Chapter 12: Epigenetics

Epigenetics is defined as a heritable change in phenotype without a change in genotype. Although epigenetic mechanisms vary, this chapter focuses on the most common mechanism: chromatin. Changes in chromatin/epigenetics accompany many steps in transcription, replication, and recombination. However, the aspects of highest interest and relevance involve examples where epigenetic factors and enzymes drive differentiation decisions, and where misregulation/mutation of these factors drives pathologies, such as hematologic malignancies. This decision making must be precise, as differentiation along the lymphoid and myeloid lineages involves the regulated generation of multiple cell types in temporal order and proper proportion. Decisions are arrived through collaboration among signaling systems, transcription factors, and chromatin regulators-which together regulate the key genes governing self-renewal, differentiation, and survival. This chapter focuses on chromatin factors with central roles in these processes: ATP-dependent remodelers, DNA methylation (DNAme)/demethylation enzymes, and histone modification enzymes. As a complete treatment is beyond the scope of this chapter, the focus here will be conceptual, with particular examples provided to create a framework for understanding the many instances where chromatin factors influence decision-making.

Beyond their roles in normal blood cell development, misregulation of chromatin factors is now known to be common in hematologic malignancies. Indeed, high-throughput whole-genome and/or exome sequencing of leukemias and lymphomas has revealed mutations in many types of chromatin regulators, including mutations in chromatin remodelers, DNA methyltransferases (DNMT), and histone modification enzymes, as well as revealing fusion proteins that involve chromatin regulators.¹ In certain instances, modeling in the mouse supports these epigenetic mutations as the main drivers of the cancer phenotype. In other instances, epigenetic mutations cooperate with (and likely enable) additional genetic mutations, which cooperate to impact proliferation, survival, and plasticity, which can enable both cancer progression and therapy resistance. However, as many chromatin regulators are enzymes, they may be more targetable than mutations in DNA binding transcription factors, providing new therapeutic approaches.² This chapter expands on these concepts, addressing the mechanistic basis of chromatin misregulation in hematologic malignancies, as well as emerging therapeutic approaches.

CHROMATIN REGULATES TRANSCRIPTION FACTOR BINDING

Chromatin has a major impact on gene expression, mediated through interplay with transcription factors. Sequence-specific DNA-binding transcription factors are the most important factors in defining whether and when a gene is transcribed, and also define the locations and character of chromatin regions, as they target chromatin remodeling and modifying proteins. However, the initial chromatin landscape can control whether transcription factors have access to the DNA at a particular gene/region. Access to DNA is deterred by nucleosomes, the main repeating unit of chromatin structure, which can block the binding sites of transcription factors to chromatin.³ Likewise, DNAme can also block the binding of transcription factors, many of which will not bind DNA if the cytosine in their binding site is methylated (DNAme is discussed more extensively in the section “DNA Methylation and Demethylation”). Thus, the nucleosome and DNAme landscape together define the initial “open versus closed” chromatin state of a region with which the current repertoire of transcription factors within that cell type must contend. However, this landscape is dynamic, as signaling systems can modify transcription factors and chromatin components, altering their activity and the landscape both through their binding, and through their recruitment of nucleosome remodelers and chromatin modifiers.^3,4,5

CHROMATIN REMODELERS CONTROL GENOME ACCESS

ATP-dependent chromatin remodeling complexes (termed hereafter remodelers) conduct central roles in regulating nucleosome occupancy and positioning (Fig. 12–1).^3,4,5 For example, remodelers specialized for chromatin assembly (such as imitation SWI remodeler [ISWI]-family and chromodomain remodeler [CHD]-family remodelers) utilize ATP hydrolysis to facilitate tight-packed nucleosomes that lead to the occlusion of sites for site-specific DNA binding proteins, such as transcription factors. Access to chromatin at enhancers, promoters and other loci can be enabled by remodelers such as the switch and sucrose non-fermenting remodeler (SWI/SNF) complex, also termed the BRG/BAF-associated factors (BAF) complex, which can slide or eject the histone octamer, using the energy of ATP hydrolysis (Fig. 12–1). Notably, SWI/SNF can interact with (and facilitate the binding of) DNA-binding activators or repressors and can, therefore, help facilitate either activation or repression.^3,4,5 Here, the ability of activators or repressors to interact with SWI/SNF complexes can be influenced by signaling cascades, which impart covalent modifications that enable or disable protein interactions. Taken together, ISWI and CHD remodelers often act to silencing genes via site blockage at enhancers and promoters, whereas SWI/SNF remodelers promote site exposure at those locations (Fig. 12–1), which is important for gene activation.

