ALBERT

Description: Abstract 2467 NUP98-fusions although rare, have been associated with de novo acute myeloid leukemia (AML), chronic myeloid leukemia, myelodysplastic syndrome, T-cell acute lymphoblastic leukemia (T-ALL) and therapy-related myeloid malignancies. The NUP98-PHF23 (NP23) gene fusion was cloned from an acute myeloid leukemia (AML) patient with a t(11;17)(p15;p13) chromosome translocation. The nucleoporin 98 protein (NUP98), normally a component of the nuclear pore complex, is known to be fused to at least 28 different fusion partners as a result of structural chromosomal rearrangements associated with hematological malignancies. PHF23 encodes the Plant homeodomain (PHD) finger 23 protein. PHF23 is largely uncharacterized, but the PHD finger motif has been shown to act as a reader of di- and tri-methylated histone 3 lysine 4 (H3K4me2/3) marks. This suggests that PHF23 may function in chromatin regulation and that the NP23 fusion protein may play a role in aberrant chromatin modification at domains of active gene transcription. To determine the oncogenic potential of NP23, we generated a transgenic mouse model that expressed the human fusion gene in hematopoietic tissues. We have characterized two founder lines (C10 and B10) expressing the NP23 fusion in hematopoietic tissue. Most of the offspring from the C10 line developed an AML that closely resembled the human disease, with increased blasts in the blood or bone marrow, widespread organ infiltration, and myeloid immunophenotype. Onset of disease was as early as 4.5 months, and 70 percent of the NP23 mice succumbed to leukemia by 12 months of age. Of note, an independent line (B10) developed a wider spectrum of leukemias but with similar age of onset and penetrance. The B10 mice predominantly developed T-ALL and AML, and four cases of B-ALL and one erythroleukemia, indicating that the NP23 protein was oncogenic in several different hematopoietic cell types. AMLs typically demonstrated an aberrant Mac-1+/B220dim phenotype, which has previously been recognized in leukemias caused by overexpression of the Hoxa cluster genes Hoxa5,7,9,10,11. Microarray gene expression analysis identified the Hoxa cluster genes to be markedly overexpressed, and validation by RQ-PCR demonstrated that Hoxa5, a7, a9 and a10 overexpression ranged from 10- to greater than 1000-fold increased in both the AML and T-ALL samples compared to wild type hematopoietic tissues; these Hoxa cluster genes were also overexpressed in hematopoietic tissues from clinically healthy NP23 transgenic mice. We also identified a novel transcript, Gm525 (homologue of H. sapiens C17orf67), that is markedly (100x) elevated specifically in the T-ALL samples. Most T-ALL samples have HD or PEST domain Notch1 mutations, and Notch1 mRNA levels, as well as its downstream target Hes1, are elevated in the T-ALLs compared to WT thymus. Immortal cell lines were established from two of the NP23 T-ALLs, and ChIP-seq was used to assay the genome wide pattern of H3K4me3 and H3K27me3 histone marks. Results show abundant levels of H3K4me3 at the Hoxa locus which tightly correlates to the increased Hoxa cluster gene expression seen in the cell lines. The NP23 model will be useful for identifying oncoproteins involved in leukemic transformation, particularly those oncoproteins which play a role in chromatin modification or are downstream targets of the HOXA genes. Disclosures: No relevant conflicts of interest to declare.

