Among the biggest surprises from cancer genome sequencing has been the extent to which mutations affecting metabolic enzymes appear as drivers of specific types of cancers. Isocitrate dehydrogenases (IDH), enzymes that convert isocitrate to α-ketoglutarate, are prime examples. Specific IDH mutations are highly prevalent features of certain subsets of cancers, including 60–90% of gliomas and secondary glioblastomas, 20% of late-stage acute myelogenous leukemia, ~50% of central and periosteal cartilagenous tumors, and 10–20% of interhepatic cholangiocarcinoma (Parsons et al., 2008; Molenaar et al., 2014). The high frequency and patterns of IDH mutations suggest that they are important in early tumor development. In the case of acute myelogenous leukemia, IDH mutations also contribute to progression of these tumors (Molenaar et al., 2014; Reitman and Yan, 2010). With rare exception, IDH mutations in tumors occur exclusively as heterozygous missense mutations. The most frequently documented IDH mutations occur as substitutions of arginine 132 in IDH1 or the equivalent arginine (172) in IDH2. The specific amino acid substitution varies across and within tumor subtypes.

Tumor-associated IDH mutations appear to abolish the conversion of isocitrate to α-ketoglutarate, and instead promote a partial reverse reaction in which α-ketoglutarate is reduced to form D-2-hydroxyglutarate (D2-HG) (Dang et al., 2009; Rendina et al., 2013). D2-HG is a metabolite that is typically maintained at low levels in human cells through the activity of the enzyme D-2-hydroxyglutarate dehydrogenase (D2HGDH) (Achouri et al., 2004). Cellular D2-HG levels are commonly elevated in IDH-mutant tumor cells, sometimes to millimolar levels (Gross et al., 2010; Dang et al., 2009).

The structural similarity between D2-HG and α-ketoglutarate inspire models for D2-HG’s role in tumor formation centered on antagonism of α-ketoglutarate-dependent enzymes. Indeed, the α-ketoglutarate-dependent TET family of DNA demethylases and Jumonji histone demethylases are inhibited by high levels of D2-HG in vitro and in cell culture, resulting in DNA and histone hypermethylation and accompanied by changes in gene expression (Figueroa et al., 2010; Xu et al., 2011; Chowdhury et al., 2011; Lu et al., 2012). There is strong circumstantial evidence from acute myeloid leukemias (AMLs) that DNA demethylases are the targets of the D2-HG produced by the IDH mutations: AMLs tend to have either the mutations in IDH or mutations in the genes encoding TET enzymes, but not both (Figueroa et al., 2010; Gaidzik et al., 2012). In the case of gliomas and other types of tumors, there is no such simple dichotomy, suggesting the possibility of other important targets of D2-HG in these tumors such as histone demethylases. Indeed elevated bulk histone methylation has been reported for some tumors producing D2-HG, but whether critical changes in gene expression result from inhibition of DNA demethylation or inhibition of histone demethylation in these tumors is an important and unresolved question (Chowdhury et al., 2011; Gaidzik et al., 2012; Lu et al., 2012). Further, tumors with elevated D2-HG levels show bulk increases in histone methyl marks typically associated with both transcriptional activation (H3K4, H3K36) and repression (H3K9, H3K27) (Turcan et al., 2012; Lu et al., 2012). The increase of global histone methylation marks could conceivably lead to architectural changes in chromatin that alter heterochromatin/euchromatin states, or change the capacity to initiate and undergo transcription, or some combination of both. Which processes underlie the critical transcriptional changes that promote development and progression in D2-HG-accumulating tumors is unknown. Here, the causality of transcriptional changes due to elevated D2-HG was tested in Saccharomyces cerevisiae. Elevated D2-HG levels led to increased gene repression at a heterochromatic locus. The mechanism behind this effect on transcription was pinpointed to elevated H3K36 methylation as a result of D2-HG inhibition of Jumonji-domain-containing demethylases Rph1 and Gis1.

To study the impact of IDH mutations on gene expression, an analogous tumor-associated IDH mutation was made in Saccharomyces cerevisiae (IDP2-R132H) at the IDP2 locus. The yeast and human proteins share 61% amino acid identity (Figure 1—figure supplement 1). This mutation is analogous to the most common mutation found in low-grade gliomas and secondary glioblastomas in the human ortholog IDH1 (reviewed in Waitkus et al., 2016). Heterochromatic gene silencing at the HML locus was used as the assay for detecting changes in gene expression of a locus for which the transcriptional state is epigenetically inherited using the Cre-Reported Altered States of Heterochromatin (CRASH) assay. In this assay, loss of silencing allows for transient expression of a cre-recombinase gene inserted within HML, leading to a permanent and heritable switch from RFP expression to GFP expression (Figure 1A) (Dodson and Rine, 2015). The frequency of switches from RFP to GFP expression within a colony serves as the readout of the strength/stability of gene silencing in heterochromatin at HML. A decrease in gene silencing leads to more cells switching from RFP to GFP. Enhanced gene silencing leads to a reduction in switches from RFP to GFP.

Figure 1 with 2 supplements see all

Download asset Open asset

Mutations expected to produce D2-HG resulted in enhanced gene silencing

Because IDP2 is expressed only on non-fermentable carbon sources (Haselbeck and McAlister-Henn, 1993), cells were plated on solid medium containing glycerol as the sole carbon source to induce IDP2 expression. The frequency of loss-of-silencing events, detected as green spots or sectors within a colony, was quantified using MORPHE (Liu et al., 2016). The IDP2-R132H mutation increased the stability of heterochromatin (p<0.0001; Student’s t test) compared to wild-type IDP2, resulting in fewer switches from RFP to GFP expression (Figure 1B). This effect was dependent on IDP2 expression, as no effect was observed when cells were plated on medium containing glucose, which represses IDP2 expression (Figure 1—figure supplement 2). Deletion of IDP2 did not affect the stability of heterochromatin suggesting that IDP2-R132H results in a gain-of-function. These results demonstrate that tumor-associated IDH mutations led to increased stability of gene silencing in a heterochromatic domain in Saccharomyces.

