The epigenetic and morphogenetic effects of molecular oxygen and its derived reactive species in development

Abstract

The development of multicellular organisms involves the unpacking of a complex genetic program. Extensive characterization of discrete developmental steps has revealed the genetic program is controlled by an epigenetic state. Shifting the epigenome is a group of epigenetic enzymes that modify DNA and proteins to regulate cell type specific gene expression. While the role of these modifications in development has been established, the input(s) responsible for electing changes in the epigenetic state remains unknown. Development is also associated with dynamic changes in cellular metabolism, redox, free radical production, and oxygen availability. It has previously been postulated that these changes are causal in development by affecting gene expression. This suggests that oxygen is a morphogenic compound that impacts the removal of epigenetic marks. Likewise, metabolism and reactive oxygen species influence redox signaling through iron and glutathione to limit the availability of key epigenetic cofactors such as α-ketoglutarate, ascorbate, NAD⁺ and S-adenosylmethionine. Given the close relationship between these cofactors and epigenetic marks it seems likely that the two are linked. Here we describe how changing these inputs might affect the epigenetic state during development to drive gene expression. Combined, these cofactors and reactive oxygen species constitute the epigenetic landscape guiding cells along differing developmental paths.

Introduction:

Development of a single cell to a multicellular organism requires the unfolding of an intricate genetic code to produce an array of cellular phenotypes. Propelling these changes is a complex, interconnected gene regulatory network consisting of hierarchical transcription factor activation, and a chromatin architecture, or epigenome, that continually interpret an inherited genome to produce phenotype-specific changes in gene expression. For the most part, a cell’s genetic program remains unchanged during development, while the epigenome interpreting it continually changes. The plasticity exhibited by epigenetics affords a multitude of cellular phenotypes from a single genotype and orchestrates subtle or overt changes in gene expression. Over the past decade our understanding of epigenetics in development has been well established. These studies have described the epigenetic fingerprint of stem cells, the global changes that occur during lineage commitment, and those of terminally differentiated cells. Likewise, we have begun to understand how epigenetics governs the reverse process during the creation of induced pluripotent stem cells. Aristotle’s original theory of epigenesis postulated that the development of organs and individuals proceeded in a series of steps from undifferentiated matter [1]. Although this might seem obvious today, this was highly controversial from Aristotle’s time up through the late 18^th century. Conrad Waddington adapted this terminology and first conceived epigenetics as the “science concerned with the causal analysis of development” in the 1930’s he was unaware of the exact nature by which the epigenome directs these changes [2]. Today, we know epigenetics is comprised of four separate but often interconnected processes: DNA methylation, histone modifications, long non-coding RNA, and higher-ordered chromatin structure. While the function of epigenetic processes in development is now well established, the mechanisms guiding their interpretation of the genome remain obscure.

Changes in developmentally associated gene expression coincide with alterations in cellular metabolism, oxygen, and redox state. Likewise, manipulating these factors in model systems can elicit developmental pathways, and drive cell and tissue-specific associated changes in gene expression [3, 4]. These observations led to speculation that developmentally regulated gene expression and redox biology are linked, compelling Allen and Balin to put forth the Free Radical Theory of Development in 1989 [5]. Likewise, we have previously expounded on this concept by describing the nexus of epigenetics, redox biology and gene expression in development [6]. This relationship is supported by the understanding that enzymes initiating and perpetuating epigenetic events are influenced by oxygen, metabolism and redox state [6–8]. Furthermore, the flux of these variables within and around cells during development coincides with differing epigenotypes [9, 10]. From this, it seems fitting that Conrad Waddington chose to depict the concept of epigenetics as one in which cell fate is determined by the shifting conditions of a landscape [11]. The subtle message of this model translates well to our message of how extraneous factors affect the genome’s interpretation by effecting the epigenome. Here, we reexplore and expand upon our original hypothesis that modifications of the epigenome are driven, at least in part, by metabolism, oxygen, and redox state.

The Epigenetic State and Free Radical Theory of Development:

The central canons of the Free Radical Theory of Development center around how metabolism and oxygen influence gene expression through the production and removal of free radicals and reactive oxygen species (ROS) [3–6]. Allen and Balin suggested that oxygen drives development by existing as a metabolic gradient in which varying partial pressures influence gene expression in the developing embryo. Sohal and others have also suggested that oxygen’s concentration in development should be synchronized with coordinating changes in the production of ROS such as superoxide (O₂^•−) and hydrogen peroxide (H₂O₂) [3, 4]. Combined, the Free Radical Theory of Development suggests the balance between the production of ROS, and their subsequent removal by antioxidant systems is responsible for directing developmentally associated changes in gene expression. When this was originally suggested the mechanism(s) by which oxygen, free radicals, and ROS influence gene expression were unknown. Here, we describe how oxygen and redox biology participate in forming the epigenetic landscape during development by influencing the activities of enzymes that modify the epigenome.

Epigenetic regulation of gene expression: Enzymes

The epigenetic state is adept at determining cell fate by regulating specific gene expression programs of the genome primarily through DNA methylation and the posttranslational modification of histone proteins. Inscribing these marks is a diverse group of enzymes that maintain, support, or opposes each other’s influence on a cell’s genome. Many of these enzymes share cofactors that interact with each other or lie on the same metabolic track (Fig. 1). The interconnectivity between the enzymes that establish and maintain epigenetic marks directly relates them to oxygen, redox biology and metabolism. The mechanisms by which these enzymes impress their influence on the epigenome, as well as their relationships with one another as been exhaustively reviewed elsewhere [12, 13]. Here we will focus on how these enzymes relate to each other based on their catalytic activities.

Figure 1. Transit map depicting the various “routes” by which the epigenetic landscape can be traversed.

The epigenetic landscape is composed of various transit routes by which an epigenetic destination can be reached. The major routes revolve around a redox central hub (O₂^•−). Each line (Cofactors, Methylation and Metabolism) has a terminus relating to an epigenetic process, and an opposite terminus related to redox and metabolism. Along each line are cofactor (circles) and enzyme (squares) “stops” to signify the interconnectivity of routes taken to reach epigenetic destination on the epigenetic landscape. α-ketoglutarate (αKG), S-adenosylmethionine (SAM) SAH (S-adenosylhomocysteine) HMT (histone methyltransferase) DNMT (DNA methyltransferase) HATs (histone acetyltransferase) FAD (flavin adenine dinucleotide) LSD1/KDM1a (lysine specific demethylase 1) ASC (ascorbate) Fe²⁺ (iron II) Fe/S (iron sulfur cluster) TET (Ten-eleven translocation methylcytosine dioxygenase) KDM (lysine demethylase), NAD⁺ (oxidized nicotinamide adenine dinucleotide), NADH (reduced nicotinamide adenine dinucleotide) P-SH (protein sulfhydryl)

Transmethylation in epigenetics

DNA methylation in multicellular animals occurs almost exclusively at CpG dinucleotides and is associated with the presence of condensed, transcriptionally inert chromatin. Furthermore, the CpG dinucleotide itself is unique, occurring at an unusually low frequency, while simultaneously converging into “CpG islands” within the regulatory elements of genes [14, 15]. DNA methylation occurs post-replication and is catalyzed by three distinct DNA methyltransferases in humans: DNMT1, 3a, and 3b [16, 17]. All three require S-adenosylmethionine (SAM) as a cofactor to transfer a methyl group to the 5-position of cytosine’s pyrimidine ring to produce 5-methylcytosine (5mC) within CpG doublet and S-adenosylhomocysteine (SAH) as a byproduct. Likewise, histone and non-histone proteins are subject to posttranslational methylation by lysine methyltransferases (KMTs) and protein arginine methyltransferases (PRMTs) [18]. Like DNMTs, they influence epigenetic regulation of gene expression, utilize SAM as a cofactor, and produce SAH as a byproduct [18, 19]. Histone methyltransferases (HMTs) differ from DNMTs in the fact that the transcriptional apparatus can respond manifold ways to the epigenetic marks they produce (For review see: [20]). Histone methylation can be associated with activated open chromatin, and transcriptionally inert heterochromatin. Methylation of DNA and histones is essential for creating the epigenotypes of mammals as they develop [21, 22]. We will discuss below how the activity of methyltransferases is influenced by perturbations in redox during development.