Description: NUP98 gene fusions, created by non-random chromosomal translocations, are associated with a wide spectrum of high risk hematologic malignancies, and typically lead to overexpression of abdominal-b HOXA genes, a common theme shared by ∼50% of AML patients. We have generated a transgenic mouse model of the NUP98-PHF23 (NP23) gene fusion, initially identified in patients with AML, which develop AML, erythroleukemia, pre-T lymphoblastic leukemia (pre-T LBL), and a novel pre-B1 B cell acute lymphoblastic leukemia (pre-B1 ALL). A common theme in the leukemias and the premalignant hematopoietic tissues, is the overexpression of a Hoxa/b +Meis1 stem cell-like gene expression signature. GSEA analysis reveals this signature to be enriched in both human AML and ALL malignancies, and in human HSPC profiles. In addition, we found Bahcc1, a gene not previously associated with malignancy, to consistently segregate with the Hoxa/b+Meis1 signature in the NP23 leukemias and the premalignant tissues, independent of hematopoietic cell lineage. Furthermore, data-mining revealed BAHCC1 to be markedly overexpressed in AML patients with HOXA9/MEIS1 overexpression, and in a subset of MLL rearranged pre-B-ALL patients, suggesting BAHCC1 may be a previously unsuspected marker of leukemic transformation. NUP98-PHF23 belongs to a subset of fusion oncoproteins (including some MLL- and NUP98-fusions) that are potently tumorigenic and act by abrogating the normal reading, writing and erasure of histone methylation. Wild type PHF23 binds H3K4me3 residues via a PHD domain, therefore we used ChIP-seq to characterize global chromatin H3K4me3 and NP23 enrichment in NP23 leukemia derived cell lines. The vast majority (88%) of NP23 binding sites were enriched for H3K4me3 binding. Conversely, the NP23 protein co-localized at only 1.6% of all H3K4me3 enriched sites (including Hoxa, Hoxb and Meis1 loci) identifying these sites as direct targets of the NP23 fusion protein. Given that the NP23 fusion appears to function, at least in part, via binding to H3K4me3 sites at specific loci, we hypothesized that NP23 cells would be sensitive to disruption of the H3K4me4 binding by the NP23 PHD domain. Treatment of leukemic NP23 cells with Tetraethylthiuram disulfide (Disulfiram), a small molecule shown to inhibit PHD domain binding of H3K4me3 marks in vitro, rapidly and selectively killed NP23 myeloblasts but not control myeloblast cell lines (188G3, 189E6 and 32D) at 2 µM. Cell death was rapid, being 100% complete within 24 hours. Cell death was preceded by decreased levels of NP23 protein bound at target loci and decreased expression of these loci (e.g., Hoxa7/9/10, Hoxb5 and Meis1). We conclude that inhibitors of H3K4me3 PHD domain readers are promising therapeutic compounds that can kill leukemic cells driven by proteins that aberrantly read or write the histone code. The NP23 model provides a robust platform on which to identify and improve such compounds. Disclosures: Denu: Sirtris-GSK: Consultancy.

Description: Structural chromosomal rearrangements of the Nucleoporin 98 gene (NUP98), primarily balanced translocations and inversions, are associated with a wide array of hematopoietic malignancies. NUP98 is known to be fused to at least 28 different partner genes in patients with hematopoietic malignancies, including acute myeloid leukemia, chronic myeloid leukemia in blast crisis, myelodysplastic syndrome, acute lymphoblastic leukemia, and bilineage/biphenotypic leukemia. NUP98 gene fusions typically encode a fusion protein that retains the amino terminus of NUP98; in this context, it is important to note that several recent studies have demonstrated that the amino-terminal portion of NUP98 exhibits transcription activation potential. Approximately half of the NUP98 fusion partners encode homeodomain proteins, and at least 5 NUP98 fusions involve known histone-modifying genes. Several of the NUP98 fusions, including NUP98-homeobox (HOX)A9, NUP98-HOXD13, and NUP98-JARID1A, have been used to generate animal models of both lymphoid and myeloid malignancy; these models typically up-regulate HOXA cluster genes, including HOXA5, HOXA7, HOXA9, and HOXA10. In addition, several of the NUP98 fusion proteins have been shown to inhibit differentiation of hematopoietic precursors and to increase self-renewal of hematopoietic stem or progenitor cells, providing a potential mechanism for malignant transformation.

