ALBERT

Description: Mutations in the DNA methyltransferase 3A (DNMT3A) gene are the most common cause of age-related clonal hematopoiesis (ARCH) in older individuals, and are among the most common initiating events for acute myeloid leukemia (AML). The most frequent DNMT3A mutation in AML patients (R882H) encodes a dominant-negative protein that reduces methyltransferase activity by ∼80% in cells with heterozygous mutations, causing a focal, canonical DNA hypomethylation phenotype; this phenotype is partially recapitulated in murine Dnmt3a−/− bone marrow cells. To determine whether the hypomethylation phenotype of Dnmt3a−/− hematopoietic cells is reversible, we developed an inducible transgene to restore expression of DNMT3A in transplanted bone marrow cells from Dnmt3a−/− mice. Partial remethylation was detected within 1 wk, but near-complete remethylation required 6 mo. Remethylation was accurate, dynamic, and highly ordered, suggesting that differentially methylated regions have unique properties that may be relevant for their functions. Importantly, 22 wk of DNMT3A addback partially corrected dysregulated gene expression, and mitigated the expansion of myeloid cells. These data show that restoring DNMT3A expression can alter the epigenetic “state” created by loss of Dnmt3a activity; this genetic proof-of-concept experiment suggests that this approach could be relevant for patients with ARCH or AML caused by loss-of-function DNMT3A mutations.

Description: The impact of intratumoral heterogeneity (ITH) and the resultant neoantigen landscape on T cell immunity are poorly understood. ITH is a widely recognized feature of solid tumors and poses distinct challenges related to the development of effective therapeutic strategies, including cancer neoantigen vaccines. Here, we performed deep targeted DNA sequencing of multiple metastases from melanoma patients and observed ubiquitous sharing of clonal and subclonal single nucleotide variants (SNVs) encoding putative HLA class I-restricted neoantigen epitopes. However, spontaneous antitumor CD8+ T cell immunity in peripheral blood and tumors was restricted to a few clonal neoantigens featuring an oligo-/monoclonal T cell-receptor (TCR) repertoire. Moreover, in various tumors of the 4 patients examined, no neoantigen-specific TCR clonotypes were identified despite clonal neoantigen expression. Mature dendritic cell (mDC) vaccination with tumor-encoded amino acid-substituted (AAS) peptides revealed diverse neoantigen-specific CD8+ T responses, each composed of multiple TCR clonotypes. Isolation of T cell clones by limiting dilution from tumor-infiltrating lymphocytes (TILs) permitted functional validation regarding neoantigen specificity. Gene transfer of TCRαβ heterodimers specific for clonal neoantigens confirmed correct TCR clonotype assignments based on high-throughput TCRBV CDR3 sequencing. Our findings implicate immunological ignorance of clonal neoantigens as the basis for ineffective T cell immunity to melanoma and support the concept that therapeutic vaccination, as an adjunct to checkpoint inhibitor treatment, is required to increase the breadth and diversity of neoantigen-specific CD8+ T cells.

Description: Severe congenital neutropenia (SCN) is an inborn disorder of granulopoiesis characterized by severe chronic neutropenia from birth, premature death secondary to infectious complications, and transformation to myeloid malignancy. Although many cases of SCN are associated with mutations in ELANE, encoding the neutrophil elastase, roughly one-third of cases do not have an identifiable genetic cause. In collaboration with the Severe Chronic Neutropenia International Registry (SCNIR), we performed exome sequencing on 90 cases of congenital neutropenia. Heterozygous missense mutations of CLPB were identified in six patients with SCN. None of these patients had mutations in other genes known to cause SCN. A total of 5 different mutations were identified that clustered within the ATPase domain. Of note, all of these mutations were predicted to be functionally deleterious and had a frequency of

