Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

TET2- and TET3- dependence of vitamin C-induced epigenomic alterations in acute myeloid leukemia Wong, Jasper 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2020_may_wong_jasper.pdf [ 4.79MB ]
Metadata
JSON: 24-1.0388359.json
JSON-LD: 24-1.0388359-ld.json
RDF/XML (Pretty): 24-1.0388359-rdf.xml
RDF/JSON: 24-1.0388359-rdf.json
Turtle: 24-1.0388359-turtle.txt
N-Triples: 24-1.0388359-rdf-ntriples.txt
Original Record: 24-1.0388359-source.json
Full Text
24-1.0388359-fulltext.txt
Citation
24-1.0388359.ris

Full Text

   TET2- and TET3- dependence of vitamin C-induced epigenomic alterations in acute myeloid leukemia  by  Jasper Wong  B.MSc., The University of Western Ontario, 2016  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE  in  The Faculty of Graduate and Postdoctoral Studies  (Genome Science and Technology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  January 2020  © Jasper Wong, 2020   ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the thesis entitled:  TET2- and TET3- dependence of vitamin C-induced epigenomic alterations in acute myeloid leukemia  submitted by Jasper Wong  in partial fulfillment of the requirements for the degree of Master of Science in Genome Science and Technology  Examining Committee: Martin Hirst Supervisor  Connie Eaves Supervisory Committee Member  Paul Pavlidis Supervisory Committee Member     iii  Abstract Neomorphic mutations in isocitrate dehydrogenase 1/2 (IDH1/2) and inactivating mutations in ten eleven translocation dioxygenase 2 (TET2) are frequent and mutually exclusive events in de novo acute myeloid leukemia (AML). IDH1/2 mutations drive epigenomic dysfunction through production of the oncometabolite R-2-hydroxyglutarate (R-2HG), which inhibits the ability of TET2 to oxidize 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), and causes DNA hypermethylation. Vitamin C (vitC) induces epigenetic reprogramming and cellular differentiation through TET re-activation in a murine leukemia model with mutant IDH1R132H. Although TET mutations in AML occur exclusively in the TET2 gene, TET2 and TET3 are expressed at similar levels in de novo AML blasts. The role of TET3 in the regulation of myeloid activation and its role in AML is currently unknown. Based on the functional redundancy among TET enzymes, I hypothesize that the vitC-induced re-establishment of methylation homeostasis and differentiation in IDH1 mutant cells is also mediated through TET3.  To delineate the contributions of TET2 and TET3 to vitC-induced reprogramming, we inactivated TET2 (TET2KO) and TET3 (TET3KO) individually or in combination (DKO) in HOXA9/IDH1R132H (R132H) cells using CRISPR/Cas9. After 15 and 72 hours of vitC treatment, we sequenced total RNA and 5hmC/5mC-immunoprecipitated DNA. We showed that IDH1R132H inhibition of TET is incomplete and that loss of either TET2 or TET3 down-regulates a set of 136 genes related to myeloid activation. A subset of the TET2/3-dependent genes can be rescued with vitC activation of the remaining TET, from which we identified PU.1, CEBPE, and RUNX1 as putative transcription factors. We identified a hypermethylation phenotype at enhancers in the context of the IDH1R132H, which can be reversed through vitC treatment in a TET2- and TET3- dependent mechanism. Pathway analyses of nearby genes and genes up-regulated by vitC suggest a common myeloid differentiation pathway that both TET2 and TET3 can activate. To verify this, TET2KO and TET3KO but not DKO cells showed reduced proliferation and increased levels of myeloid differentiation markers from vitC treatment.  These findings support my hypothesis and suggest a model in which vitC activates both TET2 and TET3 to reprogram the enhancer landscape of IDH1R132H leukemic cells to drive differentiation.            iv  Lay Summary Acute myeloid leukemia (AML) remains a deadly disease with poor prognosis in adults over 65. It is therefore important to understand the mechanisms that contribute to the development of leukemia in order to develop improved treatments. In around 40% of AMLs, mutations in genes that cause aberrant DNA methylation occur, which contributes to block the ability of leukemic cells to differentiate into mature blood cells. We previously showed that vitamin C (vitC) can reactivate a pathway (in the form of an enzyme, TET2), which contributes to the removal of DNA methylation, and can effectively push leukemic cells to differentiate. Little is known about a similar enzyme TET3, since it is rarely mutated in leukemia. I showed here that vitC can also activate TET3 to push leukemic cells to differentiate and that TET2 and TET3 cooperate to regulate myeloid blood cell activation pathways.           v  Preface I was responsible for major areas of concept, generation and functional screening of TET2/3 CRISPR-KO cell lines, data analysis of sequencing data, and manuscript composition. Martin Hirst initially identified the research program and oversaw the progression of the project at all stages. The original HOXA9 IDH1R132H cell lines were developed by Michael Heuser’s group at the Hannover Medical School in Germany. Ping Xiang at the Terry Fox Laboratory of BC Cancer generated the first HOXA9 IDH1R132H TET2KO (TET2KO) cell line using a pre-designed crRNA sequence. I optimized the protocol and re-generated the TET2KO, as well as the TET3KO, and DKO. Qi Cao performed the initial immunofluorescence experiment on the TET2KO cell line and performed the vitamin C treatment of cells that were used for the sequencing library generation. Marcus Wong extracted the total DNA and RNA and generated the meDIP-seq and hmeDIP-seq libraries. Michelle Moksa generated the RNA-seq libraries. Mansen Yu performed the flow cytometry experiments. Misha Bilenky generated the in-house differential expression, fractional methylation caller, and peak calling program used to compare RNA-seq, meDIP-seq, and ChIP-seq data. Luolan Li designed the basis of analysis for differential 5hmC calls and also the CpG fold enrichment script. Data-sets for meDIP-seq for a HOXA9 IDH1WT cell line and H3K27ac- and H3K4me1- ChIP-seq for the HOXA9 IDH1R132H were taken from a previously published data-set1.     vi  Table of Contents Abstract ................................................................................................................................................................ iii Lay summary ........................................................................................................................................................ iv Preface .................................................................................................................................................................. v Table of contents .................................................................................................................................................. vi List of tables ....................................................................................................................................................... viii List of figures ........................................................................................................................................................ ix List of abbreviations ............................................................................................................................................xiii Acknowledgements ............................................................................................................................................. xvi Dedication .......................................................................................................................................................... xvii Chapter 1: Introduction ......................................................................................................................................... 1 1.1 Epigenetic dysregulation of acute myeloid leukemia .......................................................................................... 1 1.2 Regulation of DNA methylation homeostasis  ..................................................................................................... 2 1.3 Aberrant DNA methylation in hematological malignancies  ............................................................................... 3 1.4 TET2 mutations and IDH1/2 mutations in AML  .................................................................................................. 4 1.5 Vitamin C as a potential epigenetic regulator against hematologic cancers  ..................................................... 6 1.6 Vitamin C induces TET activity to reactivate differentiation transcriptional programs  ...................................... 7 Objectives  ............................................................................................................................................................... 10 Chapter 2: Methods ............................................................................................................................................. 11 Chapter 3: Generation of TET2, TET3, TET2/TET3 homozygous deletions in HOXA9 IDH1R132H cell lines ............... 19 3.1 Optimization of CRISPR-cas9 electroporation method for bulk knockout  ........................................................ 19 3.2 Generation and screening of TET2, TET3, TET2/TET3 HOXA9 IDH1R132H knock-out  .......................................... 22 Chapter 4: Role of TET2 and TET3 in the context of IDH1R132H neomorphic mutations in AML .............................. 26 4.1 TET2 and TET3 share overlapping functions in maintaining a myeloid expression signature  .......................... 26 4.2 TET2 and TET3 drive 5hmC gain at enhancers even in the presence of IDH1R132H  ............................................ 31 4.3 IDH1R132H triggers focal hypermethylation at CpG islands and at active enhancer elements  .......................... 33 4.4 Vitamin C induces active demethylation through TET2 and TET3 preferentially at aberrantly methylated active enhancers ...................................................................................................................................................... 35 Chapter 5: TET2 and TET3 cooperate to drive epigenetic reprogramming in response to vitamin C ..................... 42 vii  5.1 Vitamin C drives myeloid differentiation via TET2/3-dependent mechanisms .................................................. 42 5.2 Temporal TET2 and TET3 dependence of vitamin C-induced epigenomic reprogramming ............................... 46 5.3 Vitamin C drives a common myeloid cell differentiation expression signature in a process mediated by TET2 and TET3 .................................................................................................................................................................. 51 5.4 TET2 and TET3 stimulation with vitamin C can partially rescue the deficiency of TET2/TET3 inactivation  ...... 57 Chapter 6: Conclusions ........................................................................................................................................ 65 Bibliography ........................................................................................................................................................ 71               viii  List of Tables Table 1. crRNA used for targeting of TET2 and TET3............................................................................................. 19 Table 2. Top transcription factors of the vitamin C-sensitive TET2/3-dependent set of genes as predicted by ChEA3  ................................................................................................................................................................. 62                 ix  List of Figures Figure 1. Frequency of mutations in de novo AML and oncoprint .......................................................................... 5 Figure 2. A proposed model of vitamin C-induced epigenomic remodeling in IDH1R132H AML ................................ 8 Figure 3. Expression of TET1-3 in human AML blasts and in the IDH1R132H model  ................................................ 10 Figure 4. A schematic of cell lines used in the study  ............................................................................................ 19 Figure 5a. Western blot of TET2 ........................................................................................................................... 21 Figure 5b. Immunofluorescence of 5hmC following 15hr vitC treatment  ............................................................ 21 Figure 5c. Proportion of clonally-derived cells that have non-homologous end-joining events at the TET2 cas9 cut-site via amplicon libraries .............................................................................................................................. 21 Figure 5d. Plots showing the optimization process for increased transduction .................................................... 21 Figure 6a. Target regions of the CRISPR-cas9 crRNA for TET2 and TET3  ............................................................... 23 Figure 6b. Representative Agilent profile for the T7EI assay confirming successful excision  ............................... 23 Figure 6c. Proportion of cells that contained a non-homologous end-joining event at TET2 or TET3 across several cell passages ........................................................................................................................................................ 23 Figure 6d. Immunofluorescence of 5hmC following 15hr vitC treatment  ............................................................ 23 Figure 6e. Representative immunofluorescence images ...................................................................................... 24 Figure 7a. Proportion of cells that contained a non-homologous end-joining event at TET2 across passages  ..... 25 Figure 7b. Representative output from the amplicon libraries ............................................................................. 25 Figure 7c. Competition assay shows no competitive advantage over time for TET2KO or DKO relative to R132H at two different proportions .................................................................................................................................... 25 Figure 8. Schematic of experimental design for meDIP-seq, hmeDIP-seq, and RNA-seq ...................................... 26 Figure 9a. The extent of concordance between differentially expressed genes at two time-points  .................... 28 Figure 9b. The number of consistently differentially expressed genes between TET2KO, TET3KO, DKO against its parental line R132H  ............................................................................................................................................ 28 Figure 9c. Box-plot showing the extent (fold change) of differential expression  ................................................. 28 Figure 9d. Upset plot showing the overlap between differentially expressed genes  ........................................... 28 Figure 9e. Heatmap showing the expression (z-score) of the 136 TET2/3-dependent genes  ............................... 29 x  Figure 9f. Bar-plot showing the top 15 gene-ontology terms for the 136 TET2/3-dependent gene-set  ............... 29 Figure 9g. Bar-plot showing the expression levels of six representative genes at the 15 and 72hr time-points  ... 29 Figure 9h. Box-plot showing the expression levels of genes that were up-regulated in IDH1WT ......................... 29 Figure 9i. Expression levels of Elane ..................................................................................................................... 29 Figure 10a. Bar-plot showing genes that were specifically down-regulated upon TET2 loss  ................................ 30 Figure 10b. Bar-plot showing genes that were specifically up-regulated upon TET2 loss ..................................... 30 Figure 11a. Box-plot showing the distribution of 5hmC and 5mC densities at all DHRs and DMRs  ...................... 32 Figure 11b. 5hmC and 5mC fold change at all promoters, promoters of the 136 TET2/3-dependent genes, and nearest 5hmC-marked enhancer of the 136 TET2/3-dependent genes................................................................. 32 Figure 11c. Box-plot showing the fold change of 5hmC at the nearest enhancers 136 TET2/3-dependent genes  32 Figure 11d. Genome browser shot of the Pdcd4 locus and its nearby enhancer  .................................................. 32 Figure 12a. Genomic coverage of global 5mC changes in R132H, TET2KO, TET3KO, and DKO relative to the non-leukemic line IDH1WT.......................................................................................................................................... 34 Figure 12b. Proportion of hypermethylated and hypomethylated regions that overlap with CpG islands and enhancers  ........................................................................................................................................................... 34 Figure 12c. Bar-plot showing the regional enrichment of hypermethylated and hypomethylated regions at various genomic features  .................................................................................................................................... 34 Figure 12d. Genome browser shot of a consistently IDH1R132H hypermethylated region at the Csf1 locus ........... 34 Figure 12e. Top 3 most significant pathways that are enriched in the IDH1R132H hypermethylated enhancers and CpG island promoters  ......................................................................................................................................... 34 Figure 13a. Heatmap showing regional changes in 5mC and 5hmC of IDH1R132H hypermethylated enhancers following 15 and 72hr of vitamin C treatment  .................................................................................................... 37 Figure 13b. Heatmap showing regional changes in 5mC and 5hmC of IDH1R132H hypermethylated CpG island promoters following 15 and 72hr of vitamin C treatment  ................................................................................... 37 Figure 14a. Box-plot showing 5hmC density at 408 IDH1R132H hypermethylated enhancers 15 and 72hrs following vitamin C treatment  ............................................................................................................................................ 40 Figure 14b. Top 10 most significant pathways that are enriched in genes within 20kb of the 408 IDH1R132H hypermethylated enhancers that gain 5hmC in R132H, TET2KO, and TET3KO at 72hrs  ....................................... 40 Figure 14c. Top 10 most significant pathways that are enriched in genes within 20kb of the 136 IDH1R132H hypermethylated enhancers that gain 5hmC in R132H, TET2KO, and TET3KO at 15hrs  ....................................... 