Description: NUP98 gene fusions, generated by non-random chromosomal translocations, are associated with a wide spectrum of high risk hematologic malignancies and have been shown to alter normal hematopoietic stem and progenitor cell (HSPC) gene expression programs. A recurrent t(11;17)(p15;p13) translocation in patients with AML leads to the production of a NUP98–PHF23 (NP23) fusion gene. The consequent NP23 fusion protein retains the PHD domain, known to bind H3K4me3, and is thought to have aberrant chromatin regulation properties. We have generated a transgenic mouse model of the NUP98-PHF23 gene fusion which develops a range of hematologic malignancies, most commonly pre-T LBL and AML. However, approximately 10% of NP23 mice develop an aggressive B-1 progenitor acute lymphoblastic leukemia (pro B-1 ALL). B-1 and B-2 lymphocytes have distinct developmental pathways and are thought to represent arms of the innate and adaptive immune systems, respectively. Mature B-2 lymphocytes predominate in the peripheral circulation, and are characterized by expression of B220; whereas B-1 lymphocytes are more prevalent in the pleural and peritoneal cavities, and do not express B220. Murine B cell malignancies typically stain positive for B220, and represent transformed B-2 cells. In the present study, NP23 progenitor ALLs displayed an immunophenotype (Lin-B220- CD19+ AA4.1+) that was identical to that of the recently described B-1 progenitor cell. All B-1 progenitor ALLs exhibited clonal rearrangements of the IgH gene locus. Specifically, these rearrangements involve favored usage of 3’ VH regions, similar to observations with fetal B-1 progenitor cells, further supporting the notion that these are leukemias of B-1 progenitors. Using whole exome sequencing, we found acquired mutations in the BCL6 interacting corepressor (Bcor) gene in 5 out of 7 B-1 progenitor leukemias. The mutations were all frame shift or nonsense mutations, and were located within a 9 bp “hot spot” in Bcor exon 8. In addition, 4 of 7 cases had somatic mutations of Janus kinase 1 (Jak1) or 2 (Jak2), and 7/7 cases showed hyperphosphorylation of Stat3 or Stat5, consistent with the contention that the Jak1/2 mutations are activating mutations, and leading to a hypothesis that the NP23 pro B-1 ALLs which do not harbor Jak1/2 mutations may have acquired an unidentified mutation in the Jak-Stat pathway. Of note, Jak1/2 mutations have previously been identified in a subset of high-risk pediatric B-cell precursor ALL patients. The striking correlation between Bcor and Jak1/2 mutations, occurring specifically in a subset of NP23 leukemias, implies that these three mutations (NP23, Bcor, and Jak1/2) collaborate and provide the oncogenic setting for B-1 progenitor transformation. Disclosures No relevant conflicts of interest to declare.

Description: Abstract 1323 V(D)J recombination of antigen receptor loci confers diversity to the mammalian immune system. However, off-target, illegitimate V(D)J recombination between non-antigen receptor loci, leading to site-specific interstitial deletion has been reported in humans and mice with lymphoid malignancies. These recurrent, site-specific interstitial deletions are thought to be oncogenic, leading to activation of proto-oncogenes or deletion of tumor suppressor genes. Notch1 and Bcl11b are known targets for illegitimate V(D)J recombination in murine precursor T-cell leukemia/lymphoma (pre-T-LBL). The Notch1 5'-deletions remove exon1 and 2; as a consequence Notch1 transcription initiates from a cryptic promoter immediately 5' of exon 25. These alternative transcripts are translated from exon 28, and lack the N-term ligand binding domain, resulting in a Notch1 protein that is independent of ligand binding. The known Bcl11b deletions excise exon2 and 3, and result in loss of function of the Bcl11b tumor suppressor gene. The frequency with which Notch1 and Bcl11b deletions occur in concert is unknown. Here we examined both Notch1 and Bcl11b deletion mutations, as well as Notch1 mutations involving the PEST, HD, and TM domains in primary murine pre-T LBL samples (N=44), and murine pre-T LBL cell lines (N=21), that arose in specific genetically engineered mouse strains (Scl-Lmo1, NUP98-PHF23, HoxA9, Lin28B, Survivin-TCR). Analysis of 44 primary tumor samples identified 17 Notch1 5'-interstitial deletions (38.6%) and 8 Bcl11b deletions (18.1%), while analysis of 21 cell lines revealed 10 Notch1 5' deletions (47.6%) and 5 Bcl11b deletions (24%). The percent of Notch1 PEST, HD, and TM mutations were 84.1%, 13.6%, and 4.5%, respectively, in the primary tumors and 95%, 5%, and 