Clear roles for remodelers in blood cell differentiation are emerging. For example, SWI/SNF components affect the pool size of fetal hematopoietic stem cells (HSCs), and also impact HSC (and progenitor) proliferation and survival.⁶ SWI/SNF complex is also used for myeloid differentiation to granulocytes and for multiple steps in thymocyte development. More specifically, in mice SWI/SNF binds the interferon-γ (Ifng) promoter, and is required for its full transcription. Furthermore, mutations in the adenosine triphosphatase (ATPase) function of SWI/SNF are known to reduce β-globin expression and to prevent erythroid differentiation.⁷ Notably, B-cell lymphoma (BCL) factors BCL7A and BCL11B, which are considered members of SWI/SNF complex in many cell types, are common in hematologic malignancies; for example, mutations in BCL7A are found in approximately 20 percent of non-Hodgkin lymphoma and multiple myeloma cases, and mutations in BCL11B are found in 6 to 12 percent of T-cell acute lymphocytic leukemias (ALLs).⁸

Roles for ISWI- and CHD-family remodelers include roles for the well-characterized CHD-family remodeler nucleosome remodeling and deacetylation factor (NuRD), which interacts with histone deacetylase (HDAC) enzymes to silence genes. The metastasis-associated (MTA) subunits of NuRD help target NuRD subtypes to particular genes through their interaction with transcription factors and chromatin modifications.⁹ For example, in B-lymphocytes, MTA3 interacts with BCL6, a major regulator of B-cell differentiation, targeting NuRD repression and preventing terminal differentiation into plasma cells.¹⁰ Remarkably, expressing BCL6 in plasma cells while MTA3 is functional results in a reversion of the cell fate and reprogramming into B lymphocytes.¹⁰ Furthermore, the recruitment of the ISWI-family complex nucleosome remodeling factor (NURF) to the early growth response protein 1 (EGR1) locus (important for thymocyte maturation) involves interaction with the transcription factor serum response factor (SRF) by the NURF subunit BPTF, enabling its stable binding to promoters.¹¹ Notably, Ikaros (which drives lymphoid differentiation) acts to inhibit both the ATP-dependent remodeling and HDAC activities of NuRD at target genes to enable activation rather than silencing.¹² Taken together, these and other examples illustrate the use of remodeler function and recruitment to activate or repress key genes in blood differentiation.

HISTONE MODIFICATION CONCEPTS: WRITE, READ, ERASE

The process of transcriptional regulation is accompanied by the ordered placement of particular histone modifications at enhancers, promoters, and coding regions. There are dozens of different modifications that occur on histones, with the most common modifications being acetylation, methylation, ubiquitylation, and phosphorylation. An inventory and functional analysis of all of these modifications, the enzymes that place and remove these modifications, is beyond the scope of this chapter; however, more important are the concepts, which can then be applied widely to various contexts.

First, the vast majority of histone modifications occur either on the extended aminoterminal “tails” of histones, whereas a minority also occur on the histone octamer “core.”¹³ The core of the histone octamer wraps the DNA, whereas histone tails serve as platforms for the regulated binding of proteins, and covalent modifications can either enhance or deter binding of chromatin remodelers, chromatin modifiers, and transcription factors, and help to orchestrate protein associations during transcription (Fig. 12–2). For example, methylation on histone H3 (H3) H3K4me can deter interaction with DNMTs and therefore cause passive DNA demethylation; in contrast, H3K4me3 can facilitate interaction with RNA polymerase II (RNAP II) machinery.¹⁴ Second, histone modifiers are typically targeted by site-specific DNA binding proteins (see Fig. 12–2), which are themselves responsive to developmental and environmental/metabolic signaling. Third, some histone modifiers are targeted or regulated by other histone modifications, which underlies (in part) why certain sets of histone modifications are coincident in regions.¹³

