Sufficiency of hypoxia-inducible 2-oxoglutarate dioxygenases to block chemical oxidative stress-induced differentiation of human embryonic stem cells.

Hypoxia benefits undifferentiated pluripotent stem cell renewal, and 2-oxoglutarate (2OG) dioxygenases have been implicated in pluripotent stem cell induction and renewal. We show in human embryonic stem cells (hESC) that an ambient oxygen-induced oxidative stress response elicited by culture in a hypoxic atmosphere (0.5% O2) correlates with the expression of 2OG dioxygenases, which oxidise DNA (TET1, 2, 3) and histone H3 (KDM4C), the former reflected by elevation in genomic 5-hydroxymethylcytosine (5hmC). siRNA-mediated targeting of KDM4C and TET1-3 induces hESC differentiation. Under ambient atmospheric oxygen (21% O2), exposure to a low inhibitory concentration of sodium arsenite (NaAsO2, IC10), as a model of chemically-induced oxidative stress, suppresses antioxidant gene expression, reduces mitochondrial membrane potential and induces hESC differentiation. Co-administration of the antioxidant N-acetyl-L-cysteine promoted anti-oxidant, pluripotency and 2OG dioxygenase gene expression, elevated genomic hydroxymethylation and blocked induction of differentiation. Transient ectopic expression of KDM4C or TET1 in ambient atmospheric oxygen achieved the same. Our study substantiates a role for 2OG-dependent dioxygenases in hypoxia's promotion of undifferentiated hESC self-renewal.


Introduction
Efforts to exploit the potential of human embryonic stem cells (hESCs) and analogous induced pluripotent stem cells (hiPSCs) in regenerative medicine are ongoing. These include their use as source material for cell therapy products such as for the treatment of conditions 5 characterised by acute or chronic cell loss (e.g. macular degeneration, repair of spinal cord trauma, Parkinson's and Huntington's diseases, cardiomyopathy; for up to date accounting go to www.clinicaltrials.gov) and in cell banking initiatives to provide a standardised resource for discovery (e.g. [1]). Such activities necessitate the implementation of best practices in cell culture defined by balance of cost, benefit and risk to efficacy and safety. A case in point 10 is cell cultivation in an ambient atmosphere (21% O2 in air, commonly referred to as normoxia) vs hypoxic conditions (0.5 to 5% O2). Despite considerable evidence substantiating the merits of lower oxygen concentrations generally, and in hESC culture specifically (reviewed in [2] and elaborated further below), cultivation under atmospheric oxygen remains common practice. An improved understanding of the mechanisms of 15 hypoxia-mediated benefits is likely to assist in changing practices.
HESC culture in the range of 2-5% O2, commonly regarded as hypoxic relative to ambient atmosphere but also considered "physiologically normoxic", has broadly been reported to improve undifferentiated colony morphology (e.g. compactness), diminish levels of 20 spontaneous differentiation, increase cell proliferation rate, enhance expression of pluripotency markers, and is associated with a reduced incidence of chromosomal abnormalities ([3-6]). At a molecular level this has been reflected by oxygen-dependent changes in gene expression [7,8] including i) upregulation of expression of hypoxiainducible factors at an mRNA or protein level, genes of the glycolytic pathway under their 25 control, the cellular relocalisation and stabilisation of these hypoxia-inducible proteins [5], and also of the c-MYC proto-oncogene [9]; ii) upregulation of substrate attachment mechanisms mediated by specific integrin receptors and CD44 [10]; and iii) activation of Notch signalling [6]. Dioxygenases, specifically the 2-oxoglutarate (2OG)-dependent dioxygenases, are regarded as O2 sensors that can modulate cellular homeostasis in 30 response to oxidative stress and constitute candidate mediators of ambient oxygen stressinduced epigenetic changes. These include ten-eleven-translocation (TET) enzyme family members (TET1, 2, 3) which oxidise 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC) and the Jumonji domain 2 (JMJD2)-containing proteins, also known as histone lysine (K)-specific demethylases (e.g. KDM2-7), which oxidise and thereby demethylate 35 specific trimethylated histone residues. All TETs are expressed in mouse and human pluripotent stem cells, and TET1 has been shown to replace OCT4 in the induction of pluripotency in mouse embryonic fibroblasts [11]. Both the TET1 and KDM4C genes in hESCs are uniquely epigenetically-marked, with distinctive patterns of gene-associated cytosine-guanine island methylation shown to be associated with proper maintenance of a pluripotent phenotype [12]. In mouse ESCs the histone H3-specific demethylase KDM4C has also been shown to act with NANOG in an interconnected regulatory loop, with depletion 5 resulting in differentiation [13]. Genome-wide localisation analyses reveal enrichment of TET1 on regulatory regions marked with H3K4me3 alone or with H3K27me3 [14 -16].
Collectively these studies implicate TET1 and KDM4C in the regulation of pluripotency and differentiation.
We sought to investigate the relationship between ambient oxygen, oxidative stress and 10 2OG dioxygenases in hESC renewal, and the potential of 2OG dioxygenases to mediate the known benefits of hypoxia. Specifically, we assessed; 1) the response of undifferentiated hESCs to culture in a hypoxic (0.5% O2) vs ambient normoxic (21% O2) atmosphere at the level of the oxidative stress response, expression of pluripotency genes, KDM4C and TET1-