Demethylation of DNA.

Epigenetic methylation is not a fixed modification. Demethylation of DNA is facilitated by the ten-eleven translocation (TET) methylcytosine dioxygenases. The family contains three members: TET1, TET2, and TET3, all of which are capable of catalyzing the sequential oxidation of 5mC to 5-hydroxymethylcytosine (5hmC), then 5-formylcytosine (5fC), and finally 5-carboxylcytosine (5caC) [7]. These enzymes all utilize oxygen, α-ketoglutarate, Fe²⁺ and ascorbate during their activity. 5hmC can serve as an epigenetic mark in its own right. Significant levels of 5hmC have been detected in developing mouse embryo and the modification is detectable in mouse embryonic stem cells [23, 24]. TET expression is also very high in the embryo at this stage, especially during implantation [21]. There appear to be two main avenues by which TET mediated oxidation can influence DNA methylation. First is a passive mechanisms in which DNMT1 cannot recognize a CpG doublet for methylation after DNA replication if it contains 5hmC, meaning the methylated CpG mark doesn’t get perpetuated in subsequent cell divisions leading to the eventual loss of 5hmC [25]. The second is an active mechanism in which the oxidation of 5mC by TET enzymes is followed by base excision repair (BER). In this route, DNA glycosylases remove the oxidized forms of 5mC to create an abasic site that is replaced with an unmethylated cytosine during subsequent BER activities [26, 27]. It has been suggested that the passive mechanism of demethylation by TETs is the dominant method by which global demethylation is achieved during preimplantation of the developing embryo [25, 28]. This likely avoids the substantial repair activity required by BER dependent mechanism. The passive, and active mechanisms might also be linked. Recent reports demonstrate that TETs mediate site-specific demethylation of Developmental Pluripotency Associated 3 (DPPA3), activates its expression. DPPA3 expression then initiates a mechanism that displaces DNMT1 from chromatin, leading to DNA hypomethylation [29].

Demethylation of histones

Demethylation of histones is carried out by two known families of lysine demethylases (KDMs). The first, reported by Shi et al in 2004 is LSD1/KDM1. KDM1specifically targets mono, or di methylated forms lysine 4 of histone H3 (H3K4me1/me2) for demethylation [30]. Demethylation utilizes an electron acceptor, likely oxygen, and FAD coenzyme to oxidize the methylated amine of H3K4, resulting in the spontaneous release of the methyl group as formaldehyde, while producing the ROS H₂O₂ as a byproduct [31]. The second is jumonji-domain-containing demethylases (JmjCs). JmjCs require oxygen, α-ketoglutarate, Fe²⁺ and ascorbate for demethylation, just like the TET family of oxidases. The JmjC enzymes target several different methylated histone lysines, thus giving them a key role in both activating and repressing gene expression (for review see [32]). In humans, the JmjCs are a family of more than twenty genes, knockout mouse models have shown that some have roles in molding the epigenome in embryonic stem (ES) cells, while other appear to be dispensable [8, 22]. The changes in histone methylation taking place in the early embryo makes the activities of these enzymes crucial in establishing the early epigenome [32].

Histone Acetylation

Since its discovery, histone acetylation has been associated with euchromatin and active transcription [33, 34]. Acetylated histones are the products of a host of histone acetyltransferases (HATs) that use acetyl-CoA derived from metabolism as a cofactor to generate ε-N-acetyl lysine in histone tails [13]. Mammalian cells have several HATs that target proteins for acetylation [35]. Histone acetylation influences more than just transcription. The euchromatin they create can also serve as a hot spot for meiotic recombination and facilitate DNA repair [36]. Global dynamics in histone acetylation also follow the cell cycle. Histones acetylation is an epigenetic mark closely associated with metabolism. In the nucleus and cytosol the enzyme ATP-citrate lyase (ACL) converts citrate from the mitochondria into acetyl-CoA for use by HATs [37]. ACL mediated production of acetyl-CoA may also constitute a mechanism for crosstalk between DNA methylation. Expression of ACL activates microRNA 148a which silences expression of DNMT1 [38]. Histone acetylation is countered by four classes of histone deacetylases (HDACs). Classes I, II, and IV utilize a coordinated metal facilitate the removal of acetyl groups from proteins [39]. Class III HDACs encompass the NAD⁺ dependent sirtuins [40]. Using NAD⁺ allows sirtuins to sense the current metabolic state and influence gene expression. The presence of a HDAC from any family is consistent with gene silencing and the formation of condensed heterochromatin.

Oxygen

Oxygen is a powerful morphogenic compound that influences development. Oxygen’s biochemical role was long viewed as an electron acceptor to produce energy and reducing equivalents during oxidative phosphorylation. However, the level of oxygen present is crucial in regulating the maintenance of pluripotency, cell fate, and driving organ development, all of which are independent of oxygen’s role in energy production. Our earliest development stages lack vasculature, meaning oxygen reaches cells via passive diffusion. Oxygen is only capable of diffusing approximately 150 μM form its source [41]. A consequence of this system if the formation of oxygen gradients within the developing fetus [5]. This creates a hypoxic microenvironment in which undifferentiated cells wait for biochemical and environmental cues, including potentially from oxygen, to begin the process of lineage commitment. Generally speaking, the lower the oxygen tension, the less differentiated the cells within it are. Growth in low oxygen preserves pluripotency by maintaining signaling pathways that conserve the poised state [42–44]. Thus, when oxygen becomes plentiful the process of lineage commitment begins while pluripotent cells remain tucked away in their own microenvironment. How do cells sense oxygen in their landscape and utilize it to shift their epigenome and gene expression? A well-defined mechanism relating oxygen levels to gene expression is hypoxia inducible factors (HIF) transcription factors. This family consists of three members: HIF-1a, HIF-2a, and HIF-3a, all of which form a heterodimer with ARNT to bind hypoxic response elements [45, 46]. HIF stability is negatively regulated by oxygen through the activity prolyl hydroxylases (PHD). The PHD enzymes require the same quartet of cofactors required by TET and JmjC demethylases: oxygen, α-ketoglutarate, Fe²⁺ and ascorbate [47]. Under high physiologic oxygen tension these dioxygenase enzymes hydroxylate prolyl residues on HIF to flag it for ubiquitinoylation by von Hippel-Lindau (VHL) and eventually degradation in the proteosome [48]. During hypoxia, or when physiological oxygen levels are how, PHD enzymes are less active leading to increased stability for HIF and activation of hypoxic response genes.