If IDP2-R132H stabilized heterochromatin through increased D2-HG levels, then other genetic changes that independently increase D2-HG levels should do the same. In human cells, D2-HG is metabolized by D2HGDH (Achouri et al., 2004). Loss of D2HGDH causes D2-HG to accumulate to high levels. The high degree of amino acid sequence similarity between human D2HGDH and yeast Dld2 and Dld3 suggested that both were potential yeast orthologs of human D2HGDH, as recently confirmed (Becker-Kettern et al., 2016). Indeed, on medium with glycerol as the sole carbon source, when oxidative phosphorylation is the primary means of ATP production, the dld2∆ mutant phenocopied the effect of the IDP2-R132H mutant and increased gene silencing compared to wild type (p<0.0001; Student’s t test) (Figure 1D,E). Moreover, the IDP2-R132H dld2∆ double mutant resulted in a further increase in heterochromatin stability compared to wild type (p<0.0001; Student’s t test) and dld2Δ alone (p<0.05; Student’s t test) (Figure 1D,E). The stabilizing effect of a dld2Δ mutation was slightly diminished in combination with an idp2Δ deletion, suggesting that even wild-type Idp2 may contribute to the formation of D2-HG, either by directly producing D2-HG or indirectly by producing α-ketoglutarate as a substrate used by other enzymes for D2-HG formation. A dld3∆ mutation did not increase heterochromatin stability when cells were grown on glycerol medium (Figure 1F,G). However, when cells were grown on glucose, where ATP is primarily generated through sugar fermentation, both dld2∆ and dld3∆ mutations led to increased heterochromatin stability (Figure 1H,I). The increase in heterochromatin stability in the dld2∆ and dld3∆ mutants implied that D2-HG is normally produced in yeast under these and other conditions (see also Becker-Kettern et al., 2016).

If the increased gene silencing in the IDP2-R132H mutant were due to increased D2-HG levels, then the phenotype of the mutant should be dominant to wild-type IDP2. Conversely, if the effects on gene silencing were due to a reduced ability to convert isocitrate into α-ketoglutarate, the IDP2-R132H allele would likely be recessive to wild type. In an IPD2-R132H/IDP2 heterozygous diploid strain, gene silencing was increased compared to the equivalent strain with two copies of wild-type IDP2 (Figure 2A,B), demonstrating that the yeast mutation, analogous to the cancer-driver mutation in human IDH1, was dominant to wild-type IDP2. Dominance of the IDP2-R132H allele was observed both in wild-type DLD2 and dld2Δ mutant backgrounds. Interestingly, a strain homozygous for IDP2-R132H had less stable gene silencing than the heterozygous strain, suggesting that the effect of IDP2-R132H on silencing was synergistic with a wild-type copy of IDP2. These results echoed the observation in IDH1^R132H/WT tumor cell lines that knockout of the wild-type copy of IDH1 (IDH1^R132H/-)  results in decreased D2-HG formation relative to IDH1^R132H/WT cells (Jin et al., 2013).

Figure 2

Download asset Open asset

Elevated D2-HG caused enhanced gene silencing

To establish whether both IDP2-R132H and dld-mutant strains accumulate D2-HG, metabolite extracts were prepared from cells harboring these mutations and analyzed by quantitative liquid chromatography-mass spectrometry (LC-MS). D2-HG levels in wild-type strains ranged between 10 and 60 µM, in agreement with previous levels measured in yeast (Becker-Kettern et al., 2016). Consistent with a role in metabolizing D2-HG, a dld2∆ mutation led to a ~50 fold increase in cellular D2-HG levels in cells grown on glycerol (Figure 3A) and a fivefold increase in cells grown on glucose (Figure 3D). A dld3Δ mutation also led to a sevenfold increase in cellular D2-HG, but this increase was observed only in cells grown on glucose (Figure 3C,D). Thus, D2-HG is a normal feature of the S. cerevisiae metabolome and is efficiently metabolized by Dld2 and Dld3, as noted (Becker-Kettern et al., 2016). However, Dld3 activity toward D2-HG was limited to cells grown on fermentable carbon sources like glucose. Although the increased D2-HG level in the Idp2-R132H mutant inferred from the silencing phenotype was not detected in the LC-MS analysis of cells that were able to metabolize it (Figure 3A), the Idp2-R132H mutation clearly contributed to D2-HG production when measured in combination with dld2∆ and wild-type Idp2 protein (Figure 3B). Tumors with the equivalent isocitrate dehydrogenase mutation (IDH1-R132H) are exclusively heterozygous, retaining a wild-type copy of IDH1, and knock-down of wild-type IDH1 reduces D2-HG produced by IDH1-R132H (Ward et al., 2013). Because IDH is typically a homodimer, presumably the wild-type enzyme provides a local source of α-ketoglutarate for the mutant enzyme to convert to D2-HG (Ward et al., 2013; Jin et al., 2013). To test whether D2-HG production by yeast Idp2-R132H was likewise enhanced by the presence of wild-type Idp2, cells were transformed with a plasmid containing a copy of the wild-type IDP2 gene expressed from its native promoter. The presence of an extra copy of wild-type IDP2 had no effect on D2-HG levels in wild-type IDP2 cells (Figure 3B, bar 3 versus bar 4). However, D2-HG levels significantly increased (p<0.05) in mutant IDP2-R132H cells when an extra copy of wild-type IDP2 was present (Figure 3B, bar 5 versus bar 4). In the same strains, the increased levels of D2-HG corresponded to an increase in silencing (Figure 3—figure supplement 1). An extra copy of wild-type IDP2 had no effect on silencing in wild-type IDP2 cells. In parallel to the results measuring D2-HG levels, an extra copy of wild-type IDP2 enhanced silencing in mutant IDP2-R132H cells compared to strains with only wild-type IDP2 (p<0.0001; Student’s t test). These results demonstrated a strong correlation between D2-HG levels and increased silencing at a heterochromatic locus.