A major concept in histone modification biology is dynamic reversibility, termed write, read, erase.^1,15,16 “Writing” involves the enzymatic addition of a covalent modification to an amino acid, within a particular protein sequence context. “Reading” involves the ability of a second protein/domain to bind that modification, within a particular protein sequence context, defining the impact of the modification. “Erasing” involves the removal of the covalent modification, within a particular sequence context, regenerating the prior/initial state. These concepts are actually quite general, and can be applied widely in protein signal transduction biology, with this terminology simply having become popularized in the chromatin field. Nevertheless, these terms are quite useful for framing histone modification cycles that accompany transcription cycles. One illustrative example is the addition of histone acetylation by histone acetyltransferase (HAT) enzymes, the binding of acetylated histone tails by the bromodomain (present in SWI/SNF remodelers and certain chromatin modifiers),^17,18 and the removal of acetylation by HDAC enzymes.¹⁹ Finally, although this chapter discusses histone modifications and modifiers, histone modifiers often also modify additional chromatin proteins, including proteins that contain “histone mimic” regions. Although beyond the scope of this chapter, those chromatin modifications often go through similar “write, read, erase” cycles to enable additional layers of protein recruitment and release within a chromatin process.

Within the context above, a common theme in cell differentiation is waves of transcription factor–chromatin modifier interactions that define the current chromatin and transcription state, and also help prepare the enhancers and promoters of genes needed for future states/cell types (see Fig. 12–2). Signaling systems inform cellular differentiation decisions, and define the next differentiation state by affecting transcription factor activity and their interaction with chromatin factors, creating a forward loop. Quite often, the transcription factor–chromatin modifier interactions of the new state (and cell type) also feedback to inhibit the prior program, as well as alternative differentiation programs, so as to ensure the proper developmental trajectory. An example that illustrates part of this program in action involves the transition between HSCs and erythroid progenitors, which involves a switch in the abundance and activity of transcription factors (e.g., GATA2 to GATA1) and histone methyltransferase (HMT) paralogs (e.g., enhancer of zeste homologue 2 [EZH2] to EZH1).^20,21 This switch serves to repress a set of stem-related genes while activating a set of pro-differentiation genes. By extension, many studies show that loss-of-function mutations in chromatin factors can prevent developmental transitions, and if this block occurs at a highly proliferative progenitor stage, it can predispose to cancer.

Trained immunity refers to a type of memory in the innate immune system where genes that have been activated in the past (via infection or vaccination, termed stimulation) are “primed” for a more rapid and/or robust future response. Here, prior to initial stimulation, monocytes and macrophages bear “latent” enhancers neighboring proinflammatory genes, which lack histone modifications. Following stimulation, these latent enhancers acquire histone modifications (e.g., H3K4me and H3K27ac) that are correlated with gene activation and maintain those modifications for days following withdrawal of the initial stimulus.²² Importantly, the retention of these modifications is correlated with a more robust or rapid activation in response to a second stimulus. Thus, chromatin states can confer a memory of prior transcriptional states that shapes future response. Here, one can infer within Fig. 12–2 that following activation, this system does not return to the initial repressed state, but rather to an intermediate “poised” state where histone modifications are retained at the enhancer.

DNA METHYLATION

DNAme is a major component of epigenetic regulation in mammals, with central roles in gene and transposon silencing, imprinting, and X-chromosome inactivation.²³ Furthermore, DNAme can predispose to cancer by at least two routes: first, through the improper placement of focal DNAme, leading to the silencing of tumor-suppressor genes; second, through hypomethylation of the genome, causing genome instability.² Here, basic principles of DNAme and demethylation are first discussed, with a later section “Epigenetic and Hematologic Malignancies” focusing on their misregulation in hematologic malignancies.

DNAme primarily involves cytosine methylation in a CpG context, and in mammalian genomes the vast majority (>85 percent) of such cytosines are methylated. DNAme is conducted by DNMTs, involving the de novo enzymes DNMT3a and DNMT3b (which can methylate unmethylated regions) or by the maintenance enzyme DNMT1, which partners with ubiquitin-like with PHD and ring finger domains factor (UHRF1) to fully methylate hemimethylated CGs during replication.²⁴ DNAme confers silencing through two modes. First, DNAme inhibits or prevents the binding of many transcription factors with CG sites in their consensus binding sequence, including cMyb,²⁵ cMyc, E2F-family,²⁶ nuclear factor-κB,²⁷ CREB-family,²⁸ ETS-family, and AP2 factors.²⁹ Second, certain methyl-domain binding (MBD) proteins (e.g., MBD1, MBD2, MECP2) bind to methylated CpG sites and can recruit both HDAC, repressive HMTs, and CHD-family remodelers (e.g., NuRD) to establish and maintain repression.³⁰

