Activating signaling mutations are common in acute leukemia with KMT2A (previously MLL) rearrangements (KMT2A-R). These mutations are often subclonal and their biological impact remains unclear. Using a retroviral acute myeloid mouse leukemia model, we demonstrate that FLT3ITD, FLT3N676K, and NRASG12D accelerate KMT2A-MLLT3 leukemia onset. Further, also subclonal FLT3N676K mutations accelerate disease, possibly by providing stimulatory factors. Herein, we show that one such factor, MIF, promotes survival of mouse KMT2A-MLLT3 leukemia initiating cells. We identify acquired de novo mutations in Braf, Cbl, Kras, and Ptpn11 in KMT2A-MLLT3 leukemia cells that favored clonal expansion. During clonal evolution, we observe serial genetic changes at the KrasG12D locus, consistent with a strong selective advantage of additional KrasG12D. KMT2A-MLLT3 leukemias with signaling mutations enforce Myc and Myb transcriptional modules. Our results provide new insight into the biology of KMT2A-R leukemia with subclonal signaling mutations and highlight the importance of activated signaling as a contributing driver.

No abstract

Infant acute lymphoblastic leukemia (ALL) with KMT2A -gene rearrangements ( KMT2A -r) have few mutations and a poor prognosis. To uncover mutations that are below the detection of standard next-generation sequencing (NGS), a combination of targeted duplex sequencing and NGS was applied on 20 infants and 7 children with KMT2A -r ALL, 5 longitudinal and 6 paired relapse samples. Of identified nonsynonymous mutations, 87 had been previously implicated in cancer and targeted genes recurrently altered in KMT2A -r leukemia and included mutations in KRAS , NRAS , FLT3 , TP53 , PIK3CA , PAX5 , PIK3R1 , and PTPN11 , with infants having fewer such mutations. Of identified cancer-associated mutations, 62% were below the resolution of standard NGS. Only 33 of 87 mutations exceeded 2% of cellular prevalence and most-targeted PI3K/RAS genes (31/33) and typically KRAS/NRAS . Five patients only had low-frequency PI3K/RAS mutations without a higher-frequency signaling mutation. Further, drug-resistant clones with FLT3 D835H or NRAS G13D/G12S mutations that comprised only 0.06% to 0.34% of diagnostic cells, expanded at relapse. Finally, in longitudinal samples, the relapse clone persisted as a minor subclone from diagnosis and through treatment before expanding during the last month of disease. Together, we demonstrate that infant and childhood KMT2A -r ALL harbor low-frequency cancer-associated mutations, implying a vast subclonal genetic landscape.

Acquired activating mutations in kinase/PI3K/RAS signaling pathways occur in about half of pediatric acute leukemia cases with Mixed Lineage Leukemiagene rearrangements (MLL-R; MLL, also known as KMT2A). Mutations resulting in activated signaling cooperate with the MLL-R in mouse leukemia models, however, detailed insight on the effect on the transcriptional and proteomic landscape is lacking. In infant MLL-R acute lymphoblastic leukemia (ALL), a subtype with a very poor prognosis, a majority of the activating mutations are subclonal, but the biological mechanisms by which subclonal mutations affect leukemogenesis remain unclear. Here we show that NRASG12D, FLT3internal tandem duplication(ITD)and FLT3N676K,cooperates with MLL-MLLT3 in myeloid leukemogenesis using a competitive murine retroviral bone marrow transplantation model. The addition of an activating mutation remodels the gene expression patterns of the MLL-R leukemia with distinct profiles for each mutation as determined by RNA-sequencing. Gene set enrichment analysis revealed enrichment of genes involved in chromatin assembly, transcription and stemness in leukemia induced by MLL-MLLT3 and NRASG12D, FLT3ITD or FLT3N676K. Leukemia induced by only MLL-MLLT3 displayed upregulation of genes involved in signal transduction suggesting activation of such pathways by alternative mechanisms. Upon secondary transplantation, mice receiving MLL-MLLT3-only leukemic cells succumbed to disease at a similar latency to those receiving MLL-MLLT3 and an activating mutation, supporting that fully transformed MLL-MLLT3 leukemias have sustained active signaling via high expression of genes involved in signal transduction. Using the same model as above but with significantly reduced numbers of transplanted cells containing an activating mutation, we could further show that a subclonal mutation, exemplified by FLT3N676K, causes significantly reduced disease latency as compared to mice receiving MLL-MLLT3 only (34 vs 50 days). The size of the double positive (MLL-MLLT3+FLT3N676K) and single positive (MLL-MLLT3) clones within each mouse was determined by flow cytometry revealing that 8/24 mice had a double positive subclone (≤50%). The clonal evolution of the double positive cells from 22/24 mice was assessed in secondary recipients showing three distinct patterns 1) increase in size, 19/22 mice 2) maintained, 2/22 mice or 3) decreased in size, 1/22 mice. Targeted gene re-sequencing of the latter leukemia that lost its MLLT3+FLT3N676K subclone, identified a de-novo CblA308T in the SH2-like domain, in the MLL-MLLT3 onlycells that had gained clonal dominance in the secondary recipient. The decreased disease latency in mice with subclonal activating mutations raises the possibility that cells with an activating mutation support the growth of other leukemic cells by direct cell-cell contact or through secreted factors. Transcriptome and proteomic analyses identified a high expression of the macrophage inhibitory factor (MIF), a pro-inflammatory cytokine, in mice with MLL-MLLT3 and NRASG12D, FLT3ITD or FLT3N676K. Addition of rMIF increased the survival of MLL-MLLT3 murine leukemic cells in vitro. Although additional factors likely mediate pro-leukemic effects in vivo, our data suggest that MIF could be one of these factors. In summary, our data demonstrate that activating mutations cooperate with the MLL-R in murine leukemogenesis and cause widespread changes in the transcriptional landscape. In addition, our results suggest that cells containing an activating mutation in addition to an MLL-fusion positively influence the survival and likely also the growth of other leukemic cells, suggesting a pro-leukemic effect mediated by interclonal cooperation between clones carrying distinct mutational set-ups in leukemogenesis. Disclosures No relevant conflicts of interest to declare.