Description: HOX genes encode a family of homeodomain transcription factors with important roles in hematopoiesis. Expression of HOX genes is also a common feature of acute myeloid leukemia (AML), and functional studies have suggested that HOX-dependent pathways may contribute to leukemogenesis. Although HOX expression is known to correlate with specific AML mutations, the patterns of expression of all 39 HOX genes in primary AML samples, and their relationships with recurrent AML mutations, are incompletely understood. In addition, little is known about the influence of AML mutations on DNA methylation at the HOX loci, and the relationship between HOX gene expression and methylation in AML. In this study, we carried out a combined analysis of gene expression data from microarray and RNA-sequencing platforms and genome-wide DNA array-based methylation from 189 primary AML samples that have been previously characterized by either whole-genome or whole exome sequencing. We also measured expression and methylation using the same platforms from normal bone marrow subsets, including CD34+ cells, promyelocytes, monocytes, neutrophils and lymphocytes, and obtained expression data from CD34+ hematopoietic precursors generated from in vitro differentiation of human embryonic stem cells. Our analysis confirmed previous work on the general patterns of HOX expression in AML. The HOXA and HOXB genes showed variation both within each cluster and across the AMLs, although high level expression was restricted to a subset of these genes, including HOXA3, HOXA5, HOXA7, HOXA9, HOXA10, HOXB2-HOXB4, and HOXB6, as well as HOX cofactor MEIS1; HOXC and HOXD genes were minimally expressed in all of the samples. These observations were orthogonally validated by RNA-seq, and with a targeted Nanostring expression platform. Consistent with previous studies, MLL-positive AML samples (n=11) expressed only HOXA genes and MEIS1. AML samples with CBFB-MYH11 rearrangements (n=12) showed expression of only MEIS1, and HOXB2-HOXB4 at moderate levels; RUNX1-RUNX1T1 (n=7) and PML-RARA (n=19) samples did not detectably express any HOX genes. In AMLs with a normal karyotype (n=85), we observed two distinct patterns; one pattern displayed little or no HOX gene expression (7/85; 8%), and another displayed canonical expression of a specific subset of the HOXA and HOXB genes and MEIS1 (78/85; 92%) with similar relative HOX gene expression levels in all cases. Comparison of this pattern with normal bone marrow revealed the same HOX expression pattern in normal CD34+ cells; additional analysis showed that this pattern was confined to hematopoietic stem/progenitor cells, but was not seen in more mature cells, including other CD34+ subsets, promyelocytes, monocytes and neutrophils. We also measured HOX gene expression in CD34+ hematopoietic precursors generated from in vitro differentiation of human embryonic stem cells, which revealed expression of only MEIS1 and the canonical HOXB genes, suggesting that activation of these genes may represent the earliest events in the HOX pathway of hematopoietic development. Correlation of HOX expression with recurrent AML mutations by gene set enrichment analysis demonstrated a significant association with NPM1 (P

Description: Key Points Decitabine treatment of in vitro expanded primary AML samples leads to global hypomethylation. Highly methylated CpGs are most affected by decitabine-induced hypomethylation, with little influence on transcriptional activity.