JmjC demethylases may also serve as oxygen sensors. Recently there have been several reports looking into the influence of hypoxia on JmjC demethylases and how it shifts the epigenome to alter gene expression. These studies demonstrate that three KDMs: KDM4a, KDM5a and KDM6a can all serve as oxygen sensors to directly influence chromatin structure in response to changing levels of oxygen [49–51]. Biochemical characterization of KDM5a has revealed its affinity for oxygen matches that of PHD enzymes. This suggests that the activity changing oxygen levels would have a similar effect on both types of enzymes [51]. Hypoxic changes in chromatin structure seem to be specific as well. Using ChIP-sequencing it was revealed that brief periods of hypoxia can lead to changes in H3K4me3 and H3K36me3 histones at genes associated with hypoxic signaling [49]. Oxygen sensing by KDMs might also influence development [52]. The increased expression of myogenic factors commits cells towards a muscle specific lineage. Driving this change is KDM6a mediated demethylation of H3K27me3 [53]. Under hypoxic conditions KDM6a is inactive and unable to remove H3K27 methylation, blocking differentiation. The utilization of oxygen by LSD1 links it to hypoxia as well [31]. This suggests that the activity of LSD1 might also be sensitive to variable levels of oxygen encountered during development we discuss here. The TET family of demethylases are also potential oxygen sensors. Work form Alison Brewer’s group has shown that the activity of the TET family is dynamic during development and affected by oxygen levels [54]. Their observations coordinate with observed changes in 5mC and 5hmC during development [55]. Clearly the presence of oxygen impacts the early epigenome. We place oxygen atop our epigenetic landscape (Fig. 2) because it directs the erosion of the slope that is guiding cells through development. By manipulating metabolism and ROS production, oxygen facilitates the creation of cofactors and redox signaling events that affect epigenetic enzymes. Thus, upstream events forming the epigenetic landscape (i.e. oxygen) guide erosion lower on the slope to allow other factors to continue the process of guiding development.

Figure 2. Metabolites, oxygen and reactive oxygen species are major topographical features on the epigenetic landscape of development.

Conrad Waddington originally depicted the epigenetic landscape as a “symbolic representation of developmental potential for a genotype” where the conditions within the fertilized egg form slopes and canals that influence develop the eventual phenotype. We extend upon Waddington’s concept in our landscape to denote the current roles oxygen (O₂) and metabolites such as α-ketoglutarate (αKG), S-adenosylmethionine (SAM), the NAD⁺/NADH ratio have on influencing the epigenome of cells during lineage commitment. We also indicate mechanisms of redox signaling through superoxide (O₂^•−), glutathione (GSH/GSSG), ascorbate (ASC) and iron II (Fe²⁺) as fundamental components creating the epigenetic landscape of development.

Free Radicals and redox signaling in development

An accepted and general concept of free radical biology suggests that perturbations in ROS production be countered with their removal by varying antioxidant systems. Likewise, a fundamental precept of the original Free Radical Theory of Development stipulates that in order for development to proceed an imbalance must be created that signals development to begin [3–6]. How can an imbalance be created and by what means does it motivate the epigenome to unfold a genetic program? The metabolic apparatus is a likely source of ROS in early development. During development the creation of O₂^•− and H₂O₂ constitutes a redox core. From this core a series of one or two electron signaling events is transduced through progressive redox reactions to ultimately shift the epigenetic landscape (Fig. 3). Below we will illustrate these processes to visualize how redox signaling through O₂^•− and H₂O₂ can influence epigenetic processes.

Figure 3. Electron signaling can create tectonic shifts in the epigenetic landscape.

One- or two-electron signaling events are transduced from a redox core consisting of superoxide (O₂^•−) and hydrogen peroxide (H₂O₂). One electron reactions between O₂^•− and iron in Heme, iron sulfur clusters (Fe-S) and the labile iron pool (Fe²⁺/Fe³⁺) can influence oxidative phosphorylation, proline hydroxylation, lysine demethylation and DNA demethylation. By altering these functions, O₂^•− can facilitate one-electron signaling to shift the tectonic plates of acetylation and demethylation. H₂O₂ can facilitate two-electron signaling through glutathione (GSH) and alter sulfhydryl to disulfide bridges (SH/SS) in redox switches such as thioredoxin and peroxiredoxin. These redox alterations can influence the production of S-adenosylmethionine (SAM) to alter methylation by DNA (DNMT) and histone (HMT) methyltransferases.

Nitric oxide

Nitric oxide (NO) is another free radical species that is important in regulating development. NO affects cGMP signaling, is rapidly destroyed by O₂^•− and interacts with iron in multiple ways. The production of NO is carried out in several cellular compartments by nitric oxide synthases (NOS) [56]. Furthermore, NO is uncharged and relatively stable, granting it the ability diffuse long distances and affect signaling in neighboring cellular compartments [57]. Preimplanted embryos require NO and NOS activity to initiate the formation of vasculature [58]. Inhibiting NOS results in the death of two and four cell stage embryos [59]. These roles of NO can be attributed to it inducing cGMP signaling pathways. Reactions with iron are also an important means by which NO exerts its influence on biology. For example, NO reacts with iron bound by proteins to influence their function [60]. In this manner, NO can perturb DNA and histone demethylase activity in a comparable manner to O₂^•− as discussed above. NO also readily reacts with O₂^•− to form the highly reactive peroxynitrite (ONOO⁻) [61]. This reaction can influence the bioavailability of NO and the downstream reactions of O₂^•−, disrupting NO’s ability to carry out signaling [62].

Redox Metabolites influence epigenetics in development: Glutathione

Glutathione (GSH) biochemistry influences epigenetic processes by altering sulfur pools, impacting the bioavailability of one-carbon metabolites, and changing two-electron signaling (Fig. 3). Sulfur containing amino acids are critical for maintaining epigenetic regulation of gene expression. With its overall positive charge at physiologic pH, sulfur creates an excellent leaving group for biochemistry. This property of sulfur is key for epigenetic methylation reactions because it makes SAM easy to use, while creating a chemically stable byproduct SAH which is converted to homocysteine and eventually recycled back to SAM via the methionine cycle [63]. De-novo synthesis of glutathione creates an alternative route for homocysteine by shunting it into the transsulfuration pathway [64]. This process siphons sulfur away from the methionine cycle and its use for epigenetic methylation. The exit of sulfur by transsulfuration is generally limited; however, in times of oxidative stress, or when increased glutathione production is needed, this process can significantly drain metabolites from the methionine cycle [65, 66].

Redox balance via GSH production is vital during the initial states of mammalian development. As a general rule, a reducing environment favors fertilization and implantation, while the embryo as a whole decreases GSH production and becomes more oxidizing as it proceeds through embryogenesis [67, 68]. Control of redox by small molecular weight antioxidants is critical during this process. The concentration of glutathione (GSH), and its ratio to GSSG fluctuates during development. Knockout mice models lacking γ-glutamate cysteine ligase catalytic subunit (gclc) are embryonic lethal, while heterozygous littermates seem to develop normally with a 20% decrease in GSH content [69]. Depleting GSH through chemical means influences development. Exposing rat embryos to buthionine sulfoximine (BSO) creates malformation embryos [68, 70]. Methanol is a well characterized teratogen. Exposure to methanol disrupts the development of model systems through GSH that is independent of DNA damage, suggesting some other mechanism is responsible [71]. Keeping GSH levels abnormally high might prevent developmental progression. Insect metamorphosis is associated with broad ranging epigenetic changes [72]. Work in these model systems has shown that increasing GSH production potentially delays, or prevent the switches between lifecycle phases. Treating Spodoptera litura with exogenous juvenile growth hormone increases GSH production and results in supernumerary instar larvae that fail to pupate on schedule [73]. Overexpression of the catalytic subunit of glutamate-cysteine ligase (GCL) extends drosophila lifespan by increasing GSH production. However, the simultaneous overexpression of GCL catalytic and modifier subunits results in lethality during metamorphosis [74]. This situation decreases the Km of the enzyme for glutamine and ebbs the feedback inhibition of the enzyme complex by GSH, resulting in an aberrantly high GSH/GSSG ratio. This could potentially interfere with the redox switch required during this stage of development and result in the observed lethality.