Figure 3 with 1 supplement see all

Download asset Open asset

As a direct test of whether increased levels of D2-HG were responsible for the increase in gene silencing, cells were grown in the presence or absence of cell membrane-permeable octyl-D2-HG. Treatment with octyl-D2-HG had no impact on silencing in wild-type cells that readily metabolized D2-HG (Figure 4A). However, treatment of dld2Δ mutant cells with octyl-D2-HG resulted in increased gene silencing compared to the untreated cells (p<0.05) (Figure 4B) establishing that increases in D2-HG were causal for stabilizing gene silencing.

Figure 4

Download asset Open asset

D2-HG did not enhance a locus-specific gene repression mechanism

In principle, D2-HG accumulation could impact other forms of gene repression. The GAL1 promoter is repressed in media containing glucose. By placing the cre gene under control of the GAL1 promoter, it was possible to measure the impact of D2-HG on another type of transcriptional repression. When dld2∆ mutations were introduced into the pGAL1::cre strain, repression of cre was maintained to the same degree as in wild type, indicating that the impact of D2-HG did not extend to all forms of repression (Figure 5).

Figure 5

Download asset Open asset

D2-HG inhibited multiple histone demethylases

D2-HG can inhibit 2-oxoglutarate/Fe(II)-dependent dioxygenases. This class of enzymes includes TET DNA demethylases and Jumonji-C class of histone demethylases, which are both important epigenetic regulators previously implicated in progression of certain cancers. Since both families of demethylase are inhibited by D2-HG (Xu et al., 2011; Figueroa et al., 2010; Chowdhury et al., 2011; Lu et al., 2012), we tested whether the enhanced gene silencing caused by D2-HG in yeast was due to inhibition of a demethylase activity. DNA methylation does not occur in S. cerevisiae, but histone methylation impacts heterochromatin at HML (Osborne et al., 2009; van Leeuwen et al., 2002). To determine if increased D2-HG lead to increased bulk histone methylation, or whether the effects were localized to particular methylated species, histone methylation levels were analyzed from yeast extracts prepared from either wild type or dld2∆ IDP2-R132H mutant cells using antibodies against specific histone methyl marks. Bulk levels of H3K36-mono and -trimethyl and H3K4-trimethyl marks were significantly higher (p<0.01) in dld2∆ IDP2-R132H strains compared to wild type (Figure 6A and B, Figure 6—figure supplement 1). The antibody for the H3K36 dimethyl species exhibited too much background to determine the statistical significance of the apparent increase in the mutant cells. Thus, bulk levels of all three methylated species of H3K36 methylation and H3K4 methylation increased in strains that accumulated high levels of D2-HG, consistent with D2-HG inhibition of histone demethylase enzymes. The lack of change in H3K79 methylation levels was consistent with this model as there is no known H3K79 demethylase in S. cerevisiae.

Figure 6 with 1 supplement see all

Download asset Open asset

Enhanced gene silencing resulted from specific inhibition of two H3K36me demethylases

There are five known Jumonji-domain proteins in S. cerevisiae and among them Rph1, Gis1, and Jhd1 demethylate various species of methylated H3K36 (Tu et al., 2007; Kim and Buratowski, 2007) and Jhd2 demethylates mono-, di-, and trimethylated H3K4 (Seward et al., 2007). Because both H3K4 and H3K36 methylation increased, inhibition of multiple histone demethylases could potentially contribute to the increase in gene silencing. To identify the relevant histone demethylase(s) whose inhibition led to enhanced gene silencing, all five Jumonji-domain genes were deleted individually and in combinations and the resulting mutants assessed for a potential role in heterochromatin stability. The strong prediction was that deletion of the relevant histone demethylase(s) would phenocopy the effect of increased D2-HG levels if D2-HG caused increased gene silencing by inhibiting a histone demethylase. Indeed, both gis1∆ and rph1∆ single mutants enhanced gene silencing compared to wild type (p<0.0001; Student’s t test), and the rph1∆ gis1∆ double mutants resulted in even further enhancement compared to wild type (p<0.0001; Student’s t test) (Figure 7A,B). The jhd1∆ mutant had no statistically significant effect on its own or in combination with either rph1∆ or gis1∆ mutations, nor did ecm5∆ and jhd2∆ mutations (Figure 7—figure supplement 1). Therefore, elevated H3K36 methylation was the determinant for increasing heterochromatin stability, and Rph1 and Gis1 were implicated as the primary demethylases inhibited by D2-HG accumulation that led to enhanced heterochromatic gene silencing. In further support, GIS1 and RPH1 overexpressed from a galactose-inducible plasmid led to destabilization of heterochromatin (Figure 7C); in contrast, overexpression of JHD1 had no effect (Figure 7C).

Figure 7 with 1 supplement see all

Download asset Open asset

In principle, the enhanced gene silencing in the gis1∆ rph1∆ double mutant could result from a failure to demethylate H3K36 methylated species, or from a failure to demethylate a here-to-fore unrecognized methylated species of any histone or non-histone protein. These possibilities were distinguished by the triple mutant combination set2∆ gis1∆ rph1∆. Methylation of H3K36 is exclusively dependent on the methyltransferase Set2 (Strahl et al., 2002). The loss of Set2 completely abrogated the effect of gis1∆ rph1∆ on gene silencing. The extent of gene silencing in the triple mutant was no different from either wild type or set2∆ single mutants (Figure 7D,E). Because set2∆ suppressed gis1∆ and rph1∆, the enhanced gene silencing in gis1∆ and rph1∆ mutants was due exclusively to H3K36 hypermethylation.