0% in the cell lines. Two primary tumor samples showed several different 5' deletion mutations, representing multiple independent leukemic clones. Of the 13 primary tumor/cell line pairs we analyzed, 5 pairs showed identical Notch1 and Bcl11b mutations. However, for 7 pairs, the primary tumor was oligoclonal with respect to Notch1 and Bcl11b mutations, with one clone being biologically selected for growth in vitro. In 3 cases, Bcl11b deletion was not detected in primary tumor but was detected in the cell line, suggesting that the Bcl11b deletion was a secondary event in these cell lines. Notch1 PEST mutations were present in almost all cell lines, and were often accompanied by an additional Notch1 mutation (5'deletion, HD, or TM). Conversely, 5' deletion, HD, and TM mutations were mutually exclusive. These findings support the hypothesis that the 5'deletion, HD, and TM mutations lead to ligand independent activation, while the PEST domain mutations prevent proteasomal degradation leading to persistence of the transcriptionally active intracellular Notch1. In several cases, samples had an identical PEST mutation, but distinct 5' deletion or HD mutations, indicating the PEST mutation preceded the 5' deletion or HD mutation. Three of 21 cell lines had both a 5' deletion and Bcl11b deletion, however there was no statistically significant correlation between these events. One 5'-recombination recognition sequence (RSS) and 2 3'-RSSs have been reported for Notch1 5' deletions. H3K4me3 ChIP-sequence data clearly demonstrated alternative transcript initiation from exon 25 in cell lines with a 5' Notch1 deletion. Curiously, one cell line (106A) which was negative for the 5' deletion showed a marked increase of H3K4Me3 at Notch1 exon25, suggesting transcription initiation at this alternative site, and leading to the hypothesis that this cell line may have a variant form of Notch1 5' deletion which was undetectable by conventional assays. We were able to identify an alternative RSS for this cell line, leading to a novel 5' Notch1 deletion. We then reanalyzed our samples and found an additional example of this alternate 5' deletion, and a third sample which displayed yet another form of Notch1 5' deletion. Of note, this deletion did not have a cryptic RSS, and had developed in a RAG1 deficient mouse. In summary, we have identified three novel forms of Notch1 5' deletions, shown that Notch1 mutations are very common in murine pre-T LBL, shown that PEST mutations generally precede 5'deletion mutations, and shown that 5' Notch1 deletions are not consistently associated with other illegitimate VDJ recombinase-mediated events (such as Bcl11b deletions). Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 3829 Therapies for the treatment of myelodysplastic syndromes (MDS) are limited. The only known curative therapy for MDS is allogeneic hematopoietic stem cell transplantation. MDS is frequently associated with epigenetic gene silencing via methylation of cytosine residues in gene regulatory regions. Clinical trials with DNA methyl-transferase inhibitors 5'azacytidine and 5-aza-2'-deoxycytidine (Decitabine, DAC) have shown hematologic responses and a survival benefit. Both azanucleosides have recently been approved for use in the clinic. However, responses are typically not durable, thus further characterization of response to treatment is required, as is the identification of new drugs to better treat MDS. Using bone marrow cells from NUP98-HOXD13 (NHD13) transgenic mice, which have previously been shown to faithfully recapitulate key features of MDS, we co-transplanted MDS and wild type (WT) bone marrow cells into WT irradiated recipient mice. The chimaeric bone marrow transplant (BMT) produces a recipient that mimics the human condition, with patient bone marrow comprised of hematopoietic cells derived from both the MDS clone as well as normal hematopoietic precursors. WT and MDS cells in the mice can be distinguished by differential CD45 alleles (CD45.1 and CD45.2, respectively), which enables analysis and purification of the MDS and normal cells; this feat is not easily achieved with human MDS patient samples, which lack cell surface antigens specific for the MDS clone. We treated mice with 0.018mg DAC (or saline) daily for 5 days, given every 4–5 weeks, to approximate MDS patient dosing schedules. Three trial groups were divided into saline (total n=19) and DAC (total n=24) treatment cohorts. Successful treatment of the chimeric MDS mice resulted in increasing levels of WT cells in the peripheral blood, with a concomitant reduction or eradication of the MDS cells. For