It is clear that the biochemistry of de-novo GSH production via transsulfuration and epigenetic methylation are intertwined (for review see [75]). During gametogenesis cells undergo epigenome wide remodeling that results in CpG hypomethylation throughout the genome [76]. This is occurring while gametes ramp up their production of GSH [77]. The pool of GSH being created is likely priming the system to protect the embryo after fertilization and before implantation on the uterine wall. The preimplantation phase is critical for both redox biology and epigenetics [78]. For redox biology, ES cells require a robust antioxidant defense via the efficient recycling of glutathione disulfide (GSSG) back to GSH. Likewise, high glutathione levels are associated with increased ES cell stemness [79]. This is achieved by glutathione reductase using NADPH created via the oxidative pentose phosphate pathway (OPPP) [80]. Likewise the activity of the OPPP decreases after fertilization, with it being the highest at the two cell stage and falling precipitously as it differentiates into a blastocyst [81]. While activity of OPPP is high, the de novo production new GSH is low. This suggests that the developing embryo utilizes a recycling approach to maintain GSH levels as opposed to de novo production. Taken together these findings suggest the methionine and transsulfuration pathways are uncoupled at these early stages of development, potentially to preserve sulfur for proficient formation of an early stage epigenome. Indeed, as the embryo undergoes its first few cell divisions and up until a blastocyst is formed, dramatic changes in the level and placement of DNA methylation are observed [82, 83]. Meanwhile, histone methylation is setting up cells to begin the complex process of establishing cell type specific gene expression.

Folate

Folate is an essential vitamin that is required to facilitate DNA, RNA, and protein synthesis in the dividing cells of an embryo. Folate deficiency can create neural tube defects in the brain and spinal cord such as spina bifida. Supplementing folate during pregnancy can decrease the risk of these defects forming [84]. Folate has a profound influence on the epigenotype of the developing fetus. Modifying the folate intake of pregnant mice influences the coat color and body types of their unborn pups. These changes can be linked to imprinting by CpG methylation at the agouti locus [85–87]. The agouti model has also proven to be useful in deciphering how alterations in cellular redox created by ionizing radiation might influence DNA methylation. Exposing unborn pups to low dose ionizing radiation induced hormesis and increased CpG methylation at the agouti locus [88]. In the same study, administration of an antioxidant diet to pregnant dams prevented hormesis and resulted in hypomethylation at agouti. An intimate relationship exists between radiation hormesis and redox signaling via hydrogen peroxide, transition metals, and increased glutathione production [89, 90]. This creates a primed state which defrays the expenses incurred by additional oxidative load. These attributes of radiation hormesis also have the potential to influence the activity of epigenetic enzymes as we have described above. Exposure to high doses of radiation can likely influence the epigenome through transient changes in cellular redox physiology. Higher doses of ionizing radiation (>1 Gy) elicit oxidative stress in pluripotent stem cell cultures and influence stem cell fate [91, 92]. Likewise, high dose radiation exposure induces loss of DNA methylation in adult mice [93]. This further suggests that a link may exist between redox signaling and DNA methylation.

In humans, dietary folate may also influence CpG methylation. Folate supplementation by pregnant woman alters DNA methylation in newborns and affects their unborn child’s body type [84, 94, 95]. How is folate manipulating the epigenome, and is it linked to redox biology? Folate is linked to the methionine cycle and able to fluctuate the flow of metabolites through it. Folate mediated one-carbon (FOCM) is carried by analogous pathways occurring in the mitochondria, cytosol and nucleus [96]. Metabolites can freely move between these pathways to promote their respective activities depending on the cell’s current metabolic need. Formate, a key FOCM metabolite sourced from the mitochondria, influences epigenetic regulation of gene expression through SAM. Exogenous formate increases the level of DNA methylation in cells by bolstering the level of metabolites available to one-carbon metabolism [97]. Disrupting mitochondrial function also influences FOCM and the level of metabolites within its associated pathways. Depleting mitochondrial DNA (mtDNA) increases the flux of metabolites through transsulfuration, one-carbon metabolism, and methionine pathways [98–100]. These alterations lead to increased levels of SAM, and DNMT mediated methylation of CpGs which alter gene expression. [99]. The extent to which signaling between the mitochondria and nucleus influences stages of development is obscure, however changes in mitochondrial anatomy and physiology are associated with mammalian development. Early mitochondria are spheroid and underdeveloped, and as development proceeds they form into fully mature and functioning mitochondria [101]. These early embryonic cells avoid using oxidative phosphorylation (OxPhos) as means to produce energy, which possibly creates similar metabolic dynamics to what is observed during mtDNA depletion [102]. Currently such information is lacking and would fill the gap in our understanding of how mitochondria influence FOCM and methylation reactions.

Deficiencies in folate can slow the conversion of homocysteine to methionine, creating hyperhomocysteinemia [103]. This can occur through alterations in the methylenetetrahydrofolate reductase (MTHFR) gene, which can also lead to decreased levels of SAM and affect global DNA methylation patterns [104]. Such defects also influence fertility and development. The MTHFR 677C>T mutation is associated with decreased levels of DNA methylation in gametes even with folate supplementation [105]. Decreasing folate bioavailability by methotrexate, or through a deprivation diet, decreases SAM while increasing SAH [106, 107]. Folate supplementation during pregnancy in humans has the opposite effect. It decreases homocysteine levels in cord plasma and results in a concomitant increase in DNA methylation at LINE-1 repeats [108]. Combined, these illustrate the sway folate has over epigenetic methylation by limiting the availability of SAM and its regeneration from SAH or homocysteine. Folate impacts cellular redox by maintaining sulfur within the methionine cycle. Methionine synthase (MS) transfers methyl groups from the folate derivative 5-methyltetrahydrofolate to homocysteine [109]. During oxidative stress or GSH depletion folate conversion to 5-methyltetrahydrofolate is increased as a means to power MS and replenish the methionine pool [110]. Thus, this single step of the methionine cycle, mediated through folate, simultaneously affects redox biology and epigenetics. By recycling methionine, the exit of homocysteine to transsulfuration and glutathione production is prevented while a future methyl leaving group is loaded to sulfur for future transmethylations.

Ascorbate

Ascorbate is an imposing feature in the epigenetic landscape. Many mammals are capable of producing the ascorbate they need. However, humans, other primates, and guinea pigs are key examples of organisms that cannot and therefore must assimilate the ascorbate they require from dietary sources. Ascorbate is a versatile electron donor and enzymatic cofactor. Numerous studies have demonstrated that ascorbate influences the activities of both by TET and JmjC demethylases [111]. As we have discussed above, both families of demethylases have members that are critical to molding the epigenome in early development. What role is ascorbate playing in regulating the activity of these enzymes? It was originally reported that these enzymes really don’t require ascorbate in the active site to function [22]. However, further studies determined that TET and JmjC enzymes likely require the presence of ascorbate to regenerate Fe²⁺ if it gets oxidized to Fe³⁺ [112]. Studies in embryos have also shown that the levels of 5hmC within them is impacted by ascorbate [113]. Can the redox chemistry of ascorbate form the epigenetic landscape by other means?