No abstract

Genetic rearrangements involving the KMT2A gene (KMT2A-R) are seen in around 10% of acute leukemia overall. KMT2A-R occurs in all ages and usually correlates with high-risk clinical features, in particular in infants aged 0-12 months of age with acute lymphoblastic leukemia (ALL). To uncover age- and leukemia-subtype specific molecular patterns in KMT2A-R ALL and acute myeloid leukemia (AML), we performed whole genome (WGS), whole exome (WES), and RNA-sequencing on a well-annotated Nordic KMT2A-R cohort of 104 patients, including infant ALL (n=33), childhood ALL (n=18), adult ALL (n=15), and pediatric AML (n=38) patients. For 77 patients, we performed WGS (40x) at diagnosis and remission as well as WES (140x) on the diagnostic sample, and remaining patients underwent WES only (n=27). RNA-sequencing was performed on 58 cases with available RNA. Twenty-two genes were recurrently altered and remarkably, NRAS, KRAS, FLT3, PAX5, TP53, CDKN2A/B and IKZF1 accounted for 70% of mutations. The landscape of mutations suggested the presence of leukemia and age-specific associations with MYST4, PTPN11, and SETD2 uniquely altered in AML and PIK3CD, DNAH11, NOTCH1, CSMD3 and CDKN2A/B in ALL. Some genes were mutated in both KMT2A-R ALL and AML, but were more common in one disease, such as FLT3 and KRAS in AML and PAX5, TP53 and IKZF1 in ALL. Moreover, age-associated patterns were seen in ALL with NRAS more frequently mutated than KRAS in infant ALL (26% vs 15%), and KRAS more frequently mutated than NRAS in childhood ALL (24% vs 18%), with adult ALL having fewer such mutations (NRAS 13%; KRAS 7%). Alterations of CDKN2A/B and TP53 were absent in infant ALL, detected in childhood and adult ALL only. PAX5 alterations were primarily detected in childhood ALL (22%, 9% infant ALL, 7% adult ALL), with all three PAX5-altered infant cases having the KMT2A-MLLT3 fusion gene. Finally, KMT2A-R pediatric AML had the highest fraction of FLT3 mutations (24%, 9% infant ALL, 11% childhood ALL, 0% adult ALL) and all but one mutation occurred in KMT2A-MLLT3 rearranged cases and most were kinase domain point mutations. We next expanded our analysis to include non-recurrent alterations. PI3K/RAS pathway alterations were detected across ages and subtypes with the highest fraction in pediatric AML (63%) and the lowest in adult ALL (27%, 43% infant ALL, 41% childhood ALL). Further, cell cycle related genes were primarily mutated in childhood (39%) and adult ALL cases (33%) and rarer in infant ALL (12%) and pediatric AML (16%) and genes within the B-cell pathway were more commonly altered in childhood ALL (29%) than in infant ALL (9%). Finally, in line with our previous study (Andersson et al, Nat Genet 2015) epigenetic mutations were absent in infant ALL, but present in 20-35% of the other patients. RNA-sequencing identified the KMT2A-fusion in 56/58 cases, with low exonic coverage preventing detection of the fusion in two cases. The reciprocal KMT2A fusion was only expressed in 13/39 cases where it was predicted to be expressed based on karyotype or whole genome sequencing data with 11/13 cases having the KMT2A-AFF1 fusion gene. In addition, RNA-sequencing identified 6 in-frame and 12 out-of-frame fusion genes that had formed either as part of the KMT2A-R itself or that were independent genetic events. Further, a novel in-frame KMT2A-ACIN1 fusion was identified in a child aged 1 year with B-precursor ALL. ACIN1 encodes Apoptotic Chromatin Condensation Inducer 1 and the fusion was formed through an insertion of 14q11 into 11q23. ACIN1 is also rearranged as part of the ACIN1-NUTM1 that we identified in an infant with ins(15;14)(q22;q11.2q32.1) (Andersson et al Nat Genet 2015). To study the ability of KMT2A-ACIN1 to induce leukemia in mice, we injected retrovirally transduced mouse bone marrow cells containing the fusion into syngeneic mice and KMT2A-MLLT3 was used as control. All mice succumbed to disease at an average of 112 days for KMT2A-ACIN1 (n=12) and 63 days for KMT2A-MLLT3 (n=5) and mice displayed splenomegaly and leukocytosis with an immunophenotype indicative of AML. Primary leukemia cells isolated from moribund mice gave rise to leukemia in sublethally irradiated recipients with reduced disease latency. In conclusion, these results highlight the differential molecular patterns in KMT2A-R leukemia across infancy to adulthood thereby providing novel pathogenetic insight. Disclosures No relevant conflicts of interest to declare.