40 xi  Figure 14d. Genome browser screenshot of 5mC and 5hmC density of R132H, TET2KO, TET3KO, and DKO at a hypermethylated enhancer within the gene-body of Notch2  .............................................................................. 41 Figure 14e. Profile of 5hmC and 5mC average signal at IDH1R132H hypermethylated enhancers in a replicate of R132H and TET2KO treated with 15hr of vitamin C .............................................................................................. 41 Figure 14f. Genome browser screenshot of 5hmC density of R132H and the TET2KO replicate at the hypermethylated enhancer within the gene-body of Notch2  .............................................................................. 41 Figure 15a. Pictures taken of 100 000 cells in a U-shaped 96-well plate following 96 and 264 hours of vitamin C treatment  ........................................................................................................................................................... 43 Figure 15b. Proliferation assay of cells treated with daily doses of 0.345mM vitamin C for 20 days  ................... 43 Figure 16a. Wright-Giemsa stain of cells taken after 15d of vitamin C treatment  ............................................... 44 Figure 16d. FACS analysis of myeloid cell surface markers following 20d of vitamin C treatment  ....................... 45 Figure 16c. Gating strategy to identify positive and negative populations  .......................................................... 45 Figure 16d. Gating strategy to identify Mac1hi vs. Mac1lo populations  ................................................................ 46 Figure 17a. Genomic occupancy of differentially hydroxymethylated regions following 15 and 72hr of vitamin C treatment  ........................................................................................................................................................... 48 Figure 17b. Venn diagrams showing the degree of overlap between DHRs at 15hr and 72hr  .............................. 48 Figure 17c. Bar-plot showing regional enrichment of differentially hydroxymethylated CpGs  ............................ 49 Figure 17d. Venn diagrams showing the degree of overlap between differentially hydroxymethylated enhancers within the 15hr and 72hr time-points  ................................................................................................................. 49 Figure 17e. Heatmap showing 5hmC density at core enhancers and secondary enhancers  ................................. 49 Figure 17f. Violin plots showing the change (log2FC) in 5mC at core enhancers  .................................................. 50 Figure 17g. Violin plots showing basal level (log10rpkm) of 5mC signal at core and secondary enhancers  ......... 50 Figure 17h. Bar-plot showing the top 12 most significant enrichment terms for unique genes within 20kb of core and secondary enhancers  ................................................................................................................................... 50 Figure 18a. Number of differentially expressed genes at 15hr and 72hr following vitamin C treatment .............. 53 Figure 18b. Violin plot showing the extent (log2FC) of vitamin C-induced expression changes at 15hr and 72hr following vitamin C treatment  ............................................................................................................................ 53 Figure 18c. Bar-plot showing expression levels TETs and DNMTs  ........................................................................ 53 Figure 18d. Upset plot showing the total number of differentially expressed genes that overlap between all up-regulated gene-sets  ............................................................................................................................................ 53 xii  Figure 18e. Upset plot showing the total number of differentially expressed genes that overlap only in the 72hr up-regulated gene-sets  ....................................................................................................................................... 54 Figure 18f. Bar-plot highlighting 8 genes that were up-regulated in a TET2-dependent manner after 72hr of vitamin C treatment  ............................................................................................................................................ 54 Figure 18g. Heatmap showing highly significant functional pathways that are common or unique across models  ............................................................................................................................................................................ 54 Figure 18h. Expression levels of R132H-only genes S100a8, S100a9, Tifab ........................................................... 54 Figure 19a. Box-plot showing extent of 5hmC changes at promoters or nearest enhancers of up-regulated genes following  72hr of vitamin C treatment  ............................................................................................................... 56 Figure 19b. Genome browser shot of the Csf2ra promoter .................................................................................. 56 Figure 19c. Genome browser shot of the Itgam locus and its nearby enhancer  .................................................. 57 Figure 20a. Heatmap showing expression levels of the TET2/3-dependent genes with and without vitamin C  ... 59 Figure 20b. Collapsed heatmap showing k-means (k=2) clustering of the 136 genes into a group that is up-regulated by 72hr of vitamin C and a group that is not  ....................................................................................... 59 Figure 20c. Representative expression levels of 5 vitamin C-sensitive genes and 5 vitamin C-insensitive genes .. 59 Figure 20d. Boxplots showing expression values of vitamin C-sensitive genes and vitamin C-insensitive genes relative to IDH1WT  ............................................................................................................................................. 60 Figure 20e. Boxplots showing number of CpGs at the promoters of vitamin C-sensitive and vitamin C-insensitive genes  .................................................................................................................................................................. 60 Figure 20f. Boxplots showing CpG density at the enhancers of vitamin C-sensitive genes and vitamin C-insensitive genes  ................................................................................................................................................. 60 Figure 21a. Boxplots showing 5hmC density (rpkm) at the promoters of vitamin C-sensitive genes and vitamin C-insensitive genes  ................................................................................................................................................. 61 Figure 21b. Boxplots showing 5hmC density (rpkm) at the nearest differentially hydroxymethylated enhancers of vitamin C-sensitive genes and vitamin C-insensitive genes  ................................................................................. 61 Figure 22a. Top de novo motifs at regions that gain 5hmC and lose 5mC within 1.5kb after 72hr of vitamin C treatment  ........................................................................................................................................................... 64 Figure 22b. Heatmap showing 5hmC density of R132H, TET2KO, TET3KO, and DKO at RUNX1-binding sites   ...... 64 Figure 23. Proposed model of TET activity gradient  ............................................................................................ 70    xiii  List of abbreviations 2-PAA – 2-phosphate ascorbic acid 5caC – 5-carboxylcytosine 5fC – 5-formylcytosine 5hmC – 5-hydroxymethylcytosine 5mC – 5-methylcytosine ɑ-KG – ɑ-ketoglutarate  AML – Acute myeloid leukemia ANOVA – Analysis of variance ATM – Ataxia telangiectasia mutated BER – Base excision repair CAR-T – Chimeric antigen receptor T-cell Cas9 – CRISPR associated protein 9 CGI – CpG island ChEA3 – ChIP-X Enrichment Analysis 3 ChIP – Chromatin immunoprecipitation CRISPR – Clustered Regularly Interspaced Short Palindromic Repeats crRNA – CRISPR RNA CXXC-domain – Domain containing C-X-X-C (C = cysteine; X = any amino acid) DDR – DNA-damage response DEG – Differentially expressed genes DHR – Differentially hydroxymethylated regions DMR – Differentially marked regions DN(23D) – 136 down-regulated genes in TET2KO, TET3KO, and DKO relative to R132H DNA – Deoxyribonucleic acid DNMT – DNA methyltransferase DKO – HOXA9 IDH1R132H TET2-/- TET3-/- mouse bone marrow cell line xiv  ESC – Embryonic stem cell FDA – U.S. Food and Drugs Administration FDR – False discovery rate G-CIMP – Glioma CpG island methylator phenotype Gulo – L-gulono-γ-lactone oxidase H3K4me1 – Histone 3 Lysine 4 monomethylation H3K4me3 – Histone 3 Lysine 4 trimethylation H3K27ac – Histone 3 Lysine 27 acetylation HDAC2 – Histone deacetylase 2 HIF – Hypoxia-inducible factor hmeDIP – Hydroxymethylated DNA immunoprecipitation HOXA9 – Homeobox A9 HSC – Hematopoietic stem cell HSPC – Hematopoietic stem and progenitor cell IDH1 – Isocitrate dehydrogenase 1 IDH1wt – HOXA9 IDH1wt mouse bone marrow cell line iPSC – Induced pluripotent stem cell JmjC – Jumonji C LSC – Leukemic stem cell MDS – Myelodysplastic syndrome meDIP – Methylated DNA immunoprecipitation NHEJ – Non-homologous end-joining PARP – Poly-ADP ribose polymerase PHD2 – Prolyl hydroxylase domain 2 PRL-3 – Phosphatase of regenerating liver 3 R132H – HOXA9 IDH1R132H mouse bone marrow cell line R-2HG – R-2-hydroxyglutarate RAGE – Receptor for advanced glycation end products xv  RNA – Ribonucleic acid RNP – Ribonucleoprotein ROS – Reactive oxygen species RPKM – reads per kilobase per million mapped reads SVCT2 – Sodium-dependent vitamin C transporter 2 T7EI – T7 endonuclease I TCGA – The Cancer Genome Atlas TDG – Thymine-DNA glycosylase TET – Ten-eleven translocation dioxygenase TET2KO – HOXA9 IDH1R132H TET2-/- mouse bone marrow cell line TET3KO – HOXA9 IDH1R132H TET3-/- mouse bone marrow cell line TF – Transcription factor TLR – Toll-like receptor tracrRNA – trans-activating crRNA TSS – Transcription start site Vitamin C/vitC – 2-phosphate ascorbic acid         xvi  Acknowledgements I offer my gratitude to the faculty, staff, and peers at the University of British Columbia and at the BC Cancer Research Centre who have facilitated intellectual discourse and inspired my work. I extend my thanks to Dr. Connie Eaves and Dr. Paul Pavlidis for being on my thesis advisory committee and enriching my knowledge within the realms of hematology and bioinformatics. I thank Dr. Martin Hirst for his direction and insight during times of turbulence. I thank Michelle Moksa, Qi Cao, and Marcus Wong for their supervision and advice throughout my experiments in molecular and cell biology. I thank Alireza Lorzadeh and Luolan Li for their expertise in bioinformatics and statistics, and for active engagement in discussions related to my work. I would also like to thank the other members of my lab for providing an encouraging space for scientific inquiry and intellectual pursuit. I gratefully acknowledge the funding provided by the following grants: Canadian Institutes of Health Research (CIHR-120589), Canadian Foundation of Innovation (CFI-31343, CFI-31098), Canadian Cancer Society (CCSRI-7034890), and the Terry Fox Research Institute Program Project (TFF-1074, TFF-1039, TFF-122869). And special thank you to my parents and sister, who have provided me with mental and emotional rapport throughout my years of education so that I may not stray.      xvii                       To my parents and sister           Chapter 1: Introduction 1.1 Epigenetic dysregulation in acute myeloid leukemia Acute myeloid leukemia (AML) is the most common leukemia in adults, with incidence increasing with age2. Characterized by the rapid accumulation of abnormal and poorly differentiated myeloid cells termed blasts in the bone marrow and peripheral blood, AML is an aggressive malignancy, with 70% of patients 65 years and older dying within a year of diagnosis3. Cytogenetic profiles were the initial method to assess risk; ranging from favourable risks (e.g. harbouring PML-RARA, RUNX1-RUNX1T1, MYH11-CBFB fusions) to unfavourable risks (monosomy karyotype), though the majority of AML patients are classified as intermediate risk, and contain a normal karyotype4,5.  With the advent of whole-genome and whole-epigenome screens, the ability to explore cancer on a comprehensive genetic level has brought unprecedented insights. The Cancer Genome Atlas (TCGA) has generated extensive profiles of various cancers including AML, providing further classification of intermediate risk AML based on the presence of recurrent gain and loss of function genetic mutations5. AML was originally characterized by the “two-hit” hypothesis, which proposes that leukemogenesis is driven by the collaboration of two broad classes of mutations, with class I mutations (FLT3, K/NRAS, TP53, c-KIT) conferring activation of proliferation pathways and simultaneous class II mutations (NPM1, CEBPA) impairing normal hematopoietic differentiation of myeloid progenitor cells6. Since then, another group of mutations that do not conform to either class has emerged, including mutations in enzymes that alter DNA methylation homeostasis such as DNA methyltransferase 3A (DNMT3A), isocitrate dehydrogenase 1/2 (IDH1/2), and ten-eleven translocation 2 dioxygenase (TET2), which appear in more than 40% of AML cases, and implicate alterations in DNA methylation in AML progression5. Furthermore, mutations in IDH1/2 and TET2 occur recurrently in pre-leukemic patients, suggesting that these mutations may be early drivers of leukemia, though importantly, they are insufficient to drive 2  leukemogenesis on their own7,8. However, survival data associated with these epigenetic mutations are inconsistent and their prognostic significance has not been established2.  1.2 Regulation of DNA methylation homeostasis The most common DNA modification is 5-methylcytosine (5mC), existing largely in the context of CpG dinucleotides in mammalian genomes. DNA methylation of CpGs at regulatory regions including promoters and enhancers, is classically associated with transcriptional repression, whereas DNA methylation within gene-bodies is associated with transcriptional processing9. Within the somatic mammalian genome the majority of CpGs are methylated (70-80%) and are found at various densities, including at CpG islands (CGIs), defined as CG-rich (>50%) regions in the genome 1kb or greater in length10,11. Methylation of promoter-associated CGIs is associated with repression of the associated gene and can be inherited through multiple cell divisions. DNA methylation has thus been identified as a form of “epigenetic memory” linked to the maintenance and establishment of cellular identity across cell types including embryonic stem cells (ESCs), induced pluripotent stem cells (iPSCs), and hematopoietic stem cells (HSCs)12–14. CpG methylation is controlled through regulated addition and removal of methyl groups. DNA methyltransferases (DNMTs) can introduce methylation de novo (via DNMT3A and DNMT3B) and maintain existing 5mC after replication (via DNMT1)15,16. In contrast, demethylation is mediated by the ten-eleven translocation (TET1-3) dioxygenase enzymes, which iteratively oxidize 5mC to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxycytosine (5-caC), from which thymine-DNA glycosylase (TDG) of the base-excision repair (BER) pathway or passive demethylation via DNA replication results in the conversion of 5mC to 5C17–19. 5hmC is an intermediate in DNA demethylation and its abundance in the mammalian genome suggests that it may also serve as a regulatory molecule20–22. While 5hmC is detected at less than 1% of the level of 5mC, it is consistently 3  detected in a majority of somatic cell types (~0.1% of total nucleotides in brain; 0.01 – 0.05% in other tissues)19,23–25. The interplay between DNMTs and TETs maintains the genome-wide DNA methylation homeostasis, which is deeply rooted in many biological processes, including stable transcriptional regulation, cellular identity, and cellular differentiation. Though structurally similar, several distinctions separate the TET family members. Unlike TET1 and TET3, TET2 does not contain a CXXC-domain to bind methylated cytosine residues and thus, its recruitment to DNA is mediated through transcription factors, such as WT1, EBF1, NANOG, PU.1, PPARγ, and PRDM126–29. Enzymatically, TET1 and TET2 have been shown to have a higher affinity for 5mC than for 5hmC or 5fC30. In contrast, the CXXC domain in TET3 has been shown to act as a 5caC reader during BER31. However, the extent to which the TETs contribute to maintaining the methylome and hydroxymethylome of cells appears to be tissue and context dependent32–34. For example, Tet1 and Tet2 are highly expressed in mouse ESCs, while Tet3 is abundant in oocytes and zygotes and is indispensable for neonatal growth35,36. In contrast, Tet2 and Tet3 are highly expressed in neuronal and hematopoietic cells whereas Tet1 is largely not detected37,38.   1.3 Aberrant DNA methylation in hematological malignancies The critical role of DNA methylation in mammalian development is exemplified in the process of hematopoietic differentiation. HSCs differentiate to give rise to myeloid and lymphoid lineages, during which distinct DNA methylation signatures emerge that are associated with the expression patterns of key genes that control myeloid (e.g. Kcnh2) and lymphoid (e.g. Susd3) cell fate reflecting an underlying differentiation trajectory that can be measured at the epigenetic level39. HSCs have the ability to continuously self-renew and differentiate throughout the lifetime of an organism, though with ageing, HSCs gradually lose their self-renewal and reconstitution capacity. Aged murine HSCs have been observed to have a propensity towards myeloid-skewed differentiation40,41. The mechanisms underlying 4  HSC ageing have pointed to reductions in DNA damage response (DDR) functionality, hypoxic microenvironments caused by increases in reactive oxygen species (ROS), and changing epigenetic modifications including altered DNA methylation levels42–44.  