Before being identified as an epigenetic metabolite, ascorbate was appreciated as a small molecular weight antioxidant. The redox chemistry of ascorbate in biological systems is complex (for review see [114]). It is water soluble, and able to reduce other species by swiftly donating two electrons through successive one electron oxidations to produce dehydroascorbic acid (DHA). Ascorbate is also capable of autoxidation, which generates H₂O₂, and proficient at reacting with heavy metals [114]. This makes ascorbate a player in both one- and two-electron signaling to influence the methylation status of the epigenome (Fig. 3). The redox chemistry of ascorbate could also potently influence methylation. We have mentioned that the period between fertilization and implantation is a time of epigenetic flux and dynamic redox chemistry. Once ascorbate has donated electrons and becomes DHA, it can be recycled by reducing it. Glutathione can directly reduce DHA back to ascorbate or through the glutaredoxin thiol shuttling system [115, 116]. Both of these mechanisms are likely at work in the developing embryo. We can therefore consider that ascorbate recycling could affect TET or JmjC mediated demethylation. However, it is even more tantalizing to speculate that utilizing GSH to rejuvenate ascorbate for demethylation can affect SAM availability and hence methyltransferase activity. The significant demethylation observed in early development would likely require a concomitant increase in ascorbate recycling mediated by GSH. To balance this, sulfur would be shunted into transsulfuration and de novo GSH production. This could result in the perturbation of DNMT and protein methyltransferase by decreasing the bioavailability of SAM we discuss above. Thus, redox signaling through GSH and ascorbate can potentially from a redox node in development to negatively regulate transmethylation while simultaneously supporting demethylation of the epigenome. Linking GSH and ascorbate to epigenetic regulation of gene expression is the quintessence of the Free Radical Theory of Development. Further investigation is warranted to understand how this redox node influence the epigenetic framework of cells.

Nexus of metabolism and epigenetics

An organism is reliant upon its metabolism to maintain life, effect its growth and facilitate reproduction. To achieve this goal, organisms require an electron rich source paired with a commensurate metabolic apparatus capable of transferring electrons to an ultimate acceptor. Shifting microenvironments can impact the availability of food sources and the electron acceptor. Likewise, the need to grow must be weighed against sustaining life and reproduction. Similar to gauges on a dashboard, an organism must “read” the flux of metabolites and match it with an appropriate metabolic apparatus to achieve its biological goal. The integrated cellular adaptive responses to such metabolic alterations are affected in the short-term by transient transcriptional activation, but in the longer-term via heritable epigenetic alterations that remodel chromatin structure and generate stable phenotypes. A stylized depiction of the interaction between metabolism, environment, and the molecular machinery participating to drive mammalian cellular differentiation and development is shown in Figure 4. Altering gene expression to meet metabolic flux is reiterated from bacteria to complex multicellular organisms, it’s the strategies each employs which makes them differ. In mammalian development factors such as iron, oxygen, NAD⁺ and Kreb’s cycle intermediates are all employed to generate the multitude of phenotypes required.

Figure 4. Metabolism powers the epigenetic clock to drive cell-type specific gene expression and propel aspects of development.

The major by-product of respiration, and a major driver powering this epigenetic clock is superoxide (O₂^•−) (gray worm gear) in the mitochondrial matrix where it can direct one-electron signaling to increase the lability of iron from Fe/S and non-heme Fe-proteins, thus perturbing electron flow through ETC and altering levels of mitochondrial metabolites that also serve as co-factors (red gears) for epigenetic writer and eraser enzymes (yellow gears). Dismutation of O₂^•− to H₂O₂ re-routes the power to two-electron signaling processes involving thiol oxidation (SH/SS) and its resolution. The net result of these metabolically integrated power trains (blue gears) within cells and tissues is to sculpt an epigenetic landscape characteristic (orange) of a specific cell-type in a multicellular organism (green).

Iron

Iron is a transition metal that is full of dichotomies. Iron’s chemistry is essential for life by facilitating energy production, but it also wastes energy by facilitating unwanted electron transfers. Furthermore, iron is a builder, an essential component in biosynthetic processes, while simultaneously being able to destroy what it builds through the production of ROS. Because of these dichotomies iron metabolism is tightly regulated. Iron is redox active and undergoes one electron reactions to cycle between ferrous (Fe²⁺) and ferric (Fe³⁺) states; while it shifts between these two forms it is capable of producing ROS. Biological systems safely store iron by sequestering it into transferrin and ferritin [117]. These proteins prevent iron’s redox chemistry, but also limit its availability for protein biosynthesis. To enable protein and enzyme production, cells maintain a small labile iron pool (LIP) [118]. The free ferrous (Fe²⁺) iron within the LIP comprises less than 2% of total iron within cells [119]. While necessary to create the catalytic function of many proteins, the LIP is redox active and apt to undergo reactions that generate ROS and likely able to transduce one electron signaling events [118].

Iron metabolism in the mitochondria

Mitochondria are centers of iron metabolism. This requires trafficking of iron from the LIP into the mitochondrial matrix where the machinery for heme and iron sulfur cluster (FeS) biosynthesis awaits its arrival [120]. Iron sulfur clusters come in many forms, but [2Fe-2S], [3Fe-4S], and [4Fe-4S] are the most common [121, 122]. After biosynthesis FeS clusters can be exported as prosthetic groups for enzymes in DNA replication, DNA repair, protein synthesis, and ribosome assembly [123]. The FeS clusters that remain in the mitochondria power a litany of enzymes in the tricarboxylic acid cycle (TCA) and electron transport chain. Aconitase, flavoproteins, oxidoreductases, Rieske iron-sulfur proteins, and succinate dehydrogenase are key examples FeS proteins in the mitochondria [121, 123]. Iron sulfur clusters are versatile and efficient at electron transferring reactions, but they are vulnerable to redox attack by free radicals [122]. O₂^•− can inactivate aconitase and several electron transport chain enzymes by attacking these FeS cores [124, 125]. This can have a tremendous effect on metabolism by limiting the flow of metabolites through the TCA cycle and OxPhos. Is it possible for O₂^•− to influence the epigenome through one electron signaling? Selective deletion of manganese superoxide dismutase (MnSOD) in hematopoietic stem cells increases O₂^•−, which in turn inactivates mitochondrial FeS containing enzymes aconitase and ferrochelatase, disrupts heme biosynthesis, and elicits epigenetic alterations [126]. Others have suggested that O₂^•− creates perturbations in iron homeostasis during the etiology of cancer [119]. Likewise, signaling through the aberrant production of reactive oxygen has been hypothesized as critical to the initiation, promotion, and progression of cancer [119, 127, 128]. ROS produced from aberrant metabolism might create redox signaling events in cancer, but development is very different with respect to is free radical biology. How could metabolism, O₂^•−, and iron signaling manifest epigenetic alterations in development?