These results strongly predicted that set2Δ mutations would also suppress the phenotypes of dld2Δ and IDP2-R132H mutations. Indeed, gene silencing in set2Δ dld2Δ double mutants and set2Δ dld2Δ IDP2-R132H triple mutants was no different from set2Δ single mutants (Figure 8A,B). If D2-HG affects gene silencing by inhibition of H3K36 demethylases, then gis1Δ rph1Δ mutants should be epistatic to dld2Δ and IDP2-R132H mutations. This prediction was confirmed by analysis of gis1Δ rph1Δ dld2Δ and gis1Δ rph1Δ dld2Δ IDP2-R132H mutants, which displayed identical levels of silencing to the gis1Δ rph1Δ double mutant (Figure 8C,D). These results provided the functional confirmation that the elevated gene silencing observed in cells that accumulate D2-HG was due to Set2-dependent methylation of H3K36, and its subsequent hypermethylation due to inhibition of the Gis1 and Rph1 demethylases.

Figure 8

Download asset Open asset

We have recapitulated cancer-associated IDH mutations in S. cerevisiae and demonstrated that the IDH mutants accumulated D2-HG in vivo and increased heterochromatin stability as reflected by enhanced gene silencing within heterochromatin. While DNA methylation is clearly impacted by D2-HG accumulation and leads to transcriptional changes in certain IDH-mutant tumors, it has been difficult to parse which cellular changes are attributable to inhibition of DNA demethylation, histone demethylation, or other unexplored targets of demethylases. Here, we demonstrated unequivocally that enhanced gene silencing was due to histone H3K36 hypermethylation in response to the oncometabolite D2-HG inhibiting two specific histone demethylases. As S. cerevisiae lacks DNA methylation, there were no confounding effects of DNA methylation in our study.

D2-HG has only been recently reported in S. cerevisiae (Becker-Kettern et al., 2016), and several new aspects of its metabolism are established here. D2-HG was detected in metabolite extracts even from wild-type cells, suggesting it is a normal component of the yeast metabolome. Wild-type cells maintained D2-HG at low levels through the activity of D2-HG dehydrogenases. Both Dld2 and Dld3 are capable of metabolizing D2-HG in vitro, and loss of either enzyme leads to accumulation of D2-HG in cells grown on glucose, with Dld3 having a stronger effect than Dld2 (Becker-Kettern et al., 2016). However, only Dld2 had a role in metabolizing D2-HG in vivo in cells grown on glycerol, suggesting that Dld3 and Dld2 are differentially regulated in response to the mode of energy production in the cell.

IDP2-R132H mutants alone, and in combination with dld2∆, led to enhanced gene silencing. Multiple lines of evidence linked enhanced gene silencing to the production of D2-HG and inhibition of H3K36 histone demethylases Gis1 and Rph1. First, mutants that led to D2-HG accumulation phenocopied gis1Δ rph1Δ double mutants. Next, gis1Δ rph1Δ double mutants were epistatic to IDP2-R132H and dld2Δ mutations, indicating that the increase in gene silencing in IDP2-R132H and dld2Δ mutants was due to inhibition of these H3K36 demethylases. Additionally, the silencing phenotypes of H3K36 demethylase mutants, IDP2-R132H, and dld2Δ all depended on the H3K36 methyltransferase activity of Set2. Finally, external application of a cell-permeant form of D2-HG recapitulated enhanced silencing, demonstrating the causality of this phenotype. D2-HG acts as a competitive inhibitor of human Jumonji histone demethylase enzymes. Whether yeast Jumonji-domain histone demethylases are biochemically inhibited by D2-HG has not explicitly been tested, although the genetic and molecular evidence strongly implied such a mechanism.

D2-HG levels increased approximately 50-fold in dld2Δ mutants, and IDP2-R132H mutations also contribute to the pool of D2-HG, although this contribution was detectable only when a copy of wild-type IDP2 was also present in IDP2-R132H mutant cells. This result was consistent with our genetic analysis of diploid cells in which silencing was greatest in IDP2/IDP2-R132H heterozygotes, and was also consistent with previous studies using human cell lines and tumors where the equivalent human mutation (IDH1-R132H) also partially depended on the presence an intact copy of wild-type IDH1 for D2-HG production (Jin et al., 2013; Ward et al., 2013). Increases in D2-HG were readily measured in dld2-mutants, but not in cells capable of degrading D2-HG. This differs from tumor cells where IDH mutations lead to measurable increases in D2-HG despite intact D2-HG dehydrogenase activity (Krell et al., 2011). This discrepancy could either be due to a higher capacity of S. cerevisiae to metabolize D2-HG, a higher capacity for the human equivalent to the IDP2-R132H mutation to generate D2-HG, or some combination of both. We were able to detect a silencing phenotype in IDP2-R132H mutants even in cells capable of metabolizing D2-HG due to the highly sensitive nature of the CRASH assay, which converts transient lapses of silencing in individual cells into permanent and heritable effects. In contrast, physical measurements of metabolites reveal only steady-state levels averaged over millions of cells. Given that D2-HG is actively degraded when Dld2 is present, it is evident that our genetic assay was able to detect change caused by changes in D2-HG that were below our present level of physical detection.