Our understanding of how individual mutations, whether present in all or just a fraction of the leukemia cells, affect cellular responses to therapy is limited. Leukemia mouse models provide a unique possibility to explore how therapy affects the evolution of genetically distinct clones and identify mechanisms of resistance allowing transfer to human disease. Herein, we studied how different therapies influenced survival, clonal evolution, and resistance patterns in mouse KMT2A-MLLT3 leukemia with subclonal FLT3 N676K. Bone marrow (BM) from a leukemia expressing KMT2A-MLLT3-mCherry in all cells and a FLT3 N676K-GFP in 40% of cells, were re-transplanted to sublethally irradiated recipients (Hyrenius-Wittsten el al, Nat Commun, 2018). Upon engraftment, treatment was started with either chemotherapy (cytarabine for 5 days + doxorubicin for 3 days), the FLT3 inhibitor AC220, chemotherapy followed by AC220, or AC220+Trametinib, a MEK inhibitor. Targeted treatment was given for 28 days; controls received vehicle (Fig. 1a). Survival was estimated by Kaplan-Meier and the developing leukemias were analyzed by flow-cytometry, RNA-sequencing and targeted gene re-sequencing. Each treatment prolonged survival with a median latency of 30 days for chemotherapy , 37.5 days for AC220, 42 days for chemotheraphy+AC220, and 45 days for AC220+Trametenib, versus 25.5 days for the control (Fig. 1b). Most leukemia cells expressed GFP/mCherry and mice displayed splenomegaly and leukocytosis. Next, we investigate how treatment impacted evolution of the KMT2A-MLLT3+FLT3 N676K cells and while they constituted all cells in control and chemotherapy-treated mice, the other treatments impacted their evolution. Three distinct patterns were discerned with either >80% of KMT2A-MLLT3+FLT3 N676K cells, >80% of cells expressing KMT2A-MLLT3 alone, or dual similar sized clones of cells expressing KMT2A-MLLT3 alone or KMT2A-MLLT3+FLT3 N676K(Fig. 1c). Eradication of the FLT3-leukemia cells was rare, but most common in mice receiving AC220+Trametinib and the frequency of dual clones increased when mice received chemotherapy followed by AC220, in line with treatment selectively affecting evolution of genetically distinct cells (Fig. 1d). To find clues to treatment resistance, RNA-sequencing (N=44) revealed segregation into three major clusters: 1) leukemias expressing KMT2A-MLLT3 alone, 2) control and chemotherapy-treated leukemias and 3) AC220 treated leukemias. Notably, a set of AC220-treated mice clustered close to the control and chemotherapy-treated mice (Fig. 1e). Flow-cytometry data showed that similar to the control and chemotherapy-treated leukemias, the myeloid BM cells of those AC220 samples, aberrantly expressed B220 (Fig. 1f). Gene set enrichment analysis revealed enrichment of gene sets correlating with stem cells and oxidative phosphorylation in those AC220-treated leukemias, suggesting a switch in cellular phenotype and metabolic state upon treatment. By contrast, the other AC220 leukemias (cluster 3), instead showed enrichment of gene sets correlating with granulocyte/macrophage progenitors and immune regulatory pathways, indicating selective dependence of distinct cellular pathways upon resistance (Fig. 1g). Finally, acquisition of AC220 resistance mutations was rare with a FLT3 D835Y and a Ptpn11 G503V detected in two leukemias only. Taken together, these results show that the specific treatment given not only affected survival of the FLT3 N676K mutated KMT2A-MLLT3 leukemia, but also impacted how the genetically distinct cells evolved. The general lack of acquired mutations upon targeted treatment suggests that target-independent mechanisms that result in alternate activation of survival/proliferation explains acquired resistance in a majority of mice and provides novel insights into treatment resistance. Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.