During ageing, accumulation of somatic mutations confers a selective advantage in one or more hematopoietic stem and progenitor cells (HSPCs), which will result in the sustained expansion of mature blood cells derived from these selected HSPCs. Clonal expansion of HSPCs, termed clonal hematopoiesis, is associated with other age-related pathological conditions, including increased risk for hematological malignancies. From a large scale clinical study of 12380 persons, clonal hematopoiesis was observed in 10% of persons older than 65 years of age and is associated with somatic lesions in genes recurrently mutated in AML (e.g. DNMT3A, TET2, ASXL1)45. This suggests that mutations in DNMT3A, TET2, and ASXL1 are early events that may promote clonal hematopoiesis and subsequent progression to dysplasia and leukemia. In support of this view, DNMT3A loss is associated with impaired differentiation in a mouse model, while TET2 loss increased the self-renewal capacity and induced a competitive growth advantage in mouse HSPCs7,46,47.  1.4 TET2 mutations and IDH1/2 mutations in AML Inactivating heterozygous TET2 mutations and neomorphic IDH1/2 mutations are mutually exclusive in AML genomes, suggesting that these mutations act on a common pathway to promote leukemogenesis (Figure 1)48. IDH1 and IDH2 are key enzymes in the citric acid cycle, where they convert isocitrate to ɑ-ketoglutarate (ɑ-KG). Neomorphic mutations in IDH1/2 confer an acquired ability to convert ɑ-KG into the oncometabolite 2-hydroxyglutarate (R-2HG), which acts to competitively and reversibly inhibit various ɑ-KG-dependent enzymes including the Jumonji C (JmjC)-domain-containing histone demethylases, the prolyl and lysyl hydroxylases required for collagen folding and hypoxia-inducible factor (HIF) stability, and the TET family49–51. In a bone marrow-derived murine cell line 5  immortalized by overexpression of homeobox A9 (Hoxa9), R-2HG was sufficient to initiate leukemogenesis in a reversible manner52,53. Mutations in IDH1/2 and TET2 converge phenotypically in the form of reduced TET activity driving a shared DNA hypermethylation and expression signature29,48. Notably, the hypermethylation phenotype associated with neomorphic IDH1/2 mutations manifests in the glioma CpG island methylator phenotype (G-CIMP), which is a specific hypermethylation signature of CpG islands that was originally attributed to mutant IDH1 expression in glioma through its inhibition of the TET enzymes54. Since mutations in IDH1 and IDH2 are recurrent in lower-grade gliomas and in AML, this observation has commonly been extended to that of neomorphic IDH1 AML48,55.   Figure 1. Mutations in IDH1/2 and TET2 are frequently occurring (~29%) (left) and are mutually exclusive events (oncoprint) (right) in de novo cytogenetically normal AML (Data from TCGA, 2013).   However, several phenotypic distinctions differentiate TET2 and IDH1 mutations, as was observed between TET2 knock-out and IDH1 knock-in mutants in mice. In an IDH1R132H knock-in mouse model, the expression of neomorphic IDH1 was insufficient to immortalize HSCs or trigger malignancies, though it did expand the HSC compartment but not its myeloid or lymphoid progenitor compartments56. In addition, transplanted IDH1R132H knock-in HSCs showed no difference in repopulating capacity relative to wild-type HSCs. In contrast, heterozygous and homozygous knock-outs of TET2 in murine models are 6  associated with a competitive advantage in transplant experiments and a propensity towards the myeloid progenitor compartment57.  A common functional outcome of neomorphic IDH1 and inactivating TET2 mutations is dysregulation of DNA damage response (DDR) and repair pathways. In a reversible TET2 knockdown mouse model, TET2 restoration was sufficient to impair self-renewal in TET2-deficient HSPCs and promote myeloid differentiation through upregulation of poly-ADP ribose polymerase (PARP) and base excision repair mechanisms58. The role of TET in the maintenance of DNA fidelity was further supported by the observation that accumulation of DNA damage occurred in a leukemic mouse model with TET2 and TET3 simultaneously inactivated37. Similarly, IDH1R132H knock-in murine HSCs accumulated DNA damage with age through a marked downregulation of the ataxia telangiectasia mutated (ATM) gene, a sensor kinase that recruits DNA repair machinery to double strand breaks59.   1.5 Vitamin C as a potential epigenetic regulator in hematologic cancers Vitamin C (ascorbic acid) is an essential nutrient that is known for its anti-oxidant ability, and its essential and conserved role in collagen synthesis60. A large majority of vertebrate and invertebrate species are able to synthesize vitamin C endogenously. However, this trait has been lost in a number of species, including bats, anthropoid primates, and guinea pigs. In all known cases, the loss of this ability is attributed to species-specific mutations in the L-gulono-γ-lactone oxidase (Gulo) gene, which codes for the enzyme that catalyzes the final step of vitamin C biosynthesis60.  Historically, since Linus Pauling proposed the application of vitamin C in chemotherapy in 1976, the mechanism of vitamin C in cancer has been attributed to its ability to generate excessive amounts of extracellular H2O2 inducing cell death61,62. Chen et al. demonstrated that high pharmacological intravenous concentrations of vitamin C triggered cell death in mouse and human tumorigenic cells, but not normal cells61,63,64. Yet, inconsistent clinical trials across studies and across cancers suggested that 7  the effects of vitamin C may be context specific, and cannot be solely explained through extracellular oxidative stress65,66.  Notably, many intracellular functions that rely on ascorbate as a cofactor require concentrations (0.5 – 4 mM) that are much higher than can be found in plasma (40 – 60 µM)67. As ascorbate is not typically stable in media and high doses of vitamin C are cytotoxic, alternative isomers are necessary to achieve a stable and accurate dosage68. To that end, 2-phosphate ascorbic acid (2-PAA) was developed as an oxidatively stable form of vitamin C that does not contribute to extracellular hydrogen peroxide formation to examine its more biologically relevant intracellular effects69–74. In the last decade, vitamin C as a potential therapeutic has gained renewed interest with the discovery that it acts as a cofactor to facilitate TET2 activity75. Vitamin C has been shown to interact with TET as a cofactor to induce differentiation of mouse embryonic stem cells into blastocyst-like states and to augment induced-pluripotent stem cell generation75–77. In the context of leukemia, vitamin C has been shown to improve the ability of The US Food and Drug Administration (FDA)-approved hypomethylating drug 5-aza-2-deoxycitidine to inhibit cancer cell proliferation and increase apoptosis in several human cancer cell line models in part through a TET2-mediated process78. Through its ability to induce active demethylation, vitamin C has been of particular interest in studying hematologic cancers with aberrant gains in DNA methylation.  1.6 Vitamin C induces TET activity to reactivate differentiation transcriptional programs Previously, we have shown that vitamin C can stimulate TET2 activity in a murine leukemic cell line harbouring a HOXA9 vector and a neomorphic IDH1 (IDH1R132H) mutation, resulting in epigenetic reprogramming and myeloid differentiation1. Genomic regions that become demethylated were enriched in DNA-binding motifs for PU.1, a transcription factor previously shown to recruit TET279. Coupled with the observation that a genome-wide loss of PU.1 binding sites occurred after 72 hours of 8  vitamin C treatment, this suggested a model in which TET2 is recruited by PU.1 to methylated regulatory regions, which becomes inhibited by the presence of R-2HG, setting up an epigenetic block that inhibits expression of genes that drive differentiation. We showed that vitamin C can out-compete the inhibitory effects of R-2HG and epigenetically reprogram the genome, allowing for the recruitment of activating transcription factors, such as RUNX1 to demethylated regions to drive a myeloid differentiation expression pattern (Figure 2)1. This model would suggest that vitamin C and TET activation are integral to the epigenetic regulation of myeloid differentiation.   Figure 2. A proposed model of vitamin C-induced epigenomic remodelling in IDH1R132H AML (adapted from Mingay et al., 2017).   Consistent with this model, vitamin C concentrations are specifically elevated in human and mouse HSC populations, decreasing with differentiation state80. Intracellular vitamin C concentration is correlated with expression of Slc23a2, the gene encoding sodium-dependent vitamin C transporter 2 (SVCT2), one of two dedicated vitamin C transporters. Mice that were depleted of vitamin C (Gulo-/-) or 9  were deficient in Slc23a2 were found to have increased HSC frequency80. Furthermore, vitamin C deficiency promotes leukemogenesis in a leukemia mouse model harbouring TET2+/- and FLT3ITD mutations80,81. Interestingly, latency was reduced in vitamin C-deficient irradiated mice (Gulo-/-) transplanted with TET2-/- Flt3ITD leukemic stem cells (LSCs), which suggests that vitamin C depletion accelerates leukemogenesis in part through a TET2-independent manner80. In murine hematopoietic cells, TET2 is the predominant driver of DNA demethylation, though TET3 is also necessary for the maintenance and proper regulation of HSC production32,37,82. Because of their structural similarities, the TET family of enzymes are thought to display functional redundancy, though their specific functions have not been well annotated. Indeed, there are overlapping functions of different TETs, as exemplified in zebrafish, where TET2 and TET3 have a combined requirement for HSPC emergence83. Notably, TET2 and TET3 have been shown to have redundant roles in suppressing leukemogenesis, wherein acute loss of both TETs is sufficient to induce leukemia in a mouse model37. Conversely, TET1 has been implicated in the regulation of murine B cell lymphopoiesis84. Importantly, in normal mouse and human hematopoietic cells, human AML blasts, and in our HOXA9 IDH1R132H mouse model, TET2 and TET3 are expressed at similar levels (Figure 3)5,37. The impact of vitamin C on serial replating capacity in bone marrow cells of TET2-/-TET3KD is less pronounced compared to TET2-/- cells, again suggesting that vitamin C may act in part through TET3 to suppress aberrant self-renewal58.    10   Figure 3. (left) Expression of TET1, TET2, and TET3 in AML patients from TCGA data (RNA-seq by Expectation Maximization (RSEM) counts). (right) Expression of TET1, TET2, and TET3 in the mouse HOXA9 IDH1R132H cell line with and without vitamin C treatment (n = 4).   However, despite TET2 and TET3 showing comparable expression levels and sharing a tumour-suppressor function, only TET2 is recurrently mutated in de novo AML. Given the established ability of vitamin C to induce hydroxymethylation in the context of our IDH1R132H AML model, I sought to use it as a tool to understand the relationship between TET2 and TET3 in the establishment of the hypermethylated phenotype associated with AML. Based on the reported functional redundancy among TET enzymes, I hypothesize that the vitC-induced differentiation in our mouse model is mediated through both TET2 and TET3.  Objectives: 1. To generate a bulk population of TET2-/-, TET3-/-, and TET2-/-;TET3-/- in the HOXA9 IDH1R132H line. (Chapters 2.1, 2.2) 2. To determine target genes and genomic features that are under the regulation of TET2, TET3, or both TET2 and TET3. (Chapters 3.1, 3.2) 3. To examine the TET2- and TET3- dependence of vitamin C on the transcriptome, hydroxymethylome, and phenotype in the context of a leukemic model harbouring IDH1R132H. (Chapters 3.3, 3.4, 4.1, 4.2, 4.3) 4. To examine the extent to which TET2 and TET3 are functionally redundant by employing vitamin C as a tool to augment TET activity. (Chapters 4.4)   Chapter 2: Methods 2.1 Cell line generation Bulk mouse bone marrow cells were transfected with retroviral vectors harbouring HoxA9 and IDH1R132H-eGFP as previously described1,8. The bulk population was then sorted for GFP expression to generate the HOXA9 IDH1R132H cell line used here. 2.2 Cell culture Cells were maintained in Dulbecco’s modified Eagle’s medium (Stem Cell Technologies, Vancouver, BC, Canada) supplemented with 15% heat-inactivated fetal bovine serum (Life Technologies, Burlington, ON, Canada) and 6 ng/mL murine interleukin-3, 10 ng/mL human interleukin-6, 20 ng/mL murine stem cell factor (Peprotech, Hamburg, Germany), and antibiotics (penicillin/streptomycin). 2.3 CRISPR-cas9 homozygous disruption in bulk cell lines  Homozygous deletions of TET2 and TET3 were conducted using the alt-R CRISPR-cas9 system (IDT). CRISPR-RNAs (crRNAs) were designed to target a PAM site in exon 3 of TET2 or TET3, and the gRNA was generated by hybridizing the crRNA to a transactivating crRNA (tracrRNA) (Table 1). The gRNA was mixed with a Hifi cas9 nuclease to form a ribonucleoprotein (RNP). The Neon Electroporation machine (ThermoFisher) was used to transduce HOXA9 IDH1R132H cells (2 x 105 cells/mL) with the completed RNP at 1400V, 20ms, 1 pulse. Cells were then seeded immediately into a 24-well plate with media (as above) without antibiotics for 24 hours. 2.4 T7 endonuclease I assay Primers that generate 800-1000bp amplicons were designed that spanned the CRISPR cut-site so that the cut-site is within 200-400bp from the end of one primer. Amplicons were generated of cells that were electroporated with a CRISPR ribonucleoprotein. The T7 endonuclease I (T7EI) assay was conducted on these amplicons as per the kit (IDT). The PCR products were then measured on an Agilent chip to determine successful cutting and relative editing efficiency. For TET2, we used the primers 12  (forward: CATGGAGATTCAAGGCAGCTAAG; reverse: TGGCATTATCAGCATCACAGG) and for TET3, we used primers (forward: CTTCGAGGAAAGCCAAGAAAGAA; reverse: CAGGAGCAAAGGATGGATGTTC) to generate amplicons for T7EI targeting. 2.5 Amplicon sequencing Primers of which one side falls within 80bp of the CRISPR cut-site were generated with a NNNN barcode and Illumina adapters. For TET2, we used the primers (forward: GAGACACCAAGTGGCAATCT; reverse: TAATGTAACGCTCCTGCCTTT) and for TET3, we used the primers (forward: ACGCTGCTCGTCTGGAA; reverse: GTGGTTTCATGCCATGCCT). Thirty cycles of PCR were done to the gDNA of cells from multiple passages that were electroporated with a CRISPR RNP. Eight cycles of PCR were conducted to add indices with Illumina PCR primers. The libraries were then pooled and sequenced on the Illumina Miseq. Fastq files were extracted and aligned to the reference amplicon sequence using the CRISPResso suite85. Reads with an average Phred score of at least 30 were taken and indels were considered successful non-homologous end joining events if they occur within 1bp of the cut-site. Substitutions were ignored to provide a conservative estimate of knockout efficiency.  2.6 DNA/RNA extraction Frozen pellets were treated with 1% 2-mercaptoethanol in Qiagen Buffer RLT Plus to lyse the cells. DNA was separated using the Qiagen AllPrep DNA spin columns, and the flow-through was used for RNA purification. DNA extraction was completed using the Qiagen DNeasy kit. RNA was purified as per the mirVana miRNA isolation kit (ThermoFisher AM1560). 2.7 Western Blot One million cells in frozen pellets were lysed using RIPA buffer (ThermoFisher 89900) with Protease Inhibitor Cocktail (Calbiochem 539134). The supernatant was extracted and quantified using the BCA protein assay. 30µg of protein was mixed with 4x LDS protein dye and reducing agent and the 13  mixture was heated at 70°C for 10 minutes to denature the proteins. Samples were loaded on a Novex 3-8% Tris-Acetate pre-cast gel and ran at 100V for 100 minutes. Gel transfer onto a nitrocellulose membrane was conducted at 26V for 2 hours. Ponceau Red was used to ensure successful transfer, and the membrane was blocked using 3% BSA. The membrane was separated between 225kDa and 120kDa and 1:1000 rabbit anti-mouse TET2 polyclonal (Santa Cruz S-13) and 1:1000 mouse anti-Vinculin monoclonal (Invitrogen VLN01) antibodies were used to blot each piece overnight at 4°C, respectively. Secondary blotting was performed with goat anti-rabbit (LIC-926-32211) and goat anti-mouse (LIC-926-68070) for TET2 and Vinculin, respectively. ImageJ was used to quantify the relative signal of protein. 2.8 Methylated DNA immunoprecipitation (meDIP) and hydroxymethylated DNA immunoprecipitation (hmeDIP) -sequencing Genomic DNA was harvested from untreated and vitamin C treated (0.345mM daily) HOXA9 IDH1R132H cells 15hr and 72hr after initial treatment. Genomic DNA was sonicated to 200bp with a Covaris E-220 Sonicator followed by end-repair and adapter ligation. 5ng of adapter ligated DNA was used to check the quality of the library. A 1:1000 ratio of 50pg/uL 1:1:1 spike-in control of unmethylated lambda:methylated T7:hydroxymethylated M13 was included to measure immunoprecipitation specificity. 100ng and 150ng of adapter ligated DNA were immunoprecipitated overnight with an anti-5-methylcytosine antibody (Eurogenentec BI-MECY-0100) and anti-5-hydroxymethylcytosine antibody (Eurogenetec BI-HMCY-0100), respectively. 5ng of adapter ligated DNA was also ran in parallel without immunoprecipitation as an input-control. Dynabeads M-280 sheep anti-mouse (ThermoFisher 11201D) antibodies were used for the secondary immunoprecipitation. Proteinase K digestion was performed, followed by a 30% PEG SeraMag bead wash to purify the DNA. A 7-cycle PCR reaction was used to add indices with Illumina PCR primers. qPCR of lambda, M13, and T7 was performed to check the specificity of the antibodies. The indexed libraries were pooled and sequenced as paired-end 125nt reads on the Illumina Hiseq 2500 platforms (v3 chemistry).  14  2.9 RNA-seq Total RNA was extracted using a combination of mirVana miRNA Isolation kit (Thermofisher, AM1560) and All prep DNA/RNA Mini Kit (Qiagen, 80204) and then assessed for quality and quantified using Agilent Bioanalyzer (Life Technologies). Total RNA was rRNA depleted using NEBNext rRNA Depletion Kit (New England BioLabs, E6310L).  1st strand cDNA was generated using Maxima H minus First Strand cDNA Synthesis Kit (Thermo Scientific, K1652) with the addition of 1ug of Actinomycin D (Sigma, A9415).   The product was purified using in-house prepared 20% PEG, 1M NaCL Sera-Mag bead solution at 1.8X ratio and eluted in 35uL of Qiagen EB buffer.  Second Strand cDNA was synthesized in a 50uL volume using SuperScript Choice System for cDNA Synthesis (Life Technologies, 18090-019) with 12.5mM GeneAmp dNTP Blend with dUTP.  Double stranded cDNA was then purified with 20% PEG, 1M NaCL Sera-Mag bead solution at 1.8X ratio, eluted in 40uL of Qiagen EB buffer, and fragmented using Covaris E220 (55 seconds, 20% duty factor, 200 cycles per burst).  Sheared cDNA was End Repaired/Phosphorylated, single A-tailed, and Adapter Ligated using custom reagent formulations (New England BioLabs, E6000B-10) and in-house prepared Illumina forked small adapter.  20% PEG, 1M NaCl Sera-Mag bead solution was used to purify the template in-between each of the enzymatic steps.  To complete the process of generating strand directionality, adapter-ligated template was digested with 5U of AmpErase Uracil N-Glycosylase (Life Technologies, N8080096).  