Iron in epigenetics

The catalytic activity of TET and JmjC demethylases is dependent upon Fe²⁺. Likewise, iron homeostasis is important in stem cells. The nascent mitochondria of these primordial cells exhibit limited OxPhos; however they would remain active sites of iron metabolism to produce FeS clusters and heme for the rest of the cell [129]. Iron loading inhibits the self-renewal of pluripotent stem cells through ROS mediated damage to DNA [130]. While this elicits differentiation, it also results in cell death. Others have shown that removing the LIP from human pluripotent stem cells by deferoxamine chelation induces their differentiation. Analysis of surviving cells revealed that the loss of pluripotency could be attributed epigenetic repression of NANOG by increased H3K27me3 at its promoter which was attributed to decreased demethylase activity [131]. Fetal-neonatal iron deficiency (ID) is a condition associated with impaired neurodevelopment that impacts cognitive function and social/emotional delay in rodents and humans [132]. These deficits can be correlated with decreased expression of genes such brain-derived neurotrophic factor (BDNF) in the hippocampus [133]. In the earliest stages of mouse development BDNF exists in a bivalent chromatin state [134], meaning its histones are simultaneously marked by active (H3K4me3) and repressive (H3K27me3) histone methylation marks [135]. Bivalent chromatin is believed to allow genes to remain poised to allow either rapid activation or repression in response to environmental cues [136]. This suggests that removal of repressive histone methylation marks at genes such as BDNF is be a critical step to insure full activation of the neuronal program. Iron dependent JmjCs have a role in this deciding the fate of bivalent chromatin. Hypoxia has been shown to induce the formation of bivalent chromatin marks by decreasing the activity of KDM6a, and lends credence to a similar possibility for iron during development [137]. Currently, the precise role of iron in regulation the fate by influencing bivalent chromatin remains unknown.

Redox cycling of iron

Redox cycling within the LIP is another potential means to create transient alterations in Fe²⁺ availability. In this context the switching of iron between its redox states can potentially transduce various ROS and antioxidant signals to alter the epigenome via enzymes that demethylate DNA and histones. How can one electron signaling via reactions between ROS and iron induce epigenetic changes? Redox switching through iron in the LIP could potentially alter its bioavailability to TET and JmjC demethylases. Both enzyme classes are sensitive to iron’s availability within the LIP. As we have discussed above, removing iron via chelation experiments decreases JmjC activity and impacts histone methylation patterns [131]. Increases in the LIP can spur the formation of 5hmC from 5mC. Recent work from Gaofeng Wang’s group has demonstrated that cAMP can trigger the release of Fe²⁺ from endosomes to transiently increase the concentration of the LIP which ultimately increased TET oxidation of 5mC to 5hmc [138]. We have previously mentioned that knocking out MnSOD is capable of disrupting iron homeostasis and eliciting epigenetic changes in hematopoietic precursor cells [126]. This can be attributed to O₂^•− disrupting FeS clusters heme production in the mitochondria, which in turn increases the concentration of iron within the LIP [126, 139]. However, complete SOD2 knockout mice develop normally in utero and only succumb to oxygen toxicity soon after parturition [140]. This observation likely excludes O₂^•− and iron originating from the mitochondria as a potential redox signaling molecule in epigenetics. However, it leaves open the possibility that redox cycling of iron within the LIP can affect TET and JmjC demethylases. How can iron sense ROS to signal changes in the nucleus?

Iron, ROS and antioxidants play important roles in the nucleus and potentially influence development. Tradition has been held that the presence of ROS and heavy metals in the nucleus is dangerous and creates damage. However, more contemporary findings suggest this knowledge may not be complete. NAD(P)H oxidase 4 (Nox4) localizes to the nucleus and influences gene expression by producing O₂^•− [141]. The production of the O₂^•− by Nox4 in the nucleus might also serve a role in maintaining stemness and self-renewal by activating the expression of transcription factors such as Oct4, SSEA-4, and Sox2 [142, 143]. There is evidence in the literature to support the existence of a LIP within the nucleus, especially following iron’s release from lysosomes [144–146]. To regulate this redox milieu, nuclei also have a panel of redox mechanisms to potentially prevent interactions between O₂^•−, H₂O₂ and these heavy metals [147, 148]. The shuttling of electrons between species can have dramatic effects on cellular function and DNA methylation (Fig 5). The dioxygenase activity of TET demethylases require iron to remain in its Fe²⁺ redox state, however these enzymes employ a Fe⁴⁺-oxo intermediate during their catalytic mechanisms to oxidize methyl groups [149]. During subsequent steps within the mechanism iron is recycled back to its Fe²⁺ state, however, it can become aberrantly reduced to Fe³⁺. This shifting the redox state of iron from Fe²⁺ to Fe³⁺ inactivates TET and demethylases. As such, TET demethylases bind Fe³⁺ with low affinity and readily replace it with Fe²⁺ that likely originates from the nuclear LIP [150]. Newly released Fe³⁺ can be subsequently reduced to Fe²⁺ by ascorbate to regain its requisite valency and join the LIP or become sequestered by nuclear ferritin [151–153]. ROS reacting with the LIP could potentially remove Fe²⁺ from use by these enzymes and hence change methylation. Attacking iron within the dioxygenases has been shown to inactivate them. KDMA3A demethylases can also be inactivated through reactions between the catalytic iron and NO, leading to accumulation of H3K9me2 in cells [154]. H₂O₂ has been shown to directly inactivate other members of the Fe²⁺/α-ketoglutarate dependent dioxygenase family [155]. Using a cell free system, Max Costa’s group has shown that H₂O₂ can inhibit JmjC and TET demethylases; however, adding additional Fe²⁺ can reinstate activity [156]. Furthermore, they have demonstrated that oxidative stress increases the global level of H3K4me3 and H3K27me3. This lends credence to the concept that iron’s redox switching is important in regulating DNA methylation. Combined, it appears that the nucleus has the method (O₂^•−, H₂O₂) and the means (Fe²⁺ dependent demethylases) to facilitate redox signaling. H₂O₂ is also produced as a byproduct during the demethylation of histones by KDM1a/LSD1 [31]. The H₂O₂ produced by LSD1 mediated demethylation creates local areas of oxidative stress which oxidizes guanine to 8-oxo-guanine [157, 158]. Theses successive oxidization events constitute a redox signaling mechanism, grounded by epigenetic methylation, that increases expression of the targeted gene [157, 158]. LSD1 and the JmjC demethylase KDM5b can be found to interact within the same protein complex [159]. It is therefore tempting to speculate that H₂O₂ produced by LSD1/KDM1 influences the iron of JmjC demethylases. This could potentially influence gene expression by comprising an epigenetic crosstalk between the two enzymes, or by serving as a means to create localized oxidation of guanine to 8-oxogunine. Why employ a redox sensitive mechanism as a means to elicit differentiation? The production of ROS is strongly correlated with iron and oxygen. During the original rendering of the free radical theory of development it was conceived that developmentally associated gene expression occurs in response to environmental conditions such as oxygen and nutrient availability. As we discuss here, these mechanisms work to elicit lineage commitment under conditions favorable to cell division and growth. Such a sensing mechanism likely continues to function after leaving the womb. When adult stem cells leave their niche these redox mechanisms sense it to signal the development of mature function cells. We suggest that reactions between ROS and iron from a redox switch or rheostat to influence the demethylation of histones and DNA, alter the topology of chromatin, and influence gene expression as an adaptive response to such cues.

Figure 5. Redox switching of the labile iron pool (LIP) influences the activity of Jumonji C domain containing histone demethylases (JmjC) and Ten-eleven translocation methylcytosine dioxygenases (TET).

