesnikov, N., Abeygunawardena, N., Burdett, T., Dylag, M., Emam, I., Farne, A., Hastings, E., Holloway, E., Kurbatova, N., Lukk, M., Malone, J., Mani, R., Pilicheva, E., Rustici, G., Sharma, A., Williams, E., Adamusiak, T., Brandizi, M., Sklyar, N., Brazma, A., 2011. ArrayExpress update – an archive of microarray and high-throughput sequencing-based functional genomics experiments. Nucleic Acids Res. 39, D1002–1004. Pennings, J.L.A., van Dartel, D.A.M., Pronk, T.E., Hendriksen, P.J., Piersma, A.H., 2011. Identification by gene co-regulation mapping of novel genes involved in embryonic stem cell differentiation. Stem Cells Dev. 20, 115–126. Ramalho-Santos, J., Varum, S., Amaral, S., Mota, P.C., Sousa, A.P., Amaral, A., 2009. Mitochondrial functionality in reproduction: from gonads and gametes to embryos and embryonic stem cells. Hum. Reprod. Update 15, 553–572. Rehman, J., 2010. Empowering self-renewal and differentiation: the role of mitochondria in stem cells. J. Mol. Med. 88, 981–986. Robinson, J.F., Theunissen, P.T., van Dartel, D.A., Pennings, J.L., Faustman, E.M., Piersma, A.H., 2011. Comparison of MeHg-induced toxicogenomic responses across in vivo and in vitro models used in developmental toxicology. Reprod. Toxicol. 32, 180–188. Schulz, H., Kolde, R., Adler, P., Aksoy, I., Anastassiadis, K., Bader, M., Billon, N., Boeuf, H., Bourillot, P.Y., Buchholz, F., Dani, C., Doss, M.X., Forrester, L., Gitton, M., Henrique, D., Hescheler, J., Himmelbauer, H., Hubner, N., Karantzali, E., Kretsovali, A., Lubitz, S., Pradier, L., Rai, M., Reimand, J., Rolletschek, A., Sachinidis, A., Savatier, P., Stewart, F., Storm, M.P., Trouillas, M., Vilo, J., Welham, M.J., Winkler, J., Wobus, A.M., Hatzopoulos, A.K., 2009. The FunGenES database: a genomics resource for mouse embryonic stem cell differentiation. PLoS One 4, e6804. Seiler, A., Visan, A., Buesen, R., Genschow, E., Spielmann, H., 2004. Improvement of an in vitro stem cell assay for developmental toxicity: the use of molecular endpoints in the embryonic stem cell test. Reprod. Toxicol. 18, 231–240. Shyh-Chang, N., Daley, G.Q., Cantley, L.C., 2013. Stem cell metabolism in tissue development and aging. Development 140, 2535–2547. Spielmann, H., Pohl, I., Doring, B., Liebsch m. Moldenbauer, F., 1997. The embryonic stem cell test (EST), an in vitro embryotoxicity test using two permanent cell lines: 3T3 fibroblasts and embryonic stem cells. In vitro Toxicol. 10, 119–127. Stacey, D.W., 2003. Cyclin D1 serves as a cell cycle regulatory switch in actively proliferating cells. Curr. Opin. Cell Biol. 15, 158–163. Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., Mesirov, J.P., 2005. Gene set enrichment analysis: a knowledge-based approach for interpreting genomewide expression profiles. Proc. Natl. Acad. Sci. U. S. A 102, 15545–15550. Takubo, K., Nagamatsu, G., Kobayashi, C.I., Nakamura-Ishizu, A., Kobayashi, H., Ikeda, E., Goda, N., Rahimi, Y., Johnson, R.S., Soga, T., Hirao, A., Suematsu, M., Suda, T., 2013. Regulation of glycolysis by Pdk functions as a metabolic checkpoint for cell cycle quiescence in hematopoietic stem cells. Cell Stem Cell 12, 49–61. Theunissen, P.T., Pennings, J.L., Robinson, J.F., Claessen, S.M., Kleinjans, J.C., Piersma, A.H., 2011a. Time-response evaluation by transcriptomics of methylmercury effects on neural differentiation of murine embryonic stem cells. Toxicol. Sci. 122, 437–447. Theunissen, P.T., Pennings, J.L., Robinson, J.F., Claessen, S.M., Kleinjans, J.C., Piersma, A.H., 2011. Time-response evaluation by transcriptomics of methylmercury effects on neural differentiation of murine embryonic stem cells. Toxicol. Sci. . Theunissen, P.T., Schulpen, S.H., van Dartel, D.A.M., Hermsen, S.A.B., van Schooten, F. J., Piersma, A.H., 2010. An abbreviated protocol for multilineage neural differentiation of murine embryonic stem cells and its perturbation by methyl mercury. Reprod. Toxicol. 29, 383–392. van Dartel, D.A., Pennings, J.L., de la Fonteyne, L.J., Brauers, K.J., Claessen, S., van Delft, J.H., Kleinjans, J.C., Piersma, A.H., 2011a. Concentration-dependent gene expression responses to flusilazole in embryonic stem cell differentiation cultures. Toxicol. Appl. Pharmacol. 251, 110–118. van Dartel, D.A., Pennings, J.L., Hendriksen, P.J., van Schooten, F.J., Piersma, A.H., 2009a. Early gene expression changes during embryonic stem cell differentiation into cardiomyocytes and their modulation by monobutyl phthalate. Reprod. Toxicol. 27, 93–102. van Dartel, D.A., Pennings, J.L., Robinson, J.F., Kleinjans, J.C., Piersma, A.H., 2011b. Discriminating classes of developmental toxicants using gene expression profiling in the embryonic stem cell test. Toxicol. Lett. 201, 143–151. van Dartel, D.A., Pennings, J.L., van Schooten, F.J., Piersma, A.H., 2010a. Transcriptomics-based identification of developmental toxicants through their interference with cardiomyocyte differentiation of embryonic stem cells. Toxicol. Appl. Pharmacol. 243, 420–428.