Libraries were then PCR amplified and indexed using Phusion Hot Start II High Fidelity Polymerase (Thermo Scientific, F 549-L).  An equal molar pool of each library was sequenced on HiSeq2500 (Illumina) using PE75 chemistry. 2.10 Immunohistochemistry of 5hmC Untreated and vitamin C-treated (0.345mM; 15hr) cells were smeared onto a slide and air dried before fixation in 4% paraformaldehyde for 15 minutes. Cells were permeabilized with 0.5% Triton X100 for 15 minutes. Denaturation was conducted using 4M HCl followed by neutralization with 100 mM tris-15  HCl pH 8.5. 3% BSA-PBS was used to block the cells for 1 hour before overnight application of the primary antibody for 5-hydroxymethylcytosine (1:1000, Active Motif, 39769). Alexa Fluor 568-conjugated goat anti-mouse (1:1000, Life Technologies) secondary antibody was added for 2 hours. The slide was then mounted with DAPI (ProLong Diamond Antifade Mountant with DAPI, Molecular Probes) before visualization and imaging with a fluorescence microscope mounted with a CoolSNAP HQ digital camera (Roper Scientific Tucson). Normalization of image signal was done using the Slidebook software package (Intelligent Imaging Innovations Inc). ImageJ was used to measure the intensity of signals in the images. DAPI images were used to determine locations of cells by thresholding DAPI signal against its background. Particles were size selected using a pixel range (size 5 – 100 pixels). The 5hmC image was then used to measure the integrated density of signal within each of these cell particles. 2.11 Proliferation assay Cells were plated at a density of 2 x 105 cells/mL in triplicates in a flat-bottom 24-well plate and incubated at 37°C under light-protective conditions. Vitamin C (L-ascorbic acid 2-phosphate sesqui-magnesium salt hydrate) was added to culture medium daily at a concentration of 100µg/mL (0.345mM). Live cells were counted every four days with trypan blue, and cells were reseeded back at a density of 2 x 105 cells/mL in new flat-bottom 24-well plates. 2.12 Lineage staining - flow cytometry Cells were plated at a density of 2 x 105 cells/mL and treated with and without 0.345mM vitamin C every day for 20 days. 20000 cells were extracted and used for lineage staining with monoclonal antibodies Gr1-FITC (BD Pharmingen 553127; clone RB6-8C5), CD11b-APC (Biolegend 101226; clone M1/70), Sca-1-Brilliant Violet (Biolegend 108129; clone D7), and c-Kit-PE (eBioscience 12-1172-81; clone ACK2). FSC and SSC were used to separate live and dead cells, and a HOXA9 immortalized bone marrow cell line was used to determine positive signal relative to unstained controls. Two distinct populations 16  emerged within the Mac1+ cells, and so a Mac1hi and a Mac1lo population were defined. Analyses were conducted using FlowJo. 2.13 Wright-Giemsa stain 2 x 105 cells treated with daily doses of 0.345mM vitamin C for 15d were extracted and cytospun for 5 minutes onto glass slides and briefly air dried at room temperature. Wright-Giemsa stain was conducted using the Siemens Healthineers Hema-Tek 2000 slide stainer (ThermoFisher). Images were taken at 40x magnification. 2.14 Competition assay Ratios of 1:1 (50% TET2KO or DKO vs. 50% R132H) or 1:3 (25% TET2KO or DKO vs 75% R132H) cells in a total of 1 000 000 cells were seeded at 5mL and maintained at 37°C. Cells were sampled for amplicon library generation of the TET2 cut-site at 6d and 10d (n = 3). Proportions of amplicons that remained wildtype were measured.  2.15 Bioinformatics analyses 2.15.1 Differential methylation Methylation was determined by partitioning the genome into regions with CpGs (±25bp) and regions devoid of CpGs (±500bp) and measuring the meDIP-seq signal at each of these regions. A custom script was used to generate an empirical cumulative distribution function of CpG regions against CpG-empty regions. A fractional methylation value was then assigned for each CpG by using the CpG-empty regions as a background from which mC signal can be measured at CpG sites.  Differentially methylated regions were called by comparing the fractional methylation signals at each CpG between two libraries. A minimum fractional methylation of 0.75 and a minimum difference of 0.6 in fractional calls were used as a cut-off to determine differentially methylated CpGs. The differentially methylated CpGs were then stitched together dynamically within a 300bp window to form 17  differentially methylated regions, with a minimum of 4 CpGs within each region. The tool GREAT was used to identify functional pathways of genes within 20kb of IDH1R132H hypermethylated regions86. 2.15.2 Differential hydroxymethylation call MACS2 was used to call peaks (q < 1 x 10-5) for each hmeDIP-seq library. The Deeptools 2.5.0.1 suite was used to generate normalized bigwig files by partitioning the genome into 20bp bins. Differentially hydroxymethylated regions were called by generating a union peak file between two libraries and subsequent normalized hmC signal was used to quantitate the density of each peak. Peaks that had a coverage fold change of 2 and a minimum coverage of 20 was used as a threshold to select for differentially hydroxymethylated regions. The program HOMER was used to identify potential transcription factors whose DNA-binding motifs may be enriched in the underlying differentially hydroxymethylated regions87. 2.15.3 RNA-seq Sequence reads were aligned to a transcriptome reference generated by JAGuar using mm10v38 reference genome annotated with read-length-specific exon-exon junction sequences88. To quantify gene expression, reads per kilobase per million mapped reads (RPKM) were calculated. Both protein-coding and non-coding RNA were included within the analyses. Differentially expressed genes were called using a custom in-house matlab tool (DEFine, FDR cutoff = 0.05; RPKM > 0.01), which performs pair-wise comparisons by normalizing expression using GC content, followed by subsequent binning of similarly expressed genes together to determine the genes that varied the greatest within each bin. Gene ontology enrichment terms were obtained and compared using the tool Metascape89. Putative transcription factors were called from a list of enriched genes using the tool ChEA3, which leverages the Fisher’s exact test and curated ChIP-seq and RNA-seq databases to generate a weighted score90.   18  2.15.4 Regional enrichment calls Enriched genomic features that were overrepresented in the population of differentially methylated/hydroxymethylated regions were selected by isolating for individual CpGs within these region sets. The proportion of CpGs that were enriched in a particular genomic feature within the region set is compared to the proportion of CpGs that occur within that particular genomic feature across the genome, representing the ratio of proportion of CpGs observed in the region set against the proportion of CpGs expected at a given feature. A fold enrichment of 2 or higher was considered to be highly enriched. 2.15.5 Additional datasets used A previous generated meDIP-seq library for the HOXA9 IDH1wt cell line and previously generated ChIP-seq libraries of transcription factors PU.1 and RUNX1 and histone marks H3K27ac, H3K4me1, H3K4me3, H3K36me3, H3K9me3, H3K27me3 for HOXA9 IDH1R132H cells treated with or without 72 hours of vitamin C were used1. Transcription factor binding sites were defined by using MACS2 to call peaks (q < 1 x 10-5) relative to an input control. The union of peaks in untreated or vitamin C-treated cells were used to define a set of PU.1 or RUNX1 binding sites. Histone peaks were defined by taking the intersection of two peak-callers MACS2 (q < 1 x 105) and FindER (http://www.epigenomes.ca/tools-and-software/finder) (q < 0.05) to define broad peaks (for H3K4me1, H3K36me3, H3K27me3, H3K9me3) and narrow peaks (for H3K27ac, H3K4me3). We defined active enhancers as regions that had both H3K4me1 and H3K27ac in either the untreated or vitamin C-treated context.   19  Chapter 3: Generation of TET2, TET3, TET2/TET3 homozygous deletions in HOXA9 IDH1R132H cells 3.1 Optimization of the CRISPR-cas9 electroporation in HOXA9 IDH1R132H cells To understand the specific roles of TET2 and TET3 in the context of the IDH1R132H leukemic background, I employed a ribonucleoprotein transduction method to generate bulk CRISPR-KO models in the HOXA9 IDH1R132H cell line (hereafter referred as R132H) that harbour homozygous deletions of TET2 (hereafter TET2KO), TET3 (hereafter TET3KO), and TET2/TET3 (hereafter DKO) (Figure 4).   Figure 4. A schematic of cell lines used in the study and their generation. Table 1. crRNA used for targeting of TET2 and TET3.   20  I first used a pre-designed CRISPR RNA (crRNA) that targeted TET2 to assay the efficacy and efficiency of bulk inactivation (Table 1)91. A 73% reduction in TET2 was observed after transduction by western blot and quantification with ImageJ, which was corroborated by an observed functional loss of TET2 activity after stimulation by 15 hours of vitamin C (Figure 5a and b). To determine the efficacy of the cas9-induced edits, 24 clones were selected and expanded using methylcellulose and screened via amplicon libraries to confirm that cas9 does induce homozygous loss of TET2 predominantly (23/24) in the population through this method (Figure 5c). The bulk population was taken forward for further analyses to avoid any selection biases toward specific clones. Transduction efficiency was optimized using a trans-activating crRNA (tracrRNA) tagged with a fluorescent marker (ATTO550) assayed across different cell densities to reach a consensus of 200 000 cells, with greater than 95% of cells transduced (Figure 5d).    21    Figure 5. (A) Western Blot of TET2 in the TET2KO (KO) cell line relative to the R132H (WT) parental line. Jurkat and U937 are positive and negative controls for TET2, respectively. (B) Immunofluorescence of 5hmC in the TETwt (HOXA9 IDH1R132H) and TET2KO (HOXA9 IDH1R132H + TET2-crRNA) after 15 hours of vitamin C treatment. (p < 0.001). (C) Proportion of cells in populations derived from CRISPR-treated singly cloned cells (labelled A01-06, B01-06, C01-06, D01-06) that has non-homologous end-joining events at the CRISPR cut-site as screened by amplicon libraries. (n = 24) (D) (left) Boxplot showing the distribution of tracrRNA-ATTO550 signal within each cell at three different cell densities used in electroporation. (right) Bar plot showing the proportion of cells at three different densities that had a successful transduction, using the mean + 1sd in the untreated as a threshold to measure successful transduction. 22  3.2 Generation and screening of TET2, TET3, TET2/TET3 HOXA9 IDH1R132H knock-out  I next proceeded to design crRNAs that targeted regions common to all isoforms of TET2 and TET3 while reducing the potential for off-target effects using the IDT design platform (Figure 6a; Table 1). Successful cas9 excision was confirmed using PCR amplification of the CRISPR cut-site, followed by a T7 endonuclease I (T7EI) digestion assay (Figure 6b). To quantify the proportion of cells in the population with a successful edit, amplicon libraries were generated and sequenced for cells at four different passages, revealing a knock-out efficiency of greater than 90% across all cell lines (Figure 6c). A directional gain in the proportion of the population that contained a TET2 out-of-frame mutation was observed over time, perhaps through a selective advantage process that the TET2KO confers in the population, even within the IDH1R132H background. This trend was also observed in an independent TET2KO biological replicate (Figures 7a and b). However, when equal proportions of HOXA9 IDH1R132H cells were mixed with DKO or TET2KO, no significant difference between the proportions was observed over a period of 10 days (Figure 7c). Notably, disruption of TET2, TET3, or both TET2/TET3 did not result in any measurable changes in the proliferative capacity of the cell lines. I employed 2-phosphate L-ascorbic acid (hereafter referred to as vitamin C) as a tool to examine TET activity in our knockouts by conducting immunohistochemistry of 5hmC on cells treated with vitamin C for 15 hours. A reduced gain in 5hmC was observed in the TET3KO relative to R132H after 15 hours of vitamin C treatment whereas no significant gains in 5hmC were observed in TET2KO or DKO (Figure 6d and e). 23    Figure 6. (A) Cas9-guided cut-site (in red) targeting TET2 and TET3. (B) Representative Agilent profile of T7EI assay of TET3KO versus untreated R132H sample. The blue line represents the profile before T7EI treatment, and the red line represents the profile after T7EI treatment. The shift from one peak to two peaks represents successful T7EI digestion. (C) The proportion of cells in the population that contained a non-homologous end joining event (perc_NHEJ) at the CRISPR cut-site for TET2 (left) and TET3 (right) across multiple passages. (D) The distribution of 5hmC signal within each R132H, TET2KO, TET3KO, and DKO treated with and without vitamin C. 24    (E) Immunofluorescent images taken of cells treated with and without vitamin C for 15hr. (Blue DAPI; Red 5hmC; 10x magnification).    25    Figure 7. (A) The proportion of cells in the population that contained a non-homologous end joining (NHEJ) event within the TET2 gene over passages (P3, P6, P8) (n=2). Two TET2wt replicates were used as controls for comparison. (B) Representative output from the amplicon libraries of TET2KO cells at the CRISPR cut-site across multiple passages (P3, P6, P8), showing the diversity of indels within the amplicon library, sorted by frequency. (C) Competition assay shows no competitive advantage over time within the DKO (top) and TET2KO (bottom) when seeded at 1:3 (left) and 1:1 (right) ratio to R132H.  26  Chapter 4: Role of TET2 and TET3 in the context of IDH1R132H neomorphic mutations in AML 4.1 TET2 and TET3 share overlapping functions in maintaining a myeloid expression signature I next examined the DNA methylation and transcriptional changes associated with TET2 and TET3 knockout in our model. To that end, we treated the R132H, TET2KO, TET3KO, and DKO lines with 0.345mM vitamin C daily for 72 hours, and we sampled cells at 15 hours and 72 hours after initial treatment to extract RNA and DNA for sequencing of total RNA and 5hmC- and 5mC- immunoprecipitated DNA (Figure 8).  Figure 8. A schematic of the experimental design. R132H, TET2KO, TET3KO, and DKO cell lines were treated with 0.345mM vitC or untreated every 24 hours (black dot) for up to 72 hours. Cells were taken at 15 hours and 72 hours (white arrows) to extract DNA and RNA for sequencing of 5mC- and 5hmC- DNA immunoprecipitated libraries and total RNA.  To determine how TET2/3 disruption affects gene expression in the IDH1R132H leukemic background, I used an in-house script to compare expression patterns of TET2KO, TET3KO, and DKO to that of R132H across two replicates (at 15hr and 72hr). To identify a common expression signature, I only considered differentially expressed genes that were consistent at both time points (Figure 9a). TET2KO, TET3KO, and DKO had 422, 378, and 376 consistently differentially expressed genes (FDR < 27  0.05) relative to R132H across two replicates (Figure 9b). The majority of these genes were downregulated and over 66% of the downregulated genes were changed by more than twofold, as opposed to only 15% in the upregulated gene-sets (Figure 9c). Differentially expressed genes showed high overlap across the TET2KO and TET3KO lines, perhaps reflective of functional redundancies in gene activation between TET2 and TET3 (Figure 9d). A set of 136 genes were consistently downregulated upon homozygous loss of either TET2 or TET3 (Figure 9e). These genes were highly enriched in gene ontology terms related to cytokine production, myeloid leukocyte differentiation, and leukocyte migration and included genes linked to myeloid activation and differentiation including Mpo, Prtn3, Elane, Gfi1b, Irf8, and Csf1r (Figure 9f and g). Mpo (myeloperoxidase) and Elane (neutrophil elastase) encode granule proteins that are highly expressed in promyelocytes and neutrophils, and were observed to be downregulated by 20-fold from the homozygous loss of one of Tet2 or Tet3. This was also observed in Prtn3 (proteinase 3), another myelocyte-associated protease whose downregulation has been associated with doxorubicin resistance in AML92. Similarly, Gfi1b (growth factor independence 1b) and Irf8 (interferon regulatory factor 8) are key transcription factors that regulate myeloid differentiation and have been identified as key tumour suppressors in MDS and AML93,94. Csf1r (macrophage-colony-stimulating factor-1 receptor) encodes for a cell surface receptor that gradually increases in expression throughout differentiation into mature macrophages95. In contrast, the set of commonly upregulated genes did not show significant enrichment for any specific pathways. Considering that IDH1R132H is thought to inhibit both TET2 and TET3 activity, I predicted that these downregulated genes would show increased expression in a setting without suppression of TET2 and TET3 activity. We performed RNA-seq on a murine bone marrow-derived immortalized cell line overexpressing Hoxa9 and harbouring an IDH1wt vector (IDH1WT), which is non-leukemic8. Strikingly, I observed 50.7% (69/136) of the TET2/3-dependent genes to be expressed at a level that is on average 20-fold greater in a background of IDH1wt than in IDH1R132H (Figure 9h). This was exemplified in the 28  expression levels of Elane, whose expression was reduced 60-fold in the presence of IDH1R132H, and was further down-regulated by an additional 20-fold upon disruption of either TET2 or TET3 (Figure 9I). Taken together, these results suggest IDH1R132H-mediated repression of TET2 and TET3 is incomplete and that they play cooperating roles in supporting myeloid differentiation.    