The reactions between reactive oxygen species and iron influences the cellular functions in multiple cellular compartments to influence development. In the mitochondria, aberrant reduction of oxygen can create superoxide (O₂^•−) that disrupts iron sulfur clusters (Fe/S) and heme biosynthesis. In the cytosol, Fe²⁺ exists as free iron in a LIP where it can be assimilated into the catalytic core of enzymes. Fe³⁺ availability is tightly regulated by iron binding proteins such as ferritin that can transport iron into the nucleus. TETs and JmjCs require and Fe²⁺, O₂, and α-ketoglutarate (αKG) during the oxidation of 5-methylcytosine (5mc) in DNA and demethylation of methylated lysines (meK) within histone tails. Spurious events during the respective mechanisms of these enzymes can result in incorrect reduction of iron to its Fe³⁺ state, rendering them inactive (orange pentagon). To regain their activity, these enzymes likely release Fe³⁺ where it is readily reduced by ascorbate (Asc⁻) to create Fe²⁺ and ascorbate radical (Asc^•−). Fe²⁺can then either be reintroduced into enzymes or remain within the nuclear LIP where it can react with ROS originating from enzymes such as NAD(P)H oxidase 4 (NOX4). These redox reactions can then decrease Fe²⁺availblity to JmjC and TET enzymes. Free radical and ROS scavenging systems such as superoxide dismutases 1 and 3 (SOD1/3), glutathione peroxidases (GPx), peroxiredoxins (Prx) work to counter these reactions from happening. Thus, modulating the redox status of iron is one means by which the activity of TET and JmjC enzymes can be controlled.

TCA cycle metabolites influence methylation

Stem cells rely heavily upon glycolysis as a means to produce energy [102, 160]. It has been suggested they do this for two main reasons. First, stem cells niches are often low in oxygen, which naturally sets them up to utilize glycolysis. Second, it is likely to avoid the overproduction of ROS from which they cannot recover [43]. Cells in the earliest phases of development contain spheroid shaped mitochondria with underdeveloped cristae, yet they are fully capable of producing ATP by OxPhos [101]. As development proceeds, a gradual shift toward OxPhos occurs which coincides with the increased availability of oxygen [102]. As burgeoning mitochondria become increasing more metabolically active a concomitant increate in the concentration, and flux of OxPhos metabolites is expected within cells. This shifting of mitochondrial bioenergetics during early development likely has profound effects on the epigenome by influencing the availability of epigenetic metabolites such as α-ketoglutarate. Indeed, treating primed human pluripotent stem cells with α-ketoglutarate elicits their differentiation by globally decreasing DNA and histone methylation which can be attributed to increased TET and JmjC activity [161]. The relative flux of individual metabolites other than α-ketoglutarate through the TCA cycle can influence the activities of demethylase enzymes. The two Krebs cycle intermediates fumarate and succinate are competitive inhibitors of α-ketoglutarate binding to TET and JmjC demethylases [162]. Knocking MnSOD in erythroid precursor cells levels O₂^•− unchecked which abrogates succinate dehydrogenase activity and allows succinate to accumulate. This decreases TET enzyme activity diminishes the ability of these precursor cells to properly differentiate into erythrocytes [163]. Mutations in succinate dehydrogenase and fumarate hydratase are associated with many cancers. It has been suggested that these mutations assist in the etiology of cancer by inhibiting the activity of demethylases by making cells replete with succinate and fumarate [162]. Could similar TCA dynamics have a role in development? Hematopoietic stem cells (HSCs) illustrate the potential for mitochondrial metabolism to direct epigenetic alterations. HSCs are quiescent, reside in hypoxia and utilize glycolysis as a means to produce energy [102]. Loss of mitochondrial carrier homologue 2 (MTCH2) in HSCs induces OxPhos activity and ROS production in the mitochondria and prompts cell fate [164]. Increasing OxPhos activity would be associated with increased flow of metabolites through the TCA cycle such as α-ketoglutarate, fumarate, and succinate. These metabolites can also inhibit the activity of the PHD enzymes that modulate HIF stability [165]. From this example it seems probable that increasing the scope of metabolism during development can direct epigenetic alterations by influencing the activities of histone and DNA demethylases.

Sirtuins

Sirtuins are a family of NAD⁺-dependent protein deacetylases with diverse functions in distinct cellular compartments. These enzymes were originally discovered based in their role in extending lifespan and influencing gene expression in yeast and other model organisms. Furthermore, they have also garnered additional attention for their connection to caloric restriction, cancer, aging, and neurodegenerative diseases [40, 166, 167]. Humans have seven sirtuin family members: Sirt1, Sirt2, Sirt3, Sirt4, Sirt5, Sirt6, Sirt7, which utilize NAD⁺ to target a diverse set of modified proteins. Many of these enzymes have been well characterized. The sirtuins localize to several subcellular compartments and target a wide range of modifications and modified proteins [40]. The relative affinity for NAD⁺ varies among the Sirt family members. This varying affinity might allow for specialization within the sirtuin family to accommodate the spectrum of NAD⁺/NADH ratios they encounter within their respective subcellular locale [168]. Utilizing NAD⁺ as a cofactor and being amendable to its changing concentrations links sirtuins to metabolism.

The NAD⁺/NADH ratio can have profound effects on a cell. Caloric restriction influences lifespan by lowering the NAD⁺/NADH ratio [169]. The NAD⁺/NADH ratio also alters chromatin structure by influencing histone acetylation. For example, Sirt1 will differentially deacetylates acetylated H3K9 and H4K16 in response to changes in the NAD⁺/NADH ratio [170–172]. Mitochondrial sirtuins have the potential to guide development by impacting oxidative phosphorylation and free radical scavenging [40]. Deletion of Sirt3 results in the accumulation of acetylated proteins in the mitochondria, particularly in Complex I. This observation is accompanied by decreased Complex I activity and ATP production [173]. Acetylated MnSOD exhibits decreased O₂^•− dismutase activity versus its unmodified form, creating a switch to modulate oxidative stress in the mitochondria [174]. Several groups have reported that Sirt3 directly deacetylates MnSOD to increase its activity [175–177]. Disrupting Sirt3 leads to MnSOD hyperacetylation and decreased O₂^•− scavenging ability.

Sirutins might also have a role in mammalian development. It has been suggested that the activity of sirtuins is vital during gestation and regulate fertility by controlling oxidative stress [178]. All seven Sirt members are expressed in eggs and reimplanted mice embryos, however it appears that a majority of the family members do not directly influence development [179]. Sirtuin knockout mice exhibit differing phenotypes during development. The deletion of some sirtuins show no noticeable phenotype, while others can exhibit varying results depending on the system employed [180–186]. Sirt1 is the most studied in mammalian development because it shares a great deal of homology with the well-defined yeast sir2α gene [180]. There are also conflicting reports about Sirt1’s role in development. These reported differences can likely be attributed to the genetic background or the means by which the model was established [186–188]. Wang et al. reported almost all sirt1−/− pups perish in utero during early stages of pregnancy, while pups surviving to later in gestation have significant developmental defects which correlate with altered histone modifications [187]. Employing other strategies to generate Sirt1 mutant mice results in apparent normal growth in utero followed postnatal death or decreased body mass in adulthood [180].