Although most genomic CGs are methylated, mammalian genomes are punctuated by small regions (250 bp to 2 kb) where DNAme is notably absent, and these regions are strongly correlated with a high relative density of CG bases, termed CpG islands.³¹ (CpG islands are regions that have avoided the strong CG depletion that has occurred over the rest of the genome, as methylated cytosine can spontaneously deaminate to create uracil). Thus, methylated CGs are absent in regions where CGs are dense, a counterintuitive observation that underscores that CG-rich regions must attract active mechanisms to either prevent DNMT activity or remove DNAme (see section “TET Proteins and Active DNA Demethylation”). CpG islands reside in the promoters of most genes that are constitutively transcribed, such as housekeeping/metabolic genes, and these islands remain unmethylated under virtually all conditions and cell types. However, CpG islands vary in size and composition; those of intermediate CG density are often found at developmental genes; notably, these intermediate CpG islands are typically unmethylated in stem cells, but undergo developmentally regulated DNAme to confer silencing in cell types where their expression might confer alternative fates.³² Notably, CpG islands often contain binding sites for transcription factors; for those transcription factors that display methylation-sensitive binding (listed above), a lack of DNAme in these regions can permit their binding, whereas CpG island methylation can prevent binding. Taken together, proper regulation of DNAme is critical, as the improper placement of focal DNAme can lead to gene silencing.

DNAme is chemically highly stable, and therefore very useful in stable propagation of epigenetic states, even through germline inheritance. Although stable, DNA demethylation can occur, and does so mainly through two routes: (1) “passive” demethylation (dilution of DNAme by the failure to conduct maintenance DNAme after DNA replication), and (2) “active” demethylation, mainly involving proteins of the TET family. TET proteins are dioxygenase enzymes that oxidize the methyl group of 5-methylcytosine (5mC) to 5-hydroxymethylcytosine^33,34 (5hmC) and additional oxidized intermediates (not addressed further). Here, 5hmC and other intermediates cause DNA demethylation by one of two modes: first, following replication the maintenance DNMT1 and its partner, UHRF1, do not recognize these oxidized products as 5mC, leading to passive demethylation.²⁴ Second, glycosylases with roles in DNA repair (e.g., thymine DNA glycosylase [TDG]) can remove 5-hydroxymethylcytosine (5hmC) and similar oxidized intermediates, which are then replaced with an unmodified cytosine by base-excision repair systems.³⁵ An important aspect of TET-family dioxygenases is their use of iron and 2-oxoglutarate (2OG) as cofactors during the oxidation reaction; a feature that renders TET enzymes sensitive to concentrations of these metabolites and related compounds, which can act as inhibitors during metabolic dysregulation (see section “Misregulation of Dna Methylation/Demethylation In Hematologic Malignancies”).

CONCEPTS IN CANCER EPIGENETICS

Misregulation of epigenetic factors is common in cancer, and a higher level of understanding is achieved by recognizing recurring themes. Epigenetic factors are often misregulated by one of three modes: fusion, loss-of-function (via mutation or expression changes), or gain-of-function (via mutation or expression changes). Notably, each mode impacts the genome and transcriptome in a particular manner, as described below:

Theme 1: Fusion Proteins

Fusion proteins are commonly observed in hematologic malignancies, and are often the product of reciprocal chromosomal translocations. Common configurations involve fusions of DNA-binding proteins to chromatin modifiers, or proteins that interact with chromatin modifiers. This creates a dominant gain-of-function protein that targets chromatin-modifying activity to genes important for proliferation, development, or survival. A well-studied and conceptually informative example involves fusion of the aminoterminal portion of the HMT mixed-lineage leukemia (MLL) protein to other proteins that interact with chromatin modifiers.^36,37 Oncogenic MLL fusions are typically driven by the endogenous MLL promoter and retain the DNA-binding domain and additional regions present in the MLL aminoterminus, but omit the catalytic HMT domain normally present at the MLL C-terminus. MLL is normally part of a large complex that methylates histones (H3K4me3) at the promoters of active genes. Oncogenic MLL-fusion proteins can dimerize with normal full-length MLL, and retain the ability to bind DNA binding and also the ability to interact with chromatin and DNA-binding factors.