instance, in trial #1, DAC treated chimaeric WT/MDS mice showed less severe anemia (2.0 g/dL higher Hgb), normalized neutrophil counts (57– 85% of mice vs 20–60%) and a significant survival benefit (median 27 vs 20 weeks, p=0.004) compared to saline treated mice. The response to treatment varied within individual mice and between trials, analogous to the variability in response seen in human MDS patients. We purified WT and MDS cells from treated mice and evaluated CpG island (CGI) DNA methylation in Chd13, a gene aberrantly methylated in MDS patients. Chd13 was hypermethylated in the murine MDS cells compared to WT cells (42–59% vs 10–16%, respectively). Of note, there was considerable variability between mice in the response to DAC treatment; half of the DAC treated mice showed a normalization of Chd13 methylation (14–20%) and half had a modest decrease in Chd13 methylation (38–40%). We next assayed global CGI methylation using a deep sequencing technique (DREAM, Digital Restriction Enzyme Analysis of Methylation). 8.2% of all CGI sites assayed were hypermethylated in MDS cells compared to WT; this increase in methylation was reduced to only 2.0% following DAC treatment of the MDS cells. This genome wide approach documented extensive hypermethylation in the NHD13 mice, similar to findings in a subset of MDS patients. Chimaeric WT/MDS mice demonstrated variable hematologic outcomes and cytosine demethylation in response to DAC treatment, as seen in human MDS patients. Successful treatment resulted in depletion of MDS cells allowing reconstitution of the bone marrow with WT cells as would happen in the successful treatment of patients. Durable responses were seen in several recipients, including one long-term survivor with no evidence of MDS cells in the bone marrow at 24 months of age (16 months post treatment). The chimaeric WT/MDS mice therefore represent a useful pre-clinical model of human MDS, which can be used to better characterize current treatments and test novel therapies. Disclosures: Issa: Johnson & Johnson: Consultancy; Astex: Consultancy. Aplan:NIH Office of Technology Transfer: Patents & Royalties.

Description: Abstract 2784 Poster Board II-760 MDS comprises a premalignant heterogeneous group of clonal stem cell disorders that also show bone marrow dysplasia and which often evolve to acute myeloid leukemia (AML). Aplastic anemia (AA) patients also share the bone marrow failure, anemia and resulting peripheral blood cytopenias of MDS. AA is thought to be caused by an oligoclonal expansion of cytotoxic T-cells that target haematopoietic stem and progenitor cells. The severe anemia and leucopenia characteristic of both diseases is relieved in AA patients and some MDS patients by immunosuppressive therapy, supporting the role of cytotoxic T-cells in the etiology of AA. However, the role of the lymphocytes in progressive MDS remains unclear. MDS has been associated with a number of genetic aberrations, including chromosomal translocations involving the NUP98 gene. Using mice that express a NUP98-HOXD13 (NHD13) transgene, previously shown to manifest the same clinical symptoms as those of MDS patients, we have followed a cohort of NHD13/Rag1−/− mice to determine if the absence of lymphocytes, especially T cells, might 1) diminish the severity of the MDS, or 2) effect transformation and/or survival in the NHD13 mice, as would be predicted by an “immune surveillance” hypothesis of malignant transformation. Serial CBCs at two month time intervals were used to evaluate the extent of anemia and leucopenia in NHD13+ /Rag1+/+ and NHD13/Rag1−/−, as well as WT/Rag1+/− and WT/Rag1−/− control groups over a 15 month period. NHD13/Rag1−/− mice were generated by crossing the NHD13+ (C57BL/6) with the B6;129S7-Rag1tm1Mom/J mouse, and housed in a Specific Pathogen-Free (SPF) environment. Mice were euthanized and analyzed when CBCs indicated severe anemia/leucopenia or leukemic transformation, or when determined to be unwell (hunched, immobile, dyspnea) by observation. Flow cytometry, histology and genomic analyses further determined leukemia subtype, extent of infiltration and leukemia clonality. NHD13+ /Rag1+/+ and NHD13/Rag1−/− mice showed no significant differences at any two month time-point in hemoglobin (Hg), mean corpuscular volume (MCV), or platelet levels, and progressive MDS occurred in both groups. Consistent with previous studies, and excluding cases that showed evident transformation to acute leukemia, NHD13+ /Rag1+/+ mice showed low WBC, neutrophil and lymphocyte numbers, which were not significantly different from the NHD13/Rag1−/− mice. NHD13/Rag1−/− mice