Description: Mutations in DNMT3A (encoding one of two mammalian de novo DNA methyltransferases) are found in 〉30% of normal karyotype AML cases and correlate with poor clinical outcomes. Most DNMT3A mutations occur at position R882 within the catalytic domain (most commonly R882H) and are virtually always heterozygous. This over-representation suggests that mutations at R882 may result in gain-of-function or dominant-negative activity that contributes to leukemogenesis. However, how DNA methylation might be altered in DNMT3A-mutant cases of AML remains unclear, and no published study to date has addressed the effects of mixing wild-type (WT) and R882H DNMT3A. Importantly, mouse HSPCs deficient in Dnmt3a dramatically expand over time and have a concurrent defect in differentiation (Challen, GA et al. Nat Genet, 2011). Mice haploinsufficient for Dnmt3a, on the other hand, do not have a measurable defect in hematopoiesis. Collectively, these data suggest that the heterozygous R882 mutations probably cause more than a simple loss-of-function phenotype. We purified full-length, human WT and R882H DNMT3A using a mammalian tissue culture system to produce recombinant proteins for biochemical modeling of the de novo methylation potential of a DNMT3A-mutant AML cell. rhR882H DNMT3A exhibits roughly 10-20% of the de novo DNA methyltransferase activity of rhWT DNMT3A, similar to observations by other groups. We added increasing amounts of R882H DNMT3A to a fixed amount of WT DNMT3A and observed a linear increase in the net enzymatic activity, reflecting the summed activity of the two forms of DNMT3A in these 4-hour in vitro reactions. In contrast, 12-hour in vitro DNA methylation assays with mixed WT and R882H DNMT3A demonstrated net methylation less than the predicted summed activity of the two enzymes, suggesting that a dominant-negative effect of R882H DNMT3A may occur with a long equilibration time. To better simulate an AML cell with a heterozygous R882H mutation, we co-transfected HEK293T cells with equal amounts of poly-His-tagged WT and R882H DNMT3A expression vectors. Subsequently co-purified (i.e. in vivo-mixed) WT and R882H DNMT3A exhibited a striking reduction in methyltransferase activity, with total activity similar to R882H DNMT3A alone (Figure 1A). TSQ mass spectrometry allowed us to verify the presence and quantify the relative concentration of WT and R882H DNMT3A in our co-purified samples. We exploited a novel tryptic cleavage site in DNMT3A produced by the R882H mutation to generate standard concentration curves using recombinant peptides distinguishing the two protein forms. Our co-purified enzyme preparations had WT:R882H ratios ranging from 0.79 to 1.60; all demonstrated the dominant-negative effect of R882H. DNMT3A is a processive enzyme, catalyzing multiple methyl-group transfers before dissociating from target DNA. This is dependent on the ability of WT DNMT3A to form homo-oligomers (tetramers and larger), which was recently shown to be disrupted by the R882H mutation using the catalytic domain of DNMT3A produced in E.coli (Holz-Schietinger, C et al. JBC, 2012). We therefore postulated that the dominant-negative effect of R882H may be due to the disruption of WT DNMT3A oligomerization. Using a Superose 6 size exclusion column, we confirmed the tetramerization defect of R882H DNMT3A relative to WT DNMT3A. Notably, in vivo-mixed (co-purified) WT and R882H DNMT3A complexes exhibited a pattern of oligomerization identical to R882H DNMT3A alone. However, WT and R882H DNMT3A mixed in vitro exhibited a distribution of oligomers corresponding to the expected average of the WT and R882H curves (Figure 1B). These data demonstrate that production of equal amounts of WT and R882H DNMT3A within the same cell provides an environment where R882H DNMT3A can exert a potent dominant-negative effect on WT DNMT3A. Furthermore, our data suggest that this effect is associated with diminished formation of tetramers when WT and R882H DNMT3A are complexed together. Thus, the R882H mutation has two distinct consequences that affect DNMT3A activity in AML cells: 1) it severely reduces its own de novo methyltransferase activity, and 2) it disrupts the ability of WT DNMT3A to form functional tetramers. These two effects severely reduce total DNMT3A activity in AML cells, and may explain why this mutation is virtually always heterozygous in AML samples, since homozygosity would not further reduce DNMT3A activity. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 2527 Acute myeloid leukemia (AML) is a hematopoietic neoplasm with high mortality that is typically treated with daunorubicin/cytarabine induction chemotherapy. Alternative therapies with cytosine analogs such as decitabine are also used in some cases with a variable clinical response that some have estimated to be as high as 25%. The mechanism of these agents is unclear, but at low doses they produce passive DNA hypomethylation by inhibiting DNMT1. Although the impact of these drugs on cell growth and DNA methylation in AML cell lines has been evaluated1, studies using primary cells are limited; importantly, most have involved extended drug treatments that may be confounded by the differentiation of the treated cells2. In addition, some evidence suggests that decitabine has a differential effect on methylation in patients who respond to treatment2, but the utility of this phenotype as an in vitro biomarker for decitabine responsiveness is unknown. In this study, we used a novel in vitro culture system for primary leukemia cells to explore the initial genomic effects of short-term low dose decitabine on primary samples from 22 AML patients. Primary bone marrow or blood samples from these patients were cultured on HS27 stromal cells in DMEM supplemented with beta-mercaptoethanol and 15% FBS along with hSCF, hIL3, hIL-6, hTPO and hFLT3L for an initial 4-day period prior to daily treatment for 3 days with either 100 nM decitabine, 100 nM cytarabine, or vehicle controls. Cells were then evaluated for growth, cell cycle effects, and differentiation (by flow cytometry and morphologic evaluation). DNA was prepared from all samples for 5-methylcytosine content measurements by mass spectrometry, and 8 samples were selected for genome-wide methylation and gene expression profiling with the Illumina Human Methylation 450 and Affymetrix Human Exon 1.0ST array platforms. Mass spectrometry revealed a mean decrease in 5-mdC of 29% (range: 13% to 62%) in the decitabine-treated samples; in comparison, cytarabine treatment resulted in a mean increase in 5-mdC of 5% (range: −10% to 37%). Methylation arrays also showed a modest shift toward lower methylation values, but unsupervised hierarchical clustering demonstrated that methylation patterns were driven by sample-specific differences and not drug treatment. Analysis of methylation changes showed the most pronounced hypomethylation at CpGs with high baseline methylation levels, irrespective of CpG island and gene-based annotation, suggesting that the initial methylation status of each CpG is responsible for preferential effects of decitabine, rather than its genomic context. Methylation at promoter-associated CpGs showed a small but statistically significant negative correlation with change in gene expression, but expression changes at individual genes were not consistent across the samples, including genes previously shown to be regulated by methylation-dependent mechanisms (eg. CDKN2B and CDx H1). In addition to these findings, we observed that a sample from a long-term decitabine responder had an exaggerated in vitro response to decitabine (58% decrease in 5-mdC after 6 days of treatment), compared to a cohort of decitabine non-responders; a sample from a second patient also showed marked hypomethylation by both mass spectrometry and methylation array, although this patient was not treated with decitabine. While more investigation is needed, this observation might suggest that extreme in vitro hypomethylation in response to decitabine could serve as a biomarker for a clinical response. In summary, our study showed that short-term low dose decitabine treatment has modest but detectable effects on DNA methylation and gene expression, but these changes did not result in activation of any canonical gene expression pathway at this early time point. We found that the baseline methylation status of a CpG appears to be the best predictor of decitabine-induced hypomethylation, with highly methylated CpGs showing the greatest change. We also observed that hypomethylation is highly variable across primary samples and at specific genes, implying that single gene approaches for measuring decitabine effect may be problematic. Finally, extreme in vitro decitabine-induced hypomethylation should be further investigated as a biomarker for decitabine responsiveness. Disclosures: No relevant conflicts of interest to declare.