Figure 9. (A) Table indicating the concordance between the numbers of differentially expressed genes that are called between the knockouts versus the parental line at two different time points. (B) Number of consistently differentially expressed genes between TET2KO, TET3KO, and DKO relative to R132H (n = 2). (C) Boxplots showing fold change of differentially expressed genes in TET2KO, TET3KO, and DKO relative to R132H. (D) Overlap between upregulated (denoted as up) and downregulated (denoted as dn) genes in TET2KO, TET3KO, and DKO. 29     (E) Heatmap (z-score) of 136 genes that were consistently downregulated upon loss of either TET2 or TET3. (F) Pathway analyses of the TET2/3-dependent set of genes. (G) Expression of six genes related to myeloid activation that become downregulated upon loss of either TETs at 15hr and 72hr. (H) Boxplot showing expression levels (rpkm) of 69/136 TET2/3-dependent genes in IDH1WT, R132H, TET2KO, TET3KO, and DKO. Scale has been log-transformed. *adj.p < 0.0001 by ANOVA. (I) Table showing expression levels of Elane (rpkm) within every cell.  30  I next aimed to identify genes that were specifically dependent on TET2 given its selective inactivation in AML. For this I identified genes that were down-regulated upon loss of TET2 in comparison to TET3KO (n = 12). Within the gene-set, I identified Cebpe, a critical transcription factor that increases in expression and activity leading to terminal granulocytic differentiation and Ass1, the gene encoding the synthetase for arginine, whose downregulation has been shown to confer a proliferative advantage in leukemia and in solid tumours (Figure 10a)96–98. Notably, I observed a specific TET2-dependent up-regulation of Cdkn1a, the gene encoding cyclin-dependent kinase inhibitor p21, which has been shown to inhibit apoptosis and Ptp4a3, which encodes protein-tyrosine phosphatase of regenerating liver 3 (PRL-3) implicated in leukemogenesis of various leukemias (Figure 10b)99–102. In contrast, no genes were found that were distinctly TET3-dependent. Taken together, this specific TET2-dependent expression pattern may provide the slight, but sufficient selective advantage that allows TET2 mutants to thrive in a population but not TET3 mutants.  Figure 10. (A) Expression levels (rpkm) of genes that are significantly down-regulated (FDR < 0.05) upon loss of TET2 relative to TET3KO (n = 12). (B) Expression levels (rpkm) of genes that become up-regulated only in the context of TET2-loss (n = 2). 31  4.2 TET2 and TET3 drive 5hmC gain at enhancers even in the presence of IDH1R132H We next performed 5hmC- and 5mC- DNA immunoprecipitation sequencing (hmeDIP-seq and meDIP-seq) on R132H, TET2KO, TET3KO, and DKO cells to assess the effects of TET disruption on 5hmC and 5mC changes at local genomic regions. I predicted that the R-2HG inhibition of TET2 and TET3 would be incomplete and that a knock-out of TET2 and TET3 in this context would result in detectable changes in 5mC or 5hmC.  To examine the changes in 5hmC and 5mC following TET2 and TET3 knockout, I took the union of all differentially methylated regions and the union of all differentially hydroxymethylated regions, and calculated the 5mC and 5hmC signal within each region set, respectively. I observed a global drop in methylation and hydroxymethylation upon loss of either TET2 or TET3 (Figure 11a). Variance in 5hmC levels was noted between the 15 hour and 72 hour replicates, which may reflect the dynamics of the 5hmC modification. I therefore calculated the fold change of 5hmC density of TET2KO, TET3KO, and DKO relative to R132H within the same timepoint. I used the 136 TET2/3-dependent genes that are downregulated upon loss of either TET2 or TET3, and measured the average change in 5mC and 5hmC signal at their promoters (TSS ± 500bp) and the nearest 5hmC-marked active enhancer, defined as genomic regions that have been annotated by H3K27ac and H3K4me1 chromatin immunoprecipitation (ChIP-seq) data from untreated R132H cells. I observed significantly lower 5hmC (adj-p < 0.01) density at the nearest 5hmC enhancer in TET2KO, TET3KO, and DKO relative to their signal changes at their promoters (Figure 11b). In contrast, no significant differences in 5hmC and 5mC changes were observed between the promoters of these genes and changes at all promoters in TET2KO and TET3KO. This would suggest that changes in 5hmC levels in enhancers rather than gene promoters are likely to be associated with loss of gene expression in the TET2KO, TET3KO, and DKO cell lines. 32     Figure 11. (A) Boxplot showing the distribution of 5mC and 5hmC signal of each sample across the total set of differentially methylated regions (left) and differentially hydroxymethylated regions (right). (B) Boxplot showing the extent of 5hmC (left) and 5mC (right) changes at all promoters, promoters of the 136 TET2/3-dependent genes (i.e. DN(23D) promoters), and nearest active enhancer of these genes  marked with 5hmC in the R132H line (i.e. DN(23D) nearest enhancer). *adj-p < 0.01 by ANOVA and Tukey post-hoc test. (C) Boxplot showing the extent of 5hmC changes at DN(23D) nearest enhancers. ****adj-p < 0.0001 by ANOVA and Tukey post-hoc test. (D) Genome browser shot of 5hmC (top) and 5mC (bottom) signal at the promoter (right highlight) of Pdcd4 and at its nearest 5hmC-marked enhancer (left highlight).   33  Notably, the reduction in 5hmC density at enhancers in the DKO line was significantly greater (adj-p < 0.0001) than that of TET2KO or TET3KO, suggesting that TET2/3 share some functional redundancies in the maintenance of methylation levels at active enhancers (Figure 11c). This decrease in 5hmC signal was exemplified at an enhancer near the Pdcd4 locus where an attenuated 5hmC density in the TET2KO and TET3KO lines is completely depleted in the DKO context (Figure 11d). Taken together, these results are consistent with the model that R-2HG is incompletely suppressing TET2 and TET3 activity.   4.3 IDH1R132H drives hypermethylation at CpG islands and at active enhancer elements The TET family is actively inhibited by the presence of R-2HG, an oncometabolite produced by IDH1R132H 103. To study the roles of TET2 and TET3 in the presence of the IDH1R132H, I identified differentially methylated regions between R132H, TET2KO, TET3KO, and DKO lines compared to a matched cell line expressing IDH1wt (IDH1WT) (Figure 4a)1. The comparisons revealed a global hypomethylation across all cell lines relative to IDH1WT (Figure 12a). Examining regions that were hypermethylated in the presence of IDH1R132H, I observed an enrichment in CGIs and active enhancers (adj.p < 1x10-16, Fisher’s exact test) but not in the hypomethylated regions (Figure 12b). Consistent with this, I observed a greater than two-fold enrichment in the proportion of hypermethylated CpGs at enhancers and CGIs within the total set of hypermethylated regions (Figure 12c). In contrast, this enrichment was not observed in the set of hypomethylated regions. I next generated the intersection of hypermethylated regions across all IDH1R132H samples to define a consensus set of 2391 CGI promoters and 729 enhancers that were consistently hypermethylated across all cell lines harbouring IDH1R132H relative to IDH1WT. This included the alternate promoter of macrophage colony stimulating factor 1 (Csf1), a cytokine that is expressed during macrophage differentiation (Figure 12d).  34    Figure 12. (A) Genomic coverage of global changes in methylation in R132H, TET2KO, TET3KO, and DKO with respect to IDH1wt. (B) Proportion of hypermethylated (top) and hypomethylated (bottom) regions that overlap with CGI promoters, CGI genebodies, CGIs at intergenic regions, enhancers, and other genomic regions. (C) Bar plot showing regional enrichment of hypermethylated (black) and hypomethylated (grey) regions at various genomic features. Regions that exceeded a two-fold enrichment level (line) were deemed to be enriched. (D) Genome browser shot of Csf1 gene. The highlighted region represents a consistently hypermethylated promoter that can be attributed to the IDH1R132H background. (E) Top three pathway enrichment analysis results for IDH1R132H-induced hypermethylated enhancers (left) and hypermethylated CpG island promoters (right). 35  I validated this set of IDH1R132H responsive features in a replicate of R132H and an independent TET2KO line and obtained a high degree of overlap across replicates, wherein 95.5% (2250/2391 R132H, 2327/2391 TET2KO) CGI promoters and 94.1% (684/729 R132H, 688/729 TET2KO) enhancers shared this hypermethylation phenotype. Functional enrichment analysis of the genes within 20kb of these hypermethylated enhancers revealed a significant (Binomial q < 1x10-12) enrichment for cell death, positive regulation of immune system process, and leukocyte differentiation pathways (Figure 12e). In contrast, hypermethylated CGI promoters were associated with terms involving neuron differentiation and development. This latter finding is consistent with a CpG island hypermethylation signature observed in low grade glioma and primary AML genomes which share the IDH1R132H allele48,54. Thus my analysis supports the well documented CGI methylation phenotype associated with IDH1R132H and our previous reports of its additional role in driving hypermethylation at active enhancers1.  4.4 Vitamin C-induced demethylation of active enhancers is both TET2 and TET3 dependent We previously showed that vitamin C treatment leads to demethylation of hypermethylated enhancers and CGIs in the presence of IDH1R132H 1. To examine the roles of TET2 and TET3 in the demethylation of aberrantly methylated enhancers and CGI promoters, I employed our TET knockout models as a tool to parse out the individual and combined effects of TET2 and TET3 in response to vitamin C treatment. Consistent with our previous observations vitamin C treatment in our R132H model led to the increase of 5hmC at both active enhancers and CGIs. Following 15hrs of vitamin C treatment, 51% (375/729) of the IDH1R132H driven hypermethylated enhancers were hydroxymethylated in R132H. In our TET2KO model, this number dropped to 20% (145/729) supporting our previous work demonstrating TET2-dependence of vitamin C treatment1. In the TET3KO model, 41% (300/729) of hypermethylated enhancers were hydroxymethylated suggesting TET3 is also capable of responding to 36  vitamin C treatment at enhancers in our model. The DKO model showed an additive effect with only 7% (52/729) of enhancers gaining hydroxymethylation, supporting a model of both overlapping and unique enhancer targets for TET2 and TET3. 72hrs of vitamin C treatment drove further gains in hydroxymethylation across all models and these were matched with a concurrent drop in 5mC signal at 72 hours in a TET2- and TET3- dependent manner (Figure 13a). Conversely, only 55, 13, 31, and 1 out of 2391 hypermethylated CGI promoters for R132H, TET2KO, TET3KO, and DKO showed any gains in hydroxymethylation at 72 hours and no distinct changes in methylation were observed at the 15 and 72hr time points (Figure 13b). Using vitamin C to stimulate TET activity, I therefore observed that both TET2 and TET3 shared a greater affinity for targeting hypermethylated enhancers compared to CpG islands in the context of mutant IDH1. Notably, in support of the model in which TET is incompletely inhibited by R-2HG, the basal level of 5hmC observed in the untreated R132H model is attenuated in TET2KO and TET3KO and depleted in the DKO.    37    Figure 13. Heatmap showing regional changes in 5mC (top) and 5hmC (bottom) of untreated (grey) and vitamin C treated (orange) cells after 15hr and 72hr at (A) enhancer regions and at (B) CpG island promoters that are hypermethylated by IDH1R132H.   38  To determine aberrantly methylated enhancers that were likely targeted by both TET2 and TET3, I identified a common set of 408/729 enhancers that gained 5hmC in R132H, TET2KO, and TET3KO following 72hrs of vitamin C treatment. This common set of enhancers showed gains in 5hmC corresponding to the relative level of TET activity in the model, with R132H showing the greatest increase, followed by attenuated increases in TET3KO, then TET2KO, and a near depletion in the DKO (Figure 14a). Given the selective inactivation of TET2 in AML, this is consistent with the model that TET2 has a greater or more efficient role in maintaining the methylation homeostasis at enhancers. Interestingly, by 72 hours of treatment, the level of 5hmC signal in the TET2KO increased to a similar level as the TET3KO, which would suggest that vitamin C is also acting upon TET3 to target a common set of enhancers. In support of this, this degree of 5hmC gain was not observed in the DKO in which TET3 was also inactivated. This common set of enhancers was localized within 20kb of genes that were significantly (Binomial -logp > 2.5) enriched in pathways including homeostasis of number of cells, leukocyte differentiation, and regulation of leukocyte migration (Figure 14b). Restricting the regions to a set of 136/729 enhancers that gained 5hmC in R132H, TET2KO, and TET3KO following 15hrs of vitamin C treatment, I again observed a significant enrichment (Binomial -logp > 2.5) for leukocyte migration and leukocyte differentiation, suggesting that changes in methylation states at these enhancers are an early event of TET2 and TET3 activation (Figure 14c). This TET2/TET3-dependent gain in 5hmC was exemplified at an IDH1R132H hypermethylated enhancer located within the gene body of Notch2, a gene enriched in the leukocyte differentiation gene-set, where TET2- and TET3- dependent increases in 5hmC were observed at 15hrs of vitamin C treatment (Figure 14d). After 72hrs of vitamin C treatment, this gain in 5hmC was amplified in both the TET2KO and the TET3KO models to a similar level as the parental R132H model, suggesting that both TET2 and TET3 can be activated independently to oxidize this hypermethylated enhancer.  39  To validate that vitamin C-induced demethylation of hypermethylated enhancers can be reconstituted in part through a TET2-independent process, I used an independent biological replicate of the HOXA9 IDH1R132H TET2KO, and concordantly, I observed a concurrent increase in 5hmC and decrease in 5mC at this set of 729 enhancers following 15hrs of vitamin C treatment using meDIP- and hmeDIP- seq in R132H and to a lesser extent in the TET2KO (Figure 14e). Relative to the set of 136 enhancers that gained 5hmC in R132H, TET2KO, and TET3KO at 15hrs, 71% (97/136) of these enhancers also show 5hmC gains after 15hrs of vitamin C treatment in the TET2KO replicate, including at the enhancer within the gene body of Notch2 (Figure 14f). Taken together, this suggests that the active demethylation signature caused by vitamin C-induction of TET2 can also be activated by vitamin C-induction of TET3.      40     Figure 14. (A) Box-plot showing 5hmC density of R132H, TET2KO, TET3KO, and DKO at the set of 408 IDH1R132H hypermethylated enhancers with and without 15hrs (left) or 72hrs (right) of vitamin C treatment. (B) Top 10 gene ontology results using all genes found within 20kb of the set of 408 enhancers that gain 5hmC after 72hrs of vitamin C treatment in R132H, TET2KO, and TET3KO. (C) Top 10 gene ontology results using all genes found within 20kb of the set of 136 enhancers that gain 5hmC after 15hrs of vitamin C treatment in R132H, TET2KO, and TET3KO. 41     (D) Genome browser screenshot highlighting (in blue) an IDH1R132H-hypermethylated enhancer within the gene-body of Notch2 that becomes marked with 5hmC in R132H, TET2KO, and TET3KO after 15 and 72hrs of vitamin C treatment. (5hmC scale was set to 0-678 rpkm, and 5mC scale was set to 0-187 rpkm). (E) Average normalized 5hmC (top) and 5mC (bottom) signal of R132H and TET2KO (replicate) after 15 hours of vitamin C at the set of enhancers found to by hypermethylated by IDH1R132H. (F) Genome browser screenshot of hmeDIP-seq libraries of an independent biological replicate of R132H and TET2KO with 15hrs of vitamin C treatment at the same enhancer within the gene-body of Notch2 (as in Figure 14d).      42  Chapter 5: TET2 and TET3 cooperate to drive epigenetic reprogramming in response to vitamin C 5.1 Vitamin C drives myeloid differentiation via TET2/3-dependent mechanisms Our results suggest that TET3 can partially compensate for TET2 loss and induce 5hmC at methylated enhancers in response to vitamin C. To examine the phenotypic changes associated with long term daily 0.345mM vitamin C treatment in our models, I cultured 100 000 cells in in U-shaped 96-well plates over 264 hours. Distinct morphological changes were apparent in the R132H, TET2KO, and TET3KO lines after 264 hours of treatment, but not in DKO line, suggesting a morphological change in the culture that is driven by TET2 and TET3 (Figure 15a). In support of this dual dependence, daily doses of 0.345mM vitamin C induced a significant decline in the proliferative capacity of R132H, TET2KO, and TET3KO models by 16 days of vitamin C treatment, which remained consistent up until 20 days (Figure 15b). In contrast, the proliferative capacity of the DKO model was unaffected by long term vitamin C treatment. This result was replicated in two independent trials (Figure 15b (1-3)). Consistent with my molecular analysis demonstrating an increased dependence on TET2 for 5hmC gains, the latency of vitamin C response varied across our models with the shortest at 8 days in R132H, followed by TET3KO at 12 days, and finally TET2KO at 16 days. These differences support a model where TET2 and TET3 have different functional efficiencies, but share a majority of targets. 43   Figure 15. (A) Pictures taken of 100 000 cells in a U-shaped 96 well plate at 96 hours and 264 hours of vitamin C treatment. (4x magnification). (B1-3) Bar plots showing long term proliferation assay of R132H, TET2KO, TET3KO, and DKO cell lines treated with daily 0.345mM vitamin C, wherein cells were quantified and split every four days into new plates (n = 3) (*adj-p < 0.0001 in untreated vs. vitamin C-treated cells by ANOVA followed by Tukey post-hoc tests).  To examine if vitamin C-induced loss of proliferation was associated with cellular differentiation, Wright-Giemsa staining was conducted following 15 days of vitamin C treatment. Vitamin C treatment of R132H, TET3KO, and to a lesser degree TET2KO models exhibited nuclei that resembled that of neutrophils (Figure 16a). In contrast, the DKO model showed no evidence of morphological change. Vitamin C-induced cellular differentiation phenotype was supported by flow cytometry of cell surface markers following 20 days of vitamin C treatment (Figures 16c and d). FACS analyses revealed a 44  significant increase (adj-p < 0.0001 by ANOVA) in the proportion of cells that expressed differentiated myeloid markers Gr-1+ and Mac-1hi (Figure 16b). Concurrently, there was also a marked decrease in cells that expressed markers representing a more progenitor state (Mac-1lo/Gr-1-/c-Kit+) in R132H, TET2KO, and TET3KO relative to the DKO. A small effect was observed in the DKO, which may reflect a TET2/3-independent effect of vitamin C that promotes differentiation. Taken together, this analysis supports our previously reported model of vitamin C-induced differentiation in a HOXA9 IDH1R132H leukemic model and demonstrates that both TET2 and TET3 are responsive to high dose vitamin C in this context.    Figure 16. (A) Wright-Giemsa stain of cells taken after 15d of vitamin C treatment. Black arrowheads denote differentiated cells.  45    (B) Proportion of cells positive for marks representing a more differentiated phenotype (Gr1+ and Mac1hi) and a more progenitor phenotype (Mac1lo/Gr1-/c-Kit+) after 20d of vitamin C treatment. (C) Gating strategy used to select for viable cells using FSC-A. A positive control (HOXA9 immortalized cell line) (top) was used to generate gates for Mac1, Gr1, and c-kit relative to an unstained control (bottom). 