did however show a significantly reduced survival when compared with the NHD13+ /Rag1+/+ mice (Log-rank test, p = 0.0135), and survival medians of 11 and 13 months, respectively. Incidence of leukemic transformation was increased in the NHD13/Rag1−/− compared with the NHD13+ /Rag1+/+ mice (p=0.0079). A range of leukemia subtypes was observed in both the NHD13+ /Rag1+/+ and NHD13/Rag1−/− mice, including myeloid, B-cell, T-cell, and erythroid leukemias. In the SPF environment provided, the WT/Rag1+/− and WT/Rag1−/− control groups showed no significant difference in survival rates. Serial CBC data indicated that there was no significant difference in the timing or degree of peripheral blood cytopenias between the NHD13+ /Rag1+/+ and NHD13/Rag1−/− mice, supporting the conclusion that absence of lymphocytes does not lead to improvement in the peripheral blood cytopenias caused by the NHD13 transgene. This observation suggests that the NHD13 transgene does not produce MDS caused by an autoimmune phenomenon. The poorer survival and increased frequency of leukemic transformation in the NHD13/Rag1−/− mice suggests that lymphocytes might play a role in the evolution of MDS to AML in the NHD13 mouse model, and supports the ‘immune surveillance' hypothesis. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 1705 The myelodysplastic syndromes (MDS) comprise a group of hematologic disorders characterized by ineffective hematopoiesis, dysplasia, and transformation to acute myeloid leukemia (AML) in a subset of cases. The causative agents for de novo or therapy-related MDS (t-MDS) include exposure to known genotoxins, such as anti-cancer chemotherapy, or ionizing radiation. It is known that MDS is associated with chromosomal aberrations including deletions, amplifications, inversions, and translocations. However, the molecular mechanisms that lead to progression of MDS to AML have not been completely elucidated. Recently, several studies have reported that oxidative stress contributes to disease progression of MDS. Reactive oxygen species (ROS) play a role in regulating several biologic phenomena including activation of signaling pathways in response to cytokines, and the gene expression pattern induced by this signaling. ROS is also known to induce oxidative DNA damage, which can lead to activation of proto-oncogenes or inactivation of tumor suppressor genes. A mouse model for MDS, generated by expression of a NUP98-HOXD13 (NHD13) fusion gene, was exploited to investigate ROS levels. Lineage negative (Linneg) bone marrow mononuclear cell (BMNC) from NHD13 mice with MDS had a 6.0 fold increased level of ROS compared with wild-type (WT) Linneg BMNC. In order to address potential mechanism(s) leading to this increased ROS, a Real-Time Quantitative PCR array, which profiles the expression of 84 genes related to oxidative stress, was utilized and the data demonstrated down regulation of Myeloperoxidase (Mpo; −9.9 fold), Eosinophil peroxidase (Epx; −8.8 fold) and Lactoperoxidase (Lpo; −5.2 fold) in the Linneg BMNC from NHD13 mice. The increased ROS level in NHD13 Linneg BMNC was associated with a 2 fold increase in G2/M phase cells along with increased expression of p16INK4A (43 fold) and p21 (10 fold), possibly to allow for DNA repair at cell cycle checkpoints. In order to investigate genomic DNA damage by ROS, immunocytochemistry for γH2AX foci was performed and demonstrated a 2 fold increase in γH2AX foci in the NHD13 BMNC compared to WT BMNC. To determine if mis-repair of the ROS-induced DNA damage led to mutations of genomic DNA in vivo, mutation frequencies (MF) were determined using the Big Blue®cII mutation detection assay. Big Blue mice are transgenic mice that have a λLIZ shuttle vector, which is derived from the coliphage lambda. Selection of mutation is based on the ability of the λ phage to multiply through either the lytic or lysogenic cycle in E. coli host cells. Big Blue mice were crossed to NHD13 mice, and NHD13/Big Blue® mice were compared with littermate controls that were positive for the BigBlue transgene but negative for the NHD13 transgene. BMNC from the NHD13/BigBlue mice showed a 1.8-fold increased MF compared to control BMNC (p value = 0.0371). Mutations were comfirmed by sequencing the cII target gene, and revealed that the NHD13/BigBlue mice had a 3-fold higher frequency of frameshift mutations compared to the control animals. These results suggest that the oxidative stress induced by the NHD13 fusion may contribute to disease progression of MDS to AML through DNA damage and mutation. Disclosures: No relevant conflicts of interest to declare.