Mutations in the genes IDH1 and IDH2 occur in about 10% of patients with acute myeloid leukemia (AML). These mutations result in the production of an abnormal metabolite, 2-hydroxyglutarate (2-HG) that is thought to contribute to AML development by inhibiting pathways that remove DNA methylation, an epigenetic modification involved in regulating gene expression. However, the specific genomic regions that are affected by IDH mutations in AML cells are poorly understood, and how the resulting changes in DNA methylation affect gene regulation remain unclear. To identify the direct effects of these mutations on DNA methylation, gene expression, and cellular function, we analyzed DNA methylation in 15 primary AML samples with either IDH1 or IDH2 mutations and modeled the epigenetic effects of these mutations using 2 patient-derived induced pluripotent stem (iPS) cell lines. Analysis of genome-wide DNA methylation patterns in IDH-mutant AML samples from whole-genome bisulfite sequencing showed that global DNA methylation levels were similar to AMLs without these mutations, but that the IDH-mutant samples had thousands of regions with focal increases in DNA methylation. These focally hypermethylated DNA sequences occurred preferentially at regulatory enhancers that are involved in gene regulation, and included sequences known to control genes important in AML pathogenesis. We used oxidative bisulfite sequencing to show that regions with DNA hypermethylation also possessed DNA that was marked by 5-hydroxymethylation, which occurs during active removal of DNA methylation, providing direct evidence that IDH mutations impair DNA demethylation. Further analysis showed that DNA hypermethylation caused by IDH mutations was attenuated in AML samples with co-occurring mutations in the de novo DNA methyltransferase DNMT3A, indicating that the DNA hypermethylation associated with IDH mutations is mediated by DNMT3A. These studies provide the first comprehensive, genome-wide view of the DNA methylation patterns in IDH-mutant AML cells.

We next modeled IDH mutations using patient-derived iPS cells that harbored an IDH2 R140Q allele. These iPS cells were generated using primary bone marrow cells from an AML patient with an IDH2 R140Q mutation via transient expression of the four Yamanaka reprogramming factors via an episomal viral expression system. We successfully obtained 2 clonal lines with pluripotent stem cell (PSC) morphology that contained the IDH2 R140Q allele, which expressed the classic PSC surface markers SSEA4 and TRA-160 and formed teratomas in immunocompromised mice. This result demonstrates that the IDH2 R140Q allele is not an absolute barrier to somatic reprogramming. We confirmed that the IDH2 gene was expressed in these iPS lines, and that the mutation was expressed at the expected 50/50 ratio vs. the wild type allele. Mass spectrometry also showed that the IDH2-mutant iPS cells produced high levels of 2-HG, confirming that the mutant alleles were functionally active in these cells. Surprisingly, analysis of DNA methylation in these cells did not reveal appreciable hypermethylation compared to wild type iPS cells, indicating that the epigenetic effects of IDH mutations are cell-type specific. Focused analysis of regions with DNA hypermethylation that were identified in AML samples, including regulatory enhancers, showed that these regions were ‘reset’ to their normal epigenetic state in the iPS cells. This shows that changes in DNA methylation that result from IDH mutations are reversible. Finally, we performed in vitro, cytokine-mediated hematopoietic differentiation to determine the differentiation capacity of IDH2-mutant iPS cells and determine whether DNA methylation is altered in a hematopoietic cell context. This demonstrated that the differentiation potential of the IDH2-mutant iPS cells was similar to wild type iPS cells. The gene expression profiles of hematopoietic cells derived from the IDH2-mutant iPS lines were also similar to those in hematopoietic cells derived from wild type iPS cells. Interestingly, analysis of these cells after a 14-day differentiation process showed only modest differences compared to the wild type cells, with only subtle hypermethylation at regions with features of genomic enhancers. This result provides new information about the kinetics of IDH-associated DNA methylation changes, which require many cell divisions before the changes are full evident. This study therefore indicates that IDH mutations are likely early events in AML development, and their associated DNA methylation changes must accumulate over years before pre-leukemic cells transform into fully malignant AML cells.

Progress since last update:

This study focused on understanding how mutations in IDH1 and IDH2 disrupt gene expression and contribute to the development of acute myeloid leukemia (AML). These mutations are known to alter epigenetic modifications to DNA and histone proteins that are important for normal gene regulation, but the specific genes that are affected by IDH mutations in AML cells and how the resulting epigenetic changes affect gene regulation remain unclear. The centerpiece of this project is a set of induced pluripotent stem (iPS) cells from primary AML samples that harbor mutations in IDH1 or IDH2. We hypothesized that these reagents could be used to define the direct epigenetic consequences of mutant IDH alleles in human cells. Because these cell lines are pluripotent and can be induced to differentiate into hematopoietic cells in culture, we further hypothesized that directed hematopoietic differentiation of IDH mutant iPS cells could be used to identify hematopoietic-specific genomic ‘targets’ of mutant IDH alleles. We tested these hypotheses via the following specific aims:

Specific Aim 1: We will define the direct epigenetic and transcriptional effects of mutant IDH alleles in human pluripotent stem cells.

Specific Aim 2: We will define the influence of mutant IDH alleles on hematopoietic development and epigenetic regulation during in vitro hematopoietic differentiation of human pluripotent stem cells.