D.A.M. van Dartel et al. / Toxicology 324 (2014) 76–87 van Dartel, D.A., Piersma, A.H., 2011a. The embryonic stem cell test combined with toxicogenomics as an alternative testing model for the assessment of developmental toxicity. Reprod. Toxicol. 32, 235–244. van Dartel, D.A., Zeijen, N.J., de la Fonteyne, L.J., van Schooten, F.J., Piersma, A.H., 2009b. Disentangling cellular proliferation and differentiation in the embryonic stem cell test, and its impact on the experimental protocol. Reprod. Toxicol. 28, 254–261. van Dartel, D.A.M., Pennings, J.L.A., de la Fonteyne, L.J.J., van Herwijnen, M., van Delft, J.H., van Schooten, F.J., Piersma, A.H., 2010b. Monitoring developmental toxicity in the embryonic stem cell test using differential gene expression of differentiation-related genes. Toxicol. Sci. 116, 130–139. van Dartel, D.A.M., Pennings, J.L.A., van Schooten, F.J., Piersma, A.H., 2010c. Transcriptomics-based identification of developmental toxicants through their interference with cardiomyocyte differentiation of embryonic stem cells. Toxicol. Appl. Pharmacol. 243, 420–428. van Dartel, D.A.M., Piersma, A.H., 2011b. The embryonic stem cell test combined with toxicogenomics as an alternative testing model for the assessment of developmental toxicity. Reprod. Toxicol. 32, 235–244.

87

Varum, S., Rodrigues, A.S., Moura, M.B., Momcilovic, O., Easley C.A.T. RamalhoSantos, J., Van Houten, B., Schatten, G., 2011. Energy metabolism in human pluripotent stem cells and their differentiated counterparts. PLoS One 6, e20914. Wobus, A.M., Loser, P., 2011. Present state and future perspectives of using pluripotent stem cells in toxicology research. Arch. Toxicol. 85, 79–117. Yanes, O., Clark, J., Wong, D.M., Patti, G.J., Sanchez-Ruiz, A., Benton, H.P., Trauger, S.A., Desponts, C., Ding, S., Siuzdak, G., 2010. Metabolic oxidation regulates embryonic stem cell differentiation. Nat. Chem. Biol. 6, 411–417. Yoshida, Y., Takahashi, K., Okita, K., Ichisaka, T., Yamanaka, S., 2009. Hypoxia enhances the generation of induced pluripotent stem cells. Cell Stem Cell 5, 237–241. Zhou, W., Choi, M., Margineantu, D., Margaretha, L., Hesson, J., Cavanaugh, C., Blau, C. A., Horwitz, M.S., Hockenbery, D., Ware, C., Ruohola-Baker, H., 2012. HIF1alpha induced switch from bivalent to exclusively glycolytic metabolism during ESCto-EpiSC/hESC transition. EMBO J. 31, 2103–2116.