Accumulation of D2-HG led to hypermethylation of histone H3 at lysine 4 and lysine 36. Importantly, it was the loss of histone demethylases that act on H3K36 that was solely responsible for increased stabilization of heterochromatin. Inhibition of H3K4 demethylation and the resulting elevation of bulk H3K4me had no effect on gene silencing. A role of H3K36 methylation in heterochromatin stability was surprising as other described roles for this modification involve co-transcriptional methylation of H3K36 to enhance transcriptional elongation (Kizer et al., 2005; Tu et al., 2007). Methylation of H3K36 at transcribed genes prevents aberrant transcription from initiating from within the open-reading frames of actively transcribed genes (Carrozza et al., 2005; Keogh et al., 2005; Joshi and Struhl, 2005), although this process appears to be incompatible with silencing at HML for reasons detailed below. Nevertheless, other aspects of histone methylation have had unexpected impacts on heterochromatic gene silencing. For example, H3K4 methylation occurs at the 5’ ends of actively transcribed genes and is important for maintaining a local chromatin landscape conducive to transcription, and yet it also plays a role in the establishment of silencing at HML (Osborne et al., 2009). Whether methylation of H3K4 is directly involved in the establishment and H3K36 in maintenance of SIR-dependent silencing remain important unresolved questions. Precedence for a direct role of histone methylation in silencing is evident in other species in which methylation of H3K9 and H3K27 directly recruit proteins that establish heterochromatin. Although budding yeasts lack the enzymes necessary to methylate H3K9 and K27, methylation of other histone lysines directly impact silencing. The site at which Sir3 interacts with histone H3 includes H3K79, and methylation of H3K79 by Dot1 inhibits Sir3-H3 interaction (Armache et al., 2011). The majority of euchromatic nucleosomes are methylated at H3K79, thus preventing Sir protein binding, while at heterochromatic loci Sir3 binds to H3 and sterically occludes Dot1 from accessing H3K79 for methylation. The results of this study are compatible with a model in which H3K36 methylation directly enhances the maintenance of silent chromatin, perhaps by recruitment of a pro-silencing factor to silent loci.

Alternatively, global increases in H3K36 methylation could recruit factors antagonistic to silencing away from heterochromatin. For example, the Rpd3 complex antagonizes silencing at heterochromatic loci HML and HMR by limiting Sir3 and Sir4 occupancy at flanking silencer regions (Thurtle-Schmidt et al., 2016). Importantly, the deacetylase activity of Rpd3(S) is directed toward nucleosomes at actively transcribed genes in part through an interaction between the Eaf3 subunit of the Rpd3(S) complex and histones methylated at H3K36 (Carrozza et al., 2005; Keogh, 2005; Joshi and Struhl, 2005). It is possible that elevated H3K36 methylation in euchromatin recruits Rpd3(S) away from HML and HMR, and partially alleviates its antagonistic effect on silencing. Because recruitment of Rpd3 to heterochromatin appears to be independent of Eaf3 and thus H3K36 methylation (Thurtle-Schmidt et al., 2016), a global increase of H3K36 methylation could lead to relocalization of a limited pool of Rpd3 away from HML and HMR to sites with increased H3K36 methylation. Finally, H3K36 methylation prevents binding of Sir proteins to heterochromatin-adjacent and telomere-adjacent regions (Tompa and Madhani, 2007). Increased global H3K36 methylation due to D2-HG accumulation could, in principle, prevent binding of Sir proteins to non-heterochromatic regions, making more Sir protein available to bind at heterochromatic loci.

Increased silencing of tumor suppressor genes is a feature of many tumors. This study pinpointed specific histone demethylase enzymes whose inhibition by D2-HG resulted in increased gene silencing. Further, this study demonstrated that D2-HG enhanced gene silencing through stabilization of heterochromatic DNA. Similar mechanisms could promote tumor formation by silencing genes encoding tumor suppressors.

1. 1. Achouri Y
   2. Noël G
   3. Vertommen D
   4. Rider MH
   5. Veiga-Da-Cunha M
   6. Van Schaftingen E

   (2004) Identification of a dehydrogenase acting on D-2-hydroxyglutarate

   Biochemical Journal 381:35–42.

   https://doi.org/10.1042/BJ20031933

   • PubMed
   • Google Scholar
2. 1. Armache KJ
   2. Garlick JD
   3. Canzio D
   4. Narlikar GJ
   5. Kingston RE

   (2011) Structural basis of silencing: sir3 BAH domain in complex with a nucleosome at 3.0 å resolution

   Science 334:977–982.

   https://doi.org/10.1126/science.1210915

   • PubMed
   • Google Scholar
3. 1. Becker-Kettern J
   2. Paczia N
   3. Conrotte JF
   4. Kay DP
   5. Guignard C
   6. Jung PP
   7. Linster CL

   (2016) Saccharomyces cerevisiae forms D-2-Hydroxyglutarate and couples its degradation to D-Lactate formation via a cytosolic transhydrogenase

   Journal of Biological Chemistry 291:6036–6058.

   https://doi.org/10.1074/jbc.M115.704494

   • PubMed
   • Google Scholar
4. 1. Carrozza MJ
   2. Li B
   3. Florens L
   4. Suganuma T
   5. Swanson SK
   6. Lee KK
   7. Shia WJ
   8. Anderson S
   9. Yates J
   10. Washburn MP
   11. Workman JL

   (2005) Histone H3 methylation by Set2 directs deacetylation of coding regions by Rpd3S to suppress spurious intragenic transcription

   Cell 123:581–592.

   https://doi.org/10.1016/j.cell.2005.10.023

   • PubMed
   • Google Scholar
5. 1. Chowdhury R
   2. Yeoh KK
   3. Tian YM
   4. Hillringhaus L
   5. Bagg EA
   6. Rose NR
   7. Leung IK
   8. Li XS
   9. Woon EC
   10. Yang M
   11. McDonough MA
   12. King ON
   13. Clifton IJ
   14. Klose RJ
   15. Claridge TD
   16. Ratcliffe PJ
   17. Schofield CJ
   18. Kawamura A

   (2011) The oncometabolite 2-hydroxyglutarate inhibits histone lysine demethylases

   EMBO Reports 12:463–469.

   https://doi.org/10.1038/embor.2011.43

   • PubMed
   • Google Scholar
6. 1. Dang L
   2. White DW
   3. Gross S
   4. Bennett BD
   5. Bittinger MA
   6. Driggers EM
   7. Fantin VR
   8. Jang HG
   9. Jin S
   10. Keenan MC
   11. Marks KM
   12. Prins RM
   13. Ward PS
   14. Yen KE
   15. Liau LM
   16. Rabinowitz JD
   17. Cantley LC
   18. Thompson CB
   19. Vander Heiden MG
   20. Su SM