Description: Mutations in the de novo DNA methyltransferase DNMT3A are found in ~25% of patients with acute myeloid leukemia (AML) and most commonly affect codon 882 within the catalytic domain of the protein. We have previously shown that this mutation has dominant negative activity in vitro and is associated with hypomethylation at specific CpG dinucleotides in primary AML samples using array-based methylation data. However, the genome-wide extent and patterns of DNA methylation associated with this hypomethylation are currently unknown. In addition, it is unclear if the methylation differences caused by this mutation result in RNA expression changes at specific targets across the genome, or whether they are associated with altered chromatin structure. To explore the genome-wide consequences of the DNMT3A R882H mutation on DNA methylation and chromatin structure, we carried out whole-genome bisulfite sequencing (WGBS) and transposase-mediated chromatin accessibility profiling (ATAC-seq) on 3 primary normal karyotype AML samples with the DNMT3A R882H mutation and 4 matched AML samples without a DNMT3A mutation. All 7 had the NPMc mutation but lacked mutations in other genes involved in DNA methylation, including IDH1, IDH2, and TET2. WGBS produced methylation data on 〉93% of the CpGs in the human reference sequence with a median coverage of 7-13x. The overall mean methylation was not statistically different in the samples with R882H mutations, although there was a small but statistically significant difference in the methylation at CpGs in CpG islands (DNMT3A R882H mean: 18.1%, DNMT3A wild-type mean: 21.4%; P=0.02). Differential methylation analysis was performed on ~5 million CpG clusters (median of 5 CpGs per cluster; median cluster size of 202 bp) and identified 95,845 differentially methylated clusters with a mean difference 〉25% and a q-value 〈 0.01, the majority of which (88,512; 93%) were hypomethylated in the DNMT3A R882H samples. Using more strict criteria (〉50% mean difference) and merging differentially methylated clusters within 50 bp, we identified 2,782 differentially methylated regions (DMRs) with a mean size of 255 bp (median of 11 CpGs), of which 97% were hypomethylated. These DMRs were distributed across the genome and were statistically associated with CpG dense regions, including annotated CpG islands and shores (islands: 1,104 of 2,782; 29.9%; shores: 1,118 of 2,782; 30.3%; P