We completed all experiments and analysis proposed in both aims during the funding period of this award. Because we identified very few DNA methylation changes in the IDH-mutant AML iPS cells in the first part of this study, we recently expanded the scope of our analysis to include investigation of DNA methylation patterns in primary AML samples with either IDH1 R132H or IDH2 R140Q mutations. This was performed to define the genomic regions that are uniquely hypermethylated in AML cells with these mutations that we could focus on in the iPS cell line models. This work was performed using existing whole-genome bisulfite data from primary AML samples collected at our institution, plus generation of additional oxidative bisulfite sequence data from these same samples. Interestingly, this analysis found that regulatory enhancers were strikingly enriched in the regions that were uniquely hypermethylated in IDH-mutant AML cells. While this association had been briefly described in previous studies, our work provided new and more comprehensive analysis of this association and allowed us to focus on these regions in our analyses of the IDH-mutant iPS cell lines. The addition of oxidative bisulfite sequencing data to our study also provided critical insight into the mechanism by which DNA hypermethylation occurs. These data showed that the regions that become hypermethylated in IDH-mutant samples have 5-hydroxymethylcytosine, a byproduct of DNA demethylation, and thus provides direct evidence that IDH-associated hypermethylation occurs because of impairment of this pathway. In addition to these experiments, in the past year we completed the hematopoietic differentiation of the IDH-mutant iPS cells and analysis of DNA methylation and gene expression of the derived hematopoietic cells. These data and our analysis of IDH-mutant primary AML samples and iPS cell lines have been prepared in two manuscripts that are currently being finalized for publication.

This study is focused on understanding how mutations in IDH1 and IDH2 disrupt gene expression and contribute to the development of acute myeloid leukemia (AML). These mutations are known to alter epigenetic modifications to DNA and histone proteins that are important for normal gene regulation, but the specific genes that are affected by IDH mutations in AML cells and how the resulting epigenetic changes affect gene regulation remain unclear. We previously generated induced pluripotent stem (iPS) cells from primary AML samples that harbor mutations in IDH1 or IDH2, which we hypothesize can be used to define the direct epigenetic consequences of these mutant alleles in human cells. Because these cell lines are pluripotent and can be induced to differentiate into hematopoietic cells in culture, we further hypothesize that directed hematopoietic differentiation of IDH mutant iPS cells can be used to identify hematopoietic-specific genomic ‘targets’ of mutant IDH alleles that may be responsible for the initiation of AMLs with IDH mutations. We are testing these hypotheses via the following specific aims: 

Specific Aim 1: We will define the direct epigenetic and transcriptional effects of mutant IDH alleles in human pluripotent stem cells.   

Specific Aim 2: We will define the influence of mutant IDH alleles on hematopoietic development and epigenetic regulation during in vitro hematopoietic differentiation of human pluripotent stem cells. 

During the first year of this project we have made substantial progress on both of these aims. As part of Aim 1, we performed gene expression analysis of iPS cell lines with IDH1 R132H or IDH2 R140Q mutant alleles, which demonstrated gene expression profiles that are consistent with pluripotent cells, indicating that the mutant alleles do not have dramatic effects on cell function or identity. We also used these data to determine that the IDH1 and IDH2 genes in these cells are expressed at normal levels, and the mutant alleles of these genes are present at the expected abundance, which is critical for establishing these cell lines as a model system. We have also completed genome-wide DNA methylation analysis of IDH mutant iPS cell lines (along with wild-type controls) to determine whether the IDH mutations directly alter DNA methylation. Interestingly, although mutant IDH alleles are clearly associated with increased DNA methylation in a variety of cancer types, including AML, our initial analyses demonstrated only minor methylation differences between the mutant iPS cells and wild-type iPS controls. Specifically, IDH mutant iPS cells do not display an excess of hypermethylated loci compared to the control cell lines. This suggests that effects of these mutations on DNA methylation depends on factors that are present in hematopoietic cells and not pluripotent stem cells. We propose to investigate this possibility in Aim 2 using in vitro hematopoietic differentiation methods, and during the past year have optimized these approaches for our iPS cells and performed several differentiation experiments. So far, we have observed decreased differentiation efficiency in the IDH mutant iPS cells, indicating that the presence of the IDH mutations may block differentiation. Impaired differentiation is one of the cardinal features of AML, and so this finding suggests that our iPS system may provide valuable insights into how IDH mutations block differentiation and contribute to cancer.  

In the next year, we plan to finalize our analysis of DNA methylation in IDH mutant iPS cells and continue the in vitro hematopoietic differentiation experiments proposed in Aim 2. We will use our differentiation approach to obtain cells that are derived from our IDH mutant iPS cells at various stages of hematopoietic differentiation, which can then be used for DNA methylation studies to test our hypothesis that cell type specific factors are required for IDH mutations to alter DNA methylation. We will also perform gene expression analysis to determine whether changes in DNA methylation result in altered gene regulation and to identify specific genes that may be susceptible to epigenetic and transcriptional changes in IDH mutant cells. Once we have obtained these data, we will prepare a manuscript for publication that describes the epigenetic consequences of mutant IDH alleles in iPS cells, which has not been previously reported.