   (2009) Cancer-associated IDH1 mutations produce 2-hydroxyglutarate

   Nature 462:739–744.

   https://doi.org/10.1038/nature08617

   • PubMed
   • Google Scholar
7. 1. Dodson AE
   2. Rine J

   (2015) Heritable capture of heterochromatin dynamics in saccharomyces cerevisiae

   eLife 4:e05007.

   https://doi.org/10.7554/eLife.05007

   • PubMed
   • Google Scholar
8. 1. Erdeniz N
   2. Mortensen UH
   3. Rothstein R

   (1997) Cloning-free PCR-based allele replacement methods

   Genome Research 7:1174–1183.

   https://doi.org/10.1101/gr.7.12.1174

   • PubMed
   • Google Scholar
9. 1. Figueroa ME
   2. Abdel-Wahab O
   3. Lu C
   4. Ward PS
   5. Patel J
   6. Shih A
   7. Li Y
   8. Bhagwat N
   9. Vasanthakumar A
   10. Fernandez HF
   11. Tallman MS
   12. Sun Z
   13. Wolniak K
   14. Peeters JK
   15. Liu W
   16. Choe SE
   17. Fantin VR
   18. Paietta E
   19. Löwenberg B
   20. Licht JD
   21. Godley LA
   22. Delwel R
   23. Valk PJ
   24. Thompson CB
   25. Levine RL
   26. Melnick A

   (2010) Leukemic IDH1 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair hematopoietic differentiation

   Cancer Cell 18:553–567.

   https://doi.org/10.1016/j.ccr.2010.11.015

   • PubMed
   • Google Scholar
10. 1. Gaidzik VI
    2. Paschka P
    3. Späth D
    4. Habdank M
    5. Köhne CH
    6. Germing U
    7. von Lilienfeld-Toal M
    8. Held G
    9. Horst HA
    10. Haase D
    11. Bentz M
    12. Götze K
    13. Döhner H
    14. Schlenk RF
    15. Bullinger L
    16. Döhner K

    (2012) TET2 mutations in acute myeloid leukemia (AML): results from a comprehensive genetic and clinical analysis of the AML study group

    Journal of Clinical Oncology 30:1350–1357.

    https://doi.org/10.1200/JCO.2011.39.2886

    • PubMed
    • Google Scholar
11. 1. Goldstein AL
    2. McCusker JH

    (1999) Three new dominant drug resistance cassettes for gene disruption in saccharomyces cerevisiae

    Yeast 15:1541–1553.

    https://doi.org/10.1002/(SICI)1097-0061(199910)15:14<1541::AID-YEA476>3.0.CO;2-K

    • PubMed
    • Google Scholar
12. 1. Goldstein AL
    2. Pan X
    3. McCusker JH

    (1999) Heterologous URA3MX cassettes for gene replacement in Saccharomyces cerevisiae

    Yeast 15:507–511.

    https://doi.org/10.1002/(SICI)1097-0061(199904)15:6<507::AID-YEA369>3.0.CO;2-P

    • PubMed
    • Google Scholar
13. 1. Gross S
    2. Cairns RA
    3. Minden MD
    4. Driggers EM
    5. Bittinger MA
    6. Jang HG
    7. Sasaki M
    8. Jin S
    9. Schenkein DP
    10. Su SM
    11. Dang L
    12. Fantin VR
    13. Mak TW

    (2010) Cancer-associated metabolite 2-hydroxyglutarate accumulates in acute myelogenous leukemia with isocitrate dehydrogenase 1 and 2 mutations

    The Journal of Experimental Medicine 207:339–344.

    https://doi.org/10.1084/jem.20092506

    • PubMed
    • Google Scholar
14. 1. Gueldener U
    2. Heinisch J
    3. Koehler GJ
    4. Voss D
    5. Hegemann JH

    (2002) A second set of loxP marker cassettes for Cre-mediated multiple gene knockouts in budding yeast

    Nucleic Acids Research 30:e23.

    https://doi.org/10.1093/nar/30.6.e23

    • PubMed
    • Google Scholar
15. 1. Haselbeck RJ
    2. McAlister-Henn L

    (1993)

    Function and expression of yeast mitochondrial NAD- and NADP-specific isocitrate dehydrogenases

    The Journal of Biological Chemistry 268:12116–12122.

    • PubMed
    • Google Scholar
16. 1. Jin G
    2. Reitman ZJ
    3. Duncan CG
    4. Spasojevic I
    5. Gooden DM
    6. Rasheed BA
    7. Yang R
    8. Lopez GY
    9. He Y
    10. McLendon RE
    11. Bigner DD
    12. Yan H

    (2013) Disruption of wild-type IDH1 suppresses D-2-hydroxyglutarate production in IDH1-mutated gliomas

    Cancer Research 73:496–501.

    https://doi.org/10.1158/0008-5472.CAN-12-2852

    • PubMed
    • Google Scholar
17. 1. Joshi AA
    2. Struhl K

    (2005) Eaf3 chromodomain interaction with methylated H3-K36 links histone deacetylation to pol II elongation

    Molecular Cell 20:971–978.

    https://doi.org/10.1016/j.molcel.2005.11.021

    • PubMed
    • Google Scholar
18. 1. Keogh MC
    2. Kurdistani SK
    3. Morris SA
    4. Ahn SH
    5. Podolny V
    6. Collins SR
    7. Schuldiner M
    8. Chin K
    9. Punna T
    10. Thompson NJ
    11. Boone C
    12. Emili A
    13. Weissman JS
    14. Hughes TR
    15. Strahl BD
    16. Grunstein M
    17. Greenblatt JF
    18. Buratowski S
    19. Krogan NJ