Results
Our study replicated all experiments in two independent hESC lines to identify conserved outcomes irrespective of known epigenomic (e.g. [12,22] Effect of hypoxia on hESC renewal, KDM4C and TET1-3. HESCs were cultured continuously for 7 days in feeder-free conditions on Matrigel in normoxia (21% O2 in air) or hypoxia (0.5% O2), one day after plating in normoxia. Both

Interference with KDM4C and TET1,2,3 in hESCs
Given hypoxia-associated elevation in the expression of KDM4C, TET1 and 3, we next 25 assessed the consequences of interference with these genes. hESC lines were transfected with gene-specific small interfering RNA (siRNA) twice at one day (24 hour) intervals one day after plating, with cells analysed two days after the first transfection (Suppl. Fig. 15    dioxygenase alone was sufficient to mimic the protective effect of hypoxia against NaAsO2 induced differentiation.

Discussion
We investigated the possible role for the 2OG dioxygenases KDM4C and TETs as oxygen Collectively our study substantiates the sufficiency of dioxygenases to protect against chemical oxidative stress-induced differentiation and as a mechanism by which hypoxia promotes pluripotency.
Several studies have confirmed the benefits of lower ambient oxygen concentrations than 20 are present in the atmosphere (21% O2) for renewal or induction of pluripotency [4,6,7,9,33,34] . Oxygen-sensing 2OG dioxygenases such as TET1 and histone lysine-specific demethylases have been implicated in pluripotency renewal [11,[13][14][15][16]. However, to our knowledge the sufficiency of 2OG dioxygenases to mediate the benefits of hypoxia in pluripotent stem cells has not been shown. In other cell systems both TET1 and KDM4C 25 have been shown to facilitate/co-activate hypoxia-inducible factor expression, and in turn, to be regulated by them [13,[35][36][37][38][39][40][41][42][43][44][45][46]. It has also been proposed that the induction of Jumonji domain-containing histone demethylases under low oxygen conditions provides a compensatory mechanism in response to compromised oxygen availability [42]. Our work is consistent with this model and previous studies in other cell types showing that hypoxia 30 induces the transcriptional upregulation of 2OG-dioxygenases KDM4C and TET1. For example, in breast cancer cell lines cultivated in 0.5% O2, HIF-1 binds to the promoters of dioxygenases generally, and KDM4C specifically, and the expression of these dioxygenases is upregulated. KDM4C specifically catalyses the demethylation of H3K36 [47]. Upregulation of KDM4C has also been shown previously in hESCs in response to a less hypoxic atmosphere than we used in our study (5% O2), along with OCT4, NANOG and SOX2 [48].
Hypoxia-induced HIF-1-dependent upregulation of TET1 expression in association with upregulation of canonical hypoxia-responsive genes and genomic 5hmC levels have been demonstrated in neuroblastoma cells [45]. Putative HIF-1a binding sites have been identified 5 in the promoters of TET1 and 3, but not TET2 [15], consistent with our observations of a differential response to hypoxia in the relative abundance of mRNA encoding these proteins.
Our targeted interference with KDM4C and TET1-3 resulted in loss of hESC self-renewal

Determination of reactive oxygen species (ROS) production
The generation of ROS in hESC cultures was evaluated using the fluorescent dye 2', 7'dichlorofluorescein diacetate (DCF-DA; 2 µM). Briefly, hESC were seeded in 6-well Matrigel-5 coated plates and exposed to a normoxia/hypoxia atmosphere with or without NaAsO2 for 15

Determination of mitochondrial membrane potential (MMP)
The mitochondrial membrane potential in hESCs was measured using Rhodamine 123.
Briefly, cells were cultured in 96-well plates and treated with NaAsO2 under normoxic (21%

Immunocytochemistry
Indirect immunofluorescence for OCT4, NANOG, 5mC and 5hmC were performed as 15 previously described in and references therein. Primary and secondary antibody suppliers and dilutions are provided in Suppl. Fig. 13. Nuclei were labelled with DAPI. Images were captured using a Zeiss Observer fluorescence microscope and prepared using Zeiss    15). Targeted genes included KDM4C, TET1, TET2, TET3, all 3 TETs concurrently, OCT4 and YAP1 (reference genes required and not required for hESC renewal, respectively [12,64]). As a further control for non-specific effects of siRNA transfection, hESCs also were transfected with a mutated sequence based on the human IDS gene (IDS-NULL) which targets no sequence in the human genome, against which effects on gene expression could also be normalised. To assess targeted siRNA effects on gene expression, transcript-5 specific mRNA first normalised internally to GAPDH values for each siRNA treated cell sample, were then calculated as fold differences relative to IDS-NULL-treated samples.

StemCelNet Network Analysis
To identify prospective molecular interactions between dioxygenases in our study with pluripotency and non-pluripotency associated genes we interrogated StemCellNet

Supplementary Information
Primary and secondary antibodies used in the study, target, source and dilution. 15 14 Oligonucleotide primer sequences for genes assessed by Q-RT-PCR. 15 15 Schematic illustration of the siRNA-mediated knockdown experiments. 15 16 Episomal plasmids derived for the ectopic expression of epigenetically-marked 2oxoglutarate-dependent dioxygenases.