46    (D) Representative profiles of two distinct populations were observed in the Mac1+ population when comparing vitamin C-treated cells (bottom) to untreated (top). Gates were generated for Mac1hi and Mac1lo based on the highest quantifiable (105) measurement.   5.2 Temporal TET2 and TET3 dependence on vitamin C-induced epigenomic reprogramming To examine the TET-dependence of vitamin C on the hydroxymethylome, I examined the total set of differential changes in 5hmC after vitamin C treatment at 15 and 72 hours. In embryonic stem cells, we previously reported a transient hydroxymethylation gain at 12 hours followed by a return to baseline at 72 hours75. In contrast, in the IDH1R132H model, levels of hydroxymethylation continued to increase out to 72 hours. Following 15 hours of vitamin C treatment, a directional global gain of 5hmC was observed in R132H (11Mb), TET3KO (9.1Mb) and TET2KO (5.8Mb), but not in DKO (Figure 17a). The level of 5hmC occupancy for all cell lines expanded by 72 hours, increasing to 31Mb in R132H, 24Mb in 47  TET3KO, and 18Mb in TET2KO. I observed a directional gain in 5hmC (11Mb) in the DKO, perhaps reflecting residual activity of the lowly expressed TET1. There was a high degree of overlap between genomic regions that gained 5hmC at 15 and 72hrs and CpGs that gained 5hmC were significantly enriched (Fisher’s exact test; q < 1x10-6) within active enhancers compared to random expectation (Figure 17b and c). In contrast, regions that lost 5hmC showed no significant enrichment. A significant (Fisher’s exact test; p < 1x10-6) proportion of 5hmC-marked enhancers overlapped between single knockouts of TET2 or TET3 at 15 hours (42%) and at 72 hours (63%), suggesting that TET2 and TET3 are recruited to an overlapping set of enhancers (Figure 17d). Indeed, I observed that 87% (3405/3907) of the enhancers and 97.5% (13469/13813) of all regions that gained 5hmC via TET2 (i.e. TET3KO) at 15 hours showed 5hmC gain via TET3 (i.e. TET2KO) by 72 hours, which would suggest that TET2 and TET3 can target many of the same genomic features, a surprising finding given the significant structural differences between these enzymes. Since the parental line R132H showed the greatest degree of 5hmC gained at both time points, I grouped the enhancers into a set that showed a consistent gain (core enhancers; n = 5953) in 5hmC at 15 and 72 hours of vitamin C and a set that gained 5hmC only at 72 hours (secondary enhancers; n = 7030). I observed that enhancers that gained 5hmC at 15 hours of vitamin C showed a higher degree of 5hmC signal by 72 hours, which would suggest that an increasing proportion of cells in the population were gaining 5hmC at these enhancers (Figure 17e). Similarly, this gain in 5hmC was supplemented with a significantly progressive 5mC loss from 15hr to 72hr, a feature not observed in the DKO model (Figure 17f).  The discrepancy between trends observed in embryonic stem cells and in our IDH1R132H model may arise due to the presence of R-2HG, which continuously drives a hypermethylation phenotype via TET inhibition75. This would suggest that these enhancers are experiencing a high level of methylation turnover, and that vitamin C is not able to fully restore normal activity of TET2 and TET3 in the presence 48  of R-2HG. In support of this model, I observed a significantly (adj.p < 10-6) greater level of basal methylation at core enhancers relative to the secondary enhancers across our models (Figure 17g). Pathway analysis of unique genes within 20kb of core enhancers revealed significant enrichment for cell-cell adhesion, myeloid cell differentiation, myeloid leukocyte activation, and T cell activation terms for the set of enhancers that accumulated 5hmC at 15hr and 72hr, which suggests that these enhancers were the targets of an early response by vitamin C-induced activation of TET2 and TET3 (Figure 17h). Conversely, unique genes associated to the set of secondary enhancers that gained 5hmC only at 72hrs showed a significant enrichment for terms related to cytokine production, neutrophil degranulation, leukocyte migration, and response to external stimuli. This may suggest a model in which vitamin C induces an initial activation of TET2 and TET3 to demethylate a core set of enhancers that promote expression of genes that drive differentiation and cytokine production. The subsequent response to extracellular cytokines in conjunction with vitamin C may then trigger additional cell-type specific regions to become demethylated. Indeed, cytokine-induced TET2 activation has been annotated in dendritic cells and erythroid progenitors104,105.  Figure 17. (A) Barplot showing the total genomic occupancy of regions that gained and lost 5hmC upon 15hr (left) and 72hr (right) of vitamin C treatment. (B) Venn diagrams showing the degree of overlap between regions that gained 5hmC at 15hr and 72hr for each cell line.  49    (C) Bar plot showing regional enrichment of CpGs that gained 5hmC or lost 5hmC upon vitamin C treatment at various genomic features. Regions that exceeded a two-fold enrichment level (line) were deemed to be enriched.  (D) Venn diagrams showing the degree of overlap of enhancers that gained 5hmC for each cell line within at 15hr and 72hr. (E) Heatmap showing 5hmC coverage at enhancers that consistently gained 5hmC at 15hr and 72hr (top) and enhancers that uniquely gained 5hmC only at 72hr (bottom).  50    (F) Violin plots showing the change in 5mC (log2FC of vitamin C/untreated) at core enhancers. Dotted line represents a fold change of ±2. (G) Violin plots showing basal level of log10 5mC signal at enhancers of core and secondary enhancers.  (H) Barplot showing top 12 most significant pathway enrichment terms for unique genes within 20kb of core enhancers (left) and secondary enhancers (right). *adj.p < 0.0001 by ANOVA throughout figure.    51  5.3 Vitamin C drives a common myeloid cell differentiation expression signature in a process mediated by TET2 and TET3 To understand the underlying transcriptional changes driven by vitamin C, I examined the changes in RNA expression following 15 and 72 hours of vitamin C in our models. Consistent with the increased alterations in 5hmC levels, I observed the greatest number of differentially expressed genes at 72hr, with a majority (67%) being upregulated (Figure 18a). Models harbouring an active TET2 (namely R132H and TET3KO) showed the greatest magnitude of increase with over 40% of upregulated genes increased by over twofold, as opposed to only 28% in the TET2KO and DKO models (Figure 18b). An attenuated but consistent transcriptional response was observed at 15 hours, where less than 25% of genes showed twofold in change. The expression levels of TET and DNMT genes were not affected by vitamin C induction, consistent with our previous findings and the literature (Figure 18c)1,58,75. Surprisingly, a majority of genes that were upregulated by vitamin C did not overlap across models, with the greatest proportion of overlap at 72hrs (Figure 18d). At 72hrs post treatment, I observed a core set of genes that were commonly up-regulated across all samples (n = 77), up-regulated only in the context of TET2 (n = 74), up-regulated with either TET2 or TET3 (n = 60), and up-regulated only in the context of TET3 (n = 38) (Figure 18e). Within the gene-set that showed upregulation by 72hrs in all models except the DKO, the magnitude of upregulation showed TET2/3-dependency. Of note, within the 38 genes that were specifically up-regulated in R132H and TET2KO (i.e. TET3-dependent), only 17 were protein-coding and of those, only three had greater than two-fold change (Cav2, Gm19684, Cmtm3). The lack of a unique TET3-dependent transcriptional signature would suggest that transcriptional changes induced by vitamin C activation of TET3 can also be mediated through TET2. In contrast, I observed a distinct signature among the genes that were upregulated specifically in R132H and TET3KO (i.e. TET2-dependent). These included transcription factors Cebpb and Tfec, neutrophil granule proteins Prtn3, Hp, and Prg2, and macrophage inflammatory receptors Ifngr1, Fcgr4 (Figure 52  18f). Strikingly, I observed upregulation (8-fold in TET3KO, 11-fold in R132H) of cytokine Il1b, a key initiator of inflammatory processes, whose forced over-expression has been linked to increased apoptosis in CD34+/CD38- cells extracted from leukemic patients106.  I next compared common and unique functional pathways that were enriched within each gene-set. Upregulated genes across all models were significantly enriched in gene ontology terms including inflammatory response and cytokine production (Figure 18g). In contrast, downregulated genes showed an even lower overlap across models and were not significantly enriched in any gene ontology terms, consistent with our previous findings1. Though not to the same extent as all other models, vitamin C was able to increase the expression of cell surface markers and inflammatory genes even in the DKO context, which may suggest that vitamin C can activate expression of inflammatory response elements and myeloid activation genes in part through other mechanisms, such as stimulation of histone demethylases and prolyl hydroxylases77,107,108. Consistent with the functional data, I observed an upregulation of genes enriched in leukocyte differentiation, including Csf2ra, Gata2, Cebpe, and Tlr2. Genes that were only upregulated in the context of functional TET2 were highly enriched (-logq > 10) in terms related to granulocyte chemotaxis and migration. Interestingly, a set of genes related to an increased inflammatory phenotype observed in MDS and the toll-like receptor pathway was uniquely upregulated in R132H, including S100a8, S100a9, and Tifab (Figure 18h)109–111. Together with the observed time-dependency of vitamin C effect in the functional and 5hmC data, this may be indicative of an early subpopulation of cells that was beginning to display granulocytic expression patterns in TET3KO and R132H. 53   Figure 18. (A) Number of differentially expressed genes at 15hr and 72hr of vitamin C treatment (FDR < 0.05). (B) Violin plot showing the extent of vitamin C-induced up- and down- regulation of genes at 15hr (blue) and 72hr (orange). (C) Expression levels (rpkm) of Dnmt1, Dnmt3a, Dnmt3b, Tet1, Tet2, and Tet3 in all cell lines with or without vitamin C treatment. (D) Overlap plot showing the total numbers of differentially expressed genes that overlap between all up-regulated gene-sets. Coloured bars highlight gene-sets that are upregulated in all (orange), upregulated in R132H and TET3KO only (green), upregulated in R132H, TET2KO, TET3KO (blue), and upregulated in R132H and TET2KO only (red). 54     (E) Overlap plot only comparing the up-regulated gene-sets from the 72hr time point. (F) Barplot highlighting eight genes that were upregulated in a TET2-dependent manner after 72hr of vitamin C. (G) Heatmap identifying highly significant (-log q > 6) functional pathways that showed shared enrichment between the cell lines. White boxes represent terms that were not-enriched in that gene-set. (H) Barplot showing expression levels (rpkm) of S100a8, S100a9, and Tifab in all cell lines with or without 15hr or 72hr of vitamin C. 55  Since I observed that vitamin C-induced deposition of 5hmC had a specific enrichment at active enhancers, I compared the changes in 5hmC density at the promoters of upregulated genes and at their nearest differentially hydroxymethylated active enhancer for each model. Consistent with my observations in the untreated samples, genes that were upregulated by vitamin C showed a significantly (adj.p < 10-6) greater increase in 5hmC at their closest differentially hydroxymethylated active enhancers relative to their 5hmC changes at their promoters (Figure 19a). Despite this difference, 5hmC gains at the promoters of upregulated genes were still significantly (adj.p < 10-6) greater in R132H and TET3KO than at all promoters, which suggests that promoter hydroxymethylation may contribute to an increased level of expression in conjunction with enhancer hydroxymethylation. In contrast, this was not observed in cell lines lacking functional TET2.  This was exemplified at the promoter of Csf2ra, the gene encoding granulocyte-macrophage colony-stimulating factor receptor. Here, I observed a direct relationship between promoter gain in 5hmC and a relative gain in expression after vitamin C (Figure 19b). In contrast, at the promoter of Itgam, which encodes for CD11b/Mac1, I observed a marked increase in 5hmC signal at a nearby active enhancer, and a moderate increase in 5hmC at the promoter itself (Figure 19c). This contrast was apparent in the 72hr TET2KO pair, in which the 5hmC gained at the enhancer corresponds to a matched increase in expression, but not at the promoter. Taken together, these results indicate that widespread promoter and enhancer demethylation induced by vitamin C activation of both TET2 and TET3 promotes the up-regulation of genes involved in driving a differentiation signature.    56    Figure 19. (A) Boxplot showing extent of 5hmC signal changes (log2 fold-change) after 72hr vitamin C treatment at all promoters, promoters of up-regulated genes, and nearest differentially 5hmC-marked active enhancer. *ANOVA and Tukey test adj.p < 10-6. (B) Genome browser shot (left) of 5hmC signal at the Csf2ra promoter, and a bar-plot representing its matched expression (rpkm) (right). 57   (C) (Left) Genome browser shot of 5hmC signal at the Itgam promoter (right highlight) and its nearest differentially 5hmC-marked active enhancer (left highlight). (Right) Bar-plot representing the corresponding expression (rpkm) of Itgam.  5.4 TET2 and TET3 stimulation with vitamin C can partially rescue the deficiency of TET2/TET3 inactivation To further examine the degree of TET2 and TET3 functional redundancy, I selected the aforementioned 136 (Chapter 4.1) genes that were initially identified to be heavily down-regulated upon loss of either TET2 or TET3, and asked if vitamin C can sufficiently induce the remaining TET to rescue the expression of these genes. Among the 136 genes, I observed the emergence of two distinct groups in terms of their response to vitamin C (Figure 20a). Using the total gene-set that showed upregulation after 72hrs of vitamin C treatment, I divided this 136 TET-dependent genes into a cluster of 44 genes that exhibited TET2/3-dependent upregulation after 72 hours of vitamin C treatment, and 92 genes that remained unaffected by vitamin C treatment (Figure 20b). Genes that were upregulated via vitamin C 58  included genes encoding granule proteins Mpo and Elane, and myeloid cell surface markers Thbs1 and Cd33, which is also a prominent drug target in various leukemias (Figure 20c)112. Conversely, genes that were unchanged by vitamin C included transcription factor Irf8, and genes encoding myeloid migration and adhesion cell surface markers CD97 (Adgre5), Cd44, Ly6e and T-cell marker CD90 (Thy1)113,114. Notably, this division in gene responsiveness to vitamin C may be indicative of genes that were regulated directly by the ability of TET2/3 to demethylate regulatory regions, and genes that were regulated by other TET2/3-dependent functions. I predicted that the genes that were vitamin C-sensitive would therefore be highly upregulated in a setting where TET2 and TET3 were not inhibited (i.e. in the IDH1WT). Consistent with this notion, I observed that 70% (31/44) of the vitamin C-sensitive genes were upregulated in the IDH1WT line, as opposed to 41% (38/92) in the vitamin C-insensitive gene-set. Indeed, I observed a significantly lower expression level in IDH1R132H relative to IDH1wt within the TET-dependent vitamin C-sensitive gene-set that was not observed in the TET-dependent vitamin C-insensitive gene-set (Figure 20d). To distinguish characteristics that separate the vitamin C-sensitive and vitamin C-insensitive genes, I observed that the majority of vitamin C-insensitive genes has a high average CpG content (53 CpGs) at their promoters (TSS±500bp) compared to the promoters of vitamin C-sensitive genes (25 CpGs) (Figure 20e). In contrast, no significant difference was observed in the CpG densities of differentially hydroxymethylated enhancers within 20kb of the vitamin C-sensitive and vitamin C-insensitive gene promoters (Figure 20f). 59   Figure 20. (A) Heatmap showing expression levels (z-score) of the 136 TET2/3-dependent genes with and without vitamin C treatment at 15 and 72hr. (B) Collapsed heatmap showing the average expression (z-score) of the 136 TET2/3-dependent genes grouped into a set that was up-regulated by 72hours of vitamin C (vitC-sensitive) and a group that was unaffected by vitamin C (vitC-insensitive). (C) Representative expression levels from vitamin C-sensitive genes, cluster 1 (top) and vitamin C-insensitive genes, cluster 2 (bottom).    60     (D) Boxplots showing expression values (rpkm) of genes that are vitamin C-sensitive (n = 44; top) and vitamin C-insensitive (n = 92; bottom) in the IDHwt and IDH1R132H lines (R132H, TET2KO, TET3KO, DKO). Scale has been log-transformed. *Tukey adj.p < 10-6. (E) Boxplot showing the distribution of CpGs in the promoters (TSS ± 500bp) of vitamin C-sensitive genes and vitamin C-insensitive genes. **t.test; p < 10-6. (F) Boxplot showing the distribution of CpG density within differentially hydroxymethylated enhancer of vitamin C-sensitive genes and vitamin C-insensitive genes. 61  I then examined the levels of 5hmC at the promoters of these genes to test if promoter 5hmC content can explain differences in expression patterns observed between vitamin C-sensitive and vitamin C-insensitive genes. Promoter 5hmC was significantly increased within the vitamin C-sensitive gene-set in R132H, TET2KO, and TET3KO following 72hrs of vitamin C treatment (Figure 21a). This increase was not observed in the vitamin C-insensitive promoters, suggesting that TET2 and TET3 can up-regulate this set of vitamin C-sensitive genes in part through promoter hydroxymethylation.    Figure 21. Box-plots showing the level of 5hmC signal at the (A) promoters and (B) nearest DHR enhancers of the vitamin C-insensitive (n = 92) gene-set and the vitamin C-sensitive (n = 44) gene-set. The genes with the highest promoter-5hmC density at 72hr of vitC are labelled. *adj.p < 0.0001. 62  In contrast, changes in 5hmC at the nearest differentially hydroxymethylated enhancers of vitamin C-sensitive and vitamin C-insensitive genes showed significant gains in 5hmC in a TET2- and TET3- dependent manner in both groups, suggesting that the changes in expression I observed between the vitamin C-sensitive and vitamin C-insensitive group may be due to differences in promoter hydroxymethylation (Figure 21b). Using the tool ChIP-X Enrichment Analysis 3 (ChEA3), which leverages existing ChIP-seq and RNA-seq datasets to match transcription factors to an enriched gene-set using a metric based on the Fisher’s exact test, I identified PU.1, TFEC, and CEBPE as the top three putative transcription factors that regulate 19, 13, and 19 genes out of the 44 genes in the vitamin C-sensitive group (Table 2)90. Conversely, no significant enrichment for transcription factors that were highly expressed in any of the models was observed for the vitamin C-insensitive gene-set. Table 2. Top transcription factors as predicted by ChEA3 and associated genes from the set of vitamin C-sensitive TET2/3-dependent genes (n = 44).   