As previewed above, the fusion partner of MLL is often a protein that interacts with and recruits chromatin modifiers, providing a route to aberrant/constitutive recruitment of chromatin modifiers to particular loci. The most common MLL fusions involve fusion of the MLL N-terminus to the ALL1-fused (AF) genes (AF9) and AF4 proteins (partial internal tandem duplications are also leukemogenic, which likely also affect interactions with chromatin modifiers). Interestingly, the AF9 and AF4 partners are themselves members of more than one chromatin and transcription complex,^38,39 and therefore capable of recruiting a range of chromatin modifiers, including a H3 methyltransferase, DOT1 (which methylates histone H3K79),^40,41,42 or TIP60, CBP, and EP300 (which acetylate histones H3 and H4). AF4 is also a member of a complex important for transcriptional elongation,^38,39 which interacts with acetylated histone tails via a bromo and extraterminal (BET)-family bromodomain present in the BRD4 subunit. MLL fusions also involve direct fusion to a chromatin modifier, including fusion to the HAT enzymes CBP or EP300. (Fusion of MLL to TET proteins are covered separately in the context of DNAme in the section “Misregulation of Dna Methylation/Demethylation In Hematologic Malignancies”).

Regarding mechanism, current thinking supports the targeting of MLL fusions to constitutively activate key genes involved in blood development, causing a block in differentiation. This block (and continued proliferation) provides the opportunity for other genetic and epigenetic events that enhance proliferation and survival. Confirmed targets for MLL fusions include HoxA9 and the Meis1 gene in mouse, where the fusion contributes to their transcriptional activation.⁴¹ However, it is likely that a larger repertoire of target genes is involved, as ectopic expression of HoxA9 and Meis1 in mice is effective at inducing leukemias only under certain contexts. Finally, it is important to note that MLL fusions represent one particular class and mechanism; in contrast, fusions involving the retinoic acid receptor (RAR) (e.g., RAR-PLZF) are known to block differentiation through the constitutive recruitment of repressive chromatin modifiers (e.g., HDACs), conferring repression of genes important for activation.^43,44 Thus, as illustrated in Fig. 12–3, proper differentiation involves waves of transcription that involve activating a new program and silencing the former program.

The involvement of multiple enzymes in MLL fusions has inspired therapeutic approaches based on enzyme inhibition.² For example, DOT1 (a histone H3 methyltransferase) inhibitors have proven useful in mouse models of MLL-fusion–induced leukemia (and in cell lines),⁴⁵ and have entered phase I clinical trials (NCT01684150). Notably, the involvement of BRD4 in this system, along with its known importance in transcriptional activation in MYC-driven cancers, provides additional therapeutic possibilities. Here, inhibitors of BET-family bromodomains (JQ1 and others) have proven very effective in cell lines from patients, laying the foundation for clinical trials.^46,47,48,49

Theme 2: Loss-of-Function Mutations in Chromatin Modifiers

Loss-of-function mutations in chromatin modifiers are now very common in many cancers. The key concept in this theme is that the loss of epigenetic control confers both gene-specific and genome-wide epigenetic variation, eliciting transcriptome variation and plasticity. As a result, individual cells with transcriptomes that promote growth, survival, and/or metastasis can be selected from a diverse population. For example, this epigenetic variation can allow cells to sample a transcriptome that promotes invasion and later convert to a transcriptome that favors colonization. One mode involves the aberrant epigenetic silencing of tumor-suppressor proteins, either by the acquisition of “silencing” histone modifications, DNAme, or both. Epigenetic variation and selection are themselves powerful tools; however, they can also combine with genetic mutations to provide a further fitness benefit and reinforce oncogenic properties.

Examples of mutations in epigenetic factors in hematologic malignancies are numerous; even a partial list of factors and their impact is beyond the scope of this chapter.¹ However, mutations in certain factors are found in many hematologic malignancies and help to illustrate the concepts above; consequently, they are treated further here. One example that builds on an earlier section “Epigenetics and Hematologic Malignancies” involves mutations in MLL genes. MLL is actually a family of five similar genes; however, whereas MLL1 is most commonly involved in leukemogenic fusion proteins (discussed earlier), mutations in MLL2 are very common in lymphomas, with mutation rates as high as 89 percent for follicular lymphoma.⁵⁰ Other chromatin factors are also mutated at high frequency in BCLs, including the HAT complex members EP300 and CREBBP.⁵⁰

Mutations that affect the addition or removal of the repressive histone modification H3K27me3, are increasingly common in hematologic malignancies. For example, mutations in the polycomb repressive complex 2 (PRC2) complex, which adds H3K27me3, are associated with myeloproliferative diseases, myelodysplastic syndromes, and T-cell ALL.^51,52,53 In addition, mutations in the main enzyme that removes H3K27me3, known as X-chromosome encoded ubiquitously transcribed tetratricopeptide repeat (UTX), are common in multiple myeloma.⁵⁴ Furthermore, mutations in these enzymes are known to synergize with mutations in other epigenetic enzymes, such as TET proteins, in myeloid disorders.⁵¹ Notably, many genes that become improperly methylated in cancer cells were marked by H3K27me earlier in their development and were DNA demethylated. Thus, proper regulation of H3K27me3—a modification present at many silent but “poised” developmental genes—appears critical for tumor prevention.