    (2005) Cotranscriptional set2 methylation of histone H3 lysine 36 recruits a repressive Rpd3 complex

    Cell 123:593–605.

    https://doi.org/10.1016/j.cell.2005.10.025

    • PubMed
    • Google Scholar
19. 1. Kim T
    2. Buratowski S

    (2007) Two Saccharomyces cerevisiae JmjC domain proteins demethylate histone H3 Lys36 in transcribed regions to promote elongation

    Journal of Biological Chemistry 282:20827–20835.

    https://doi.org/10.1074/jbc.M703034200

    • PubMed
    • Google Scholar
20. 1. Kizer KO
    2. Phatnani HP
    3. Shibata Y
    4. Hall H
    5. Greenleaf AL
    6. Strahl BD

    (2005) A novel domain in Set2 mediates RNA polymerase II interaction and couples histone H3 K36 methylation with transcript elongation

    Molecular and Cellular Biology 25:3305–3316.

    https://doi.org/10.1128/MCB.25.8.3305-3316.2005

    • PubMed
    • Google Scholar
21. 1. Krell D
    2. Assoku M
    3. Galloway M
    4. Mulholland P
    5. Tomlinson I
    6. Bardella C

    (2011) Screen for IDH1, IDH2, IDH3, D2HGDH and L2HGDH mutations in glioblastoma

    PLoS One 6:e19868.

    https://doi.org/10.1371/journal.pone.0019868

    • PubMed
    • Google Scholar
22. 1. Liu TY
    2. Dodson AE
    3. Terhorst J
    4. Song YS
    5. Rine J

    (2016) Riches of phenotype computationally extracted from microbial colonies

    PNAS 113:E2822–E2831.

    https://doi.org/10.1073/pnas.1523295113

    • PubMed
    • Google Scholar
23. 1. Lu C
    2. Ward PS
    3. Kapoor GS
    4. Rohle D
    5. Turcan S
    6. Abdel-Wahab O
    7. Edwards CR
    8. Khanin R
    9. Figueroa ME
    10. Melnick A
    11. Wellen KE
    12. O'Rourke DM
    13. Berger SL
    14. Chan TA
    15. Levine RL
    16. Mellinghoff IK
    17. Thompson CB

    (2012) IDH mutation impairs histone demethylation and results in a block to cell differentiation

    Nature 483:474–478.

    https://doi.org/10.1038/nature10860

    • PubMed
    • Google Scholar
24. 1. Molenaar RJ
    2. Radivoyevitch T
    3. Maciejewski JP
    4. van Noorden CJ
    5. Bleeker FE

    (2014) The driver and passenger effects of isocitrate dehydrogenase 1 and 2 mutations in oncogenesis and survival prolongation

    Biochimica Et Biophysica Acta (BBA) - Reviews on Cancer 1846:326–341.

    https://doi.org/10.1016/j.bbcan.2014.05.004

    • PubMed
    • Google Scholar
25. 1. Notredame C
    2. Higgins DG
    3. Heringa J

    (2000) T-Coffee: a novel method for fast and accurate multiple sequence alignment

    Journal of Molecular Biology 302:205–217.

    https://doi.org/10.1006/jmbi.2000.4042

    • PubMed
    • Google Scholar
26. 1. Osborne EA
    2. Dudoit S
    3. Rine J

    (2009) The establishment of gene silencing at single-cell resolution

    Nature Genetics 41:800–806.

    https://doi.org/10.1038/ng.402

    • PubMed
    • Google Scholar
27. 1. Parsons DW
    2. Jones S
    3. Zhang X
    4. Lin JC
    5. Leary RJ
    6. Angenendt P
    7. Mankoo P
    8. Carter H
    9. Siu IM
    10. Gallia GL
    11. Olivi A
    12. McLendon R
    13. Rasheed BA
    14. Keir S
    15. Nikolskaya T
    16. Nikolsky Y
    17. Busam DA
    18. Tekleab H
    19. Diaz LA
    20. Hartigan J
    21. Smith DR
    22. Strausberg RL
    23. Marie SK
    24. Shinjo SM
    25. Yan H
    26. Riggins GJ
    27. Bigner DD
    28. Karchin R
    29. Papadopoulos N
    30. Parmigiani G
    31. Vogelstein B
    32. Velculescu VE
    33. Kinzler KW

    (2008) An integrated genomic analysis of human glioblastoma multiforme

    Science 321:1807–1812.

    https://doi.org/10.1126/science.1164382

    • PubMed
    • Google Scholar
28. 1. Reitman ZJ
    2. Yan H

    (2010) Isocitrate dehydrogenase 1 and 2 mutations in cancer: alterations at a crossroads of cellular metabolism

    JNCI Journal of the National Cancer Institute 102:932–941.

    https://doi.org/10.1093/jnci/djq187

    • PubMed
    • Google Scholar
29. 1. Rendina AR
    2. Pietrak B
    3. Smallwood A
    4. Zhao H
    5. Qi H
    6. Quinn C
    7. Adams ND
    8. Concha N
    9. Duraiswami C
    10. Thrall SH
    11. Sweitzer S
    12. Schwartz B

    (2013) Mutant IDH1 enhances the production of 2-hydroxyglutarate due to its kinetic mechanism

    Biochemistry 52:4563–4577.

    https://doi.org/10.1021/bi400514k

    • PubMed
    • Google Scholar
30. 1. Seward DJ
    2. Cubberley G
    3. Kim S
    4. Schonewald M
    5. Zhang L
    6. Tripet B
    7. Bentley DL

    (2007) Demethylation of trimethylated histone H3 Lys4 in vivo by JARID1 JmjC proteins

    Nature Structural & Molecular Biology 14:240–242.

    https://doi.org/10.1038/nsmb1200

    • PubMed
    • Google Scholar
31. 1. Sikorski RS
    2. Hieter P

    (1989)

    A system of shuttle vectors and yeast host strains designed for efficient manipulation of DNA in saccharomyces cerevisiae

    Genetics 122:19 LP–27.