This was consistent with our previous finding, in which we proposed that PU.1 was a key transcription factor that facilitated TET2 interaction with methylated DNA1. Leveraging a PU.1 transcription factor ChIP-seq library that was generated for R132H following 72hrs of vitamin C treatment, I identified PU.1 binding sites within 2kb of 73% (32/44) of the promoters (TSS ± 500bp) of genes that were upregulated by vitamin C, and only 41% (38/92) of the genes in the vitamin C-insensitive set1. As mentioned, I observed that Cebpe expression was modulated by the level of TET2/3 63  activity. Similarly, upon vitamin C treatment, I observed upregulation of Cebpe in a manner dependent on the level of TET2/3 activity (Figure 20c). Strikingly, among the vitamin C-sensitive gene set, I observed an exceptional increase in 5hmC at the promoter of Cebpe (Figure 21a). This specific targeted demethylation of the Cebpe promoter suggests a direct downstream target of TET2 and TET3 activity that may contribute to the myeloid differentiation phenotype observed. In contrast, Tfec, a myeloid-restricted transcription factor, was only upregulated in the context of functional TET2 (Figure 18f)115. To identify transcription factors that were affected by changes in TET activity, I identified a set of genomic regions that were likely undergoing TET-mediated active demethylation by restricting for regions that gained 5hmC at 72 hours of vitamin C and were within 1.5kb of regions that lost 5mC for each cell line. With this filter, I found 2774, 1059, 1875, and 1104 regions for R132H, TET2KO, TET3KO, and DKO, respectively. Through de novo motif analysis of the underlying sequences at these regions, I observed motifs enriched in PU.1 (35%), CEBPB/E (15%), RUNX1 (26%), and composite CEBP:AP1 (31%) in R132H (Figure 22a). These findings were largely mirrored in TET2KO and TET3KO, though only AP-1 was enriched in both groups and not the composite. Notably, RUNX1 binding motifs were not enriched in the DKO, which may suggest a unique binding motif specific to TET2/3. Indeed, TET2 and TET3 have both been shown to enhance RUNX1-mediated DNA demethylation116. I leveraged a RUNX1 transcription factor ChIP-seq library previously generated for R132H following 72hrs of vitamin C to examine this relationship. Markedly, I observed an increase in 5hmC at regions flanking RUNX1-binding sites in R132H, TET2KO, and TET3KO, but not in DKO at 72 hours of vitamin C treatment (Figure 22b). Taken together, this indicates that RUNX1 activation is a downstream effect of vitamin C treatment and suggests that both TET2 and TET3 can be recruited to RUNX1 binding sites to demethylate their flanking regions. 64    Figure 22. (A) Top hits from HOMER de novo motif analysis performed on regions that concurrently gained 5hmC and lost 5mC at 72hr of vitamin C treatment. (B) Heatmap showing the levels of 5hmC at RUNX1 transcription factor ChIP-seq regions in R132H, TET2KO, TET3KO, and DKO cells at 15hr (left) and 72hr (right) in untreated (grey) vs. vitamin C (orange).   65  Chapter 6: Conclusion In this thesis, we present evidence for an integral role of TET3 in the maintenance of a myeloid expression signature in a murine AML model. Our findings suggest a model in which both TET2 and TET3 are required to maintain the expression of 136 genes encoding transcription factors, cell surface markers, and enzymes that are implicated in myeloid differentiation. Loss of either TET2 or TET3 was sufficient to significantly reduce the expression of these genes. This loss in expression did not show a strong relationship in 5mC or 5hmC levels at their promoters. However, we did observe a significant loss of 5hmC at the nearest active enhancer marked with 5hmC in the parental line. Furthermore, 5hmC loss was exacerbated in the TET2 and TET3 double knockout, which suggests that some basal level of TET2 and TET3 activity is maintained even in the presence of the inhibitory effects of IDH1R132H. We observed that among the set of TET2/3-dependent genes, 44 of them can be rescued by vitamin C induction of the remaining TET, which suggests that those genes are directly regulated by TET’s ability to maintain a demethylated phenotype. Furthermore, these genes were expressed at a significantly higher level in a background without TET suppression (i.e. in the IDH1WT). We identified several putative transcription factors that may interact with TET2 and TET3, including PU.1 and RUNX1, consistent with our previous finding1. In addition, we identified TFEC and CEBPE as novel targets for TET2 and TET3 activity. Little is known about the myeloid-restricted transcription factor TFEC, though it is part of the Mitf-Tfe family of basic helix-loop-helix-leucine zippers including Mitf, Tfe3, and Tfeb and was shown to be up-regulated by IL-4 induction115,117. Strikingly, Cebpe expression is directly linked to the degree of TET2 and TET3 activity; knockout of either TET2 or TET3 down-regulates Cebpe expression, and vitamin C treatment reverses this effect in a TET2- and TET3- dependent manner. This is remarkable since Cebpe expression have been shown to be directly regulated by RUNX1 and is implicated as a driver of the differentiation block observed in Runx1-inactivated AML118. Furthermore, Cebpe expression has been observed to increase in expression leading to terminal granulocyte differentiation and was found to not be required 66  for initial differentiation towards a neutrophil progenitor state, but for further differentiation and maturation96,119. If TET2 and TET3 both contribute to the regulation of the myeloid differentiation pathway, it is surprising that only loss of function mutations to TET2 are recurrent in AML. We identified a set of genes (n=14) that were specifically differentially expressed in the TET2KO and DKO, but not in TET3KO. Among that list is the gene encoding argininosuccinate synthase 1 (Ass1), which is indispensable for endogenous arginine synthesis and whose deficiency has been found to confer a proliferative advantage in multiple cancers, including AML98,120. Interestingly, we observed that effects primarily mediated by TET2 can also be mediated by TET3, albeit at a later time point. This may be due to a difference in catalytic efficiencies between TET2 and TET3, in which TET2 was found to have a 2.6-fold greater efficiency in 5mC oxidation than TET3121. Furthermore, TET3 has been observed to be a caC reader in mouse neuronal cells, which may suggest that TET3 has a greater role in further oxidizing 5hmC and recruiting BER machinery. Since TET2 mutations primarily happen in a heterozygous context and we observed that TET2 plays a greater role in oxidizing 5mC compared to TET3, this would support a model in which a threshold level of TET activity is necessary to maintain proper DNA methylation homeostasis. This was supported by our finding that the IDH1R132H protein incompletely inactivates TET2 and TET3. Interestingly, we were able to observe a set of 92 genes that were affected by TET2 and TET3 in a mechanism that does not involve TET’s ability to oxidize methylated DNA. Both TET2 and TET3 have been shown to interact with other proteins such as HDACs and OGTs as scaffolds, and their absence may contribute to downregulation of these genes through changes in histone modifications122–124. This further highlights a phenotypic difference between AMLs with mutations in IDH1 and TET2, perhaps indicative of additional pathways of disruption that promotes leukemogenesis. The product of IDH1R132H is the oncometabolite R-2HG, which has been shown to act in a paracrine manner to speed up leukemic progression in mouse bone marrow cells overexpressing Hoxa9 in a reversible manner independent of 67  the mutant IDH1 protein52. Perhaps paradoxically, R-2HG’s enantiomer S-2HG has been shown to be a more potent inhibitor of various α-KG-dependent enzymes including TET2 than R-2HG103. However, S-2HG is unable to promote transformation. A key difference between R-2HG and S-2HG is in their interactions with prolyl hydroxylase domain 2 (PHD2), which is encoded by Egln1; R-2HG promotes PHD2 activity, while S-2HG inhibits it53. In contrast, TET2 has been observed to display functions beyond its role in facilitating active demethylation, including its role in resolving inflammation through the recruitment of histone deactylase 2 (HDAC2)122. TET2 inactivation in HSPCs was shown to perpetuate a high level of inflammation through release of pro-inflammatory cytokines, which induces pro-apoptotic signaling in neighbouring TET2wt HSPCs 46. Importantly, changes in 5hmC and 5mC levels were insufficient to explain all of the transcriptional differences we observed between the single knockouts of TET2, TET3, and both. Using vitamin C to induce TET2 and TET3 activity, we observed a convergent demethylation signature at a common set of enhancers that are localized around genes expressed during differentiation and inflammatory response, which supports the observed redundancies within the TET family in T-cells and B-cells32,125,126. Notably, the differences observed after vitamin C induction of TET2 and TET3 in terms of global patterns of methylation and expression may reflect a temporal effect reflecting their functional efficiencies. In support of this, the majority of regions that gained 5hmC in the TET3KO at 15hrs of vitamin C treatment showed gains in 5hmC in the TET2KO by 72hrs. This is reflected in the functional data, in which R132H showed the earliest drop in proliferation in response to repeat vitamin C, followed by TET3KO, and TET2KO, in that order.  The US Food and Drug Administration (FDA) has approved inhibitors for mutant IDH1 (ivosidenib or AG-120) and IDH2 (enasidenib) in 2017 and 2018 in the treatment of relapsed or refractory AML with mutations in IDH1/2127,128. 41.6% of ivosidenib-treated patients and 40.3% of enasidenib-treated patients showed an overall response, which was markedly high relative to the poor response observed 68  with hypomethylating agents (23%)129. Among the cohort that showed resistance to ivosidenib, mutations in genes involved in receptor tyrosine kinase pathways were highly enriched, including FLT3, KIT, KRAS, and PTPN11128. Given our understanding of the known pathways, it would be interesting to determine if vitamin C synergizes with the effects of mutant IDH1/2 inhibitors to further induce differentiation of AML blasts.  We also observed a set of genes exclusively upregulated in R132H after 72 hours of vitamin C induction, including peptides S100a8 and S100a9. Normally highly expressed in neutrophils, S100a8 and S100a9 play critical roles in activating inflammatory signaling through the toll-like receptor 4 (TLR4)- or receptor for advanced glycation end products (RAGE)- mediated pathways. These peptides have recently been given the spotlight due to their roles in leukemia pathogenesis and relationship to resistance of certain chemotherapeutics, such as the BCL-2 antagonist venetoclax111. In contrast, S100a9 has also been shown to activate TLR4 to promote differentiation in monocytic AML110.  Intriguingly, S100a8 and S100a9 have shown interactions with CD33, a receptor often expressed in myeloid-derived suppressor cells (MDSCs), an immunosuppressive population in AML patients. We observed that CD33 expression is TET2/3-dependent and that a putative nearby enhancer gain 5hmC after vitamin C induction. CD33 has been an attractive target for targeted therapy in AML, though clinical trials using the anti-CD33 drug gemtuzumab ozogamicin has shown little difference in overall survival, though it is more effective in AML blasts that express a higher level of C33130,131. Recent interest in CD33-directed therapy was renewed with chimeric antigen receptor T-cells (CAR-Ts), which was shown to demonstrate high on-target effects for CD33+ AML and limited toxicities to normal myeloid cells112. Here, we observed that CD33 expression is closely linked to the presence of both TET2 and TET3, and that vitamin C induction can increase the relative expression of CD33 in a mechanism that is TET dependent. Indeed, this observation may suggest a potential combinatorial therapy that includes the combined addition of anti-CD33 and vitamin C treatment. 69  Collectively, this study provides mechanistic insight into an overlapping function of TET2 and TET3 in the regulation of myeloid genes in AML. We identified that a low level of TET activity was still present in an AML model harbouring a neomorphic mutation in IDH1, and that TET activity has a specific propensity towards active enhancer elements. We also confirmed a novel hypermethylation signature at active enhancers in the IDH1R132H background. This was corroborated by the observation that vitamin C drives an active demethylation process at these enhancers through both TET2 and TET3 with varying efficiencies. Due to the functional redundancies of TET2 and TET3, it was previously difficult to ascribe functions that were entirely TET-dependent. Here we showed that TET2 and TET3 cooperates to regulate a core set of 136 genes, which partially depends on the ability of TET2 and TET3 to oxidize 5mC at regulatory regions including at nearby putative enhancers and at promoters. Furthermore, we observed that the majority of regions that gained 5hmC after vitamin C treatment can be targeted by both TET2 and TET3, albeit at different efficiencies. This was supported by the observation that a common set of myeloid differentiation genes were upregulated after vitamin C treatment in all models with at least functional TET2 or TET3. This was confirmed experimentally, wherein we showed that the vitamin C-induced myeloid differentiation in IDH1R132H AML is driven not only through TET2, but also through TET3, which reveals another potential target of vitamin C activation, especially given that TET2 is frequently inactivated in AML.  This data suggest a TET activation gradient model, in which a threshold of TET activity is necessary for proper differentiation to occur (Figure 23). Our data supported the finding that TET2 was more catalytically efficient than TET3121. In the presence of R-2HG, the net level of TET activity would be reduced to a point below the threshold, which incurs a differentiation block phenotype. Further inactivation of TET2 and TET3 would reduce this activity, which we observed in terms of transcriptional down-regulation and 5hmC reduction of active enhancer elements. Vitamin C induction increases activity of both TET2 and TET3 past the threshold, which was observed from transcriptional up-70  regulation of myeloid differentiation genes and an increase in differentiated cells in R132H, TET2KO, and TET3KO models. This gradient model may therefore explain the specificity of TET2 inactivation in AML.  Figure 23. A proposed model of a TET activity gradient in relation to differentiation. The dotted bars represent predicted levels of TET activity in an IDH1WT cell line with either TET2+/- or TET3-/-.     71  Bibliography 1. Mingay, M. et al. Vitamin C-induced epigenomic remodelling in IDH1 mutant acute myeloid leukaemia. Leukemia (2017). doi:10.1038/leu.2017.171 2. De Kouchkovsky, I. & Abdul-Hay, M. ‘Acute myeloid leukemia: a comprehensive review and 2016 update’. Blood Cancer J. 6, e441 (2016). 3. Meyers, J., Yu, Y., Kaye, J. A. & Davis, K. L. Medicare fee-for-service enrollees with primary acute myeloid leukemia: An analysis of treatment patterns, survival, and healthcare resource utilization and costs. Appl. Health Econ. Health Policy 11, 275–286 (2013). 4. Breems, D. A. et al. Monosomal karyotype in acute myeloid leukemia: A better indicator of poor prognosis than a complex karyotype. J. Clin. Oncol. 26, 4791–4797 (2008). 5. Cancer Genome Atlas Research Network, T. C. G. A. R. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N. Engl. J. Med. 368, 2059–74 (2013). 6. Gary Gilliland, D. & Griffin, J. D. The roles of FLT3 in hematopoiesis and leukemia. Blood 100, 1532–1542 (2002). 7. Busque, L. et al. Recurrent somatic TET2 mutations in normal elderly individuals with clonal hematopoiesis. Nat. Genet. (2012). doi:10.1038/ng.2413 8. Chaturvedi, A. et al. Mutant IDH1 promotes leukemogenesis in vivo and can be specifically targeted in human AML. Blood (2013). doi:10.1182/blood-2013-03-491571 9. Jjingo, D., Conley, A. B., Yi, S. V., Lunyak, V. V. & King Jordan, I. On the presence and role of human gene-body DNA methylation. Oncotarget (2012). doi:10.18632/oncotarget.497 10. Ziller, M. J. et al. Charting a dynamic DNA methylation landscape of the human genome. Nature 72  (2013). doi:10.1038/nature12433 11. Bird, A., Taggart, M., Frommer, M., Miller, O. J. & Macleod, D. A fraction of the mouse genome that is derived from islands of nonmethylated, CpG-rich DNA. Cell (1985). doi:10.1016/0092-8674(85)90312-5 12. Bock, C. et al. DNA Methylation Dynamics during In Vivo Differentiation of Blood and Skin Stem Cells. Mol. Cell (2012). doi:10.1016/j.molcel.2012.06.019 13. Shipony, Z. et al. Dynamic and static maintenance of epigenetic memory in pluripotent and somatic cells. Nature (2014). doi:10.1038/nature13458 14. Polo, J. M. et al. Cell type of origin influences the molecular and functional properties of mouse induced pluripotent stem cells. Nat. Biotechnol. (2010). doi:10.1038/nbt.1667 15. Goll, M. G. & Bestor, T. H. EUKARYOTIC CYTOSINE METHYLTRANSFERASES. Annu. Rev. Biochem. (2005). doi:10.1146/annurev.biochem.74.010904.153721 16. Okano, M., Bell, D. W., Haber, D. A. & Li, E. DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell (1999). doi:10.1016/S0092-8674(00)81656-6 17. Ito, S. et al. Tet proteins can convert 5-methylcytosine to 5-formylcytosine and 5-carboxylcytosine. Science (80-. ). (2011). doi:10.1126/science.1210597 18. He, Y. F. et al. Tet-mediated formation of 5-carboxylcytosine and its excision by TDG in mammalian DNA. Science (80-. ). (2011). doi:10.1126/science.1210944 19. Tahiliani, M. et al. Conversion of 5-methylcytosine to 5-hydroxymethylcytosine in mammalian DNA by MLL partner TET1. Science (80-. ). (2009). doi:10.1126/science.1170116 73  20. Bachman, M. et al. 5-Hydroxymethylcytosine is a predominantly stable DNA modification. Nat. Chem. (2014). doi:10.1038/nchem.2064 21. Madzo, J. et al. Hydroxymethylation at gene regulatory regions directs stem/early progenitor cell commitment during erythropoiesis. Cell Rep. 6, 231–244 (2014). 