Theme 3: Gain-of-Function Mutations in Chromatin Modifiers

Gain-of-function of epigenetic enzymes typically occurs either through upregulation of expression (through copy number variation or promoter fusions) or via mutations that upregulate the activity of the enzyme. The main concept in this theme is that high levels and/or the inability to turn off an epigenetic enzyme can lead to sustained activation or silencing of target loci, depending on the main function of the modification. Among many examples is EZH2, the main enzyme for H3K27me addition, which is either greatly overexpressed or hyperactivated (via mutation) in various hematologic malignancies.⁵⁵ For example, EZH2 is highly expressed in mantle cell lymphomas, whereas activating mutations (conferred by amino acid substitutions in the catalytic domain) are common in diffuse large B-cell and follicular lymphomas.⁵⁶ This hyperactivity has led to therapeutic strategies involving selective competitive inhibitors that mimic the cofactor S-adenosyl-methionine, the methyl donor for the EZH2 enzyme, which have proven effective in mouse xenografts.⁵⁶

Over the past several years, several studies have made striking links between hematologic malignancies and the misregulation of DNAme. High-throughput sequencing of leukemias and lymphomas have revealed three different types of mutations: (1) loss-of-function mutations in the de novo DNMT3a enzyme; (2) loss-of-function mutations in particular TET proteins; and (3) gain-of-function mutations in the metabolic enzymes isocitrate dehydrogenase (IDH)1 and IDH2, which create small molecule inhibitors of TET proteins (expanded below in this section). To begin, hypomorphic mutations in DNMT3a are common in acute myeloid leukemias (AMLs),⁵⁷ although precisely how a reduction in DNMT3a activity promotes leukemia is not yet understood. Of the three TET-family genes, particular TET genes are mutated at high frequency in hematologic malignancies. Strikingly, mutations in TET2 are found in almost half of chronic myelomonocytic leukemias (CMML),^58,59 and are also common in certain T-cell lymphomas. The key emerging concept is that defects in DNA demethylation by TET2 mutations may lead to increases in DNAme at certain CpG island regions (including those bearing H3K27me),⁶⁰ which may confer gene silencing of developmental and/or tumor-suppressor type genes; however, clear cause-and-effect links have not yet been made. Consistent with this mechanism, TET2 mutations are often found with mutations in EZH2, which catalyzes H3K27me addition.^51,59

Of particular interest are recent links between metabolic dysregulation, hematologic malignancies, and gliomas. Remarkably, gain-of-function mutations in the TCA cycle enzymes IDH1 and IDH2 are oncogenic.^{61,62,63,64,65,66} Normally, IDH1/2 convert isocitrate to 2OG (also known as α-ketoglutarate), a cofactor for both TET enzymes and JmjC-class lysine demethylases. Notably, oncogenic IDH1/2 mutations create proteins that additionally convert 2OG to (R)-2-hydroxyglutarate (R-2HG).^66,67 This both depletes the normal cofactor for TETs and JmjC demethylases and further creates a potent “oncometabolite” inhibitor of TET enzymes and prolyl hydroxylases (which regulate hypoxia-inducible transcription factor [HIF] proteins). Additionally, mutations in other Krebs cycle enzymes (e.g., succinate dehydrogenase [SDH]) accumulate succinate, which can likewise inhibit TET, JmjC, and prolyl hydroxylases⁶⁸; notably, SDH mutations and HIF mutations are both common in neuroendocrine tumors.⁶⁸ These observations have inspired multiple therapeutic approaches, including: (1) reversing the impact of these oncometabolites on the epigenome (e.g., via DNMT, HDAC or HMT inhibitors); (2) implementing selective inhibitors of these gain-of-function IDH mutant proteins to prevent (R)-2HG production; (3) using selective inhibitors of prolyl-hydroxylases; and (4) the use of ascorbic acid (vitamin C), which can enhance TET protein activity by affecting the reduction-oxidation (redox) state of iron, an essential cofactor for TET enzymes.