Sirt1 might also have a role during the exit of pluripotency and establishing lineage commitment. Sirt1 is critical for maintaining cells in a naive pluripotent state by actively deacetylating Oct4 in epiblasts. When Sirt1 activity decreases, Oct4 acetylation increases, and is able to facilitate the creation of a primed pluripotency state that is ready for lineage commitment [189]. Sirt1 mediated deacetylation globally impacts histone acetylation patterns in mesenchymal stem cells (MSCs). Subjecting MSCs to caloric restriction activates Sirt1 to lock in their developmental fate [190]. The plant polyphenol resveratrol is capable of increasing sirtuin activity in vivo and mimic caloric restriction [191, 192]. This has prompted studies looking at whether resveratrol can impact cell fate and lineage commitment. Resveratrol can increase the rate at which adipose stem cells differentiate into chondrocytes, suggesting the potential of Sirt1 to activate an epigenetic switch to assist in this transition [193]. What could be influencing sirtuin activity during development in utero? A distinct microenvironment exists within the preimplanted embryo and during gestation. Even after parturition stem cell niches persist through adulthood [194]. Above we discussed oxygen as a key factor in these microenvironments, but so is glucose and how it is consumed in these cells. Given that stem cells rely heavily upon glycolysis as a means to produce energy they would likely consume high levels of glucose [101, 102, 160]. This would also coincide with a low NAD⁺/NADH ratio coupled with a lower yield of ATP. While glycolysis is inefficient at producing energy, it excels at making metabolic intermediates for the genesis of various biomolecules. Could sirtuins be a critical component in cellular stratagems to integrate metabolism accommodate microenvironmental influences? Sirutins indirectly, yet simultaneously, sense the current states of oxygen, glucose consumption, oxidative phosphorylation and metabolic flux within a cell through the NAD⁺/NADH ratio. They can then directly influence the metabolic apparatus to match these states and modify gene expression through the deacetylation of histone and non-histone proteins alike. This means that entry and exiting a specific microenvironment constitutes as an epigenetic switch for cells to propel lineage commitment.

Induced Pluripotent stem cells

The creation of induced pluripotent stem cells (iPSC) has opened a new therapeutic avenue for disease. In a first, Yamanaka and colleges were able to create a pluripotent stem cell like phenotype in fibroblasts through ectopic expression of the transcription factors Oct3/4, Sox2, c-Myc and Klf4 fibroblasts [195]. Almost immediately, it was determined that this reprogramming was concomitant with and was reliant upon comprehensive restructuring of the fibroblast cell’s epigenome [196–198]. This included reactivation of the silent X chromosome in female cells, global changes in DNA methylation and histone modifications [198]. Producing iPSC creates a poised phenotype capable of becoming any cell lineage through the acquisition of new epigenetic changes. Generating iPSCs is a straightforward process, but still fraught with challenges. These includes oncogenic transformation, heterogeneity, and the inability to successfully reprogram [199]. Many of these challenges are attributable to aberrant epigenetic reprogramming during iPSC generation. Progress has been made in understanding the iPSCs process, along with variations in the factors used since the original Oct3/4, Sox2, c-Myc and Klf4 fibroblasts [197]. Given that metabolism is closely linked to epigenetic alterations during development, could it have a role in creating iPSCs?

As we discussed above, the poised state of embryonic stem cells is amendable to oxygen, redox and metabolic intermediates. Manipulating the conditions under which iPSC are grown influences their viability and ability to reprogram. [160, 200]. Just like in utero development GSH levels are important. Conversion of cysteine to cystine in media can lead to GSH depletion in cells and affect their stemness and viability [201]. Demethylation reactions mediated by ascorbate might also have a role in reprogramming. The addition of ascorbate promotes the transitions of pre-iPSCs to a fully reprogrammed state [202]. The metabolism of iPSC might also be critical in their reprogramming. Reprogrammed IPSC have increased glycolytic activity and decrease their use of OxPhos [203]. Fibroblasts can also be “primed” to increase reprogramming efficiency by growing them at physiological concentrations of oxygen [204]. Both of these factors would have dramatic effects on the activities of epigenetic enzymes. Thus, by adjusting growth conditions an epigenetic landscape is synthesized to match the undifferentiated developmental state we are trying to reprogram fibroblasts into (Fig. 2). Overall, these observations further suggest that redox, oxygen and metabolism are critical components of the epigenetic landscape.

Summary

Epigenetics is constantly exerting influence over the genetic program. Here we have described how the free radical theory of development’s central canons of redox, oxygen, and metabolism form a device to drive the epigenetic processes guiding development [3, 4, 6] (Fig. 4). The central mechanism of this device is formed by oxygen, iron, ascorbate, Acetyl-CoA, α-ketoglutarate, NAD⁺, and SAM. These cofactors are essential to the activity of enzymes that initiating and perpetuating epigenetic regulation of gene expression. Furthermore, as we have discussed all of these factors are amendable to the temporal changes in metabolism and redox observed during the earliest stages of development through birth. Our desire to crack the mechanisms controlling gene expression led to our understanding of how DNA methylation and histone modifications control lineage commitment. Likewise, when the dynamics of epigenetic marks was observed in development, we sought out the responsible enzymes and their associated cofactors. Today we seek to understand how metabolism, redox and oxygen change with development to influence these cofactors. Given this, it’s perhaps apt that Conrad Waddington chose a map to depict the interactions between the epigenome and environment. The epigenetic landscape has evolved from a conceptual map to a literal one, guiding us to further explore what’s underneath the topographical features laying on top. It’s up to us to follow.

Oxygen and its derived species are potent morphogenic agents.

TCA metabolites and dietary nutrients influence the epigenome during development.

Labile iron is an important determinant in demethylase activity.

Developmental redox conditions manipulate the methionine cycle and the availability of SAM.

List of Abbreviations:

5mC

5-methyl cytosine

5hmC

5-hydroxymethyl cytosine

5fC

5-formyl cytosine

ACL

ATP citrate lyase

ARNT

arylhydrocarbon nuclear receptor translocator

ATP

adenosine triphosphate

BER

base excision repair

ChIP

chromatin immunoprecipitation

CpG

cytosine-guanine dinucleotide

DHA

dehydroascorbate

HIF

hypoxia-inducible factor

DNA

deoxyribonucleic acid

DNMT

DNA methyltransferase

ES

embryonic stem

FAD

flavin-adenine dinucleotide

FOCM

Folate mediated one-carbon

GSH

reduced glutathione sulfhydryl

GSSG

oxidized glutathione disulfide

HAT

histone acetyltransferase

HDAC

histone deacetylase

HMT

histone methyltransferase

IPSC

induced pluripotent stem cell

JmjC

Jumonji C domain containing histone demethylase

KDM

lysine demethylase

KMT

lysine methyltransferase

LINE

long interspersed nuclear elements

LIP

labile iron pool

LSD

lysine-specific demethylase

MnSOD

manganese superoxide dismutase

MS

methionine synthase

MSC

mesenchymal stem cell

mtDNA

mitochondrial DNA

MTHFR

methylene tetrahydrofolate reductase

NAD⁺

oxidized nicotinamide adenine dinucleotide

NADH

reduced nicotinamide adenine dinucleotide

NADPH

nictotinamide adenine dinucleotide phosphate

NO

nitric oxide

NOS

nitric oxide synthase

OPPP

oxidative pentose phosphate pathway

OxPhos

oxidative phosphorylation

PHD

prolyl hydroxylase domain containing protein

PRMT

protein arginine methyltransferase

RNA

ribonucleic acid

ROS

reactive oxygen species

SAH

S-adenosyl homocysteine

SAM

S-adenosyl methionine

TCA

tricarboxylic acid cycle

TET

Ten-eleven translocation methylcytosine dioxygenase