22. Han, D. et al. A Highly Sensitive and Robust Method for Genome-wide 5hmC Profiling of Rare Cell Populations. Mol. Cell (2016). doi:10.1016/j.molcel.2016.06.028 23. Kriaucionis, S. & Heintz, N. The nuclear DNA base 5-hydroxymethylcytosine is present in purkinje neurons and the brain. Science (80-. ). (2009). doi:10.1126/science.1169786 24. Li, W. & Liu, M. Distribution of 5-hydroxymethylcytosine in different human tissues. J. Nucleic Acids (2011). doi:10.4061/2011/870726 25. Globisch, D. et al. Tissue distribution of 5-hydroxymethylcytosine and search for active demethylation intermediates. PLoS One (2010). doi:10.1371/journal.pone.0015367 26. Nakajima, H. & Kunimoto, H. TET2 as an epigenetic master regulator for normal and malignant hematopoiesis. Cancer science 105, 1093–1099 (2014). 27. Delatte, B., Deplus, R. & Fuks, F. Playing TETris with DNA modifications. EMBO J. (2014). doi:10.15252/embj.201488290 28. Wang, Y. et al. WT1 recruits TET2 to regulate its target gene expression and suppress leukemia cell proliferation. Mol. Cell 57, 662–673 (2015). 29. Xu, Y. et al. Genome-wide Regulation of 5hmC, 5mC, and Gene Expression by Tet1 Hydroxylase in Mouse Embryonic Stem Cells. Mol. Cell (2011). doi:10.1016/j.molcel.2011.04.005 30. Hu, L. et al. Structural insight into substrate preference for TET-mediated oxidation. Nature 74  (2015). doi:10.1038/nature15713 31. Jin, S. G. et al. Tet3 Reads 5-Carboxylcytosine through Its CXXC Domain and Is a Potential Guardian against Neurodegeneration. Cell Rep. (2016). doi:10.1016/j.celrep.2015.12.044 32. Li, C. et al. Overlapping Requirements for Tet2 and Tet3 in Normal Development and Hematopoietic Stem Cell Emergence. Cell Rep. 12, 1133–1143 (2015). 33. Fang, S. et al. Tet inactivation disrupts YY1 binding and long-range chromatin interactions during embryonic heart development. Nat. Commun. (2019). doi:10.1038/s41467-019-12325-z 34. Huang, Y. et al. Distinct roles of the methylcytosine oxidases Tet1 and Tet2 in mouse embryonic stem cells. Proc. Natl. Acad. Sci. U. S. A. (2014). doi:10.1073/pnas.1322921111 35. Tsukada, Y. I., Akiyama, T. & Nakayama, K. I. Maternal TET3 is dispensable for embryonic development but is required for neonatal growth. Sci. Rep. (2015). doi:10.1038/srep15876 36. Ito, S. et al. Role of tet proteins in 5mC to 5hmC conversion, ES-cell self-renewal and inner cell mass specification. Nature (2010). doi:10.1038/nature09303 37. An, J. et al. Acute loss of TET function results in aggressive myeloid cancer in mice. Nat. Commun. 6, 10071 (2015). 38. Montalbán-Loro, R. et al. TET3 prevents terminal differentiation of adult NSCs by a non-catalytic action at Snrpn. Nat. Commun. (2019). doi:10.1038/s41467-019-09665-1 39. Farlik, M. et al. DNA Methylation Dynamics of Human Hematopoietic Stem Cell Differentiation. Cell Stem Cell 19, 808–822 (2016). 40. Xing, Z. et al. Increased hematopoietic stem cell mobilization in aged mice. Blood (2006). doi:10.1182/blood-2005-12-010272 75  41. Rossi, D. J. et al. Cell intrinsic alterations underlie hematopoietic stem cell aging. Proc. Natl. Acad. Sci. U. S. A. (2005). doi:10.1073/pnas.0503280102 42. Rossi, D. J. et al. Deficiencies in DNA damage repair limit the function of haematopoietic stem cells with age. Nature (2007). doi:10.1038/nature05862 43. Porto, M. L. et al. Reactive oxygen species contribute to dysfunction of bone marrow hematopoietic stem cells in aged C57BL/6 J mice. J. Biomed. Sci. (2015). doi:10.1186/s12929-015-0201-8 44. Sun, D. et al. Epigenomic profiling of young and aged HSCs reveals concerted changes during aging that reinforce self-renewal. Cell Stem Cell (2014). doi:10.1016/j.stem.2014.03.002 45. Genovese, G. et al. Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N. Engl. J. Med. (2014). doi:10.1056/NEJMoa1409405 46. Cai, Z. et al. Inhibition of Inflammatory Signaling in Tet2 Mutant Preleukemic Cells Mitigates Stress-Induced Abnormalities and Clonal Hematopoiesis. Cell Stem Cell (2018). doi:10.1016/j.stem.2018.10.013 47. Challen, G. A. et al. Dnmt3a is essential for hematopoietic stem cell differentiation. Nat. Genet. (2012). doi:10.1038/ng.1009 48. Figueroa, M. E. et al. Leukemic IDH1 and IDH2 Mutations Result in a Hypermethylation Phenotype, Disrupt TET2 Function, and Impair Hematopoietic Differentiation. Cancer Cell 18, 553–567 (2010). 49. Ward, P. S. et al. The Common Feature of Leukemia-Associated IDH1 and IDH2 Mutations Is a Neomorphic Enzyme Activity Converting ??-Ketoglutarate to 2-Hydroxyglutarate. Cancer Cell 17, 225–234 (2010). 76  50. Dang, L. et al. Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature 462, 739–44 (2009). 51. Cairns, R. A. & Mak, T. W. Oncogenic isocitrate dehydrogenase mutations: Mechanisms, models, and clinical opportunities. Cancer Discovery (2013). doi:10.1158/2159-8290.CD-13-0083 52. Chaturvedi, A. et al. Enantiomer-specific and paracrine leukemogenicity of mutant IDH metabolite 2-hydroxyglutarate. Leukemia 30, 1708–1715 (2016). 53. Losman, J.-A. et al. (R)-2-Hydroxyglutarate Is Sufficient to Promote Leukemogenesis and Its Effects Are Reversible. Science (80-. ). 339, 1621–1625 (2013). 54. Turcan, S. et al. IDH1 mutation is sufficient to establish the glioma hypermethylator phenotype. Nature (2012). doi:10.1038/nature10866 55. Dang, L., Jin, S. & Su, S. M. IDH mutations in glioma and acute myeloid leukemia. Trends in Molecular Medicine (2010). doi:10.1016/j.molmed.2010.07.002 56. Sasaki, M. et al. IDH1(R132H) mutation increases murine haematopoietic progenitors and alters epigenetics. Nature (2012). doi:10.1038/nature11323 57. Moran-Crusio, K. et al. Tet2 Loss Leads to Increased Hematopoietic Stem Cell Self-Renewal and Myeloid Transformation. Cancer Cell 20, 11–24 (2011). 58. Cimmino, L. et al. Restoration of TET2 Function Blocks Aberrant Self-Renewal and Leukemia Progression. Cell 1–17 (2017). doi:10.1016/j.cell.2017.07.032 59. Inoue, S. et al. Mutant IDH1 Downregulates ATM and Alters DNA Repair and Sensitivity to DNA Damage Independent of TET2. Cancer Cell (2016). doi:10.1016/j.ccell.2016.05.018 60. Drouin, G., Godin, J.-R. & Page, B. The Genetics of Vitamin C Loss in Vertebrates. Curr. Genomics 77  12, 371–378 (2011). 61. Chen, Q. et al. Ascorbate in pharmacologic concentrations selectively generates ascorbate radical and hydrogen peroxide in extracellular fluid in vivo. Proc. Natl. Acad. Sci. U. S. A. 104, 8749–8754 (2007). 62. Cameron, E. & Pauling, L. Supplemental ascorbate in the supportive treatment of cancer: Prolongation of survival times in terminal human cancer. Proc. Natl. Acad. Sci. U. S. A. 73, 3685–3689 (1976). 63. Chen, Q. et al. Pharmacologic ascorbic acid concentrations selectively kill cancer cells: action as a pro-drug to deliver hydrogen peroxide to tissues. Proc. Natl. Acad. Sci. U. S. A. 102, 13604–9 (2005). 64. Chen, Q. et al. Pharmacologic doses of ascorbate act as a prooxidant and decrease growth of aggressive tumor xenografts in mice. Proc. Natl. Acad. Sci. U. S. A. (2008). doi:10.1073/pnas.0804226105 65. Wilson, M. K., Baguley, B. C., Wall, C., Jameson, M. B. & Findlay, M. P. Review of high-dose intravenous vitamin C as an anticancer agent. Asia-Pacific Journal of Clinical Oncology 10, 22–37 (2014). 66. Mojić, M. et al. Extracellular iron diminishes anticancer effects of vitamin C: an in vitro study. Sci. Rep. 4, 5955 (2014). 67. Evans, R. M., Currie, L. & Campbell,  a. The distribution of ascorbic acid between various cellular components of blood, in normal individuals, and its relation to the plasma concentration. Br. J. Nutr. 47, 473–82 (1982). 68. Hata, R. ???I & Senoo, H. L???ascorbic acid 2???phosphate stimulates collagen accumulation, cell 78  proliferation, and formation of a three???dimensional tissuelike substance by skin fibroblasts. J. Cell. Physiol. 138, 8–16 (1989). 69. Fujiwara, M. et al. Enhanced protection against peroxidation-induced mortality of aortic endothelial cells by ascorbic acid-2-O-phosphate abundantly accumulated in the cell as the dephosphorylated form. Free Radic. Res. 27, 97–104 (1997). 70. Takamizawa, S. et al. Effects of ascorbic acid and ascorbic acid 2-phosphate, a long-acting vitamin C derivative, on the proliferation and differentiation of human osteoblast-like cells. Cell Biol. Int. 28, 255–265 (2004). 71. Takamizawa, S. et al. Effects of ascorbic acid and ascorbic acid 2-phosphate, a long-acting vitamin C derivative, on the proliferation and differentiation of human osteoblast-like cells. Cell Biol. Int. 28, 255–265 (2004). 72. Austria, R., Semenzato, A. & Bettero, A. Stability of vitamin C derivatives in solution and topical formulations. J. Pharm. Biomed. Anal. (1997). doi:10.1016/S0731-7085(96)01904-8 73. Fujiwara, M. et al. Enhanced protection against peroxidation-induced mortality of aortic endothelial cells by ascorbic Acid-2-O-phosphate abundantly accumulated in the cell as the dephosphorylated form. Free Radic. Res. (1997). doi:10.3109/10715769709097842 74. Hitomi, K. & Torii, Y. Increase in the Activity of Alkaline Phosphatase by L-Ascorbic Acid 2-Phosphatein a Human Osteoblast Cell Line, HuO-3Nl. J. Nutr. Sci. Vitaminol. (Tokyo). (1992). doi:10.3177/jnsv.38.535 75. Blaschke, K. et al. Vitamin C induces Tet-dependent DNA demethylation and a blastocyst-like state in ES cells. Nature 500, 222–226 (2013). 76. Chen, J. J. et al. H3K9 methylation is a barrier during somatic cell reprogramming into iPSCs. Nat. 79  Genet. 45, 34–42 (2013). 77. Esteban, M. A. et al. Vitamin C Enhances the Generation of Mouse and Human Induced Pluripotent Stem Cells. Cell Stem Cell 6, 71–79 (2010). 78. Liu, M. et al. Vitamin C increases viral mimicry induced by 5-aza-2’-deoxycytidine. Proc. Natl. Acad. Sci. U. S. A. 113, 10238–44 (2016). 79. de la Rica, L. et al. PU.1 target genes undergo Tet2-coupled demethylation and DNMT3b-mediated methylation in monocyte-to-osteoclast differentiation. Genome Biol. (2013). doi:10.1186/gb-2013-14-9-r99 80. Agathocleous, M. et al. Ascorbate regulates haematopoietic stem cell function and leukaemogenesis. Nature (2017). doi:10.1038/nature23876 81. Shih, A. H. et al. Mutational cooperativity linked to combinatorial epigenetic gain of function in acute myeloid leukemia. Cancer Cell 27, 502–515 (2015). 82. Pulikkottil Jose, A. et al. The Methylcytosine Dioxygenase TET3 Is Aberrantly Expressed in Acute Myeloid Leukemia and Promotes AML Growth. Blood 128, (2016). 83. Cheung, N. et al. Targeting Aberrant Epigenetic Networks Mediated by PRMT1 and KDM4C in Acute Myeloid Leukemia. Cancer Cell 29, 32–48 (2016). 84. Cimmino, L. et al. TET1 is a tumor suppressor of hematopoietic malignancy. Nat. Immunol. 16, 653–662 (2015). 85. Pinello, L. et al. Analyzing CRISPR genome-editing experiments with CRISPResso. Nature Biotechnology (2016). doi:10.1038/nbt.3583 86. McLean, C. Y. et al. GREAT improves functional interpretation of cis-regulatory regions. Nat. 80  Biotechnol. (2010). doi:10.1038/nbt.1630 87. Heinz, S. et al. Simple Combinations of Lineage-Determining Transcription Factors Prime cis-Regulatory Elements Required for Macrophage and B Cell Identities. Mol. Cell (2010). doi:10.1016/j.molcel.2010.05.004 88. Butterfield, Y. S. et al. JAGuaR: Junction alignments to genome for RNA-seq reads. PLoS One (2014). doi:10.1371/journal.pone.0102398 89. Zhou, Y. et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat. Commun. (2019). doi:10.1038/s41467-019-09234-6 90. Keenan, A. B. et al. ChEA3: transcription factor enrichment analysis by orthogonal omics integration. Nucleic Acids Res. (2019). doi:10.1093/nar/gkz446 91. Wang, H. et al. One-step generation of mice carrying mutations in multiple genes by CRISPR/cas-mediated genome engineering. Cell (2013). doi:10.1016/j.cell.2013.04.025 92. Wu, C. H., Gordon, J., Rastegar, M., Ogretmen, B. & Safa, A. R. Proteinase-3, a serine protease which mediates doxorubicin-induced apoptosis in the HL-60 leukemia cell line, is downregulated in its doxorubicin-resistant variant. Oncogene (2002). doi:10.1038/sj.onc.1205639 93. Thivakaran, A. et al. Gfi1b: A key player in the genesis and maintenance of acute myeloid leukemia and myelodysplastic syndrome. Haematologica (2018). doi:10.3324/haematol.2017.167288 94. Sharma, A. et al. Constitutive IRF8 expression inhibits AML by activation of repressed immune response signaling. Leukemia (2015). doi:10.1038/leu.2014.162 95. Bartelmez, S. H. & Stanley, E. R. Synergism between hemopoietic growth factors (HGFs) detected 81  by their effects on cells bearing receptors for a lineage specific HGF: Assay of hemopoietin‐1. J. Cell. Physiol. (1985). doi:10.1002/jcp.1041220306 96. Paul, F. et al. Transcriptional Heterogeneity and Lineage Commitment in Myeloid Progenitors. Cell (2015). doi:10.1016/j.cell.2015.11.013 97. Miraki-Moud, F. et al. Arginine deprivation using pegylated arginine deiminase has activity against primary acute myeloid leukemia cells in vivo. Blood (2015). doi:10.1182/blood-2014-10-608133 98. Huang, H. Y. et al. ASS1 as a novel tumor suppressor gene in myxofibrosarcomas: Aberrant loss via epigenetic DNA methylation confers aggressive phenotypes, negative prognostic impact, and therapeutic relevance. Clin. Cancer Res. (2013). doi:10.1158/1078-0432.CCR-12-2641 99. Roninson, I. B. Oncogenic functions of tumour suppressor p21Waf1/Cip1/Sdi1: Association with cell senescence and tumour-promoting activities of stromal fibroblasts. Cancer Letters (2002). doi:10.1016/S0304-3835(01)00847-3 100. Zhou, J. et al. The pro-metastasis tyrosine phosphatase, PRL-3 (PTP4A3), is a novel mediator of oncogenic function of BCR-ABL in human chronic myeloid leukemia. Mol. Cancer (2012). doi:10.1186/1476-4598-11-72 101. Grönroos, T. et al. Overexpression of PTP4A3 in ETV6-RUNX1 acute lymphoblastic leukemia. Leuk. Res. (2017). doi:10.1016/j.leukres.2016.12.005 102. Zhou, J. et al. PRL-3, a metastasis associated tyrosine phosphatase, is involved in FLT3-ITD signaling and implicated in anti-aml therapy. PLoS One (2011). doi:10.1371/journal.pone.0019798 103. Xu, W. et al. Oncometabolite 2-hydroxyglutarate is a competitive inhibitor of α-ketoglutarate-dependent dioxygenases. Cancer Cell (2011). doi:10.1016/j.ccr.2010.12.014 82  104. Vento-Tormo, R. et al. IL-4 orchestrates STAT6-mediated DNA demethylation leading to dendritic cell differentiation. Genome Biol. (2016). doi:10.1186/s13059-015-0863-2 105. Jeong, J. J. et al. Cytokine-regulated phosphorylation and activation of TET2 by JAK2 in hematopoiesis. Cancer Discov. (2019). doi:10.1158/2159-8290.CD-18-1138 106. Yang, J., Ikezoe, T., Nishioka, C., Nobumoto, A. & Yokoyama, A. IL-1β inhibits self-renewal capacity of dormant CD34 +/CD38- acute myelogenous leukemia cells in vitro and in vivo. Int. J. Cancer (2013). doi:10.1002/ijc.28198 107. Nytko, K. J. et al. Vitamin C is dispensable for oxygen sensing in vivo. Blood 117, 5485–5493 (2011). 108. Jeong, Y. J. et al. Vitamin C-treated murine bone marrow-derived dendritic cells preferentially drive naïve T cells into Th1 cells by increased IL-12 secretions. Cell. Immunol. (2011). doi:10.1016/j.cellimm.2010.10.005 109. Varney, M. E. et al. Loss of Tifab, a del(5q) MDS gene, alters hematopoiesis through derepression of Toll-like receptor-TRAF6 signaling. J. Exp. Med. (2015). doi:10.1084/jem.20141898 110. Laouedj, M. et al. S100A9 induces differentiation of acute myeloid leukemia cells through TLR4. Blood (2017). doi:10.1182/blood-2016-09-738005 111. Karjalainen, R. et al. Elevated expression of S100A8 and S100A9 correlates with resistance to the BCL-2 inhibitor venetoclax in AML. Leukemia (2019). doi:10.1038/s41375-019-0504-y 112. Borot, F. et al. Gene-edited stem cells enable CD33-directed immune therapy for myeloid malignancies. Proc. Natl. Acad. Sci. U. S. A. (2019). doi:10.1073/pnas.1819992116 113. Xu, X. et al. IFN-Stimulated Gene LY6E in Monocytes Regulates the CD14/TLR4 Pathway but 83  Inadequately Restrains the Hyperactivation of Monocytes during Chronic HIV-1 Infection. J. Immunol. (2014). doi:10.4049/jimmunol.1401249 114. Jin, L., Hope, K. J., Zhai, Q., Smadja-Joffe, F. & Dick, J. E. Targeting of CD44 eradicates human acute myeloid leukemic stem cells. Nat. Med. (2006). doi:10.1038/nm1483 115. Rehli, M., Lichanska, A., Cassady, A. I., Ostrowski, M. C. & Hume, D. A. TFEC is a macrophage-restricted member of the microphthalmia-TFE subfamily of basic helix-loop-helix leucine zipper transcription factors. J. Immunol. (1999). 116. Suzuki, T. et al. RUNX1 regulates site specificity of DNA demethylation by recruitment of DNA demethylation machineries in hematopoietic cells. Blood Adv. (2017). doi:10.1182/bloodadvances.2017005710 117. Rehli, M. et al. Transcription Factor Tfec Contributes to the IL-4-Inducible Expression of a Small Group of Genes in Mouse Macrophages Including the Granulocyte Colony-Stimulating Factor Receptor. J. Immunol. (2005). doi:10.4049/jimmunol.174.11.7111 118. Ng, K. P. et al. Runx1 deficiency permits granulocyte lineage commitment but impairs subsequent maturation. Oncogenesis (2013). doi:10.1038/oncsis.2013.41 119. Shyamsunder, P. et al. Identification of a novel enhancer of CEBPE essential for granulocytic differentiation. Blood (2019). doi:10.1182/blood.2018886077 120. Haines, R. J., Pendleton, L. C. & Eichler, D. C. Argininosuccinate synthase: At the center of arginine metabolism. International Journal of Biochemistry and Molecular Biology (2011). 121. Sudhamalla, B., Dey, D., Breski, M. & Islam, K. A rapid mass spectrometric method for the measurement of catalytic activity of ten-eleven translocation enzymes. Anal. Biochem. (2017). doi:10.1016/j.ab.2017.06.011 84  122. Zhang, Q. et al. Tet2 is required to resolve inflammation by recruiting Hdac2 to specifically repress IL-6. Nature (2015). doi:10.1038/nature15252 123. Chen, Q., Chen, Y., Bian, C., Fujiki, R. & Yu, X. TET2 promotes histone O-GlcNAcylation during gene transcription. Nature (2013). doi:10.1038/nature11742 124. Deplus, R. et al. TET2 and TET3 regulate GlcNAcylation and H3K4 methylation through OGT and SET1/COMPASS. EMBO Journal (2013). doi:10.1038/emboj.2012.357 125. Yue, X., Lio, C. W. J., Samaniego-Castruita, D., Li, X. & Rao, A. Loss of TET2 and TET3 in regulatory T cells unleashes effector function. Nat. Commun. (2019). doi:10.1038/s41467-019-09541-y 126. Lio, C. W. et al. Tet2 and Tet3 cooperate with B-lineage transcription factors to regulate DNA modification and chromatin accessibility. Elife (2016). doi:10.7554/eLife.18290 127. Stein, E. M. et al. Enasidenib in mutant IDH2 relapsed or refractory acute myeloid leukemia. Blood (2017). doi:10.1182/blood-2017-04-779405 128. DiNardo, C. D. et al. Durable remissions with ivosidenib in IDH1-mutated relapsed or refractory AML. N. Engl. J. Med. (2018). doi:10.1056/NEJMoa1716984 129. Roboz, G. J. et al. International randomized phase III study of elacytarabine versus investigator choice in patients with relapsed/refractory acute myeloid leukemia. J. Clin. Oncol. (2014). doi:10.1200/JCO.2013.52.8562 130. Bross, P. F. et al. Approval summary: Gemtuzumab ozogamicin in relapsed acute myeloid leukemia. Clin. Cancer Res. (2001). 131. Khan, N. et al. Expression of CD33 is a predictive factor for effect of gemtuzumab ozogamicin at different doses in adult acute myeloid leukaemia. in Leukemia (2017). doi:10.1038/leu.2016.309 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0388359/manifest

Comment

Related Items