    • PubMed
    • Google Scholar
32. 1. Strahl BD
    2. Grant PA
    3. Briggs SD
    4. Sun ZW
    5. Bone JR
    6. Caldwell JA
    7. Mollah S
    8. Cook RG
    9. Shabanowitz J
    10. Hunt DF
    11. Allis CD

    (2002) Set2 is a nucleosomal histone H3-selective methyltransferase that mediates transcriptional repression

    Molecular and Cellular Biology 22:1298–1306.

    https://doi.org/10.1128/MCB.22.5.1298-1306.2002

    • PubMed
    • Google Scholar
33. 1. Thurtle-Schmidt DM
    2. Dodson AE
    3. Rine J

    (2016) Histone deacetylases with antagonistic roles in saccharomyces cerevisiae heterochromatin formation

    Genetics 204:177–190.

    https://doi.org/10.1534/genetics.116.190835

    • PubMed
    • Google Scholar
34. 1. Tompa R
    2. Madhani HD

    (2007) Histone H3 lysine 36 methylation antagonizes silencing in Saccharomyces cerevisiae independently of the Rpd3S histone deacetylase complex

    Genetics 175:585–593.

    https://doi.org/10.1534/genetics.106.067751

    • PubMed
    • Google Scholar
35. 1. Tu S
    2. Bulloch EM
    3. Yang L
    4. Ren C
    5. Huang WC
    6. Hsu PH
    7. Chen CH
    8. Liao CL
    9. Yu HM
    10. Lo WS
    11. Freitas MA
    12. Tsai MD

    (2007) Identification of histone demethylases in saccharomyces cerevisiae

    Journal of Biological Chemistry 282:14262–14271.

    https://doi.org/10.1074/jbc.M609900200

    • PubMed
    • Google Scholar
36. 1. Turcan S
    2. Rohle D
    3. Goenka A
    4. Walsh LA
    5. Fang F
    6. Yilmaz E
    7. Campos C
    8. Fabius AW
    9. Lu C
    10. Ward PS
    11. Thompson CB
    12. Kaufman A
    13. Guryanova O
    14. Levine R
    15. Heguy A
    16. Viale A
    17. Morris LG
    18. Huse JT
    19. Mellinghoff IK
    20. Chan TA

    (2012) IDH1 mutation is sufficient to establish the glioma hypermethylator phenotype

    Nature 483:479–483.

    https://doi.org/10.1038/nature10866

    • PubMed
    • Google Scholar
37. 1. van Leeuwen F
    2. Gafken PR
    3. Gottschling DE

    (2002) Dot1p modulates silencing in yeast by methylation of the nucleosome core

    Cell 109:745–756.

    https://doi.org/10.1016/S0092-8674(02)00759-6

    • PubMed
    • Google Scholar
38. 1. Wach A
    2. Brachat A
    3. Pöhlmann R
    4. Philippsen P

    (1994) New heterologous modules for classical or PCR-based gene disruptions in saccharomyces cerevisiae

    Yeast 10:1793–1808.

    https://doi.org/10.1002/yea.320101310

    • PubMed
    • Google Scholar
39. 1. Waitkus MS
    2. Diplas BH
    3. Yan H

    (2016) Isocitrate dehydrogenase mutations in gliomas

    Neuro-Oncology 18:16–26.

    https://doi.org/10.1093/neuonc/nov136

    • PubMed
    • Google Scholar
40. 1. Ward PS
    2. Lu C
    3. Cross JR
    4. Abdel-Wahab O
    5. Levine RL
    6. Schwartz GK
    7. Thompson CB

    (2013) The potential for isocitrate dehydrogenase mutations to produce 2-hydroxyglutarate depends on allele specificity and subcellular compartmentalization

    Journal of Biological Chemistry 288:3804–3815.

    https://doi.org/10.1074/jbc.M112.435495

    • PubMed
    • Google Scholar
41. 1. Xu W
    2. Yang H
    3. Liu Y
    4. Yang Y
    5. Wang P
    6. Kim SH
    7. Ito S
    8. Yang C
    9. Wang P
    10. Xiao MT
    11. Liu LX
    12. Jiang WQ
    13. Liu J
    14. Zhang JY
    15. Wang B
    16. Frye S
    17. Zhang Y
    18. Xu YH
    19. Lei QY
    20. Guan KL
    21. Zhao SM
    22. Xiong Y

    (2011) Oncometabolite 2-hydroxyglutarate is a competitive inhibitor of α-ketoglutarate-dependent dioxygenases

    Cancer Cell 19:17–30.

    https://doi.org/10.1016/j.ccr.2010.12.014

    • PubMed
    • Google Scholar

Author details

1. Ryan Janke

   1. Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States
   2. California Institute for Quantitative Biosciences, University of California, Berkeley, Berkeley, United States

   Contribution

   RJ, Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Writing—original draft, Writing—review and editing

   Competing interests

   The authors declare that no competing interests exist.

   "This ORCID iD identifies the author of this article:" 0000-0002-5755-8019

2. Anthony T Iavarone

   California Institute for Quantitative Biosciences, University of California, Berkeley, Berkeley, United States

   Contribution

   ATI, Data curation, Methodology, Writing—original draft

   Competing interests

   The authors declare that no competing interests exist.

3. Jasper Rine

   1. Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States
   2. California Institute for Quantitative Biosciences, University of California, Berkeley, Berkeley, United States

   Contribution

   JR, Conceptualization, Supervision, Funding acquisition, Writing—original draft, Writing—review and editing

   For correspondence

   jrine@berkeley.edu

   Competing interests

   The authors declare that no competing interests exist.

   "This ORCID iD identifies the author of this article:" 0000-0003-2297-9814